Family focus or career focus: controlling for infertility

Family focus or career focus: controlling for infertility

Social Science & Medicine 49 (1999) 1615±1622 www.elsevier.com/locate/socscimed Family focus or career focus: controlling for infertility Christine ...

104KB Sizes 0 Downloads 31 Views

Social Science & Medicine 49 (1999) 1615±1622

www.elsevier.com/locate/socscimed

Family focus or career focus: controlling for infertility Christine Siegwarth Meyer* Bentley College, Department of Economics, 175 Forest Street, Waltham, MA, USA

Abstract In order to shed light on the direction of causality between fertility timing and earnings, this paper uses medical diagnoses of infertility as instruments for age at ®rst birth (for those women who did give birth) and childlessness among married women. Although multivariate ordinary least squares regression results ®nd a positive correlation between childbirth at later ages and higher wages as well as between childlessness and increased wages, delays in childbearing due to infertility do not signi®cantly increase a woman's wages. Thus, data from the 1995 National Survey of Family Growth (NSFG) indicate that delaying childbirth does not, by itself, guarantee higher wages in the labor market. Therefore, this study does not support the conventional notion of the `mommy track' in which career success and motherhood are incompatible. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: Infertility; Childbearing; Wages; Labor market; United States

Introduction Women in the US are delaying their childbearing, with the ®rst birth rate for 20±24 year olds falling from 78.2- women in 1970 to 53.4 in 1992 and the corresponding ®rst birth rates of women over the age of 25 rising during the same time period (National Center for Health Statistics, 1995). Whether increased maternal age causes detrimental e€ects to the woman or her child is very much in debate. The label `high risk' is often applied to pregnant women over the age of 35 because medical practitioners are worried about complications during labor and delivery as well as birth defects and other medical problems for the child (Blum, 1979). One reason for delayed childbearing among American women may be the perception that choosing to combine childrearing and work outside the home

* Tel.: +1-781-891-2476; fax: +1-781-891-2896. E-mail address: [email protected] (C.S. Meyer)

means sacri®cing career success. The question of whether or not delays in childbearing are necessary and sucient for a woman who wishes to advance in the labor market is central to the ongoing debate about the `mommy track,' an expression ®rst popularized after the publication of `Management Women and the New Facts of Life' by Schwartz (1989). Feminists have vehemently denied that such a tradeo€ is necessary (Bureau of National A€airs, 1989) but the controversy over the career consequences of delayed or foregone childbirth continues (Korzec, 1997). Research to understand whether early childbearing hinders a woman in the labor market is especially challenging. While data from the National Survey of Family Growth (NSFG) show that women in the sample who gave birth to their ®rst child after the age of 30 earn in excess of US$10,000 more than their counterparts who gave birth at earlier ages, the causality between women's choices about when to have children and their labor market outcomes is dicult to determine (Cramer, 1980). On the one hand, a woman who chooses both to have children early and to pro-

0277-9536/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 9 9 ) 0 0 2 1 0 - 5

1616

C. Siegwarth Meyer / Social Science & Medicine 49 (1999) 1615±1622

vide home care of these children may make labor market decisions which are consistent with those choices. For example, the woman may forgo higher education since the cost of schooling is not justi®ed if she does not receive the high wages associated with this education. Even after her children are grown and no longer living at home, she may choose not to reenter the labor force because her lack of experience and tenure leads to a low market wage. On the other hand, the labor market may also in¯uence a woman's fertility choices. For example, a woman who is on a career track in which early career experience and networking is essential may delay childbearing until she is established in her occupation. One important question is whether delayed childbirth is an important factor for women's labor market success. Women who give birth at later ages and women who remain childless tend to have higher levels of education (Rindfuss et al., 1996) and higher earnings than do women who give birth early in their life (Blackburn et al., 1993). One might conclude that delaying childbirth is a necessary precondition for high earnings. However, it might be that women are making fertility timing choices based on their expected wages in the market and that the observed correlation between delayed fertility and high wages re¯ects that choice. In other words, it may be that high expected wages lead some women to choose to delay childbirth. Methods The technique of instrumental variables is ideal for analysis in which the direction of causality between two outcomes is in question. The essential element necessary for such an analysis is one or a set of instrumental variables, observable conditions which are correlated with delayed childbearing and childlessness, but are otherwise uncorrelated with unobservable determinants of wages (Greene, 1990). The instrumental variables used are medical diagnoses that indicate infertility. Such diagnoses will cause delays in childbearing but need not lead to childlessness since advanced reproductive techniques (in vitro fertilization, for example), gamete donation, surrogacy and adoption provide alternative ways in which a family can be created. Conventional multivariate regression analysis in which wages are the dependent variable and age at ®rst birth is included in the set of independent variables is ¯awed since women who delay childbearing are more likely to be career oriented, due in part to unobservable background and family factors, than are women who give birth at early ages. Therefore, age at ®rst birth is not simply picking up fertility di€erences, but also unobservable factors correlated with one's

wage. Using infertility diagnoses as instruments, however, overcomes this problem, since infertility strikes women randomly and is not (with several exceptions discussed below) correlated with unobservable factors in¯uencing earnings. Whether medical diagnoses of infertility are uncorrelated with earnings, aside from their in¯uence on the timing of childbirth, is not a straightforward issue. Infertility is de®ned as the inability of a couple to become pregnant after 12 months of unprotected intercourse. In the United States, 7.1% of married couples (2.1 million couples) were infertile in 1995 (Abma et al., 1997). Research for developed countries shows that in 37% of infertile couples, the infertility can be traced solely to female factors; in 8% of cases, male factors are the only source; and in 35% of cases, both partners have conditions contributing to the infertility (World Health Organization, 1992). These percentages do not add up to 100% since in 5% of cases, no cause is found and in 15% of cases, the couple became pregnant during the investigation. The underlying infertility diagnosis falls into one of several categories. First, the female partner may have problems with ovulation in which the ovaries are not producing eggs normally, blocked or damaged fallopian tubes, other pelvic problems, or endometriosis, a disorder that may cause in¯ammation and scarring in the pelvic cavity. Second, infertility may be caused by sperm or semen problems in the male partner. Third, it may be the result of other, less common factors such as DES exposure. In the 1950s and 1960s, diethylstilbestrol (DES) was given to some women to prevent miscarriages. Exposure to DES in the womb has been correlated with infertility, speci®cally tubal and uterine abnormalities (Kaufman, 1980). Fourth, infertility may be unexplained. The likelihood of a successful pregnancy depends, to some extent, on the underlying diagnosis. For instance, in a 1995 study of clinics performing advanced reproductive techniques (including IVF (in vitro fertilization), intracytoplasmic sperm injection (ICSI), GIFT (gamete intrafallopian transfer) and ZIFT (zygote intrafallopian transfer), the rate of live births per cycle ranged from a high of 21.4% for couples with male factor infertility to a low of 17.2% for couples with unexplained infertility. These numbers should be compared with a pregnancy probability of 25% for a non-impaired couple having regular unprotected intercourse (Hull et al., 1992). Older women are more likely to report themselves as being infertile. This can stem from two causes. First, older women may be more likely to be attempting a pregnancy and therefore, may be less likely to be using contraception. Thus, unlike in younger women, birth control methods are not masking impaired fertility (Kalmuss, 1987). Second, fertility of women has been shown to decline after the age of 35 (Hansen, 1988)

C. Siegwarth Meyer / Social Science & Medicine 49 (1999) 1615±1622

1617

Table 1 Descriptive statistics by infertility status standard deviations in parentheses

Age at ®rst birth Childless Number of children in the household Northeast Midwest South Residence in an SMSA Black Hispanic Age Earnings Husband's earnings Years of schooling Experience Tenure Number of observations

All women

Fertile or sterile women

Infertile women

23.57 (4.881) 0.2343 (0.4236) 1.332 (1.137) 0.1790 (0.3834) 0.2503 (0.4333) 0.3305 (0.4705) 0.7883 (0.4086) 0.1366 (0.3435) 0.1292 (0.3355) 33.75 (6.511) 22,077 (16,638) 34,251 (20,932) 13.37 (2.611) 10.34 (6.877) 5.185 (5.342) 3939

23.53 (4.832) 0.2180 (0.4129) 1.369 (1.137) 0.1794 (0.3838) 0.2514 (0.4339) 0.3306 (0.4705) 0.7888 (0.4082) 0.1322 (0.3387) 0.1303 (0.3367) 33.71 (6.550) 22,004 (16,483) 34,206 (20,797) 13.39 (2.615) 10.24 (6.853) 5.148 (5.318) 3684

24.48 (5.775) 0.4706 (0.5001) 0.7922 (0.9801) 0.1725 (0.3786) 0.2353 (0.4250) 0.3294 (0.4709) 0.7804 (0.4148) 0.2000 (0.4008) 0.1137 (0.3181) 34.30 (5.901) 23,124 (18,739) 34,899 (22,824) 13.07 (2.538) 11.81 (7.071) 5.720 (5.663) 255

and there is some evidence that advancing age may also have an impact on male fertility (Murray and Meacham, 1993). The largest drop in fertility among women seems to occur between 30 and 40 year of age, with very low fertility rates after the age of 40. Studies of both `natural' populations, i.e. groups in which no contraception occurs (Eaton and Mayuax, 1953) and of women undergoing advanced reproductive therapies (Schwartz and Mayuax, 1982) show decreased pregnancy rates and increased rates of miscarriage in older women, particularly in women over the age of 40. The risk of miscarriage associated with delayed childbearing is the subject of some controversy. Mans®eld (1988) surveys the existing literature relating advanced maternal age with negative pregnancy outcomes and ®nds little empirical support for the relationship between the two events. The main factor in this decreased fertility and increased rate of spontaneous abortion among older women seems to be lower ovarian responsiveness, oocyte (or immature eggs) reservoir and oocyte quality. A woman begins her reproductive life with a ®xed number of oocytes, of which a certain number are lost each month. Although the relationship between age and a woman's oocyte quality and ovarian reserve is not clearly understood, it may be that poor nuclear development or abnormalities in the chromosomes are part of the answer (Planchot et al., 1988). One piece of evidence supporting the hypothesis that the oocyte is the major cause of decreased fertility at advanced ages is that women who receive donor oocytes from younger women see fertility rates of 20±36%, higher than women using other assisted reproductive techniques (Sauer et al., 1991). Infertility that occurs when a woman has chosen to

delay childbearing is not exogenous to her labor market outcomes. In other words, infertility which stems from a woman's choice to postpone her ®rst pregnancy may be in¯uenced by factors related to her wage in the labor market, namely a focus on career advancement and success. Therefore, the infertility diagnoses used as instruments to identify exogenous delays in childbearing will not include those most associated with advanced female age, namely ovulatory problems and unexplained infertility. Only the following four diagnoses will be considered valid instruments: blocked tubes, other tubal or pelvic problems, endometriosis and male factors. The data used in this paper are taken from the National Survey of Family Growth (NSFG) Wave V, a cross-section of 10,847 married women in 1995. Only 3939 women have non-missing data for all the variables used in the model. Out of this group, 225 women (roughly 6%) report having impaired fertility not due to sterilization. The NSFG is unique in that it reports detailed data on both infertility status as well as income, thus providing the data necessary for this study. Results Table 1 reports means and standard deviations for all variables used in the analysis by fertility status. For purposes of this analysis, I compare infertile women with all other women, a group which includes both fertile and sterilized women, since sterilization is a choice made by the woman and her partner. As expected, both age at ®rst birth (for those women who did give birth) and the percent of women who are childless is

b

a

Signi®cant at the 10% level. Signi®cant at the 5% level.

Instruments R2 Number of observations

Age at ®rst birth Childless Years of schooling Number of children in the household Northeast Midwest South Residence in an SMSA Black Hispanic Age Age squared Husband's earnings (10ÿ5) Constant 0.0153 (0.0843) 0.0413 (1.01) 0.0637 (0.0391) ÿ0.0262 (0.238) 0.00877 (0.0541) ÿ0.131b (0.0309) ÿ0.0750b (0.0313) 0.123b (0.0456) 0.0603 (0.0735) 0.0596a (0.0315) 0.0681b (0.0240) ÿ0.000845b (0.000264) 0.674b (0.0638) ÿ0.502 (0.308) 2 Diagnoses 0.2491 3939

0.2500 3939

Model 1

Instrumental variables

0.0135b (0.00249) 0.313b (0.0668) 0.0659b (0.00424) ÿ0.0392b (0.0108) 0.0831 (0.0294) ÿ0.131b (0.0274) ÿ0.0773b (0.0258) 0.124b (0.0234) 0.0558b (0.0284) 0.0577a (0.0301) 0.0697b (0.0141) ÿ0.000881b (0.000214) 0.673b (0.0489) ÿ0.473b (0.225) None

OLS

0.1654 3939

ÿ0.0189 (0.0852) 0.0201 (1.00) 0.0794b (0.0392) 0.0652 (0.245) 0.0270 (0.0557) ÿ0.137b (0.0323) ÿ0.0692b (0.0329) 0.138b (0.0463) 0.0331 (0.0741) 0.0615a (0.0331) 0.0757b (0.0243) ÿ0.000824b (0.000277) 0.691b (0.0657) ÿ0.426 (0.315) 4 Diagnoses

Model 2

0.2225 3935

0.0365 (0.0395) 0.628 (0.564) 0.0552b (0.0196) ÿ0.0880 (0.123) ÿ0.00324 (0.0367) ÿ0.130b (0.0288) ÿ0.0789b (0.0286) 0.113b (0.0302) 0.0763a (0.0448) 0.0526a (0.0315) 0.0637b (0.0177) ÿ0.000874b (0.000259) 0.662b (0.0542) ÿ0.544b (0.266) 4 Diagnoses and mom's fertility

Model 3

Table 2 Multivariate ordinary least square (OLS) and instrumental variables regressions dependent variable: natural log of the hourly wage standard errors in parentheses

1618 C. Siegwarth Meyer / Social Science & Medicine 49 (1999) 1615±1622

C. Siegwarth Meyer / Social Science & Medicine 49 (1999) 1615±1622

higher for the infertile group. Additionally, this same group has, on average, fewer children in the household. Consideration was given to using infertility diagnoses as instruments for the number of children in the household. However, in the ®rst stage regressions, infertility diagnoses were not signi®cant predictors making them poor instruments. In my subsample of 3939 women, black women are more likely to be infertile than white women are, which is consistent with the distribution of infertility status by race across the entire sample of NSFG women. Labor market earnings for the infertile population are higher than for the fertile/sterile women. The goal of this paper is to determine whether the higher earnings are directly related to the delayed childbearing experienced by women who are diagnosed with a medical condition leading to infertility. In looking for factors which contribute to the higher earnings of the infertile population, population means suggest that the higher level of labor market experience and tenure in one job seen in the infertile population lead to the increased earnings. Estimation results from multivariate ordinary least squares (OLS) and instrumental variables regressions of the natural log of hourly wages for women with positive labor market earnings are found in Table 2. The OLS regression model, in which fertility timing is treated as being exogenous or not in¯uenced by career aspirations or labor market success, indicates that all of the variables are signi®cant (with a p-value of less than 0.01) with one exception. Wages of women living in the Northeast region of the United States are not signi®cantly di€erent from those living in the Western region (the region). Overall, however, region is an important factor in wage determination, since an F-test ®nds that jointly, the three regional dummy variables are signi®cant with a p-value of less than 0.0001. The positive relationship between delayed or foregone childbearing and a woman's earnings are seen with the positive and signi®cant coecients on age at ®rst birth and the indicator variable for childlessness in the OLS model. Two variables, namely years of schooling and age, are included in the regression to indicate the additional productivity (or human capital) of a worker with additional education or with additional experience in the labor market (which is approximated by the worker's age). The signi®cant coecients on these variables re¯ect the positive e€ect that additional human capital has on earnings, either directly through increased productivity or through other economic e€ects such as a signaling of the worker's ability to learn new tasks (in the case of years of schooling) or a compensation system in which wages increase over time with the same employer (in the case of age). Because age (potential labor market experience) is used instead of actual years of experience, the coecient on age at ®rst birth

1619

and childlessness may overstate the direct e€ect of these variables on earnings, since a portion of the e€ect may come indirectly through labor market experience. Furthermore, the coecient on age squared is negative indicating a declining marginal e€ect of age of earnings. Controlling for other factors, black and Hispanic women in this sample have higher earnings than whites. Within the entire NSFG sample, the average black woman earns 96% of the average white woman's salary. However, within the sample of women for whom all data was non-missing, the average black woman actually earned 3% more than her white counterpart, indicating that our regression sample is not fully representative of the racial distribution of the NSFG or that other variables included in the multivariate regression explain the lower compensation received by black women. Residence in the Midwest (relative to the West) and in an SMSA signi®cantly a€ect annual earnings, most likely due to di€erences in the cost of living. There are two possible e€ects of husband's earnings on the wages of the wife. The ®rst, called the income e€ect, states that with an increase in other household earnings, a woman may choose to work fewer hours per week. Since part-time workers generally receive lower hourly wages than do their full-time counterparts, the income e€ect would imply that an increase in the earnings of the husband would decrease the wages of the wife. On the other hand, men and women tend to positively assortatively match on wages, meaning that high wage earning men tend to marry high wage earning women. This e€ect would imply that an increase in the earnings of the husband would be associated with a wife who has higher wages. Results from the multivariate regression indicate that the husband's earnings a€ect a wife's wages positively, indicating that the e€ects of assortative matching in marriage outweigh the income e€ect. The last three columns of Table 1 report results from instrumental variable regressions of annual earnings. In these models, instruments are used to identify age at ®rst birth and the indicator variable for childlessness. Three models rely on di€erent sets of instruments to check for robustness. In model 1, only two diagnoses of infertility, endometriosis and male factor infertility, are used as instruments. In model 2, four diagnoses of infertility are used to identify the e€ect of an exogenous delay in childbearing on earnings: endometriosis, male factor infertility, blocked tubes and other tubal and pelvic problems. In model 3, those same four diagnoses are included as well as a variable that is not a diagnosis: the number of children the woman's mother had. The fertility of one's mother does not directly in¯uence a woman's current earnings, but it is possible that it is

1620

C. Siegwarth Meyer / Social Science & Medicine 49 (1999) 1615±1622

Table 3 Auxiliary OLS regressions dependent variable: age at ®rst birth standard errors in parentheses. F-test reports the p-value of a test with a null hypothesis that the coecients on the instruments are all equal to zero

Years of schooling Number of children in the household Northeast Midwest South Residence in an SMSA Black Hispanic Age Age squared Husband's earnings (10ÿ5) Constant Infertility diagnosis: endometriosis Infertility diagnosis: male factors Infertility diagnosis: blocked tubes Infertility iagnosis: other tubal and pelvic problems Number of children mother had F-test a b

Model 1

Model 2

Model 3

0.748b (0.0242) ÿ0.265b (0.0597) 0.361a (0.190) ÿ0.288 (0.181) ÿ0.225 (0.166) 0.816b (0.151) ÿ1.69b (0.186) 0.216 (0.183) 1.22b (0.0979) ÿ0.0167b (0.00147) 2.00b (0.298) ÿ8.32b (1.57) 2.13b (0.439) 1.60b (0.428)

0.737b (0.0242) ÿ0.263b (0.0599) 0.362a (0.190) ÿ0.287 (0.181) ÿ0.217 (0.166) 0.825b (0.151) ÿ1.70b (0.186) 0.209 (0.183) 1.22b (0.0979) ÿ0.0167b (0.00147) 2.00b (0.298) ÿ8.32b (1.57) 1.92b (0.454) 1.51b (0.431) 0.594 (0.421) 0.407 (0.427)

< 0.00001

< 0.00001

0.738b (0.0247) ÿ0.262b (0.0601) 0.355a (0.190) ÿ0.296 (0.181) ÿ0.221 (0.166) 0.826b (0.151) ÿ1.70b (0.190) 0.199 (0.188) 1.22b (0.0981) ÿ0.0167b (0.00147) 2.00b (0.300) ÿ8.35b (1.57) 1.92b (0.454) 1.51b (0.431) 0.593 (0.421) 0.413 (0.430) 0.00104 (0.0243) < 0.00001

Signi®cant at the 10% level. Signi®cant at the 5% level.

correlated with a woman's attitude regarding the relative importance of career and family. It could be argued that the mother's fertility might be correlated with unobservable characteristics associated with a woman's earnings; however, a test reported below fails to reject its exogeneity. Results from all three models indicate that exogenous delays in childbearing and exogenous childlessness do not signi®cantly increase a woman's earnings in the labor market. In models 2 and 3, the number of instruments is greater than the number of endogenous variables, allowing for a test of whether the additional instruments included in the model are indeed exogenous (not in¯uenced by current market wages). For both models, this test fails to reject the null hypothesis of exogenous instruments (with p-values of 0.139 and 0.315, respectively). This test is conducted by regressing the residuals of the earnings equation on all exogenous variables including the instruments. The test statistic is calculated as N (the number of observations) multiplied by R2 from this regression of the residuals. The test statistic is distributed w 2 with degrees of freedom equal to the number of overidentifying restrictions (two in model 2 and three in model 3). It is possible that the statistical insigni®cance of the endogenous variables (age at ®rst birth and childlessness) in the instrumental variables models occurs simply because the instruments have poor predictive power, thus leading to predicted values in the second

stage which are signi®cantly collinear with the other exogenous variables. A series of F-tests conducted after the ®rst stage regressions (exogenous variables regressed on each of the endogenous variables) rejects the hypothesis that the coecients on the instruments are jointly equal to zero (with p-values less than 0.0001) in each model and for each endogenous variable. Results of these F-tests along with the results of the auxiliary IV regressions of age at ®rst birth as a function of the explanatory variables and the instruments can be found in Table 3. Discussion Economic analyses of fertility outcomes look at supply side factors (in¯uences on the couples' ability to bear children) and demand side factors (in¯uences on the couples' desire for children). Several previous studies have focused on the supply side of the fertility outcome to identify exogenous fertility changes and their e€ect on labor market outcomes to test whether a woman's earnings are a€ected by uncontrollable events in her life which a€ect the timing of childbearing or the number of children she has. To test whether unanticipated extra children a€ect labor supply decisions, Rosenzweig and Wolpin (1980) used the occurrence of twins in a woman's ®rst pregnancy as an exogenous fertility change. They ®nd that, while lifetime labor

C. Siegwarth Meyer / Social Science & Medicine 49 (1999) 1615±1622

supply, as measured by labor force participation at age 35±44, seems to be una€ected by the occurrence of twin births, there is a temporary reduction in labor supply immediately following the birth. Rosenzweig and Schultz (1985) estimate the e€ects of contraception on the monthly probability of conception; this allows them to separate the behavioral (referring to the use of contraceptives) and biological components of fertility. The biological component, then, serves to identify exogenous fertility supply di€erences between couples and their e€ect on labor supply and earnings. Their ®ndings support the hypothesis that exogenous variations in birth supply negatively a€ect labor supply and earnings of the mothers. Hotz et al. (1997)identify the e€ect of teen childbearing on a woman's subsequent education and earnings by using women who become pregnant as teenagers but experience a miscarriage as a control group. They ®nd little di€erence in subsequent education and labor market outcomes between women who got pregnant as teenagers but later miscarried and those who had a teen birth. Their analysis, therefore, does not support the theory that early childbearing places teen mothers on a path of low earnings and public assistance. Although these previous studies have examined the e€ects of exogenous shocks to child quantity and teenage fertility on labor market outcomes, similar analyses have not been conducted to examine whether the timing of childbearing among women over the age of twenty directly a€ects their labor market outcomes. This paper ®lls this void by using medical diagnoses of infertility to identify the e€ect of unplanned delays in childbearing on the wages of women. OLS estimates con®rm a positive correlation between delayed or foregone childbirth and a woman's labor market earnings. However, when medical diagnoses of infertility are used as instrumental variables for delayed and foregone childbirth, the positive association is no longer present. This indicates that a postponed ®rst birth or the lack of any birth due to medical reasons does not lead to higher labor market wages, despite the potential for the woman to have received additional education and early experience. Thus, the empirical results suggest that the `mommy track' typically does not involve a severe trade-o€ with the career track.

Acknowledgements Lucia Nixon, Professor Rex Santerre and participants in the Bentley College Department of Economics seminar series provided critical advice for this paper.

1621

References Abma, J., Chandra, A., Mosher, W., Peterson, L., Piccinino, L., 1997. Fertility, Family Planning, and Women's Health: New Data from the 1995 National Survey of Family Growth. National Center for Health Statistics. Vital Health Stat. 23(19). Blackburn, M.L., Bloom, D.E., Neumark, D., 1993. Fertility timing, wages and human capital. Journal of Population Economics 6, 1±30. Blum, M., 1979. Is the Elderly Primapara Really at High Risk? Journal of Perinatal Medicine 7, 108±112. Bureau of National A€airs, 1989. The `Mommy Track' Debate and Beyond: Public Policy? Corporate Reality? Special Report 16 (April). Cramer, J.C., 1980. Fertility and female employment: problems of causal direction. American Sociological Review 45, 167±190. Eaton, J.W., Mayuax, M.J., 1953. The social biology of very high fertility among the Hutterites: the demography of a unique population. Human Biology 25, 206. Greene, W.H., 1990. Econometric Analysis. Macmillan, New York. Hansen, J.P., 1988. Older maternal age and pregnancy outcome: a review of the literature. Obstetrical and Gynecological Survey 41, 276. Hotz, V.J., Williams McElroy, S., Sanders, S.G., 1997. The costs and consequences of teenage childbearing for mothers. In: Kids Having Kids: Economic Costs and Social Consequences of Teen Pregnancy. Urban Institute Press, Washington, DC. Hull, et al., 1992. Expectations of Assisted Conception for Infertility. British Medical Journal 304, 1465±1469. Kalmuss, D.S., 1987. The use of infertility services among fertility-impaired couples. Demography 24 (4), 575± 585. Kaufman, R.H., et al., 1980. Upper genital tract changes and pregnancy outcome in o€spring exposed in utero to diethylstilbestrol. American Journal of Obstetrics and Gynecology 137, 299±308. Korzec, R., 1997. Working on the `mommy-track': motherhood and women lawyers. Hastings Women's Law Journal 8 (1), 117±140. Mans®eld, P.K., 1988. Midlife childbearing: strategies for informed decisionmaking. Psychology of Women Quarterly 12, 445±460. Murray, M.J., Meacham, R.B., 1993. The e€ect of age on male reproductive function. World Journal of Urology 11, 137±140. National Center for Health Statistics, 1995. Vital Statistics of the United States, 1992, Vol. I, Natality. Public Health Service, Washington. Planchot, M., DeGrouchy, J., Junca, A.M., et al., 1988. Chromosomal analysis of human oocytes and embryos in an in vitro fertilization program. Annals of the New York Academy of Science 541, 384. Rindfuss, R.R., Morgan, S.P., O€utt, K., 1996. Education and the changing age pattern of American fertility: 1963± 1989. Demography 33 (3), 277±290. Rosenzweig, M.R., Wolpin, K.I., 1980. Life-cycle labor supply and fertility: causal inferences from house-

1622

C. Siegwarth Meyer / Social Science & Medicine 49 (1999) 1615±1622

hold models. Journal of Political Economy 88 (2), 328± 348. Rosenzweig, M.R., Schultz, T.P., 1985. The demand for and supply of births: fertility and its life cycle consequences. American Economic Review 75 (5), 992±1015. Sauer, M.V., Paulson, R.J., Macaso, T.M., et al., 1991. Preembryo donation to women with ovarian failure. Fertility and Sterility 55, 39. Schwartz, D., Mayuax, M.J., 1982. Female fecundity as a

function of age: results of arti®cial insemination in 2193 nulliparous women with azoospermic husbands. New England Journal of Medicine 306, 404. Schwartz, F., 1989. Management women and the new facts of life. Harvard Business Review 67 (1), 65±76. World Health Organization, 1992. Recent Advances in Medically Assisted Conception. In: WHO Technical Report Series, 820. World Health Organization.