Women’s work plans: contrasting expectations and actual work experience
Lois B. Shaw
Women’s work plans: contrasting expectations and actual work experience
The human capital literature in the past decade has emphasized that women’s early work expectations affect their subsequent earnings and occupations.1 If women expect to withdraw from the labor market when they have children, they may have little incentive to invest in work-related skills early in their working lives. They may look for jobs that pay well initially but offer few prospects for on-the-job training and advancement. They may also choose occupations in which skill depreciation will be limited during periods of labor market withdrawal.2 These considerations lead to the prediction that the earnings of women who plan for continuous labor force participation will increase more rapidly than those of women who expect to experience work interruptions.
We use questions from the National Longitudinal Surveys of Labor Market Experience of Young Women (NLS) to examine how young women’s plans affect their subsequent work experiences and earnings.3 We find that those young women who planned to be in the labor market at age 35 were indeed more likely to be employed when they reached that age. More importantly, planning to work yielded a signficant net wage advantage: among women in their midthirties, those who, throughout their twenties, had consistently planned to work had wages that were nearly 30 percent higher than those of women who had never planned to work, even after controlling for work experience and other determinants of wage rates. This wage advantage was even greater for those women who were employed in occupations in which they had expected to be employed.
Beginning in 1968, young women, ages 14 to 24, in the National Longitudinal Survey sample were asked at each interview whether they wanted to be working when they reached age 35. By 1980, the oldest women in the sample had actually reached age 35, making a comparison of plans and actual work behavior possible. Using data from the first 5 years of the NLS, Steven Sandell and David Shapiro showed that young women were considerably underestimating the likelihood that they would work outside the home in the future.4 Sandell and Shapiro also presented evidence indicating that women who had work plans found jobs with more potential for training and advancement, albeit lower initial wages, than women who did not expect to be working. We followed the NLS young women over 7 more years to determine how well their plans were realized, why plans were sometimes not realized, and the extent to which early work plans contributed to subsequent wages. If young women underestimate their future employment and have lower wages as a result, this factor will contribute to malefemale earnings differences among adults. To the extent that this is indeed the case, it suggests that young women need to be provided with better information about the likelihood of future employment and the importance of planning ahead for their working lives.
In the next section, we describe and provide an overview of the NLS data on work expectations. Then, we examine the association between early work plans and actual adult work behavior, and later extend that analysis and explore the factors that influence whether or not a young woman’s work expectations are realized. Finally, we analyze the relationship between early work plans and subsequent earnings of women, and follow with a summary and conclusions.
Work expectations of young women
Two versions of a question concerning women’s work expectations have been used in the NLS. In the initial round of the surveys (1968), respondents were asked “Now I would like to talk to you about your future plans. What would you
like to be doing when you are 35 years old?’ In the following years, the question was changed to read “Now I would like to talk to you about your future job plans. What kind of work would you like to be doing when you are 35 years old?’ (Emphasis added to indicate changes from the previous version.) In the second version as well as the first, keeping house or raising a family was a possible response. Because the questions mention both plans and preferences, which may not always reflect actual plans, respondents may have interpreted the questions in more than one way. However, interpreting their actions as reflecting their plans, as we do here, appears to produce plausible results.
As mentioned previously, evidence presented by Sandell and Shapiro suggested that young women, as a group, underestimated their future work activity. Table 1 provides further evidence of this phenomenon. Focusing on white women respondents who were ages 34-36 in 1980, the table shows the percentage who reported plans to work at age 35 in each survey year for 1968-78.5 Except for the 1977 and 1978 surveys, young women consistently underpredicted (in the aggregate) the likelihood that they would be working at age 35. It should be noted that their initial expectations were also lower than the actual labor force participation of women who were age 35 in 1968. Young women either expected to work less than their elders or were poorly informed about the likelihood that a 35-year-old woman would be employed. There is a clear trend for work expectations to approximate more closely actual work activity as the women approach age 35.6 Hence, it appears that expectations regarding future market work activity were essentially adaptive, beginning at a low level when respondents were in their early twenties and rising (more or less) steadily over time. Looking at future work plans by educational attainment in 1980, we see that women who had attended college were consistently more likely than their noncollege counterparts to anticipate working at age 35, and the differences in work expectations between college and noncollege women appear to have widened somewhat over time.
The aggregated data on future work expectations reveal a fairly steady trend. A somewhat different perspective is provided by comparing the responses of individuals over time. Table 2 shows the number of times in the first seven interviews (covering the period from 1968 to 1975) that respondents ages 34-36 in 1980 indicated they planned to be working in the labor market at age 35. Overall, the frequency distribution is fairly evenly divided across the four work-plans groups. However, the distribution of responses for the total sample conceals large differences in responses by educational attainment. Although a great deal of variation occurred in both groups, better educated women were most likely to have indicated plans for work at least six times and least likely to have planned to work only once or not at all; for women who had never attended college, the reverse was true.
Just over half of the total group responded consistently from year to year (they had plans to work in at least 6 years out of 7 or did not express plans to work in at least 6 years out of 7). Those individuals whose responses to questions about plans were more mixed over the 7 years fall into two broad groupings: women who shifted from having no future work plans to having such plans and all other women. Examination of the detailed data on the sequencing of work expectations of individuals over time (not shown here) indicated that approximately 12 percent of the total group in table 2 fell into the former category, while more than one-third were in the latter category.7 Thus, four principal patterns of work expectations emerged: women who consistently anticipate working at age 35 throughout their twenties; those who consistently indicate no plans for future work; women who shift to having future work expectations at some time during their twenties; and those who give highly variable responses over time.
Linking work plans to work behavior
In this section, we examine the association between early work plans and subsequent work behavior: expectations about future work reported in the first seven rounds of the National Longitudinal Surveys are linked to labor force participation in 1980 and to cumulative work experience (weeks worked) between 1976 and 1980. This juxtaposition enables us to determine the extent to which women’s early work plans are realized, and thereby provides useful information about the predictive reliability of the responses.
Table 3 shows the labor force participation rates and average number of years worked between 1976 and 1980 for NLS respondents ages 34-36 in 1980, according to the number of times between 1968 and 1975 that respondents indicated they had plans for work at age 35. Perhaps most striking is that nearly half of the women who had consistently indicated no plans for work at age 35 were in the labor force in 1980. At the same time, however, the data in table 3 reveal a strong association between early work plans and subsequent work activity. This association reflects not simply an independent effect of work expectations on labor force activity–it also reflects the relationship between work expectations and other factors that directly influence work activity. For example, better educated women report greater expectations of work at age 35, and it is well known that better educated women have higher labor force participation rates than their lesser educated counterparts. Hence, part of the association between early work plans and subsequent work activity evident in table 3 results from the intervening effect of educational attainment: the data reflect the fact that women with work plans also tend to have more schooling than those who do not plan to work.8
To isolate the net association between early work expectations and later labor force activity, we have estimated reduced-form labor-supply equations for the NLS respondents ages 34-36 in 1980. Table 4 reports the results of a probit analysis of the factors influencing labor force participation in 1980. The explanatory variables include the respondent’s educational attainment, other family income, marital status, number and ages of children, and early plans for work at age 35. Results of estimating an ordinary least squares regression for work experience between 1976 and 1980 are reported in table 5.
In both equations, early plans for work are significantly related to work activity at age 35. The equations imply that, compared to women with no plans, women who indicated in each year from 1968 to 1975 that they would like to be working at age 35 have a labor force participation rate more than 30 percentage points higher and have worked about nine-tenths of a year more between 1976 and 1980. These results indicate that work activity of consistent planners is approximately 50 percent greater than that of women with no work plans, other things equal.9 Coefficients of the other variables in the equations are almost all statistically significant with the expected signs.
The evidence present in this section suggests a fairly strong association between early work plans and later work behavior. Even after controlling for other factors that affect female work activity, early work expectations are significantly related to the subsequent work behavior of white women. This correlation suggests that the association between actual work behavior and the work expectations measure reflects not only the effects of readily quantified variables such as educational attainment, but that it also reflects the effects of unmeasured variables. For example, even at the same educational level, some women are more career-oriented than others, and some enjoy homemaking activities while others do not.
Factors affecting work participation
In this section, we examine the factors associated with the realization of earlier plans for each of three groups of women: those who consistently planned to work at age 35; those who consistently indicated no work plans; and all others. Table 6 summarizes the results of two probit equations for labor force participation in 1980 for the three groups. The first equation for each group uses the basic labor supply model used earlier in the article. For purposes of considering deviations of actual participation from expected participation, however, it would be desirable to have measures of change in variables that are known to influence labor force participation. For example, women who expected to be married or to have (or not have) children at age 35, may not have had their expectations realized. Similarly, some women may have hoped to devote themselves exclusively to their families at age 35, but their husbands’ unemployment or slow advancement may have precluded the women remaining at home. Other women may have misperceived the relative rewards of career and housewife roles. Unfortunately, the National Longitudinal Surveys contain data on expectations in only one of these areas: number of children. In the second equation for each group, then, we use two dummy variables indicating whether the respondent had more or fewer children in 1980 than she had expected to have in 1973; these variables are substituted for the age of children variables of the first equation.
Examination of the coefficients in each equation in table 6 indicates that these three groups of women differed significantly with respect to the responsiveness of their labor supply to schooling, income, and children.10 Differences in educational attainment were most important as a determinant of labor force participation among women who did not consistently express plans for work at age 35. Among women who did consistently express such plans, schooling was not significantly related to labor force participation. With schooling viewed as a proxy for a woman’s potential wage rate, these results suggest that women with strong expectations of market work manifest a rather inelastic labor supply–not unlike that of adult men. Women with weaker work expectations, by contrast, exhibit more elastic labor supply.
Other family income is inversely related to the probability that a woman will be in the labor force, and the effect of other income becomes weaker as one moves from women who had no work expectations to women with consistently strong expectations. Other things equal, the absence of a husband appears to contribute to an increased likelihood of being in the labor force only for women who consistently expressed no work plans.11
Having more children reduces labor force participation for all three groups of women. The presence of preschoolage children is not significantly related to the labor force participation of women without work plans, but it is a highly significant deterrent to labor force participation among women with intermediate and strong early work plans. While having fewer children than expected did not contribute to higher labor force participation among women, having more children than expected did result in a significantly lower likelihood of being in the labor force among women with intermediate work plans. A smaller effect among women with strong work plans was not quite statistically significant at conventional levels.
Overall, then, among women who had consistently expressed no plans for work at age 35, the principal determinants of labor force participation at age 35 were schooling, other family income, number of children, and marital status. However, among women who had consistently indicated plans for work at age 35, fertility (including both ages and numbers of children) was the sole significant determinant of labor force participation. The labor force participation of women in the intermediate group was responsive to their educational attainment, other family income, and fertility.
Meeting expectations?
These results are useful in understanding what determines whether or not a young woman’s early work expectations are realized. Half of the women who consistently indicated no plans for work at age 35 were nonetheless in the labor force at that age. The results in table 6 suggest several possible reasons for “unexpected’ labor force participation. First, economic insecurity, whether due to the absence of a husband or to low income, caused some women to work although they had not planned to do so. Second, higher levels of education (and hence greater earnings potential) led to changes in plans, either because the earnings forgone by remaining at home were too great or because the housewife role was less satisfactory than the women had anticipated, especially for women with access to good jobs.12 Third, women with small families could more easily change their plans and go to work if either of the first two factors came into play. However, it is interesting to note that the age of the youngest child was not a significant influence for this group of women who were strongly committed to home-oriented activities.
About 1 in 6 women with strong early work expectations were not in the labor force at age 35; these women were primarily those who had a preschooler or perhaps, in some cases, had more children than originally expected. Some women with large families may also have found combining work and childraising more difficult than they had anticipated.13 This family-work conflict appears to cut across all education and income levels. However, it should be noted again that the great majority of women who were strongly committed to work were able to realize their plans; and some of those who did not were likely to have been out of the labor force only temporarily.
Relating work plans to higher wages
We have seen that many women in their mid-thirties were working in 1980 even though they had not been consistently planning to work. How important an effect does planning ahead have on women’s wages? To answer this question, we estimated wage equations for the National Longitudinal Surveys’ respondents ages 34-36 in 1980. The dependent variable was the natural logarithm of the hourly wage rate, and explanatory variables were years of school completed, years of work experience (the number or years that the respondent was employed for 6 months or more), years of tenure at the current job, residence in a Standard Metropolitan Statistical Area, and residence in the South. In addition, we also included a variable measuring the number of times between 1968 and 1975 that the respondent indicated plans for work at age 35, as well as interaction terms between this work-plans variable and years of work experience and job tenure. The human capital hypothesis that women with greater expectations of market work would have experience-wage profiles that are steeper and begin lower than those of women without strong work expectations implies that the coefficient of the plans-experience interaction term should be positive while the coefficient of the work-plans variable itself should be negative.14
Equation 1 of table 7 reports the estimated coefficients of the wage equation described in the preceding paragraph. Schooling, work experience, tenure, and residence in a Standard Metropolitan Statistical Area are all significantly related to wages. However, the plans-experience interaction term has a negative and insignificant (rather than the hypothesized positive) coefficient, and the work-plans variable is positively (rather than negatively) and significantly related to wages. The plans-tenure interaction is also insignificant. Excluding the interaction terms (equation 2) yields similar results: women with early plans for work are paid significantly more, other things equal, than their counterparts without early expectations of future market work. The coefficient of the work-plans variable in equation 2 implies that a woman with plans to work at age 35 in each of the first 7 years of the NLS would at age 35 be paid almost 30 percent more, other things equal, than a woman with no work plans.15
The evidence in table 7 thus provides support for the human capital hypothesis that early work plans will result in higher future wages. We find that, among white women in their mid-thirties (with an average of more than 10 years of work experience), those with stronger early expectations of adult work activity have significantly higher wage rates. However, no evidence was found for the difference in experience-wage profiles implied by human capital theory and found earlier by Sandell and Shapiro.16 The difference between our present findings and this earlier result may be due in part to the narrower age range used in our analysis.17 Our results lend support to the finding of Reuben Gronau18 that the skill requirements of women’s jobs and hence the amount of on-the-job training they receive are unrelated to their plans for quitting. Gronau argues that regardless of their work plans, women’s opportunities to obtain skillintensive jobs may be limited because of employers’ misconceptions about women’s work attachment.
In an effort to explore further the link between early work expectations and subsequent earnings, we examined the occupations that women with plans for work indicated they would like to be engaged in at age 35. In equation 3, we replaced the single work-plans variable with two variables measuring the number of times the respondent expressed plans for work in the specific (3-digit) occupation in which she was employed in 1980 and the number of times she indicated plans for work in other occupations. To the extent that job skills are occupation-specific, early plans for work in the 1980 occupation should have a greater impact on wages than plans for work in other occupations. To the extent that job skills are transferable across occupations, however, early work plans for other occupations should still contribute to higher wages in 1980. The results indicate that both plans for the specific 1980 occupation and plans for other occupations are significantly related to 1980 wage rates, and the coefficient of the specific-occupation plans variable is almost twice as large as the coefficient of the other-occupation plans variable.19 Apparently, women who make realistic plans and acquire necessary skills fare best in the labor market.
Conclusions
When they were in their early twenties, young women in the National Longitudinal Surveys sample greatly underestimated their future work involvement. Expectations for working at age 35 gradually increased, but up until about 3 years before the women actually reached that age, women in the sample continued to underestimate the likelihood of working.
Plans for working were significant independent predictors of actual work behavior. After controlling for other factors affecting labor force participation, a woman who consistently planned to work had a probability of working that was about 30 percentage points higher than did a woman who consistently planned not to work. More than 80 percent of the women who answered the plans to work question positively at least 6 times out of 7 were actually in the labor force in 1980. However, nearly half of the women who had expressed no intention to work in at least 6 of 7 interviews were nevertheless actually in the labor force when they reached age 35. For this latter group, economic factors had the largest impact on actual labor force participation; these women apparently worked although they planned not to because divorce or their husbands’ low income made working necessary or because their own level of education made the earnings forgone by staying at home too large to ignore. Women who had planned to work at age 35 were likely to do so unless they had large families or a preschool child.
Planning to work yielded a significant wage advantage. Women who had consistently planned to work had wages that were nearly 30 percent higher than those of women who had never planned to work. However, unlike the earlier results reported by Sandell and Shapiro, ours showed no evidence of greater returns to work experience among women who had planned to work. Our research implies that planning their working lives is important for women’s wages, but that this effect must operate, not through faster wage growth, but through their having better paid jobs at all levels of work experience.
Young women in their late teens and early twenties today appear to have much stronger work expectations than the women we studied here.20 Therefore, in the future fewer women may find themselves in low-wage jobs because they had never planned to work.21 Whether young women have also become more adept at occupational planning is not certain. Our research shows that women whose occupational plans are realized have higher wages than their counterparts who do not achieve their occupational goals. This finding suggests that providing young women with information about labor market trends and prospects has a potentially high payoff.
1 See, for example, Jacob Mincer and Solomon W. Polachek, “Family Investments in Human Capital: Earnings of Women,’ Journal of Political Economy, March/April 1974 supplement, pp. S76-S108; Jacob Mincer and Soloman W. Polachek, “An Exchange: Theory of Human Capital and the Earnings of Women: Women’s Earnings Reexamined,’ Journal of Human Resources, Winter 1978, pp. 118-34; Solomon W. Polachek, “Differences in Expected Post-School Investment as a Determinant of Market Wage Differentials,’ International Economic Review, June 1975, pp. 451-70; Solomon W. Polachek, “Discontinuous Labor Force Participation and Its Effects on Women’s Market Earnings,’ in Cynthia B. Lloyd, ed., Sex Discrimination and the Division of Labor (New York, Columbia University Press, 1975), pp. 90-122; and Yoram Weiss and Reuben Gronau, “Expected Interruptions in Labour Force Participation and Sex-Related Earnings Growth,’ Review of Economic Studies, 1981, pp. 607-19.
2 Polachek has been the leading proponent of this view. In addition to the papers cited in the previous footnote, see also Solomon W. Polachek, “Occupational Segregation: An Alternative Hypothesis,’ Journal of Contemporary Business, 1976, pp. 1-12; Solomon W. Polachek, “Sex Differences in College Major,’ Industrial and Labor Relations Review, July 1978, pp. 498-508; and Solomon W. Polachek, “Occupational Segregation Among Women: Theory, Evidence and a Prognosis,’ in Cynthia B. Lloyd, Emily Andrews, and Curtis Gilroy, eds., Women and the Labor Market (New York Columbia University Press, 1979), pp. 137-57. The ability of the human capital approach to account for observed male-female differences in wages and occupations has been questioned in Steven H. Sandell and David Shapiro, “The Theory of Human Capital and the Earnings of Women: A Reexamination of the Evidence,’ Journal of Human Resources, Winter 1978, pp. 103-117; Mary Corcoran and Greg. J. Duncan, “Work History, Labor Force Attachment, and Earnings Differences Between the Races and Sexes,’ Journal of Human Resources, Winter 1979, pp. 3-20; Paula England, “The Failure of Human Capital Theory to Explain Occupational Sex Segregation,’ Journal of Human Resources, Summer 1982, pp. 358-70; and Andrea H. Beller, “Occupational Segregation by Sex: Determinants and Changes,’ Journal of Human Resources, Summer 1982, pp. 371-92.
Table: 1. Percent of women planning to work, 1968-78, and percent actually in the labor force in 1980, by educational attainment
Table: 2. Percent distribution of women by the number of times they indicated plans to work in first seven surveys, by educational attainment
Table: 3. 1980 labor force participation rates and 1976-80 work experience, by number of times respondent indicated plans to work
Table: 4. Probit maximum likelihood estimates for labor force participation equation, controlling for work plans
Table: 5. Work experience equation, controlling for work plans
Table: 6. Probit maximum likelihood estimates for labor force participation equations, by number of times respondent indicated work plans
Table: 7. Wage equations, controlling for work plans
3 A complete description of the NLS young women’s sample may be found in Center for Human Resources Research, The National Longitudinal Surveys Handbook (The Ohio State University, 1986).
4 Steven H. Sandell and David Shapiro, “Work Expectations, Human Capital Accumulation, and the Wages of Young Women,’ Journal of Human Resources, Summer 1980, pp. 335-53.
5 In a preliminary analysis, black women’s work plans were only weakly associated with later work behavior. Either black women were more likely to encounter obstacles to the realization of their plans or they interpreted the question differently from white women. Because sample sizes are small and the results are not clear cut, analyses throughout the entire article are limited to the white sample.
6 It seems plausible to suggest that part of the increase between 1968 and 1969 in the percentage of women planning to work at age 35 resulted from the change in the wording of the question. The trend toward greater work expectations over time is, nonetheless, readily apparent even without the data for 1968.
7 Among those women in the latter category, just over half manifested no clear pattern over time in their work expectations. The remainder of these women fell into three groups: those who initially had no plans for work at age 35, then for a time expressed such plans before finishing with no future work plans; a converse group, consisting of women who began with expectations of future work, then for a time expressed no such plans before returning to their initial expectations; and a small group of women who began with plans for future work but eventually abandoned those plans.
8 Regressions of educational attainment on family background variables plus a dummy variable for plans to work at age 35 as expressed in 1968 or 1969 indicate that young women with expectations of future work complete an average of roughly half a year of schooling more than those without future work plans, other things equal. Because schooling undoubtedly influences work expectations, as well as vice versa, this half-year difference should not be regarded as the “effect’ of work expectations on schooling.
9 These results were obtained by evaluating the labor force participation and weeks worked equations at the mean of all explanatory variables except work plans. Number of times planning to work was first set equal to seven and then equal to zero and the results compared to obtain the reported measures of the effect of work plans.
10 A test for the significance of differences between equations is not readily available for probit analysis. However, we used the Chow test to determine whether the comparable ordinary least squares equations for the three groups were significantly different from each other. Each pairwise comparison revealed significant differences. Because the probit and ordinary least squares results were similar, we conclude that equations for the three groups are probably significantly different. To obtain slope coefficients at different values of the dependent variable in these equations, the probit coefficients may be multiplied by the following: at the maximum, which is also approximately the mean for 0-1 plans, by .399; at the mean for 2-5 plans, by .368; and at the mean for 6-7 plans, by .245.
11 The coefficients on this variable do not consistently attain conventional levels of statistical significance. Note, however, that most of the total effect of not having a husband will be reflected in lower other family income, and the largest (in absolute value) effects of this variable are for women without future work plans. Overall, then, it appears that the absence of a husband contributes indirectly and, to a lesser degree, directly to higher labor force participation, with both effects being strongest among women with the weakest attachment to the work force.
12 Labor market conditions for well-educated women may well have improved during the 1970’s, leading to job opportunities that were better than women had earlier anticipated. Elsewhere we have presented evidence that the payoff to education among women in their early thirties was greater in 1978 than in 1967, and that women’s labor force participation was more responsive to their earnings potential at this later date. See David Shapiro and Lois B. Shaw, “Growth in the Labor Force Attachment of Married Women: Accounting for Changes in the 1970s,’ Southern Economic Journal, October 1983, pp. 461-73.
13 Women with strong work plans had slightly larger families, but fewer preschool children, than other women in 1980. The birth expectations of all three groups were similar as of 1973.
14 The plans-tenure interaction term tests for differential investment behavior in the current spell of employment. However, because average tenure was 5 years, many women may have changed their work plans after their current employment began. In this case, any effect of early plans on investment would be attenuated. See Sandell and Shapiro, “Work Expectations,’ for further discussions.
15 Because the wage equation is semilogarithmic, the effect of 7 versus 0 years of plans for work is calculated as (e7(.036) – 1) = .287.
16 See Sandell and Shapiro, “Work Expectations.’ To test more fully for the presence of differences in the slopes and starting points of the experience-wage profiles, we examined the robustness of our results in several ways. We first estimated wage equations with a quadratic specification of work experience and with the corresponding interaction terms. The quadratic terms and corresponding interactions were not significant. Similarly, elimination of either interaction term from equation 1 of table 7 did not change the basic results. Finally, we used the lambda probit technique (James J. Heckman, “Sample Selection Bias as Specification Error,’ Economietrica, January 1979, pp. 153-61) to correct for possible sample selection bias. The coefficient on lambda was insignificant and the coefficients on other variables in the analysis were affected very little. In addition, we examined the frequency distributions of work experience for women with differing work expectations. Our thought was that the absence of significant differences in the experience-wage profiles might reflect a paucity of observations of women who combined strong expectations of future work with very limited work experience, and of women without work plans, but with extensive work experience. However, this was not the case; considerable variation in accumulated work experience was evident within each work-plans group.
17 When the regressions in table 7 were repeated for the entire NLS sample, ages 26-36, the results were similar to those of Sandell and Shapiro, “Work Expectations,’; women with work plans had lower initial wages, but steeper wage growth paths than women with no plans for working. This finding suggests that combining the experiences of different cohorts may sometimes produce spurious wage-growth profiles that do not describe the experience of any one age group.
18 Reuben Gronau, Sex-Related Wage Differentials and Women’s Interrupted Labor Careers–The Chicken or the Egg, Working Paper No. 1002 (Cambridge, MA, National Bureau of Economic Research, October 1982).
19 It should be noted that our occupational work-plans variables undoubtedly contain a good deal of measurement error, particularly given our use of 3-digit occupational codes. For example, a respondent might have plans for the same occupation each year, but slight variations in how she describes the occupation could lead to coding errors. Also, we have not made any allowance here for the existence of “families’ of occupations (clusters of 3-digit occupations that share common skill requirements to a high degree).
20 David Shapiro and Joan E. Crowley, “Aspirations and Expectations of Youth in the United States: Part 2. Employment Activity,’ Youth and Society, September 1982, pp. 33-58.
21 Of course, the extent to which women will be disproportionately concentrated in low-wage jobs will depend not only on their work plans, but also on employers’ perceptions of women’s work commitment and their willingness to hire women for high-skill jobs.
COPYRIGHT 1987 U.S. Bureau of Labor Statistics
COPYRIGHT 2004 Gale Group
