As we noted in the discussion of fertility, the theory of the allocation of time provides a conceptual framework that suggests that in industrialized countries in particular fertility and female labor force participation will likely be inversely related. Further, the theory implies that the two activities will be jointly determined, in the sense that potential earnings (the wage rate) and employment possibilities in the labor market will influence decisions about childbearing and about participation in the labor market.
In this context, then, both the secular decline in fertility and the secular increase in the labor force attachment of married women in the U.S. are seen as consequences of increased real wages and market work opportunities available to women. Further, technological change is an important contributing factor in both cases: it has facilitated growth in married women's labor force participation by providing substitutes for the wife's time at home, and it has lowered the costs of fertility control.
The lectures in this section will focus on labor force participation and fertility. As a starting point, we need to consider the economic framework for analysis of female labor force participation: neoclassical labor supply theory. In this framework, a woman's decision regarding labor force participation in a particular period, t, will depend on a comparison of the marginal benefit and the marginal cost of having a job.
The marginal benefit of having a job is the market wage rate, Wt. More formally, as noted by Klerman and Leibowitz (K & L), one would want to look at the net wage rate -- i.e., the after-tax wage rate taking into account income tax credits for child care costs. The net wage rate should also take into consideration costs associated with having employment.
The marginal cost of having a job is the value of the woman's time spent in nonmarket activities. This value is called the shadow price of time or shadow wage, and we'll denote it as Wt*. The shadow price of time, then, measures the value the woman places on marginal units of her time in nonmarket production and consumption. It is in fact equal to the absolute value of the slope of an indifference curve in the income-nonmarket time space.
Since we are focusing here on the decision to participate in the labor force, we are interested in the value of Wt* when hours of market work equal zero (i.e., the absolute value of the slope of the indifference curve at the "corner", where Hw = 0). This is known as the reservation wage (Wr).
If Wt > Wr, the woman will work in the labor market. However, if this inequality is not satisfied (i.e., if Wr is greater than or equal to Wt), she will remain out of the labor force. Fig. I.D.1 provides a diagrammatic representation of this decision that also shows determination of equilibrium hours of work for labor force participants (equilibrium being where Wt = Wt*).
An individual's (potential) market wage rate is determined by personal characteristics, such as schooling, work experience, and health status, and by characteristics of the labor market in which she is located (rural/urban, region, industrial mix). The shadow price of a married woman's time (and hence the reservation wage) tends to be positively related to husband's income (or other family income, more broadly), household assets, the woman's education level, and the number of children, and negatively related to the ages of children.
Consider in particular the impact of the age of the youngest child on a woman's reservation wage. There is a high time intensity of newborns in particular and also infants (under age one), which in our terminology could be described as high productivity of mother's time in child care (steep indifference curves). This translates into a high reservation wage.
However, as children age they become less time intensive with respect to mother's time (K & L: "As the child grows, the marginal product of the mother's home time falls"), and market substitutes for the mother's time become increasingly feasible. Hence, the reservation wage falls as the child ages, and given the market wage, this means that the likelihood of employment increases as the child gets older.
Traditionally, the birth of a first child was a key life cycle point associated with a period of extended withdrawal from the labor market by women to engage in child care (K & L: "In 1960, most women left the work force during pregnancy and remained out of the labor force until their youngest child entered school.") However, this pattern has eroded substantially over the course of the past 30-35 years, such that now more than half of mothers of infants under the age of 1 are employed in the labor market.
K & L argue in their paper that the dramatic growth in married women's labor force attachment following the birth of a child is a consequence of three factors: improved wages and market opportunities for women, poorer opportunities for their spouses, and more generous tax treatment of child care expenses. I want to turn now to look at a body of empirical work that Frank Mott and I have done that is longitudinal in nature, and sheds additional light on linkages between fertility and female labor force participation. We also look at links between labor force activity at different life cycle points, and the longer-term consequences for wages. In essence, there are three "generations" or distinct time frames to our work.
We used data from the NLS of Young Women, aged 14-24 in 1968. Our sample consisted of women in the panel who gave birth to their first child between the 1968 and 1973 interviews. We began by examining the labor force participation and employment of these women during the pregnancy and the infancy of that first child.
Nearly two thirds of the women worked well into their pregnancy (i.e., they were employed within the six-month period prior to the birth). Perhaps more surprisingly, in view of the traditional pattern, over a third of the women were employed soon (within six months) after the birth. We argued that the strong work attachment of these prospective and new mothers was both a consequence of the long-term increase in female work attachment (i.e., a reflection of increased real wages for women), and also a portent of strong lifetime ties to the labor market.
We also found some support for the economic framework for analyzing labor force activity even in the months surrounding the first birth. That is, in a reduced-form analysis, higher educational attainment was associated with a greater likelihood that pregnant young women and recent mothers would be employed (note dual effect of education and implication), while higher family income and being married were associated with lower probabilities of employment.
At the same time, there was considerable variation in labor force participation that could not be accounted for by the variables used in our operationalization of the economic approach. We suggested that much of this variation reflected unmeasured differences in lifetime work attachment, and we hypothesized that these differences would manifest themselves in later years.
Our second generation constituted a follow-up study as of 1978, 5-10 years following the first birth. During this period most of the women went on to have additional children. Our focus was on examining the link between early work attachment (i.e., labor force participation in the period surrounding the first birth) and later work attachment (5-10 years after the first birth).
We expected to find that the women who remained in the labor force in the months just before and just after the first birth would be less likely to go on to have additional children or would have fewer additional children, and consequently we expected them to be more likely to be working later. This hypothesis simply reflected the view that fertility and labor force participation would be inversely related.
However, there was little evidence of such a phenomenon: work behavior around the time of the first birth was not a good predictor of subsequent fertility. At the same time, first-birth employment activity was an excellent predictor of subsequent employment activity, even after controlling for measured characteristics like schooling that are good predictors of work activity both around the time of the first birth and subsequently.
Note that this net association may reflect either or both of two kinds of effects: state dependence or heterogeneity. A state dependence effect exists when being in a state this period influences the likelihood of being in the state in one or more subsequent periods. Human capital accumulation associated with increased work experience and job tenure, by raising a woman's market wage, clearly provides one mechanism whereby we might expect true state dependence in this setting.
In the case at hand, heterogeneity simply refers to the idea that there are underlying, unmeasured (unobservable) differences across individuals that influence their propensity to be employed. These differences would reflect things like a taste for market work or a career orientation.
Our third generation study follows these women for an additional nine years, up through 1987 -- 14 to 19 years following the birth of their first child. By this time, they had mostly finished their childbearing, and more than three quarters of them were employed. Again, we wanted to see the linkages between work at the time of the first birth and subsequent employment. In addition, we wanted to see how first-birth employment was related to wages.
(What follows are highlights of class discussion of overheads of some of the tables and graphs in our 1994 JHR paper plus some additional tables and graphs. If you missed this class please come and see me so that you can take a look at the overheads that I showed in class.)
Our focus is on the different patterns of work activity surrounding the first birth, and how these are related to subsequent fertility, employment experience, and wages. Table 1 of our paper provides an overview of these patterns. It is convenient to contrast the women who were employed both soon before and soon after the first birth (those most strongly attached to the labor market) with those who were not employed in the 12-month period surrounding the birth (those least strongly attached).
The weakness of first birth employment activity as a predictor of subsequent fertility mentioned earlier is still apparent for whites for the longer period under consideration here, as indicated in Table 3 (there is but a modest difference between the two extreme groups). By contrast, the inverse association that we had expected initially is evident among blacks.
Tabular data not included in our published paper (overhead Table 3) indicate fairly clearly that even as of 1987, both childbearing and work activity surrounding the first birth are related to the likelihood that a woman is employed. The effects of additional children on work behavior over time are also evident from overheads that document some racial differences as well.
A longitudinal perspective is provided by overhead Figures 1, 2, and 3, which show employment activity in the 14 years following the first birth for white women according to both their subsequent parity and their work attachment as of the time of the first birth. In general, it is clear that there is a tendency for employment activity to increase over time so that the differences existing initially diminish. However, the most strongly attached (as of the first birth) manifest the greatest employment activity while the least strongly attached show the lowest employment activity.
Overhead Table 4 documents the significant gross association between employment just before and just after the first birth and employment activity 14-19 years later. There is also a significant relationship net of schooling, income, fertility, and labor market characteristics between measures of employment activity as of 1987 and employment immediately after the first birth.
Examination of the employment activity of these women over the long haul as a function of relevant variables (from the economic framework for analyzing labor supply) plus the "Before" and "After" variables highlights the significance of these measures of first-birth work activity net of measurable characteristics (overhead Table 6). Note that the effect of the "After" variable (when we expect normally that the reservation wage will be extremely high) is more than three times as large as the effect of the "Before" variable.
Finally, overhead Table 8 links up first-birth work behavior, total work experience, and wages in the sort of empirical wage equations that I've been discussing at various points in the course. Several points are relevant here.
First, note that the various human capital variables (schooling, total experience, and job tenure) are generally highly significantly related to wages in the expected direction. Likewise, the labor market characteristics (labor force size and South) also have the expected signs.
Second, once these variables are taken into consideration, the "Before" and "After" variables are insignificant -- i.e., they have no net association with wages. A strong human capital view would lead one to expect some relationship here (reflecting greater investment in on-the-job training by more strongly attached women), but we were unable to find one.
However, we saw earlier that the strongly attached women have greater work experience than other women, ceteris paribus. Hence, there is a payoff (in the form of higher wages) to greater work attachment, with the mechanism being through accumulation of work experience.
As discussed in the conclusion to our paper, these results have interesting implications for various public policy issues such as maternity/family leave provisions and favorable tax treatment for or subsidization of child care. Further, the growth in women's work attachment, as reflected either by observed increases in employment activity or by young women's statements about their future work plans, means that the "most attached women" in our sample have gone from being an important minority to being the majority.