Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
The Significance of Sex Segregation in the Workplace Women are a large and growing portion of the labor force, ant] paid employment is clearly of growing importance in many wom- en's lives. More women work outside the home and for longer portions of their lives than ever before. Women's employment, like men's, plays a vital role in our economy. Nearly 50 million women were in the labor force in 1984 and constituted 43 percent of the labor force. Of all women ages 18-64 in April 1984, 63 percent were in the labor force. Nearly all women work at some point in their lives, and the average woman today is expected to spend 12 more years working than did women in her mother's generation. In fact, the labor force participation patterns of women ant] men appear to be converging, as women's participation has increased and that of men has decreased somewhat in re- cent years. Despite increasing similarities in wom- en's and men's work lives, significant areas of difference remain in particular, earn- ings and occupations. Although for most women as for most men, their earnings are crucial to their own support and to the fi- nancial support of their families, women's earnings are substantially less than those of 1 men. For as long as data have been available for the United States, women's average earnings have been about 60 percent of men's for full-time, year-round workers. Women also often work in different kinds of jobs. Ibe majority of women work in a small num- ber of occupations, particularly in occupa- tions in which the workers are predominantly women. Men work primarily in occupations that are predominantly male, although the number of occupations is larger. Ike concentration of women and men in different jobs that are predominantly of a single sex has been labeled sex segregation in the labor market. The overall degree of sex segregation has been a remarkably stable phenomenon; it has not changed much since at least 1900. This stability is surprising in light of the enormous changes that have taken place in the structure of the economy: the turnover in occupations as obsolete occu- pations disappear and new ones develop; the narrowing of e(lucational differentials be- tween men and women, particularly since World War II; and, most recently, the in- creasing similarity in the work patterns of men and women over their lifetimes. It is this stable phenomenon-the concentration
2 of men and women in different jobs that is the subject of our report. In the past women have not had equal opportunity in the labor market, and they have faced discrimination in hiring, pay, and advancement. To some extent the differ- ences in women's and men's earnings and in the occupations they hold reflect that past discrimination; to some extent they reflect current discrimination; and to some extent they reflect a host of other factors, such as differences between women anti men in their preferences, attitudes, values, experience, education, training, and so on. And it is highly likely that all these factors are inter- related. In this report we attempt to unravel the various causes of sex segregation in the workplace, to understand its extent, future direction, and remarkable persistence. To the extent that it reflects the preferences, values, and attitudes of women themselves, it may not be an appropriate object of public concern. But to the extent that it reflects restrictions on women's choices that result from discriminatory practices in the labor market or various other barriers, it is a mat- ter of grave public concern. Women have the right to participate in the labor market, as they choose, without social or legal coer- cion and without unfair treatment in pay or other working conditions. Equal employ- ment opportunity is an established goad of national policy: it contributes not only to the better utilization of the country's human re- sources and to economic growth, but also to the filll participation of all members of so- ciety in the nation's political, social, and eco- nomic life. In this report we seek to deepen under- st~nding of the processes that give rise to sex segregation in the workplace, to assess the aspects of sex segregation that are harm- fi~l, and to offer guidance on how to amel- iorate those aspects. Our method has been to gather and assess the available research literature on these issues. Any literature re- view is necessarily selective, and ours is no exception. We have tried to identify signif WOMEN'S WORK MEN'S WORK leant research from a variety of perspectives, however, and to assess the major alternative explanations that have been offered for the persistence of job segregation. In the remainder of this chapter we pro- vide further description of the situation of women in the labor market; discuss the con- cepts of segregation in general and sex seg- regation in the workplace; and briefly review the literature on the consequences of the latter. We find that those consequences are several and significant and we believe they warrant the committee's effort to better un- derstand sex segregation in the workplace and the ways in which it can be affected. In subsequent chapters, we look at the recent past and likely fixture of sex segre- gation, identify its causes, and assess a va- riety of interventions that have been imple- mented to re(luce segregation. In Chapter 2, in order to better understand what it is that requires explanation, we review esti- mates of the current extent of segregation and identify the changes that have occurred over the past decade among certain groups of women and within certain occupations. We also present projections of the extent of segregation for the rest of this decade. In Chapter 3 we review the evidence for the most important explanations of labor market segregation, assess their relative strengths, and give our view of the most likely cause of continued segregation. In Chapter 4 we assess the evidence regarding the effective- ness of federal regulations and legislation prohibiting discrimination and mandating affirmative action in employment, employ- ment training, vocational education, and general education. Finally, in Chapter 5 we present a summary of our findings and make recommendations for strategies to reduce job segregation by sex and increase equal opportunity in the workplace. WOMEN IN THE LABOR MARKET Women's Participation in the Labor Force The majority of adult women are in the labor force, and their rate of participation
SIGNIFICANCE OF SEX SEGREGATION IN THE WORKPLACE 3 has been steadily increasing throughout the century (Waite, 1981~. In contrast, men's labor force participation rates have been slowly declining. In 1950, 86.4 percent of men ages 16 and over were in the labor force; by 1984 the percentage had dropped to 76.4 (U. S. Department of Labor, Bureau of La- bor Statistics, 1985a). The rate of increase for women has been substantial since the 1950s and has not yet tapered off, as can be seen in Figure 1-1. Between 1950 and 1984 the labor force participation rate of women ages 16 and over increased from 33.9 to 53.6 percent. Labor force participation rates for women vary by age, marital status, and race and ethnicity. And more women work at some time during the year than are in the labor force at any one time. The 1982 Work Expenence Survey of the Current Popula- tion Survey indicated that 58 percent of women ages 16 and over worked some time during 1981 (U.S. Department of Labor, 54f LL3 I - u~ ~ 48 ~C.) ON 46 ~0 11 .. _ (D ~ 42 it, it LL UJ CL 52 50 40 __ 38 36 _ / 34 White Black and Other Hispanica _ _- - _ ~/ - 1955 1960 1965 1970 1975 1980b YEAR aData for Hispanics were not available until 1973. b1980 data are for civilian women age 20 and over. FIGURE 1-1 Labor force participation rates of women ages 16 and over, based on annual averages for selected years between 1955 and 1980. SOURCES: U.S. De- partment of Labor, Bureau of Labor Statistics (1980: Tables 65 and 66; 1981c:Table 44~. Women's Bureau, 19831. Annual averages for 1984 indicate that 53.8 percent of all women were in the labor force, with black women most likely to be in the labor force (55.5 percent), white women next most like- ly (53.4 percent), followed by Hispanic women (49.8 percent; U.S. Department of Labor, Bureau of Labor Statistics, 1985b Table 21. Women's increasing rates of labor force participation are reflected in their growing work-life expectancies. In 1979-1980 a 20-year-old woman could expect to work 27.2 years, compared with 14.5 years in 1950; the comparable estimate for a 20-year-old man in 1979-1980 was 36.8 years (S. Smith, 19851. ~ Of all women in the labor force, sev- en-tenths hold full-time jobs. The disparity between the labor force par- ticipation rates of married and unmarried women has declined, as the labor force par- ticipation rates of married women have in- creased rapidly. In 1984, 63.3 percent of never-married women age 16 and over were in the labor force, compared with 52.8 per- cent of married women with husbands pres- ent, 61.1 percent of married women with husbands absent, and 74.3 percent of divorced women. For younger married women the rates are quite high. Increased rates of em- ployment by married women with children have contributed substantially to the grown in women's labor force participation. In 1950 about 12 percent of women with a child un- der 6 years old were in the labor force; in 1980 the ratio was 52.1 percent (Hayghe, 1984). Figure 1-2 illustrates the historical change in the age-specific labor force participation rates for women who were born between 1926 and 1960. In 1980, for the first time, the labor force participation rate did not de- cline for women ages 25-29, a peak child- bearing group. In fact, the job-leaving rates ~ These estimates are based on age-specific proba- bilities of movement into and out of the labor force (S. Smith, 1982:1~11).
WOMEN'S WORK, MEN'S WORK 70 - O ~ I C: ~ O O O Lum 45 llJ a: LL 65 60 55 50 40 35 FIGURE 1-2 Women's age-specific labor force par- ticipation rates by birth co hors. SOURCE: Hartmann and Reskin (1983~. for women at all ages under 55 dropped be- tween 1970 and 1977 (S. Smith, 1982~. Women's Earnings In 1981 the median earnings of women who worked fills time year-round were $12,001, or 59 percent of what men earned, $20,260. For workers over age 18, the earn- ings ratio for white women and men was 60 percent; for black women and men, 76 per- cent; and for Hispanic women and men, 73 percent. The ratio of black women's earnings to those of white men was 54 percent; of Hispanic women's earnings to those of white men, 52 percent (U.S. Department of La- bor, Women's Bureau, 19831. Most women who work contribute sub- stantially to or finely support themselves and their dependents. In 1981, about one out of five women workers maintained families on their own. U. S. Department of Commerce data as of March 1984 indicate that one-sixth of all U. S. families (about 9.9 million) were maintained by women with no husband present; they were never married, separat- ed, divorced, or widowed (U. S . Department - 1956 60 Birth Cohort 1951 - 1955 Birth Cohort / 194~50 Birth Cohort :' 1941~5 Birth Cohort /193600 Birth Cohort / 1931-35 Birth Cohort /2~30 Birth Cohort 30 . . . . . . . 2~24 25-29 3~34 35-39 40-44 45 49 5~54 AGE of Commerce, Bureau of the Census, 1984b). Most of these families depend principally on the earnings of women. Ibe incomes of mar- ried women living with their husbands are also important for their families' economic well-being. In general, the lower a hus- band's income, the more likely it is that his wife works (Sweet, 19731. The earnings of women who are married are especially likely to be important to families when the hus- bands' earnings are low. In 1981, the median percentage of family income contributed by married women Swim husbands present) was 26. 7. Ibe percentage increases to 69 percent if annual family in- come is less than $10,000; 56 percent if it is between $10,000 and $14,999; and 46.6 per- cent if it is between $15,000 and $19,000 (U. S. Department of Labor, Women's Bu- reau, 19831. Minority women make larger economic contributions to their families than white women. In 1980, minority women's incomes represented one-third of their fam- ily income, compared with one-quarter for white women (personal communication, Harriet Harper, Women's Bureau, U. S. De- partment of Labor, 1982~. And wives' earn
SlGNIFlCANCE OF SEX SEGREGATION IN THE WORKPLACE 5 ings take on particular importance during periods of high unemployment. In sum, even though women earn sub- stantially less than men, their earnings are a significant source of support for themselves and their families. Women are more likely than ever to be in the labor force, and they can expect to spend a substantial portion of their adult lives doing paid work. For these reasons the consequences of sex segregation are significant and enduring. SEX SEGREGATION IN THE WORKPLACE The term segregation has been used to connote many different phenomena. It has often been used to describe situations in-. valving the physical and social separation of members of different socially identifiable groups, particularly the isolation of a mi- nority group from the majority group. Its Latin roots are se, meaning apart, and grew, meaning flock. For example, apartheid in South Aflica today like past racial segre- gation in the southern Unite] States- in- volves the physical separation of the races in neighborhoods, schools, and public ac- commodations, often by law and sometimes by social custom. Even when the races are in close proximity, as they often are in em- ployment situations, social norms enforce social distance. When segregation is the re- sult ofthis type of legal ant} social restriction, it usually connotes the inferiority of the mi- nority group and can be an important means of maintaining its minority status. Segregation can also be a voluntary mat- ter. For example, many neighborhoods in cities and regions of the country that are ethnically identified are so primarily as a matter of choice and not compulsion, though compulsion ant! limited opportunities may have played some role in their initial estab- lishment. Many people like to live among their kin or coreligionists and close to churches, stores, and schools that cater to their ethnic group. Others do not. Observed patterns of segregation may also be partly coerced and partly voluntary, brought about by a combination of social pressure, lack of knowledge of alternatives, socialization, and choice. Sex segregation in the workplace, which takes both physical and social forms, is al- most certainly the result of both restriction and choice, although we have come to the considered conclusion, based on the evi- dence we have reviewed (which is presented in the following chapters), that restriction plays the more important role. The measure of sex segregation in employment most com- monly used in this report, and in other social science research, measures the degree,of segregation against a standard of total in- tegration. The index of dissimilarity, often called the segregation index, measures the degree to which the distributions of the groups being studied (women and men here) across a set of categories (occupations or jobs here) differ from each other.2 Such a mea- sure implies a goal of complete integration, with the proportions of women and men within every occupation identical to their representation in the labor force as a whole. There is no reason to believe, however, that if all barriers to the Dee and informed ex- ercise of choice by women in the labor mar- ket were removed, the distributions of wom- en and men across all occupations wouIcl be identical. They might be, but they might equally well not be. Some differences be- tween women and men are deeply rooted in culture and may last for decades; some, though perhaps not many besides the most obvious, are rooted in biology and may last 2 The index of segregation, I.S., is defined as n I.S. = 1 ~ | xi ~ Yi I i=1 where xj = the percentage of one group (e. g., women) in the ith category of a classification (e.g., a particular occupation), and Yi = the percentage of the other group (e.g., men) in that same category (Duncan and Duncan, 1955~.
6 WOMEN'S WORK, MEN'S WORK . .. .. ~4 CC - C.) ~ Cal C) ~ ~ C) x cn en C) Ct U' o - C~ o ._ en o . o I' sly a) o o ._ ._ CO ._ en o Ct - C) o - 1 - ~3 ~ CID ~ ~ _ ~ 0 00 Cal 0 of Us O _ Cal CC Cal _ ~ CO ~ co ~ 0 o ~ Cal ~ Go ~ _ Go ~ U. ~ ~ ~ CD ~ ~ ~ ~ CD - (D ~ ~- - Ct .- ho ~C) = ~ 00 ~ CD ~ Cal ~ 0 ~ O CD 0 0 0 00 U) O I) ~ O CD ~ ~ ~ O ~ Cal Cal - - - O 0 _ _ Cal Cal Cal C ~ ~C ~0) 00 C~ CD ~4 C~ ~ 0 U.00 ~ C~ O ~ ~ ~i ' ~ C~ ~ (D' - - O ~ C ~- ~- C~ - -O Co _ 0 C~, O (D c~ _ _ ~ _ 2O c~ ~ c~ _ o0 co c~ O U ~ C~ 00 ~ ~ C~ CC C~ 00 C ~CS C~ 0 ~_ _ O - C ~_ ~- C~ - -O C53 _ ~ - - O 0 CD 1~ ~ - - C] 00 (D ~ C~ 0 0 0 ^ . =4 ~CD U: _ ~ oo 00~ ~ ~ C~ _ _ _ ~ O C~ __ __ - C~ - ~ - O - ~ ~ O 0 0O 00 ~ ~ - CO O ~ O CC Ca, 00 ~ O. C' 1O C*~ ~ O CJ _ U~ OC) ~ CS) O O ~ ~ ~ U) O CS: C~ - - C:4 _ C~ C~ O ~ CD O ~ CD a, _ ~ ~ _ u: ~ c~ _ 0 ~ ~ _ 0 ^ . . . . . . . . . . . . . . . . . . - =4 C~ - 0 C~ ~ ~ 0 C~ ~ O ~ 0O CC CD U. O ~ C~ - - - _ C~ C~ O O _ a, ~!~= i=~.~= ~ ji ~ ~ ~ ~ ~ ~ ~ ~ ~ 3~4 Y _ C~ C~ C~ C~ C) 2 .. 00 0 _ _ C~ _ .- _ C~ C~ O C~ O _ CC C) ce o :: L. c: c~ o c~
SIGNIFICANCE OF SEX SEGREGATION IN THE WORKPLACE 7 even longer. The appropriate policy goal is not therefore the complete elimination of segregation as measured by the index of dis- similarity, but rather the elimination of bar- riers to women's full exercise of their em- ployment nghts. We have not estimated how much sex segregation would be reduced if equality of opportunity were achieved or how much would remain out of choice, but we believe, on the basis of our review of the evidence, that the reduction would be sub- stantial. The segregation of the sexes is a basic feature of the world of work. The strikingly different distributions of women and men across occupations can be seen in the dis- tribution of the sexes across major occupa- tional categories. Table 1-1 provides com- parisons for black and white women and men across 13 broac! occupational categories in 1984. When one looks at detailed occupa- tional categories, sex segregation is still clearer. In 1980 among 503 occupational cat- egories, the most detailed level at which census data is tabulated, workers in 187 cat- egories were at least 90 percent members of one sex; 275 occupations were composed of at least 80 percent female or male workers (computed from U. S. Department of Com- merce, Bureau of the Census, 1983b:Table 11. Almost half of all employed women work in occupations that are at least 80 percent female (Rytina, 1981), which include librar- ians, health technicians, secretaries and typ- ists, data-entry keyers, nurses, bank tellers and bookkeepers, telephone operators, sew- ers and stitchers, child care workers, and Tenth assistants (U. S. Department of Com- merce, Bureau of the Census, 1983b). The occupation of most women not in the labor force, homemaker, is one of the most see regaled occupations. Slightly over half of all men work in occupations that are at least 80 percent male. Among these 229 predomi- nantly male occupations are engineers, ar- chitects, natural scientists, physicians and dentists, lawyers, nonretail sales represen- tatives, mail carriers, electrical and elec tronic equipment repairers, construction workers, machinists, motor vehicle opera- tors, and freight, stock, and material han- dlers. There is some division of labor by sex in most societies (Burton et al., 1977~. Across all societies, moreover, there is a pattern to this division of labor. Women generally do those tasks that are compatible with child care" tasks that are not dangerous, do not take them far from home, do not require close attention, and are readily interrupted a Brown, 19701. As consistent as this pat- tern is, it is not unmodifiable. In societies in which women must do work incompatible with breast-feeding, for example, babies are started on breast milk substitutes earlier (Nerlove, 1974~; where women's work re- quires them to travel distances, as they do to gather vegetable food in hunting/gath- ering economies, or to participate in long- distance trading networks, they leave chil- dren with substitute caretakers. Ibe only universal with no exception seems to be that everywhere, it is primarily women who mother. Within the limits of female-assigned chilc! care and sexual dimorphism in strength and energy, there is a great deal of variability across societies as to which gender is ex- pected to do what job, even in the West. For example, dentists are primarily female in Denmark, Poland, and the Soviet Union, in contrast with the United States, where dentistry is 93 percent male (computed from U. S. Department of Commerce, Bureau of the Census, 1983b: Table 11. In the Soviet Union, both physicians and street cleaners are usually female (Lapidus, 19781. Beyond industrial societies, there is yet more vari- ability. Household servants, predominantly female in the West, are typically male in India (Blumberg, 1978), and construction la- bor is shared by the sexes (Eloserup, 1970~. West African women engage in highly or- ganized long-distance trading that is else- where an exclusively male occupation (Ham- mond and Jablow, 19761; and, as do women
8 WOMEN'S WORK MEN'S WORK in other horticultural economies, they hoe the fields. In hunting/gathering societies, women commonly do virtually all the gath- ering of vegetable food, which is the dom- inant source of subsistence in halfthe known ethnographic cases (FriedI, 1975:13~. This degree of cross-societal variation in the sex division of labor, and even reversal of what is traditionally considerec! men's or women's work from society to society, suggests that most occupational sex typing is highly influ- enced by cultural constructions of gender. The degree of cross-national agreement that is also observed suggests that many cultural values are shared. Ike division of labor and sex segregation in work changes with time. lIistorical evi- dence shows change in the sex typing of many specific occupations (Davies, 197S; Tyack and Strober, 1981; Kessler-Harris, 1982~. Since World War II several occupa- tions in the United States have changed sharply in their sex composition; for exam- ple, bank teller, insurance adjuster, and real estate agent have all changed from male to female. Sex segregation in employment, however, seems to be deeply ingrained in cultural beliefs and well established in the organization of work. Occupations change their sex typing, but segregation remains. The aggregate amount of sex segregation across occupations, as measured by the in- dex of dissimilarity, has been virtually stable since 1900 until 1970. Sex segregation in the workplace takes many forms. In addition to the most obvious form, occupational segregation, men and women in the same occupation often work in different industries or for different em- ployers.3 Establishment and industry seg- regation are common, and they occur even when occupations are integrated. For ex- ample, the occupation waiter includes men and women, but many restaurants hire all 3 See Blau (1977) for a thorough discussion of different types of sex segregation in employment. men or all women. Industries have also been found to be more segregated than would be expected from their occupational mix, in- dicating additional segregation beyond that measured by occupational segregation alone. For example, clerical workers, a large and diverse category that is approximately two- thirds female, are more likely to be male in some industries than in others. Because the sex composition of occupations differs in dif- ferent establishments and industries, aggre- gate measures of occupational segregation underestimate the degree of segregation in the world of work. Aggregate measures of occupational seg- regation underestimate segregation for an- other reason as well. Occupational cate- gories are themselves aggregates, composed of smaller categories, some of which may be even more segregated. For example, wom- en were 59 percent of all workers in service occupations in 1980, but they made up 95 percent of all private household workers and 12 percent of all workers in protective ser- vice occupations, both of which were sub- categories of service occupations (computed Tom U. S. Department of Commerce, Bu- reau of the Census, 1983b:Table 1~. Cad- culations based on even the most detailed census occupational classification (the three- dig~t level) underestimate the amount of seg- regation because each category sometimes combines occupations with widely different sex ratios.4 The Dictionary of Occupational Titles(U.S. Department ofLabor, 1977~lists 4 The Census Bureau categorizes occupations at vary- ing levels of detail. The broadest classification includes 13 major categories (see Table 1-1), recently modified from 11 (see Rytina and Bianchi, 1984, and Bianchi and Rytina, 1984, for a discussion of comparability with earlier census years). A somewhat more detailed clas- sification, sometimes referred to as the two-digit Cen- sus Bureau categories, included 44 occupations in 1970. The classification referred to as detailed or three-digit included 441 occupations in 1970 and 503 in 1980. When we refer in the text to "detailed census occu- pational categories" or "three-digit census occupations" we mean this refined classification.
SIGNIFICANCE OF SEX SEGREGATION IN THE WORSE . _ over 12,000 unique job titles, which rep- resent an aggregation of perhaps 1 million jobs done by 115 million members of the labor force (Miller et al., 1980~. Thus, even the 500 detailed occupations classified in the census or the Current Population Survey involve substantial aggregation. Because measures of occupational segre- gation underestimate segregation in work, it would be very desirable to have data for jobs, rather than occupations, in order to be able to assess the extent of and changes in segregation accurately. A job can be defined as a particular task within a particular work group in a particular company or establish- ment perfor~ned by one or more individuals (Bridges and Berk, 19781. Examples are check-out clerk at the Indianapolis Speed- way K-Mart store or upholsterer at the Boeing plant in Renton, Washington. Distinguishing between job segregation and occupational segregation is critical for reasons other than the tendency of occu- pationally based measures to underestimate the true amount of segregation. Most im- portant, the processes that contribute to oc- cupational segregation may differ from those that produce job segregation. Theories that focus on workers' choices are concerned with occupational outcomes, but hiring decisions occur at the establishment level and must be explained with data on men's and wom- en's access to jobs. In addition, focusing on occupational segregation may imply differ- ent remedies than those suggested by an emphasis on jobs. These differences and the committee's ultimate concern with the total amount of sex segregation in the workplace dictate focusing on jobs when the data per- mit. Often, however, constraints imposed by available data or research limit our focus to occupations. THE CONSEQUENCES OF SEX SEGREGATION IN EMPLOYMENT lbe consequences of sex segregation in the workplace extend beyond the symbolic Act of its existence. Society, the economy, 9 and individuals aD lose when workers are allocated to jobs on the basis of character- istics such as gender, race, or age rather than on their ability to perform the work. Seg- regation necessarily restricts individuals' chances for self-fi~fi~Iment. When jobs are classified as men's work or women's work, neither men nor women are Dee to do the jobs that might best suit them. Because it has made substantial investments in devel- oping its members' abilities, society as well as its individual members lose when workers are assigned to jobs on the basis of their gender rather than their talents. To the ex- tent that involuntary job segregation re- stricts employment opportunities for oth- erwise qualified workers, it represents the failure of the economy to make use of the available labor supply most appropriately. Ike misallocation of human resources in the work force necessanly depresses national productivity, and the loss in productivity that job segregation entails will increase if it persists at current levels at the same time that more women attain advanced education and their expected work life increases. To the extent that cieclines in the sizes of ad- olescent and young adult cohorts (U. S. De- partment of Commerce, 1981) and the num- bers of high school graduates (U.S. De- partment of Education, 1981) reduce the traditional supply of new workers for skilled and technical jobs, labor shortages may well occur unless these jobs are open to talented inclividuals irrespective of their gender. Although gender affects what jobs are available to persons of both sexes, segre- gation is more harmful to women primarily because the occupations held predominant- ly by women are less clesirable on various dimensions than those held predominantly by men. In particular, segregation contrib- utes to women's lower wages. Female-clom- inate~l occupations also provide less on-the- job training and fewer opportunities for mo- bility. Ilese and other consequences of sex- segregated work careers also follow women into retirement. Because wage conse- quences are so important, we enumerate
10 WOMEN'S WORK MEN'S WORK several ways in which wages are affected by job segregation; we then take up other con sequences. Wage Consequences Occupational sex segregation, sex segre- gation across firms, and job segregation within firms all recluce women's earnings relative to men's. Occupational Segregation and Wage Disparity In the United States, Airtime, year-round white female workers on the average earn approximately 60 percent as much as fi~- time, year-round white male workers, a ratio that has been almost constant for at least the last 25 years. According to the report of the National Research Council's Committee on Occupational Classification and Analysis (Treiman and Hartmann, 1981), a substan- tial part of this overall earnings differential can be attributed to the low wages women earn in predominantly female occupations. For 499 detailed occupational categories in the 1970 census classification, the correla- tion between median annual wage and salary earnings (adjusted for time spent working) and the percentage female among occupa- tional incumbents is -.45: the higher the percentage female, the less an occupation paid. Employment in a female-dominated occupation depressed wages of workers of bow sexes; each additional percentage point female in an occupation was associated with $42 less in median annual earnings. The ex- pectec! median wage in an occupation filled exclusively by women was $3,946, less than half the $8,185 median in exclusively mate occupations (Treiman and Hartmann, 19811. Differences in occupational characteristics (as measured by the Dictionary of Occu- pational Titles) accounted for about 35 per- cent of the gross association with percentage female. The report also estimated the pro- portion of the gross male-female earnings differential that could be attributed to sex segregation among detailed occupations. The analysis indicated that the segregation of men and women into different occupations ac- counts for about 3540 percent of the sex difference in average earnings. The remain- der is due to the fact that within each ~de- tailed occupation men tend to earn more than women. These and other data lecI that committee to the conclusion that occupa- tional segregation has a substantial effect on women's earnings and that, in particular, the wages of female-dominated occupations are depressed relative to what they would be in the absence of segregation. Although the 1980 census data have not yet been analyzed to assess whether the ef- fect of occupational segregation on earnings differentials has changed since 1970 (the year to which the Committee on Occupational Classification and Analysis's estimates per- tain), a crude assessment is possible through the use of published data from the Current Population Survey (CPS). The main defi- ciency of the CPS data is that the relatively smaller sample size requires considerable aggregation of occupational categories. From 1970 through 1974, data on the mean earn- ings of full-time, year-round workers by sex are presented for 24 occupational categories; in 1975 the number of categories was ex- panded to 51. Table 1-2 compares 1970 with 1979 data using the 24L-category classification and 1975 with 1979 data using the 51-cat- egory classification. Probably because of the highly aggregated classifications, only 22 percent of the gender difference in earnings in 1970 can be attributed to occupational segregation, compared with the figure of 35- 40 percent based on the 499 categories of the 1970 census. Ibe comparison across years suggests a slight decrease in the effect of segregation between 1970 and 1979. In 1970 the average earnings of women would have been 90 percent of those of men had women earned as much as men in the same occu- pation; by 1979 this had increased to 92 per- cent. The 1975-1979 comparison based on
SIGNIFICANCE OF SEX SEGREGATION IN THE WORKPLACE 11 TABLE 1-2 Decomposition of Earnings Differentials Between Men and Women Into Within-Occupation and Between-Occupation Components, for Full-Time Year-Round Workers in Selected Years, 1970-1979 24-Category 51-Category Classifications Classifications 1970 19796 1975 1979 Earnings differentials Mean earnings of men $9,918 $19,109 $14,029 $19,109 Expected mean earnings of women if they earned the male average in each occupations 8,975 17,583 12,550 17,230 Mean earnings of women 5,675 10,876 7,930 10,876 Earnings differentials expressed as a percentage of male earnings Mean earnings of men 100% 100% 100% 100% Expected mean earnings of women 90 92 89 90 Mean earnings of women 57 57 57 57 Decomposition of earnings differentials Percent due to occupational segregation Percent due to within-occupation pay differences 22 78 18 82 24 76 23 77 a Each classification was based on the most disaggregated set of categories available. b lbe 51-category 1979 classification was aggregated to 24 categories to match the 1970 classification by taking a weighted average of the mean earnings of men or women in each of the component categories, weighted by the number of men in each component category. The number of women in the aggregated category is just the sum of the number of women in each component category. c lye means for male workers had to be estimated for three categories (secretaries and stenographers, typists, and private household workers, aggregated to two in the 24-category classification) for which there were too few incumbents for the CPS to be willing to report a mean. For 1979 and for the first two categories in 1975, means were calculated directly from the published distribution by scoring the income categories at their midpoint. In the mice of secretaries and stenographers in 1979, an outlier an estimated 2000 cases with annual earnings of $60,000-$75,00() was omitted from the computation. For 1970, and for private household workers in 1975, means were estimated by assuming that the category mean to be estimated bore the same ratio to the mean earnings of the total male labor force as it did in 1979. SOURCE: Current Population Reports, Series P-60: No. 80, Table 55; No. 105, Table 52; No. 129, Table 58. 51 occupational categories yields similar re- sults although the changes are not as large, perhaps because of the shorter time period. Job Segregation and Wage Disparity Sex segregation within occupations fur- ther contributes to the earnings gap. As we noted above, sex segregation occurs within occupations because men and women who perform the same occupation may be seg- regated by firm or enterprise, and because within firms men and women in the same occupation may do different jobs. That ex perlsive restaurants almost always employ men to wait on tables, while inexpensive restaurants and coffee shops are much more likely to hire women, is an example of be- tween-firm segregation. Since expensive es- tablishments pay better and provide larger tips, male waiters earn more than female waiters. An example of within-firm segre- gation contributing to the income gap is the assignment of men to higher-paid night work. The evidence regarding the consequences of job segregation for earnings differentials is sparse. The small number of studies must be regarded as suggestive rather than de
- WOMEN'S WORK MEN'S WORK finitive, especially since none estimates the proportion ofthe earnings differential within occupations that can be attributed to job segregation. Segregation Across Firms Although no economy-wide quantitative estimates of the effect on earnings differences of the segre- gation of men and women in the same oc- cupation in different firms are available, ex- a~nples suggest that such segregation is an important source of the wage gap. Several studies show that within specific occupations women and men tend to be employed in different establishments, with better-paying firms disproportionately employing men. Blau (1977) has shown that within sex-in- tegrated occupations, such as accounting clerk, men tend to earn more than women because they tend to work in higher-paying firms. She found that more of the male-fe- male wage differential within each occupa- tion was between rather than within firms. That is, men were overrepresented in high- er-paying firms, which hired fewer women across ad the occupations she studied. Buck- ley (1971) and McNulty (1967) found similar results for clerical jobs, as did Talbert and Bose (1977) for retail clerks, Allison (1976) for beauty salon operators, and Johnson and Stafford (1974) and DarIand et al. (1974) for college and university faculty members. Other evidence of the importance of this type of segregation for earnings comes from industry data; industries provide a crude proxy for firm differences. Calculations by Malveaux (1982a) show that industrial sex segregation accounted for 13.5-27.5 percent of the wage gap within broad (one-digit) oc- cupational categories when industrial dis- tribution was also controlled at the one-digit level. Segregation Within Firms It is not possible to estimate the overall wage elect of the segregation of men and women into different jobs within the same occupation and firm, but some evidence suggests that this type of segregation also contributes to earnings differences because men and women filling the same occupation within firms are as- signed different specific jobs at different pay rates. The most direct evidence comes from the work of Bridges and Berk (1978), who found that white-colIar female-dominated jobs in Chicago financial firms paid less largely because they did not compensate in- cumbents' qualifications and job character- istics at the same rate as did male-dominated jobs. Almost three-quarters of the $2,250 annual wage disparity was due to differential payment for qualifications and job charac- teristics, while differences in the mean qual- ifications of workers in male- and female- dominated jobs accounted for a little more than $300 of the wage gap. Similarly, Talbert and Bose (1977) found that male retail sales clerks were more likely to be assigned to "big-ticket" departments of stores (e.g., filr- niture, large appliances) and hence earned more on average than female clerks. (Inter- estingly, however, there was a strong in- teraction between gender and department in determining earnings; department mat- tered more for men than for women, so that the earnings gap between the sexes was greater in the big-ticket departments.) Using 1960 data, Halaby (1979b) found that female managers in a large public utility firm earned on average 64 percent as much as male managers. Only 9.7 percent of the earnings gap could be explained by differ- ences in levels of education, seniority, and previous work experience, while 75.3 per- cent was explained by the difference in re- turns to human capital for men and women. In analyzing the source of differential re- turns to human capital, Halaby found that men and women were largely segregated into different managerial "ranks." While more than 94 percent of women were in ranks V and VI (the lowest ranks), more than 85 percent of men were in rank IV or above. When rank was entered as an explanatory variable, the difference in distribution of men and women across ranks alone explained 65 percent of the wage gap, reducing the effect of differences in human capital levels and
SIGNIFICANCE OF SEX SEGREGATION IN THE WORKPLACE 13 returns to 27.3 percent. lIalaby concludes that rank segregation severely restricts women from transforming their stocks of hu- man capital into higher salaries. Between 45 and 55 percent of the men's earnings ad- vantage among professional employees in a large research organization that Malkie} and Malkie} (1973) studied could be attributed to the greater tendency for men to be as- signec] or promoted to higher rank, even if attributes thought to be related to produc- tivity (post-high-school education, college field of study, job-related labor market ex- perience, rate of absenteeism, and personal productivity as measured by number of pub- lications) are controlled for. The remainder of the earnings gap stemmed from sex dif- ferences in these variables. An important mechanism in producing wage differences is the propensity of firms to assign men ini- tiaDy to higher ranks (Cabral et al., 1981; Newman and Wilson, 1981; Harlan and 0'- Farrell, 1982) or to promote them more rap- i~y than similarly qualified women (Com- mittee on the Education and Employment of Women in Science and Engineering, 1983~. Judging from the available evidence, within-firm pay differences for workers in the same occupation appear mainly to reflect differences in rank. Progress in studying the effect of job seg- regation on earnings differentials will de- pend on the availability of much more de- tailed data than is available from the census. Such data are nearly always limited to stud- ies of single industries or single enterprises. Despite their limited generalizability to the labor force as a whole, studies of industries and enterprises that make use of very de- tailed job classifications should be encour- aged because they illuminate the processes of segregation. Other Consequences of Sex Segregation Wages are but one aspect of the negative consequences for women of sex segregation in the labor market. lob segregation also contributes to sex differences in retirement income, susceptibility to unemployment, on- thejob training, occupational and status mo- bility, prestige, stress, power, and the di- vision of labor within the household. Retirement Income Women are less likely than men to be covered by private pension plans 40 per- cent offull-time women workers are covered by such plans, compared with 55 percent of similar men (D. Beller, 1981 - partly be- cause they are concentrated in low-wage firms and occupations and less profitable in- dustries that are less likely to provide pen- sion coverage (Benson, 1980~. Although the sex difference in pension coverage is also partly due to women's shorter average ten- ure in their current jobs, it remains consicI- erable even after controlling for years em- ployed narrowed the coverage gap between the sexes (D. Belier, 1981~. Twenty-five per- cent of all women Chaff of those employed in the private sector) work in retail and ser- vice industries, which have the lowest pen- sion coverage (U.S. Department of Labor, Labor-Management Services Administra- tion, 1980~. Only 10.5 percent of women over 65 received money from private pen- sions, compared with 27.7 percent of men (Moss, 1983~. Of course, because women earn less than men, both their social security and retirement benefits are lower (Moss, 19831. Susceptibility to Unemployment The link between occupational sex seg- regation and unemployment is not straight- forward. Until 1981 women's unemploy- ment rates typically exceecled men's by 1-2 percentage points (Lloyd and Niemi, 1979; U. S. Department of Labor, Bureau of Labor Statistics, 1981a, 1981b). However, in late 1981 the adult male unemployment rate sur- passed that of women by 0.1-0.5 percentage points (U. S. Department of Labor, Bureau of Labor Statistics, 1982a). Their different occupational distribution exposes men to a
14 WOMEN'S WORK, MEN'S WORK greater risk of cyclical unemployment than women (Barrett and Morgenstern, 1974; Niemi, 1974; R. Smith, 1977; Urquhart and Hewson, 19831. Women are concentrated in clerical and service occupations and indus- tries, which are less cyclically sensitive than the predominantly male blue-collar occu- pations in manufacturing and construction. Indeed, Cornfield (1981) and Blau and Kahn (1981a) found that women's occupational and industrial distributions contribute substan- tially to their Tower layoff rates relative to those of men. Much of the beneficial effect of women's concentration in occupations and industries that are less vulnerable to cyclical unem- ployment, however, is cancelled out by their greater propensity to be labor force en- trants, which subjects them to high unem- ployment rates. Moreover, although wom- en's concentration in certain occupations or sectors recluces their aggregate risk of un- employment, women who work in female- dominated occupations are unemployed sig- nificantly longer than are other women (Bar- rett and Morgenstern, 19771. In addition, women's occupation-specific unemploy- ment rates continue to exceed men's within many occupations (Urquhart and lIewson, 1983~. With the exception of the 1980 reces- sion, women in manufacturing and blue-col- lar occupations have been more likely to be laid off then men during a recession (Terry, 1982). Women who have recently entered male- dominated occupations are especially vul- nerable to layoffs during economic down- turns (Kelley, 19821. During the recent recessions in the United States and Europe this was true for women in certain blue- colIar occupations, such as durable goods manufacturing (O'Neill and Smith, 1976; R. Smith, 1977~. In recent federal "reductions in force," women in positions with ratings of GS 12 or above were laid oE at a rate 2.3 times the average rate, presumably primar- ily because they had less seniority, although veterans' preference also protected men (Federal Government Service Task Force, 19811. Thus, in the short run, reducing seg- regation would place women in more cycI- ically sensitive sectors or occupations, but in the long run it would probably increase their labor force attachment and thereby re- duce both the male-female unemployment clifferential and the overall sex difference in labor force participation. On-the-Iob Training On-thejob training offers workers the op- portunity to acquire skills that facilitate oc- cupational mobility and wage increases (Mincer, 1962b). Thurow's (1975) charac- terization of the labor market as a training market in which training slots are allotted to workers recognizes the importance of ac- cess to training. But women tend to receive less training than equally experienced men, their jobs involve shorter training periods (Duncan and Hoffman, 1978, 1979), and, among federal employees at least, their training costs less per hour (Taylor, 1985~. Evidence that sex segregation accounts for these differences is indirect and thus only suggestive. For example, since training is usually reflected in more rapid wage gains over time, the flatter experience-wage pro- files ZelIner (1975) observed for female- dominated occupations is consistent with the finding of less on-thejob training. Occupational and Status Mobility Research on sex differences in occupa- tional mobility suggests that, in part because of occupational segregation, women expe- rience less career mobility than men. For example, Rosenfeld and Sorensen (1979) found that most of the difference in men's and women's chances to move between par- ticular sets of occupations was due to dif- ferences in their distribution over occupa- tional categories. Using Duncan's socioec
_IGNlFlCANCE OF SEX SEGREGATION IN THE WORKPLACE 15 anomie index (SEIJ5 to measure occupational status, Wolf and Rosenfeld (1978) found that women experienced less upward SKI mo- bility than men over a five-year period. Leaving the male sector increaser! the like- lihood of a prestige loss for both sexes. Nei- ther men nor women who shifted from one female-dominated job to another were likely to experience upward SKI mobility, al- though starting in a female-dominated oc- cupation did not reduce the mobility chances of either sex, provided they moved to a non- female-dominated occupation.6 However, 80 percent of the men and only 31 percent of the women made such moves. Research- ers who assessed mobility in terms of wage changes found that men's earnings rose fast- er than women's (Rosenfeld, 1980; Blau and Kahn, 1981b). Because femaTe-dominated occupations have characteristically shorter career lad- ders, i.e., opportunities to advance in pay and status from entry-level positions, wom- en often attain their maximum level within a few years. Typically female entry-level jobs, such as telephone operators or stitchers (Grinker et al., 1970) tend to be on shorter ladders Can typically male entry-level jobs (Blau, 1977; Kanter, 1977; Stevenson, 1977; New York State Commission on Manage 5 The socioeconomic index (Duncan, 1961) was con- structed to measure occupational prestige. It computes occupational prestige on the basis of the salaries and the educational attainment of incumbents of occupa- lions. England (1979), Boos (1981), and others have criticized the use of the SKI to compare the sexes be- cause it does not take into account differences in the kinds of occupations women and men typically hold. 6 Although Wok Ed Rose~eld (1978) found no evi- dence that men changing jobs within the "female sec- tor" had more SKI mobility than similar women, men may have an advantage in some [emale-dominated oc- cupabons. For example, Grimm and Stern (1974) found that men were overrepresented in higher-status and administrative jobs in teaching, nursing, academic li- brarianship, and social work, and Fox and Hesse-Biber (1984) confirmed this Ending for a larger number of professions more recently. ment and Productivity, 1977; Peterson-Hardt and PerIman, 1979; C. Smith, 1979; Ratner, 1981; EIaignere et al., 1981), and women in typically male occupations may be assigned to jobs that offer few promotion opportun- ities (Martin, 1980; Hochschild, 1975; Ep- stein 1970b). As a result, short-term comparisons un- derestimate long-term differences in the probability of upward mobility. When Sew- ell et al. (1980) observed occupational mo- bility over a longer period, the women be- gan in occupations with higher SKI scores, but 18 years later the men had surpassed ~em, and married women with children had actually lost ground. Even childless women gained little occupational status over the course of their working lives, and never- married women gained only one-third as much as did men (Sewell et al., 1980~. Mar- ini (1980) also found that after controlling for education and labor force experience, women showed very small gains in occu- pational status between their first and a sub- sequent job, while men's occupational status increased over time. Some of the sex dif- ference was due to the differential ability of men and women to benefit from their ed- ucation and employment experience. This difference stems from both their different concentrations in occupations that reward these personal resources differently and the tendency of some employers to hold women to higher promotion standards than men (Ol- son and Becker, 1983~. Occupational Prestige Several studies (see Bose and Rossi, 1983; Jacobs and Powell, 1983, for reviews) sug- gest that workers in sex-atypical occupations do not have the occupational prestige ac- corded sex-typical incumbents of the same occupations. For example, Jacobs and Pow- ell (1983) found that the more an occupation was dominated by one sex, the greater the discrepancy between the prestige that raters accorded to sex-typical and sex-atypical job
16 holders. Given the differing amounts of prestige accorded to male and female in- cumbents in the same occupation, a move that would represent upward prestige mo- bility for men might mean downward mo- bility for women. According to the prestige ratings their respondents assigned sex-typ- ical and sex-atypical workers, florist to plumber and typist to electrician represent such moves (Powell and Jacobs, 19841. lob Stress Across-the-board comparisons inclicate that women and men find their jobs equally satisfying (U. S. Department of Labor, Em- ployment and Training Administration, 1979c). But sketchy evidence suggests that some female-dominated occupations may be more stressful. Secretaries, for example, had the seconcI-highest incidence of stress-re- lated diseases among workers in 130 occu- pations studied by the National Institute of Occupational Safety and Health (NIOSlI) in 1975. A 1980 study revealed that data-entry clerks who operated video display terminals fills time exhibited the highest stress levels of any occupational group NIOSH had ever studied, including air traffic controllers (cit- ed in Working Women Education Fund, 19811. Haynes anc! FeinIeib (1980) found Hat coronary heart disease among participants in the longitudinal Framingham Heart Study was about twice as common among female clerical workers who had children as among other women workers or housewives. Sup- pressed hostility, a nonsupportive boss, lit- tle job mobility, and a blue-collar husband were all associated with coronary heart dis- ease among clerical workers, presumably because they contributed to increased stress. However, knowledge of the effects of oc- cupational segregation on workers' levels of psychological stress is very sketchy at this time. WOMEN'S WORK MEN'S WORK Power and Work Within the Family Job segregation and the resulting differ- ences in earnings may influence women's home lives by affecting the distribution of power between marriage partners and the division of househoIc! labor. No studies have tested these suppositions directly, however. McDonald's (1980) review of studies of fam- ily power notes that resource theory pro- vides the principle framework for such stud- ies. Most posit a material base for marital power, supplemented by ideology or psy- chological factors. Some evidence supports a connection be- tween wives' employment and material power within the family; Rainwater (1979) suggests that wives' employment influences family consumption patterns, away from "male" goods such as sporting equipment and toward "female goods" such as home appliances. In contrast, time use studies based on data from the late 1960s (Meissner et al., 1975; Walker and Woods, 1976; Va- nek, 1980) show little if any increase in hus- bands' contribution to household work when their wives are employed. More recent time budget studies based on data from the mid- 1970s (fleck with Rustad, 1981; Berk, 1979; Stafford! and Duncan, 1979) reveal a slight convergence in the amount oftime husbands and wives spent in family roles and in total work time (both paid and family). But the slight increase in husband's family time is not linked to wives' employment, since hus- bands' time in family roles does not vary with their wives' work time. Although work- ing wives have reduced their family time, particularly housework, substantially in re- cent years, women still do the vast majority of housework. As Moore and Sawhill (1978), Hartmann (1981), and others have notecl, women have talcen on a new set of activities without forgoing their traditional responsi- bilities. The household division of labor appears to share with job segregation a resistance to
SIGNIFICANCE OF SEX SEGREGATION IN THE WORKPLACE - change, and the two are likely to be mutually reinforcing. The failure of husbands' house- hold time to respond to their wives' paid work may contribute to their wives' choices regarding paid work. Women's choices both contribute to and result from occupational segregation, and segregation reduces the re- sources women bring to the marital unit and thus, potentially, their power in the house- hold. CONCLUSION We have reviewed evidence that shows that sex segregation in employment has sig- nificant consequences for women, men, families, and society- but particularly for women. It contributes to women's low wages and lesser employment-related benefits of all kinds, and some have argued that it con- tributes to a household division of labor that also seems to disadvantage women. The neg- ative consequences of sex segregation in em- ployment are likely to increase, if sex seg- regation does not decline as more women work for wages and families come to in- creasingly rely on their earnings. Sex seg- regation in employment, as we use it in this report, generally refers to any observed dif 17 ference in the distributions of women and men across job categories. Some of the dif- ference observed may not be problematic for women or society, because it results from a voluntary sorting out of people and jobs. To the extent that the difference is volun- tary, it may not be an appropriate object of public policy. Sex segregation is only one manifestation of unequal opportunity in the workplace. Women's lower earnings and such phenom- ena as sexual harassment and unequal fringe benefits are others. In our view, job seg- regation is among the most significant. And perhaps most important, to the extent that sex segregation in the workplace connotes the inferiority of women or contributes to maintaining women as men's inferiors, it has great symbolic importance. We believe that sex segregation is fundamentally at odds with the established goals of equal opportunity and equality under the law in American so- ciety. Therefore we focus on the factors af- fecting the occupational outcomes of wom- en. Our emphasis is on why women end up in a small number of less remunerative oc- cupations and how to alter these outcomes in order to improve women's occupational opportunities.