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.
Trends In Occupadonal Segreggdon by Sex and Race, 1960 1981 ANDREA H. BELLER Interest among economists in occupa- tional sex segregation stems from the fairly well established relationship between the sex differential in earnings and women's concentration in a small number of occu- pations. It also stems from a family-based analysis of women's roles, although this con- nection continues to be controversial. Such an analysis says that because of their family roles, women invest less in market-oriented human capital than men do (Becker, 1981), and this includes choosing traditionally fe- male occupations (Polachek, 19791. Recent empirical studies tend to refute this expla- nation of sex differences in occupational choices (Belier, 1982b; Corcoran and Dun- can, 1979; England, 19821. While untan- gling the causes of occupational sex segre- gation has proved an ambitious challenge, measuring its trends is no less difficult. This paper assesses the trends in occu- pational segregation of the sexes during the 1970s and compares them with those of the 1960s. A number of studies have examined changes in occupational segregation be- tween census years: 1900-1960 (Gross, 1968), 1950-1970 (Blau and Hendricks, 1979), 1960- 1970 (Economic Report of the President, 11 1973), and 1950-1970 among professional oc- cupations (Fuchs, 1975~. Using the index of segregation from the Duncan Index (Dun- can and Duncan, 1955), these studies con- cur.in the relative lack of change noted in occupational segregation through 1960 and the small decline during the 1960s. (The decline of sex segregation in the professional occupations during the 1960s was somewhat greater than that for all occupations.) With the strengthening of equal employ- ment opportunity (EEO) legislation and the promulgation of equal educational oppor- tunity legislation in 1972, one might have anticipated an accelerated decline in occu- pational segregation during the 1970s. Moreover, there is a general perception that many women are becoming increasingly ori- ented toward nontraditional family roles and nontraditional jobs in the workplace.) Sur- prisingly, the index of segregation remained unchanged through 1976 or 1977, according to two recent works (Lloyd and Niemi, 1979; ~ See, for example, Cherlin and Walters (1981) and Mason et al. (1976~.
12 ANDREA H. BELLER U. S. Commission on Civil Rights, 19781. The segregation indexes computed by these studies (as well as by Blau and Hendricks and by Fuchs) are presented in Appendix B. Table By-. But findings from other studies seem to conflict with these reports of no change. Beller (1982b) showed that EEO laws reduced occupational segregation by 1974 and that EEO laws combined with equal educational opportunity legislation in- creased the effects of years of college com- pleted on women's entry into nontraditional occupations between 1971 and 1977, espe- cially among new entrants (Belier, 1982a). The reason these studies detected no change is a lack of comparability between the two data sets they used to compute the segre- gation indexes. In fact, I have found that the index of segregation declined from 68.32 in 1972 to 64.65 in 1977 and 61.66 in 1981, a rate of decline almost three times as large as that during the 1960s. In the next section, the trends in occu- pational segregation from 1971 to 1981 are documented and compared with those of the 1960s. Trends in segregation among all oc- cupations, among professional occupations, and among college majors are discussed. An analysis of cohort differences in occupational segregation during the 1970s follows. I then compare and contrast changes in the sex composition of detailed occupations during the 1960s and the 1970s. Finally, race dif- ferences in trends in occupational segrega- tion are presented. MEASUREMENT AND DATA Trends in occupational segregation are commonly measured by the index of seg- regation (Duncan and Duncan, 19551. The index is defined as follows: St = i/22 mitfit | where mi' = the percentage of the male labor force employee! in occupation i in year t, and fit = the percentage of the female labor force employed in occupation i in year t. The index may take on a value between O and 100, where zero represents perfect integration and 100 represents complete segregation. The number tells the propor- tion of women (or men) that would have to be redistributed among occupations for the occupational distribution to reach complete equality between the sexes. The index of segregation has two com- ponents, labeled the mix effect and the com- position effect by Blau and Hendricks (1979~. The value of the index depends on both the relative size of various occupations and the sex composition within occupations.2 Changes in the index thus derive from two sources: changes in the occupational distribution and changes in the entry of the sexes into various occupations. (It also depends on the inter- action of the two.) These changes may be in reinforcing or opposing directions. Signs of progress within occupations, for example, could be masked by unusual growth in oc- cupations that are predominantly single sex. A standardization procedure can be used to determine the influence of each of these two effects. For example, to determine the ef- fect of changes in the sex composition within occupations on the changes in the segre- gation index from year t - 1 to year t, the index of segregation for year t can be com- puted standarclizing the size of occupations to year t - 1. Thus, the employment stand- ardized index of segregation holds constant the distribution of employment across oc- cupations (occupational mix) and enables one 2 The value of the index may also depend on the degree of aggregation of the occupations. Typically, the greater the degree of aggregation, i.e., the fewer the occupations, the lower the level of measured segre- gation. For this reason, in comparing indexes over time, one should use the same number of occupations at the same degree of aggregation. This methodological issue is discussed in England (1981).
TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 13 to observe the effects of changes in the sex composition within occupations alone.3 Similarly, to observe the effects of changes in the occupational structure alone, the pro- portion standardized index of segregation can be computed holding constant the sex com- position at year t - 1 and using the em- ployment distribution of year t. (Standard- izing by the size of occupations in year t - 1 arbitrarily assigns the interaction term in one direction. Standardizing by the size of occupations in year t would assign the in- teraction term in the other direction. ~ These standardization procedures can also be ap- plied to a given year to determine how the index of subgroup j differs from that of the population as a whole. This allows us to de- compose the index of the subgroup into the effects of occupational mix and sex compo- sition. For example, by standardizing the segregation index of the youngest cohort to the occupational mix of the whole labor force, it can be seen how the sex composition within occupations for the youngest cohort differs from that of the rest of the labor force. To assess trends in occupational segre- gation during the 1970s, I used data from the Current Population Survey (CPS) con- ducted monthly by the U.S. Bureau of the Census. Data for the years 1971-1974 and 1977 used for the detailed analyses pre- sented in this paper come from the March Annual Demographic Files (ADF) of the CPS. Me ADF data are supplemented here with more recent data from the Bureau of Labor Statistics' (BLS) annual averages (AA), tab- 3 The employment standardized index of segregation is defined as follows: Si*t = i/22 | m* - ~ i, where ml*` = (M`,lT`~) (Tan) (100~/~ (MiiTi`) (Ti`-~), fit = (F`,lT`~) (T`~_~) (100~/~ (F,jTit) (Tit_~), Fit = the number of females in occupation i in year t, M`' = the number of males in occupation i in year t, and T`` = F.` + M`` = total employment in occupation i in year t. ulated by the BLS from the monthly CPS.4 More detail on these sources and on issues of comparability and the choice of occupa- tions included in the sample are discussed in Appendix A. As discussed there, the CPS occupational data collected during the 1970s are not comparable to the 1970 census data even though the same occupation codes are used, because the Census Bureau changed its method of assigning individuals to oc- cupations in December 1971. Hence, sta- tistics based on these two sources should not be compared. Although their reliability dif- fers (see Appendix A), I make some com- parisons between the two different sources of CPS data in order to include 1981 data in the analysis. The 1960 and 1970 census occupational data are used to show trends during the 1960s; these data were made comparable by the Census Bureau's recod- ing of the 1960 data according to the 1970 occupation codes (U. S. Bureau of the Cen- sus, 1972~. TRENDS IN OCCUPATIONAL SEGREGATION, 1960-1981 Occupational segregation of the sexes de- clined continuously during the 1970s at a 4 Data on occupations for 1971-1974 and 1977 come from the 1972-1975 and 1978 Annual Demographic Files, which are available on public-use computer tapes. These files contain considerable demographic detail, making it possible to cross-classify occupation by such char- acterisffcs as labor market experience, which is done later in this paper. These were the only years for which I had these data at the time of this writing. To incor- porate more recent data than 1977, I obtained from the BLS unpublished tabulations of annual averages (AA) for 1981; to ascertain comparability between the AA and the ADF data, I also obtained these tabulations for 1977 and for 1972, the earliest year for which they are available. (The cooperation of Elizabeth Waldman and Jack Bregger of the BLS, who made these data available expeditiously, is gratefully acknowledged.) These data are not cross-classified by demographic characteristics. Thus, while overall trends can be as- sessed through 1981, cohort trends can be assessed only through 1977.
14 ANDREA H. BELLER rate that exceeded the decline during the 1960s. Ike index of segregation declined Tom 68.32 in 1972 to 61.66 in 1981, according to the BLS AA data. According to the ADF, it declined Tom 68.14 In 1971 to 64.15 In 1977, and the decline occurred continuously over the intervening years. These indexes, which are computed over a common group of 262 three-digit census occupations, appear in Table 2-1, lines 2 and 3. For comparison purposes, indexes of segregation for 1960 and 1970 computed from the decennial cen- sus over the same 262 occupations are in- cluded. According to Census Bureau data, the index declined from 68.69 in 1960 to 65.90 in 1970. As pointed out earlier, the index levels are not comparable across data sets, but trends in one data set should be comparable to trends in the other as long as compara- bility has been established within each data set. Between 1972 and 1981 the index of segregation declined at an average annual rate nearly three times as high as during the 1960s, i. e., - 0.74 compared with - 0.28. These figures appear in the bottom part of Table 2-1. The annual rate of decline in the segregation index appears to have acceler- ated slightly in the mid-1970s and to have remained steady through 1981. To decompose the change in the segre- gation index during the 1970s, I have stand- ardized it to the employment my at the beginning of the decade. As mentioned above, we want to make comparisons only within a data set, so we standardize the Cen- TABLE 2-1 Segregation Indexes, 1960-1981, All Occupations Census 1960 1970 1971 1972 1973 Index Level CPS 1974 1977 1981 Unstandardized Census 68.69 65.90 Annual Demographic File (ADF) - 68.14 67.36 67.09 66.39 64.15 Annual Averages (AA) - 68.32 64.65 61.66 Employment Standardized Census (1970) 68.06 65.90 ADF (1972) 67.99 67.36 66.97 66.64 64.49 AA (1972) 68.32 65.18 62.88 Change in Index Average Annual Rate of Change 1960-1970 1971-1974 1974-1977 1977-1981 1972-1981 -0.28 Unstandardized Census ADF AA Employment Standardized Census (1970) ADF (1972) AA (1972) . NOTE: These indexes are based on a common group of 262 three-digit census occupations. SOURCE: U.S. Commerce Department, Bureau of the Census, U.S. Census of Population: 1970, Detailed Characteristics, Final Report PC(l)-Dl, U.S. Summary (Washington, D.C.: U.S. Government Printing Office, 1973), Table 221; Annual Demographic Files of Current Population Survey, 1972-1975 and 1978, computer tapes; and Bureau of Labor Statistics, annual averages of monthly Current Population Surveys, 1972, 1977, and 1981, unpublished tabulations. -0.58 -0.75 Total Change 1972-1981 _ _ -0.75 -0.74 -6.66 -0.22 -0.45 -0.72 -0.56 -0.60 -5.44
TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 15 sus Bureau data to 1970 and the ADF and AA data to 1972. Employment standardized indexes reveal the amount of the change in segregation that is clue to changes in sex composition within occupations of a fixed size, the composition effects. As shown in Table 2-1, the employment standardized segregation index declined from 68.32 in 1972 to 62.88 in 1981, or by almost as much as the unstandardized index declined. Thus, most of the decline in the segregation index during the 1970s was due to changes in the sex composition within occupations, but the employment distribution also shifted slightly toward a less segregated work force. Professional occupations continued dur- ing the 1970s to be less segregated than the work force as a whole ant! to experience a somewhat larger clecline in segregation. The segregation index for 59 professional occu- pations declined from 59.44 in 1972 to 50.55 in 1981, according to the annual averages data. This yields an average annual rate of decline of nearly 1 percentage point, 0.99. Since these occupations are composed pri- marily of individuals with a college degree, it is instructive to examine an inclex of seg- regation for earned bachelor's degrees con- ferred on men and women by field of study. These data, published annually by the Na- tional Center for Education Statistics (NCES), are based on the distribution of all degrees granted by all accredited degree-granting institutions in the Unitec! States during a specific academic year.5 The segregation in- dex computed over college majors declined s These data are taken from National Center for Ed- ucation Statistics, Projections of Education Statistics to 1988-89, by Martin M. Frankel and Debra E. Gerald, April 1980. flue original sources for the major portion of these data are the annual NCES reports on Earned Degrees Conferred. Further information came from ". . . education and professional associations, experts in other academic areas, and other agencies in the fed- eral government . . ." (as cited in NCES, April 1980, p. 49~. The numbers in this report differ slightly from the ones for the same year published in Earned Degrees Conferred. from 46.08 in 1969 to 35.62 in 1978. The average annual rate of decline in this index is 1.16 per year.6 Thus, segregation by field of study among bachelor's degree recipients declined rapidly during the 1970s, followed by the professional occupations, and, finally, the work force as a whole. COHORT DIFFERENCES IN OCCUPATIONAL SEGREGATION, 1971-1977 Was the decline in occupational segre- gation by sex during the 1970s distributed throughout the labor force or concentrated in groups most able to take advantage of improved access to nontraditional jobs and opportunities for advancement? Beller (1982a) found the effect of equal employment op- portunity legislation between 1971 and 1977 to be largest among college-educated new and recent entrants into the labor market. That is, compared to 1971, the chances of women with 1-10 years of potential labor market experience (new entrants) in 1977 or with 7-16 years (recent entrants) finding em- ployment in a nontraditional occupation in- creased more than for older cohorts in 1977. Recent entrants in 1977, who were new en- trants in 1971, found increased opportuni- ties to move into nontraditional occupations as they aged over this period. To examine changes in occupational seg- regation by cohort, I stratified the labor force by potential work experience, defined as AgeEducation6, as in this previous study. Using the ADF data for 1971 and l9T7, I stratified women and men into groups with the following years of experience (EX- PER): 1-10, 7-16, and 11-40+. I believe that new and recent labor market entrants are best able to benefit from improved op- portunities, and I hypothesize that young 6 The index of segregation did not decline for post- graduate degrees; however, it is at a lower level than for bachelor's degrees (Belier and Han, in press).
6 ANDREA H. BELLER cohorts will show greater changes than older cohorts. If access to nontraditional occupa- tions increases, new entrants will have more opportunities to enter the occupational structure at preferred points than older co- horts with the same education. Since ad- justments in education can only occur with some lag, new entrants also have the great- est opportunities to acquire more education and to alter their field of study in response to perceptions of improved opportunities in the labor market. In general, the educa- tional attainment of younger cohorts of women is higher than of older cohorts, and women are increasingly likely to obtain ad- ditional degrees at all degree levels (Belier and Han, 1984~. Recent entrants in the early stages of their careers can take advantage of new opportunities for advancement.7 Thus, I compared new entrants into the labor mar- ket (EXPER = 1-10) with the rest of the labor force (EXPER = 11-40+), new en- trants in 1971 with new entrants in 1977, and new entrants in 1971 (EXPER = 1-10) with themselves 6 years later in 1977 (EX- PER = 7-161. We can see how segregated the youngest cohort is compared with the rest of the labor force, how segregated the entering cohort is at the beginning com- pared with the mid-1970s, and how much change in occupational segregation the 1971 entering cohort experienced as it aged. The segregation indexes for these expe- rience cohorts appear in Table 2-2. The youngest cohort is less segregated than the remainder of the labor force in both 1971 and 1977, and segregation declined for all 7 While these arguments and data strongly suggest that the results should be stronger for young cohorts, a potential bias in our results exists in that the sex difference in actual experience probably widens with potential experience. Ibus, stronger results for younger cohorts might be related to the actual versus potential experience issue. In the absence of comparable cohort data prior to 1971, it wasn't possible to assess the effect of such a potential bias. TABLE 2-2 Segregation Indexes by Experience Cohort Experience Cohort Unstandardized 1-10 67.47 7-16 69.94 11-40 + 69.36 Standardized to Employment of Whole Labor Force 1-10 67.44 7-16 69.07 11-40+ 69.13 Standardized to Proportions of Whole Labor Force 1-10 7-16 11-40 + NOTE: The indexes are computed on the basis of 258 occupations. Occupations with no employment in any experience cohort were dropped from all groups. SOURCE: Current Population Survey, Annual De- mographic Files, computer tapes. 1971 1977 Change 1971-1977 62.51 64.03 66.31 61.96 64.89 66.65 67.78 68.54 67.84 64.35 63.93 63.69 -4.96 -5.91 -3.05 -5.48 -4.18 -2.48 -3.43 -4.61 -4.15 groups over this period. The differential be- tween the youngest cohort and the rest of the labor force widened over time, how- ever, as segregation decreased more rapidly in the youngest cohort as hypothesized. The segregation index for this group (EXPER = 1-10) declined from 67.47 in 1971 to 62.51 in 1977, or by 0.83 percentage points per year, while the index for the remainder of the labor force (EXPER = 11-40+) de- clined from 69.36 to 66.31, or by 0.51 points per year. During this period the decline is greatest not for the youngest cohort but for the group with 7-16 years of potential work experience. For this group (EXPER = 7- 16) the segregation index declined from 69.94 in 1971 to 64.03 in 1977, or by 0.99 points per year. If we follow the entering cohort in 1971 for 6 years to 1977 (EXPER = 7- 16), we find that the segregation index de- clined within this cohort by 3.44 percentage points from 67.47 to 64.03, or by 0.57 points per year. The entering cohort in 1971 be- came less segregated as it aged through 1977
TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE . . 17 when the entering cohort is less segregated than in 1971. This implies that each entering cohort is less segregated than in the past and experiences a greater decline in seg- regation as it ages. To decompose these cohort changes in occupational segregation, I standarclized each subgroup to the whole labor force in each year. To determine how segregated this group would be if it had the same occupational mix as the labor force as a whole but maintained its own sex composition within occupations, I standardized the segregation index to the occupational mix of the whole (employment standardized). To determine how segre- gated this group would be if it had the same sex composition within occupations as the whole but its own occupational mix, I stand- ardized to the sex composition within oc- cupations of the whole (proportion stan(l- arclized). As it turns out, the occupational mix is quite similar across cohorts; the employ- ment standardized indexes are nearly iden- tical to the unstandardized indexes. What this implies is that, while the occupational distribution does not differ between older and younger generations as a whole, the sex composition within occupations differs sub- stantially between recent and older cohorts. Thus, for example, while approximately the same proportion of the youngest and the older cohorts are accountants, a higher pro- portion of youthful than of older accountants are women. The proportion standardized in- dexes for 1977 show that the youngest co- hort (EXPER = 1-10) would be much more segregated if it had the same sex composi- tion within occupations as the labor force as a whole (64.35 compared to the actual 62.511; symmetrically, the remainder of the labor force would be less segregated if it had the sex composition of the whole (63.69 com- pared to the actual 66.311. Although as a group new entrants have the same occu- pational distribution as everyone else, new female entrants (as well as new male en- trants) are in different occupations than their older counterparts. MAJOR COMPONENTS OF CHANGE IN TlIE OVERALL INDEX, 1972-1981 The specific detailec! occupations contrib- uting the largest amounts to the decline in the index of segregation between 1972 and 1981 based on the AA data are the following: accountants; elementary school teachers; bank officers and financial managers; sales clerks, retail trade; secretaries, not else- where classified (n.e.c.~; telephone opera- tors; typists; sewers and stitchers; delivery and route workers; janitors and sextons; cooks, except private household; child care workers, private household; and maids and servants, private household. Any difference in contribution can be due to either a change in the size of a segregated occupation or a change in the sex compo- sition within an occupation. Two of the tra- ditionally female occupations, private household maids and servants and sewers and stitchers, showed a large decrease in size over the period. Each of these declines took over 1 percentage point off of the seg- regation index in 1981. Other traditionally female occupations that decreased in size are telephone operators and private house- hold child care workers. One traditionally male occupation, delivery and route work- ers, also decreased in size over the period. Although the numbers of secretaries, n.e.c., and elementary school teachers increased between 1972 and 1981, the proportion of the female labor force that crowded into these traditionally female occupations declined from 9.2 to 8.7 percent and from 3.6 to 2.8 percent, respectively. A smaller proportion of the female labor force crowded into the constant-sized female occupations of retail sales clerk and typist in 1981 than in 1972, 4.2 percent as opposed to 5.4 percent, and 2.4 percent as opposed to 3.3 percent, re- spectively. But the female share in these
18 ANDREA H. BELLER occupations did not decline. Women also entered three rapidly growing traditionally male occupations: accountants, bank officers and financial managers, and janitors and sex- tons. The female share in these occupations increased from 21.7 to 38.5 percent among accountants, from 18.7 to 37.4 among bank officers, and from 10.5 to 19.0 among jani- tors. Cooks is a rapidly growing occupation that men are entering in greater numbers than previously; the percentage of females in this occupation declined from 62.4 to 52.3 percent over this period. The segregation index declined despite the fact that some occupations contributed more to segregation in 1981 than in 1972. The occupations that contributed more to the index are primarily rapidly growing fe- male occupations. The largest increases came from registered nurses; office managers, n.e.c.; bank tellers; computer and periph- eral equipment operators; and miscelIa- neous clerical workers. Nurses and bank tellers are both rapidly growing predomi- nantly female occupations. The field of mis- cellaneous clerical workers is both growing and becoming increasingly female as are the fields of office managers and computer and peripheral equipment operators. Changes in segregation during the 1970s may be summarized as follows. While women continued to enter some of the traditionally female occupations in large numbers, such as registered nurses, they decreased their rate of entry into others, such as secretaries. While many nontraditional occupations be- came slightly less male dominated, large de- clines in segregation occurred in only a few, e.g., accountants. Also contributing to a de- cline in segregation were the dramatic de- clines in the size of the traditionally female occupations of sewers and stitchers and tele- phone operators, presumably the first clue to a declining industry and the second! to rapid mechanization, eliminating the need for as many telephone operators. These changes suggest that women are working in many different nontraditional places in the labor force, changes which bear a closer look. CHANGES IN THE SEX COMPOSITION OF OCCUPATIONS, 1960-1977 This section examines changes in the sex composition of size-standardized occupa- tions, assuming that all are of equal size. It also summarizes material presented in greater detail in Beller (1981~. Occupations are cat- egorized according to their sex label and broad occupational group. Changes in the sex composition of detailed occupations dur- ing the 1960s are contrasted with changes between 1971 and 1977. These analyses are bases! on data for 262 occupations from the decennial censuses and the ADF and focus exclusively on changes in the sex composi- tion within occupations, a variable amenable to alteration through public policy. Each detailed occupation is assigned a sex label defined by deviations in its sex com- position of + .05 from that of the labor force as a whole. According to this definition, oc- cupations are categorized as male if in 1960 the percentage of males equaled or ex- ceeded .722; in 1970, .669; in 1971, .668; and in 1977, .640. Table 2-3 shows the num- ber and percentage of occupations that are male, female, and integrated in each year.8 Although a majority of occupations continue to be male dominated, the percentage de- clined during the 1970s, though it had in- creased during the 1960s; a number of oc- cupations changed from male to integrated, while the percentage that was female re- mained unchanged. A comparison of changes in women's share of employment by occupation from 1971 to fi Although the choice of the value + .05 is somewhat arbitrary, it has little effect on substantive conclusions in this paper. It simply affects how wide a segment of the occupational distribution we choose to call inte- grated.
TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 19 TABLE 2-3 Sex Label of Detailed Occupations Census CPS Sex Labela 1960 1970 1971 1977 Male- 159 165 157 140 Integrated 17 19 15 32 Female 86 78 90 90 Total 262 262 262 262 Percentage Male 60.7 63.0 59.9 53.4 Integrated 6.5 7.3 5.7 12.2 Female 32.8 29.8 34.4 34.4 Total 100.0 100.0 100.0 100.0 a Defined relative to the sex composition of the labor force in the year given. SOURCE: U.S. Bureau ofthe Census, U.S.CensusofPopulation:1960.FinalReportPC(2~-7A. Subject Reports. Occupational Characteristics. (Washington, D.C.: U.S. Government Printing Office, 1963), Table 1; U.S. Bureau of the Census, U.S. Census of Population: 1970. Final Report PC(2~-7A. Subject Reports. Occupational Charac- tenstics. (Washington, D.C.: U.S. Government Printing Office, 1973), Table 1; 1972 and 1978 Current Population Survey, Annual Demographic Files, computer tapes. 1977 with those from 1960 to 1970 reveals the following. Women's share of employ- ment increases! absolutely in a majority of occupations in both periods: in 77 percent of occupations between 1960 and 1970 and in 71 percent between 1971 and 1977. But women's share of employment relative to their share of the labor force increased in many more occupations during the 1970s than curing the 1960s: in 45 percent of oc- cupations between 1971 and 1977 as com- pared to 26 percent between 1960 and 1970. Women's relative share in male occupations also increased much more widely during the 1970s: in 46 percent of male occupations compared to 25 percent during the 1960s. These changes were most pronounced among the white-colIar\ occupations, especially professional and managerial, and little or no change occurred among the blue-collar oc- cupations. Women's share continued to grow both absolutely and relatively in the already predominantly female clerical occupations. I have also examined the magnitude of change in the sex composition of the average occupation. Between 1960 and 1970 the av- erage occupation became 2.8 percent rela- tively more male, while between 1971 and 1977 it became 0.6 percent relatively less male. Thus, while women had become more occupationally concentrated during the 1960s, they began entering nontraditional occu- pations at a greater rate than the labor force as a whole during the 1970s. Male occupa- tions also became relatively more male on average during the 1960s in every broad oc- cupational category with the exception of clerical. In the 1970s, with the exception of crafts and operatives, in every broad occu- pational category male occupations became relatively less male. For example, the av- erage male managerial occupation, which became 2.5- percent more male cluring the 1960s, became 4.9 percent less male be- tween 1971 and 1977. In summary, the contrast in changes in the sex composition of occupations between the 1960s and 1970s shows that a new pattern of female entry has emerged. Rather than con- tinue to crowd into a limited subset of occu- pations, women are entering a wide variety of nontraditional occupations. These changes are most prominent at the white-collar level, especially among professional ant] managerial occupations. But little such change appeared for the blue-collar occupations.
ANDREA H. BELLER TABLE 2-4 Sex Segregation Indexes for All Occupations and Professional Occupations by Race All Occupations Professional Occupations 1972 1977 1981 1972 1977 1981 Index Level Unstandardized White 68.39 64.96 62.08 60.05 54.68 50.75 Nonwhite 68.00 63.29 59.39 51.58 49.95 48.88 Employment Standardized to 1972 White 68.39 65.35 63.07 60.05 55.20 50.89 Nonwhite 68.00 65.43 63.52 51.58 48.58 50.59 Change in Index, 1972-1981 Unstandardized Annual Average Total Annual Average Total White - 0.70 - 6.31 - 1.03 - 9.30 Nonwhite - 0. 96 - 8.61 - 0.30 - 2.70 Employment Standardized to 1972 White - 0.59 - 5.32 - 1.02 - 9.16 Nonwhite - 0.50 - 4.48 - 0.11 - 0.99 NOTE: These indexes are based on a common group of 262 three-digit census occupations of which 59 are professional occupations. SOURCE: U.S. Bureau of Labor Statistics, annual averages of monthly Current Population Survey, 1972,1977, and 1981, unpublished tabulations. TRENDS IN TlIE SEX SEGREGATION OF OCCUPATIONS BY RACE, 1972-1981 Occupational segregation by sex declined continuously for both whites and nonwhites between 1972 and 1981. While the index of segregation was approximately the same for both races in 197268.39 for whites and 68.00 for nonwhites it declined relatively faster for nonwhites during the 1970s. For whites it declined to 62.08 in 1981, while for nonwhites it declined to 59.39. These indexes appear in Table 2-4. The figures in the bottom part of this table show the av- erage annual rate of decline in the index of segregation between 1972 and 1981, 0.96 for nonwhites compared to 0.70 for whites.9 To identify the portion of the overall change attributable to changes in the sex compo- 9 Although in this sample many occupations contain only a few nonwhites, the level of the segregation in- dexes changes very little when occupations with fewer than 10 nonwhites are excluded from the computations. sition within occupations, the indexes were standardized to the occupational distribu- tion of employment in the initial year, 1972. By contrast to the unstandardized indexes, the standardize`] indexes, shown in Table 2-1, declined slightly more for whites than for nonwhites, from 68.39 to 63.07 for whites and from 68.00 to 63.52 for nonwhites. Incus, the greater decline for nonwhites than whites in the unstandardized index can be attrib- uted to shifts in the nonwhite occupational distribution from heavily single-sex occu- pations toward less segregated ones. As a matter of fact, nearly one-half of the decline in sex segregation among nonwhites was due to such changes in their occupational dis- tribution ~ - 4.13) toward less segregated oc- cupations as compared with changes in their sex composition within size-standardized occupations (-4.481. I conclude that changes in the sex composition within occupations was about the same for both races over the clecade, but the occupational distribution of nonwhites also shifted toward less sex seg- regation.
TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 21 The picture for professional occupations contrasts dramatically. On the one hand, while the levels of the segregation indexes are Tower for both races for professional than for all occupations in 1972, these indexes are much lower for nonwhites. As shown in the second pane} of Table 2-4, the 1972 seg- regation index was 60.05 for white profes- sionals and 51.58 for nonwhite professionals in contrast to 68.39 and 68.00, respectively, for all occupations. On the other hand, non- whites experienced little decline in sex seg- regation among professional occupations during the 1970s, while whites experienced larger declines than among all occupations. By the end of the decade, white profession- als had become slightly less sex segregated than nonwhite professionals had been at the beginning of the decadean index value of 50.75 compared to 51.58 - while nonwhites had become somewhat less segregatedan index value of 48.88. The annual average rate of decline in the segregation indexes for professional occupations over the decade was 1.03 for whites and 0.30 for nonwhites, in contrast to 0.70 and 0.96, respectively, for all occupations. According to the employ- ment standardized indexes, this entire de- cline for whites resulted from changes in their sex composition within occupations (1.02), while for nonwhites most of this de- cline came from shifts in their occupational distribution toward less (nonwhite) sex-seg- regated professional occupations (0.111. The question naturally arises as to what proportion of each racial group constitutes the professional occupations. In 1972, 14 percent of white men and 15 percent of white women were in professional occupations, while only 8 percent of nonwhite men and 11 percent of nonwhite women were. by 1981 the proportions had risen for all groups, with the largest increase occurring for non- white women to 15 percent, the next largest increase for nonwhite men to 11 percent, and identical increases for whites of both sexes to 16 percent for men and 17 percent for women. These percentages reflect growth rates of professional employment of 90 per- cent for nonwhite women, 69 percent for nonwhite men, 54 percent for white women, and 26 percent for white men. Neverthe- less, professional occupations still com- prised a smaller proportion of nonwhite than white employment in 1981. If sex segregation declined as much for nonwhites as for whites over all occupations but not among professional occupations, then most change among nonwhites must have occurred at other levels of the occupational distribution. MAJOR COMPONENTS OF CHANGE IN THE OVERALL INDEX BY RACE The occupations contributing the most to decreasing the segregation index among nonwhites between 1972 and 1981 differ considerably from the ones for whites, al- though some similarities exist. The occu- pations that contributed the most to de- clines in the index of sex segregation among nonwhites only are primarily laborer and service worker occupations. Frequently, they are typically male occupations that declined in size over this period and in which non- whites are represented disproportionately. The larger decline for nonwhites than for whites in the index of segregation over all occupations can be traced to this source. The specific detailed occupations tatting over one-half a percentage point off the segre- gation index for nonwhites only between 1972 and 1981 are the following: storekeepers and stock clerks; clothing ironers and pressers; construction laborers, except carpenters' helpers; freight and material handlers; gar- deners and groundskeepers, except farm; miscellaneous laborers; unspecified labor- ers; farm laborers, wage workers; chamber- maids and maids, except private household; cleaners and charwomen; nursing aids, or- derlies, and attendants; and practical nurses. Among these, nearly all laborer occupations declined in size, while nearly all service worker occupations increased in size. Non-
22 ANDREA H. BELLER white women increased their share of store- keeper jobs from 20 to 41 percent between 1972 and 1981. Ike segregation index declined for both races because of the dramatic decline in the size of the private household maids and servants occupation, but the decline was much greater for nonwhites. The decline in the number of nonwhite females in this oc- cupation took an exceptional 8.04 percent- age points off the segregation index for non- whites. Ike specific detailed occupations contrib- uting the largest amounts to the decline in the index of segregation among whites be- tween 1972 and 1981 are mostly the same as for the whole population presented ear- lier. Nevertheless, some occupations con- tributed to the decline in the index for whites only: bookkeepers, garage workers and gas station attendants, waiters and waitresses, and hairdressers and cosmetologists. Al- though the number of whites employed as bookkeepers increased between 1972 and 1981, the proportion of the white female labor force that crowded into this tradition- ally female occupation declined from 5.2 to 4.6 percent. While men entered the two expanding traditionally female occupa- tions waiters and waitresses, and hair- dressers and cosmetologists at an increas- ing rate over this period, the male share of white employment increased from 7.2 per- cent to 9.2 percent in the former and from 9.2 to 11.1 percent in the latter. The tra- ditionally male occupation of garage workers and gas station attendants showed a decline in size over the period. Comparable declines in segregation for whites and nonwhites occurred for the fol- lowing occupations: the clerical occupation, telephone operators; the operative occupa- tion, sewers and stitchers; the laborer oc- cupation, delivery and route workers; ant! the three service worker occupations, jani- tors and sextons, cooks (except private household), and child care workers (private household). By contrast, sex segregation among whites also declined among the fol- lowing white-collar occupations: account- ants; elementary school teachers; bank of- ficers and financial managers; sales clerks, retail trade; bookkeepers; secretaries, n.e.c.; and typists. Two of these occupations- bookkeepers and secretaries, n.e.c. ac- tually became more segregated among non- whites because women but not men entered these fields. The occupations that contributed more in 1981 than in 1972 to the segregation index for whites are identical to those for the pop- ulation as a whole reported earlier. Occu- pations that became more segregated among nonwhites only include bookkeepers; sec- retaries, n. e. c.; investigators and esti- mators, n.e.c.; statistical clerks; electricians; and assemblers. These were all sex-segre- gated occupations in which the numbers of nonwhites employed grew. The typically fe- male occupations among them also became increasingly female. While the number of electricians grew rapidly, the female share of nonwhite employment increased from O in 1972 to 3.9 percent in 1981. Contributing toward increasing the segregation index by comparable amounts for both races are reg- istered nurses, bank tellers, computer and peripheral equipment operators, and mis- celIaneous clerical workers. In summary, much of the decline in oc- cupational segregation by sex during the 1970s occurred for both races; however, ma- jor differences exist. The major exodus of nonwhite females from the occupation of private household maids and servants and the decline in size of a number of laborer occupations in which nonwhite males dom- inated shifted the nonwhite occupational distribution toward a greater reduction in occupational sex segregation during the 1970s than for whites. On the other hand, white women reduced their rate of entry into a number of traditionally female white-collar occupations that nonwhite women contin- ued to enter, and white women increased their entry into a number of traditionally
TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 23 male white-colIar occupations more than nonwhite women did. CONCLUSION Occupational segregation of the sexes di- minished significantly during the 1970s, as measured by the index of segregation. Most of the decline was due to changes in the sex composition of traditionally male occupa- tions, particularly at the professional and managerial levels. Declines in segregation among new and recent job market entrants were greater than for the rest of the labor force. While nonwhites experienced a greater decline in occupational sex segregation than whites over the decade, about the same amount was due to changes in the sex com- position of traditionally male occupations. The marked declines in sex segregation in professional occupations apparent among whites dice not hold for nonwhites, but non- white professionals were much less segre- gated than white professionals at the start of the decade. Continued declines in oc- cupational segregation by sex depend on the apparent momentum for change continuing into the next decade. ACKNOWLEDGMENTS This research was supported in part by funds from the University of Illinois at Urbana-Champaign, Research Board. Help- ful comments on an earlier drab were pro- vided by Marianne Ferber, Victor Fuchs, Barbara Reskin, and several reviewers. Ex- cellent research assistance was provided by Kee-ok Kim Han. Computational assistance by John Boyd and Alex Kwok is gratefully acknowledged. APPENDIX A Data Sources To assess trends in occupational segre- gation for intercensual years, a data set other than the decennial census must be selected. The monthly Current Population Survey (CPS) collects detailed three-digit census occupation data from a random cluster sam- ple of (initially around 50,000) around 60,000 households (1/1500) designed to represent the civilian noninstitutional population of the United States. Two sources provide detailed occupational data from this survey. The first is the Bureau of the Census's March Annual Demographic Files (ADF), available on public-use tapes since 1968. The question on "longest job held last year" should pro- vide reasonably reliable estimates of the previous year's occupational distribution. The second is the monthly statistics compiled by the Census Bureau for the Bureau of Labor Statistics (BLS) from which the latter com- pute, ant! since 1974 publish in Employment and Earnings, the annual averages (AA). AA data were used in the two studies referred to in the text (Lloyd and Niemi and U.S. Commission on Civil Rights), which com- putec] segregation indexes for more recent data than the 1970 census. Unpublished tab- ulations of AA, which include data for smaller occupations than the published data (50,000 incumbents), are available directly from the BLS for 1972 on. In comparing the CPS with the decennial census, the primary disadvantage of the CPS is its smaller sample size. The AA data are somewhat more reliable on these grounds than the ADF data. For total labor force data other than agricultural employment and un- employment, the sampling error of the an- nual averages is 0.67 times the sampling er- ror of the monthly data (Employment and Earnings, May 1982, Table I). To improve reliability, the smallest occupations should be excluded. For purposes of this paper, all tabulations excluded occupations with fewer than 25 survey respondents in either the 1975 or the 1978 ADF data set (representing occupations with fewer than approximately 40,000 incumbents). Out of the 441 detailed three-digit 1970 census occupations, this left 267 in 1974 and 280 in 1977, 262 of which
24 ANDREA H. BELLER are common to both years. All tabulations in this paper include only those 262 occu- pations, or fewer where noted. Comparability of Data In attempting to assess trends in occu- pational data over time, two comparability problems arise, depending on the period of interest. The first is changes in the Census Bureau's occupation codes with each decen- nial census. A variety of techniques for deal- ing with this problem are discussed in Eng- land (1981~. The 1960 census data were recoded according to the 1970 census cocles by John A. Priebe (U. S. Bureau of the Cen- sus,1972~. These data are published in U. S. Bureau of the Census (1973, Table 221~; the 1960 and 1970 census data in this paper, as well as in Blau and Hendricks (1979), come from this source. The 1980 census used a substantially revised set of occupation codes, and thus its occupation data will not be com- parable to earlier census occupation data un- less the Census Bureau double codes them with the 1970 and 1980 occupation codes. Another comparability problem arose in that the segregation indexes computed us- ing the CPS data were inconsistent with the one computed using the 1970 census data despite the fact that both used 1970 census codes. In an attempt to find out why, the following information was discovered. In APPENDED B TABLE B-1 Indexes of Segregation From Other Studies December 1971 a question eliciting infor- mation on major activities or duties was added to the monthly CPS in order to determine more precisely the occupational cIassifica- tion of individuals. According to the BLS (`Employment and Earnings, January 1979, p.207), "this change resulted in several dra- matic occupational shifts, particularly from managers and administrators to other groups. Thus, meaningful comparisons of occupa- tional levels cannot always be made for 1972 and subsequent years with earlier periods." For this reason, the 1970 census data are not comparable with the CPS data after 1971. The two studies that found no change in segregation in the 1970s relied on such a comparison. The earliest comparable data would be from the 1972 ADF on the longest job held last year, 1971. For these reasons, presented here are census data for 1960 and 1970 comparisons; data from the ADF for 1971-1974 and 1977, the years for which we have the data tapes; and the unpublishe~l annual averages data for 1972, 1977, and 1981 for more recent data. It was found that the computations based on the ADF and the AA data sets are quite similar, although in- dividual occupations can differ. To include data as current as 1981, comparisons across these two data sources were sometimes made, although their reliability diners. Additional comparability problems will arise when the CPS converts to the new 1980 occupation codes, beginning with the 1982 clata. Census 1960 1970 CPSa Projected 1976 1977 1985 All Occupations Blau and Hendricks (1979) (N = 280) Lloyd and Niemi (1979) (N = 236) U.S. Commission on Civil Rights (1978) (N = 441) Professional Occupations Fuchs (1975) (N = 33) Unstandardized Standardized to 1960 a Computed from the BLS's AA data. 68.33 66.2 66.2 65.77 60.10 64.5 65.8 64.3 66.1 59.2 62.7
TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 25 TABLE B-2 Percentage of Occupations With Changes in Sex Label by Initial Sex Label and Period 1960-1970 1971-1977 1971-1974 1974-1977 All 9.5 11.1 11.5 11.1 Male 3.1 10.8 8.9 7.5 Integrated 53.0 33~3 53~3 37~5 Female 12.8 7.8 8.9 9.9 SOURCE: Same as for Table 2-3. REFERENCES Becker, Gary S. 1981 A Treatise on the Family. Cambridge, Mass.: Harvard University Press. Belter, Andrea H. 1981 "Changes in the Sex Composition of U.S. Oc- Fuchs, Victor R. cupations, 1960-77." Unpublished paper. 1982a "The Impact of Equal Opportunity Policy on Sex Differentials in Earnings and Occupa- tions." The American Economic Review, Pa- pers and Proceedings (May):171-75. 1982b "Occupational Segregation by Sex: Determi- nants and Changes." The Journal of Human Resources 17 (Summer):371-92. Beller, Andrea H., and Kee-ok Kim Han 1984 "Trends in Major Fields of Study Among Women in Higher Education." In Kathryn Rettig and Mohamed Abdel-Ghany, eds., Eco- nomic Decisions of Families: Security for the Elderly and Labor Force Participation of Women. Washington, D.C.: American Home Economics Association. Blau, Francine D., and Wallace E. Hendricks 1979 "Occupational Segregation by Sex: Trends and Prospects." The Journal of Human Resources 14 (Spring):197-210. Cherlin, Andrew, and Pamela B. Walters 1981 "Trends in United States Men's and Women's U.S Sex-Role Attitudes: 1972 to 1978." American 1963 Sociological Review 46 (August):453-60. Corcoran, Mary, and Gregory J. Duncan 1979 "Work History, Labor Force Attachment, and Earnings Differences Between the Races and Sexes." The Journal of Human Resources 14 (Winter):3-20. Duncan, Gus Dudley, and Beverly Duncan 1955 "A Methodological Analysis of Segregation In- 1973a dexes." American Sociological Review 20 (April):210-17. Economic Report of the President 1973 Washington, D. C.: U. S. Government Printing 1973b Office, pp. 155-59. England, Paula 1981 "Assessing Trends in Occupational Sex Seg- regation, 1900-1976." Pp. 273-95 in Ivar Berg, ea., Sociological Perspectives on Labor Mar- kets. New York: Academic Press. 1982 "The Failure of Human Capital Theory to Ex- plain Occupational Sex Segregation." The Journal of Human Resources 17 (Summer): 358-70. 1975 "A Note on Sex Segregation in Professional Occupations." Explorations in Economic Re- search 2 (Winter):105-11. Gross, Edward 1968 "Plus ca change . . .? The Sexual Structure of Occupations Over Time." Social Problems 16 (Fall): 198-208. Lloyd, Cynthia, and Beth Niemi 1979 The Economics of Sex Differentials. New York: Columbia University Press. Mason, Karen D., et al. 1976 "Changes in U. S. Women's Sex-Role Atti- tudes, 1964-1974." American Sociological Re- view 41 (August):573-96. Polachek, Solomon W. 1979 "Occupational Segregation Among Women: Theory, Evidence and a Prognosis." Pp. 137- 157 in Cynthia B. Lloyd et al., eds., Women in the Labor Market. New York: Columbia University Press. Bureau of the Census Census of Population: 1960. Subject Reports. Final Report PC(2~-7A. Occupational Charac- teristics. Washington, D.C.: U.S. Govern- ment Printing Office. 1972 "1970 Occupation and Industry Classification Systems in Terms of their 1960 Occupation and Industry Elements," by John A. Priebe. Tech- nical Papers, No. 26. Census of Population: 1970, Detailed Char- acteristics. Final Report PC(l)-Dl. U.S. Sum- mary. Washington, D.C.: U.S. Government Printing O~ice, Table 221. Census of Population: 1970. Subject Reports. Final Report PC(2~-7A. Occupational Charac- teristics. Washington, D.C.: U.S. Govern- ment Printing Office. U.S. Commission on Civil Rights 1978 Soci~l Indicators of Equality for Minor~ties and Women. August.
26 ANDREA H. BELLER U.S. Department of Labor, Bureau of Labor Statistics 1979 Employment and Earnings 26~1~:207. 1980 Employment and Training Report of the Pres- ident. Employment and Training Administra- tion. 1982a Employment and Earnings 29~5~:Table J. 1982b Labor Force Statistics Derived from the Cur- rent Population Survey: A Databook 1 (Bul- letin 2096~:Table C-23. U.S. National Center for Education Statistics 1980 Projections of Education Statistics to 1988-89. April.