An empirical analysis of adoption.

Medoff, Marshall H.

I. INTRODUCTION

Adoption is as old as recorded history. It was practiced by the ancient Egyptians, Greeks, and Romans. Technically, adoption is a legal procedure by which a couple, or, less commonly, an individual, takes someone else's child into their family and raises it as their own. The child may be unrelated to either adoptive parent, but is often the child of one member of the couple or related in some other way to the adoptive parents. Adoption severs all legal ties between the birth parents and the child and replaces them by a legal relationship between the child and the adoptive parents. In most cases of adoption, the birth parents voluntarily surrender their parental rights so that an adoption can take place.

It seems highly likely that the U.S. Supreme Court will allow further restrictions on the availability of abortion. As a consequence, considerable attention is certain to be focused on the adoption option and its social implications. Prior to 1970, fertility behavior was little studied by economists, partly because it was thought that the determinants were largely noneconomic. During the past two decades, however, economists using a choice-theoretic framework have examined a variety of fertility issues such as marriage (Becker |1974~), procreation and children (DeTray |1973~), birth control (Willis |1973~), and abortion (Leibowitz, Eisen, and Chow |1986~). Surprisingly, no economic research exists on the issue of adoption, despite the fact that it represents a natural extension of the traditional theory of consumer choice. Most adoption research focuses on sociological issues such as infertility or adoption procedures. This paper empirically estimates the supply of adoptions using the decision-making economic framework of completed fertility developed by Becker |1960~ and extended by Michael |1973~. I believe their model's determinants of desired fertility and family size will also be significant in explaining a household's decision to give up a child for adoption. Section II outlines the adoption model and relates it to the economic theory of desired fertility and family size. The third and fourth sections discuss the empirical estimates of the adoption model. The fifth section examines the impact of various state regulations on the adoption decision, and the last section discusses public and social implications of the empirical results.

II. THE ADOPTION MODEL

Michael argues that households' decisions concerning desired family size and the timing of childbearing can be explained in terms of a household production function. Michael's model is based on a comparison of the benefits and costs from an additional child over time. His theory of completed fertility purports to be over a household's full reproductive age span, but it is in actuality a period by period analysis. Households are myopic in that, although they consider the future costs and benefits of a child, an independent decision is made in each period regarding whether to have a child. A household's optimal family size may be, for example, two children. But if when the first child is conceived the household perceives the timing of the birth as undesirable, then, at that point in time, the net benefit from the additional child is negative. The consequence is that the child is "unwanted," and the household will take steps to reduce the discrepancy between the desired and actual stock of children.

One possible solution is an adoption. Once a child is born, a household has two choices: keep the child or put it up for adoption. Michael's framework implies that the likelihood that a household will place a child for adoption is positively related to the (negative) discrepancy between the household's desired and actual stock of children at that time. The factors affecting a household's desired family size--income, time values, relative productivities, and revealed tastes--are reflected in the discrepancy between the desired and actual stock of children. These factors, which determine the desirability of childrearing, should also figure in the adoption decision. The adoption supply equation is modeled in terms of these factors.

The adoption supply equation to be estimated is

(1) ||ADOPT.sup.s~.sub.i~ = |b.sub.0~ + |b.sub.1~ Female |Income.sub.i~

|b.sub.2~Male |Income.sub.i~

+ |b.sub.3~ |Single.sub.i~ + |b.sub.4~ Labor Force |Participation.sub.i~

+ |b.sub.5~ |Unemployment.sub.i~ + |b.sub.6~ |Education.sub.i~

+ |b.sub.7~ Aid Families Dependent |Children.sub.i~

+ |b.sub.8~ |Fundamentalism.sub.i~ + |b.sub.9~ |Black.sub.i~.

The dependent variable is the adoption rate (the number of unrelated adoptions of healthy infants as a percentage of live births) for women of childbearing age (defined as fifteen to forty-four years old) in state i during the 1982 calendar year. The dependent variable examines only healthy infants placed for adoption and excludes the adoptions of children with special needs or by foster parents. It is generally agreed that there is a severe shortage of adoptable healthy infants. The 1982 National Survey of Family Growth estimated that there were approximately one million childless couples, who constitute the group most likely to consider an adoption. However, in 1982 there were less than eighteen thousand healthy infants available for adoption. Consequently, the market for adoptable infants has chronic excess demand and equilibrium is reached solely on the basis of the supply curve. Any exogenous change will shift the adoption supply curve, and a new equilibrium will be achieved at essentially the same price, but at a different level of output (adoptable infants). In effect, the paper examines how various independent variables shift the supply of adoptable infants in the face of chronic excess demand.(1)

The independent variables in equation (1) are restricted principally to women rather than households since it is women who bear the children and who primarily incur the explicit and implicit costs of childrearing.(2) Furthermore, through a succession of opinions, the U.S. Supreme Court has ruled that for unmarried couples the biological mother has primary parental decision rights as long as the biological father's due process is recognized.

The female income variable is the average income of women aged fifteen to forty-four, and the male income variable is the average income of males. The sum of these measures the budget constraint on a household's ability to purchase goods and services. Changes in a woman's income will affect the adoption decision in two ways, with opposite effects. A rise in a woman's income increases the opportunity cost of childrearing in terms of the earnings that a woman must forego if she leaves the workforce to raise a child. This will, therefore, tend to affect the adoption option positively (substitution effect). However, the increase in a woman's income also relaxes the budget constraint (income effect). The income effect on the adoption decision may be either positive or negative. Becker suggests two possible reasons why the income effect may be negative. First, the probability of an unwanted child may fall because the effectiveness of contraceptive methods may increase with income. Second, households may increase the amount spent per child, substituting an increase in the quality of children for the quantity of children. Thus the impact of a change in female income on the adoption decision depends on the relative magnitudes of the substitution and income effects, and the predicted effect is ambiguous. Similarly, the effect of male income on the adoption decision is uncertain.(3) The empirical research on whether the income elasticity of demand for quantity of children is positive or negative is contradictory.(4)

Another determinant of the adoption rate is a woman's marital status. It is hypothesized that single women will be more likely to place a child for adoption because the private and social costs of an additional child are greater for them than for married women. Married women, who are more likely to already have at least one child, benefit from economies of scale and greater experience in childrearing, and have a spouse to help them, and so are likely to have lower outlays for childrearing than single women.(5) The social cost of a child is greater for unmarried women than married women due to the stigma usually associated with unmarried motherhood. The predicted effect of the percentage of women aged fifteen to forty-four who are single is positive.

Also relevant in a woman's adoption decision is her labor force status. Women in the labor force should be more likely to place a child for adoption because of the greater cost incurred as a result of the loss in competitive skills and on-the-job training and experience should they leave the labor force to raise a child. However, a number of researchers have found that working women are much more likely than nonworking women to have planned both the timing of childbirth and the number of children they want, and thus are less likely to have unwanted births.(6) Accordingly, if one accepts both possibilities as potential determinants of the adoption rate, then the predicted effect of the variable the labor force participation rate of women aged fifteen to forty-four is uncertain.(7)

Another determinant of the adoption rate is the unemployment rate of fertile women. Economic theory suggests that as the unemployment rate rises, the value of a woman's time falls, and, concomitantly, so do the costs of rearing an additional child. This implies that the adoption rate will fall. However, this assumes that a woman's labor supply is procyclical. It is possible that a woman's labor supply is countercyclical (the added worker effect) to reduce the variability of household income.(8) If this were the case, then as the unemployment rate rises the value of a woman's time and costs of rearing an additional child would rise and the adoption rate would increase. Thus the expected sign of the unemployment variable is an empirical question.

Michael suggests that education may have two separate and opposite effects on a household's desired fertility objective. On the one hand, education may reduce the likelihood of adoption since the more educated the woman the less likely she will have unwanted births (both in timing and number) because of greater knowledge about and use of effective birth control techniques. It is also more likely the household will be induced to substitute toward quality and away from quantity of children. On the other hand, education, through its effects on an individual's human capital, may also increase the likelihood of adoption since it raises the opportunity cost of a household's time and, since children are time-intensive relative to other goods, increase the relative price of an additional child (Willis |1973~). The education variable is the percentage of women aged fifteen to forty-four in each state who have completed twelve years of school, and its expected effect is indeterminate.

The adoption rate should also be dependent upon the level of financial assistance available from social welfare programs. Ellwood and Bane |1985~ and Garfinkel and McLanahan |1986~ found that welfare programs did not influence the childbearing decisions of women but did have an impact on living arrangements (i.e., female-headed households). Welfare payments may enable a woman in poverty to keep her baby since the level of financial assistance increases for each additional child. Aid to Families with Dependent Children is the guaranteed payment for one child in each state. I hypothesize that it will have a negative influence on the adoption rate.

The last two variables control for demographic differences in tastes for adoption. Religion is a powerful moral and social force in household choices. Research by Goldscheider and Mosher |1991~ found that Christian fundamentalists, as compared to other religious groups, were less likely to use contraception and those using contraceptives were more likely to use a relatively ineffective method. Also, fundamentalism has been found to have a significant effect on sexual morality and abortion.(9) This suggests that fundamentalist women may also be more likely to have more unplanned or unwanted pregnancies. If fundamentalist women are more likely to value large families or less likely to conceive out of wedlock, then they would be more likely to keep the child. Alternatively, fundamentalist women with unwanted pregnancies might choose the adoption option in order to avoid their church's and community's moral sanctions against abortion. The fundamentalism variable is the percentage of each state's population that claims membership in a denomination that professes a belief in the literal interpretation of the Bible.

The other taste variable is the percentage of fertile women in each state who are black. Sociologists and demographers (e.g., Westoff and Ryder |1977~) have suggested that black women practice less family planning and have more unwanted births than white women, which would imply a higher adoption rate for black women. However, black women may be reluctant to put a child up for adoption because the available pool of black adoptive families is smaller than for white families since black women are less likely than white women to have ever married (National Committee for Adoption |1985~).(10)

III. THE EMPIRICAL RESULTS

The adoption equation was first estimated using ordinary least squares. A plot of the residuals against the estimated values of the dependent variable suggested the presence of heteroscedasticity.(11) Using the procedure proposed by Glejser, I determined that the variance of the residuals decreased with the square of the number of black female teenagers in each state.(12) In order to achieve efficient estimates, I reestimated equation (1) using generalized least squares.(13) The generalized least squares estimates of equation (1) appear in Table I, column 1. Female income is positive, but not statistically significantly different from zero. This does not mean that a woman's income is not a determinant of her placing a child for adoption. Rather it suggests that the substitution effect was offset by an equal but opposite pure income effect that resulted in female income being statistically and numerically insignificant.(14) Similarly, male income is negative, but not a significant determinant of the adoption decision.(15)

TABULAR DATA OMITTED

The marital status variable is significantly positive, which is consistent with the hypothesis that unmarried women, due to their higher private and social costs, are more likely to place a child up for adoption. The variable for labor force participation is also significantly negative. This result is consistent with previous research that found that working women are less likely to have unwanted births than nonworking women. The higher the level of Aid to Families with Dependent Children the more likely women in poverty will keep an additional child. The higher the unemployment rate the less likely women are to put a child up for adoption, presumably because women's labor supply is procyclical with a resultant lower opportunity cost of childrearing. Fundamentalist women are found to be significantly more likely to place a child for adoption. This finding is consistent with the contention by sociologists that the social stigma of illegitimate births to fundamentalist women and the religious prohibition of abortion increase the likelihood of a child being given up for adoption. The percentage of black women in the population of fertile women is found not to be a significant determinant of the adoption rate. Black women are not more or less likely to place a child for adoption.

One particularly interesting result was the finding that the percentage of women who have completed high school has a statistically significant positive effect on the adoption rate. There are several possible, not necessarily competing, explanations consistent with this finding. First, high school dropouts may be less likely to offer a child for adoption because of a lower opportunity cost of childrearing. Second, female high school dropouts are more likely to have grown up in disadvantaged socioeconomic circumstances (broken families, on public assistance, or mother was teenager at first birth) and as a consequence may be less likely to consider the benefits of adoption.(16) Third, women with at least a high school degree may be more motivated to avoid unplanned births and childrearing because such an event would interfere with their educational, occupational, and economic goals.(17)

To test for the robustness of the model, I reestimated equation (1) with several other independent variables included. A regional dummy variable equal to one for states in the West was added to equation (1) to determine if there were regional differences in social mores or attitudes regarding adoption. The western states variable was not statistically significantly different from zero. A similar result was obtained for states in the East. The percent of women of childbearing age who are between the ages of 15-19, 15-17, and 18-19 was included separately in equation (1) to test whether any of these age groups were more likely to place a child for adoption. However, each variable was found not to have a significant influence on the adoption rate. In all the estimates, the coefficients of the other variables in the model remained virtually identical to those previously reported.(18)

IV. ADOPTION AND ABORTION

The theoretical model of desired family size suggests that when excess fertility occurs, a child is "unwanted" and a household will attempt to reduce the discrepancy between actual and desired family size at that time. I have focused on the decision by women to reduce the discrepancy by placing the child for adoption. However, relative to adoption, the method most frequently used to reduce unwanted births is abortion.

It is of interest to empirically examine what impact the abortion option has on the adoption decision.(19) In order to test this, equation (1) was reestimated with the dependent variable redefined as the number of unrelated adoptions of healthy infants as a percentage of pregnancies of women of childbearing age in state i in 1982 (ADOPT/|PREG.sub.i~). In addition to the independent variables already specified in equation (1), two other independent variables were included. The price of abortions is the average cost of an abortion in each state in 1982.(20) If abortion is a substitute for adoption, then the predicted effect of the price of an abortion is positive. In addition, in 1982 federal funding of abortions through the Medicaid program was prohibited. However, fourteen states continued to provide unrestricted funding of abortions through state medicaid programs. Such expenditures represent a subsidy to abortion and clearly will have a price effect. The effect is likely to be particularly strong for young, unmarried, low-income women. For this group of women the availability of state-funded abortions may be a more important price determinant than the actual cost of an abortion. Thus state funding of abortions, by reducing the cost of an abortion relative to an adoption, would be expected to have a negative impact on the adoption decision. The medicaid variable is a dummy variable equal to one for those states that continued abortion funding. The empirical results are given in Table I, column 2.

The estimated coefficients of the independent variables previously reported remained virtually unchanged. The price of an abortion is negative, but not statistically significantly different from zero. This result suggests that abortions and adoptions are not substitutes. Even though both methods reduce excess fertility, women with unwanted pregnancies apparently do not perceive adoption and abortion as equivalent options. The coefficient of the state medicaid funding variable is negative and statistically significantly different from zero. This result which in conjunction with the negative coefficient of the Aid to Families with Dependent Children variable suggests that poor women regard adoption as a less desirable alternative than keeping the child or having an abortion.

The finding that the price of an abortion has no statistically significant impact on the adoption option does not necessarily mean that the availability of abortions has no effect on the supply of adoptive children. A pregnant woman has three mutually exclusive options available to her. She can have an abortion, bear the child and keep it, or give it up for adoption. Thus a pregnant woman faces three alternatives and must choose one of them. Such a choice model can be estimated using the multinomial logit model. The advantage of this direct approach is that the choice probabilities are dependent on individual characteristics only.(21) The functional form of the multiple logit model for the equation of particular interest is log(Prob adopt/Prob abort) = f(X) where Prob adopt and Prob abort are the proportion of pregnancies that result in an adoption or abortion, respectively, and X is the vector of all the independent variables used in the previously estimated equation in this section.(22) The estimated parameters show the probability that an individual with a specified set of personal characteristics will choose an adoption relative to an abortion. The estimated coefficients are given in Table I, column 3.

Essentially, what these results show is that a single woman or a woman in the labor force is more likely to chose an abortion relative to adoption, whereas a fundamentalist woman is more likely to choose the adoption option. The variable TABULAR DATA OMITTED of particular interest, the price of an abortion, is negative, but not statistically significantly different from zero. Again the empirical results suggest that abortion availability has had very little effect on the supply of adoptive children. The reason may be that women who obtain legal abortions would undertake other alternatives if abortion was not legal. Such possibilities include becoming a single parent, keeping the child and marrying the father, or obtaining an illegal abortion.

V. THE IMPACT OF STATE REGULATIONS

Adoption in the United States is regulated by the states, subject to state laws and under the jurisdiction of state courts. Of particular interest is the impact various state regulations have on the supply of adoptive children.

There are two types of adoptions: agency and private. All states permit children to be placed for adoption by state social service agencies or state-licensed adoptive agencies (e.g., United Way, Catholic Services). Some states also allow private adoptions which are arranged by prospective adoptive parents through intermediaries such as lawyers or doctors. Many states prohibit private adoptions, arguing they increase the possibility of extortion in unduly pressuring a mother to surrender her child. Proponents argue that private adoptions facilitate the adoption process by providing flexibility in finding adoptive parents quicker and more efficiently in finding adoptive children. To investigate the effect of private adoptions, I added a dummy variable equal to one for states which allow private adoptions to equation (1). The empirical results, which appear in Table II, column 1, show that private adoptions have no statistically significant effect on the supply of adoptive children. States which allow private adoptions do not have more adoptive children.

Adoption procedures emphasize the confidentiality of the adoption parties. There has been a strong tradition of protecting the privacy rights of those involved in adoption. Virtually all states required that adoption records be sealed. During the mid-1970s opposition mounted to sealed adoption records. By 1980, because of pressure from adoptee lobbying groups, a number of states enacted laws allowing adoptees and/or biological parents to register their desire to meet as well as laws that provided for contacting the biological parents in order to obtain consent for a meeting. Of particular interest is whether allowing adoption records to be opened deters placing a child for adoption. To test this, I added a dummy variable equal to one for states which mandate the confidentiality of adoption records to equation (1). The empirical results from Table II, column 2 show that the confidential records variable is not statistically significantly different from zero. Keeping adoption records confidential does not foster the adoption process, or conversely, allowing open adoption records does not inhibit the supply of adoptive children.

Many states allow prospective adoptive parents to pay certain expenses of the birth mother (e.g., medical, legal, and counseling costs) presumably to encourage adoptions. Do such subsidies encourage birth mothers to give up a child for adoption or are they irrelevant to the adoption decision? To examine the effect of expense subsidization on the adoption decision, a dummy variable equal to one for states that permit the biological mother's expenses to be paid by the adoptive parents was added to equation (1). The empirical results, from Table II, column 3, find the expense variable is not statistically significantly different from zero. Subsidization of the biological mother's expenses by adoptive parents does not have any statistically significant impact on the supply of adoptive children.

In order for an adoption to take place, consent of the biological mother is required, however, do allow the birth mother time to revoke her consent before she terminates her parental rights. An interesting issue, particularly to adoptive parents, is whether the length of time allowed for withdrawal of consent has any impact on the birth mother's decision to give up a child for adoption. Are biological mothers more likely to change their mind the longer the time period allowed, or does the decision tend to be irrevocable? This question was examined by adding the length of time a state allows for withdrawal of consent to equation (1). Table II, column 4 shows that the time variable was negative, but not statistically significantly different from zero. A birth mother's decision to give up a child tends to be invariant to the time period allowed for withdrawal of consent.(23)

I. CONCLUSION

Opponents of abortion have often argued that adoption is a viable alternative if the U.S. Supreme Court were to prohibit legal abortions. All abortions, however, would not be eliminated since there would still be the possibility of obtaining an illegal abortion. Based on the approximately 1,500,000 abortions done annually in the United States, Medoff |1988~ estimated that if abortions were prohibited there would still be between 600,000 and 900,000 illegal abortions performed annually.

This study estimated the supply of adoptions. The results found that the adoption rate was negatively related to a woman's participation in the labor force, the size of Aid to Families with Dependent Children payments, and the unemployment rate. Single mothers, fundamentalist women, and a high school education were found to have a positive impact on the adoption rate. The empirical results suggest that societal changes and trends in women's attitudes, economic status, and feminist views have caused adoption to be considered a less desirable option than abortion or childrearing. Even if abortions were made illegal, only a small number of women, those who could not obtain an illegal abortion or who decided not to become single mothers, would surrender their child for adoption. Furthermore, state regulations, which presumably are designed to encourage the adoption option, are found not to have any impact on the decision to relinquish a child.

MARSHALL H. MEDOFF Department of Economics, California State University, Long Beach. Funding for this research was provided by California State University, Long Beach through the Scholarly and Creative Activity Program.

1. Unlike a conventional supply curve which shows at each price how many units of output suppliers would be willing to offer for sale, the adoption supply equation does not have a price variable which would induce women to offer a child for sale since every state imposes criminal penalties for child selling and/or child buying.

2. It might be argued that the independent variables should be restricted to mothers of newborn infants rather than all fertile women. There is no bias in the estimated coefficients since the factors affecting a household's desired family size are the same for both groups of women.

3. In 1982, approximately 82 percent of births were to married women.

4. For two viewpoints see Dooley |1982~ and Cain and Weininger |1973~.

5. In 1980, 50.5 percent of all married women (spouse present) had at least one child, versus 14.2 percent for unmarried women (U.S. Bureau of the Census |1983~).

6. On this point see Westoff and Ryder |1972~ and Ward and Butz |1980~.

7. It might be argued that income and labor force participation are endogenous to fertility. However adoption is an ex post event--the decision by a women to give up her child. Thus all the variables are exogenous at the time of the adoption decision.

8. In 1982, two out of three unemployed persons lived in a household with another working member. I am indebted to an anonymous referee for suggesting this possible relationship.

9. See Jones |1983~ and Medoff |1989~.

10. The data on all economic variables were obtained from the U.S. Bureau of the Census, State Reports, Detailed Characteristics |1983~. The data on adoptions and adoption law were obtained from the National Committee For Adoption, Adoption Factbook |1985~ and Sloan |1988~. The data on religious affiliation was from the National Council of Churches survey, Churches and Church Membership in the United States: 1980 (Quinn |1982~). The mean and standard deviation (in parentheses) for the dependent variable and independent variables in equation (1) are |ADOPT.sup.s~: .61(.40); Female Income: 6407.3(936.34); Male Income: 14612.4(1767.7); Single: 34.16 (3.96); Labor Force Participation: 61.99(4.61); Unemployment: 7.63(1.93); Education: 71.88(4.39); Aid Families Dependent Children: 255.42(104.61); Fundamentalism: 13.72(12.8); Black: 9.47(9.79).

11. A Goldfeld-Quandt procedure on equation (1) showed that the calculated F-value exceeded the critical F-value at the .05 level of significance and thus the null hypothesis of homoscedasticity was rejected.

12. Sixteen states had a total population which was less than 3 percent black.

13. A complete description of the Glejser procedure and generalized least squares estimation is detailed in Maddala |1984~.

14. Mincer |1963~ suggests that since a husband and wife's earnings tend to be positively correlated, failure to account for this influence biases downward estimates based on cross-sectional data of the effect of income on fertility.

15. When the income variables were omitted from equation (1) the empirical results were (absolute value of t-statistics in parentheses):

|ADOPT.sup.s~ = -5.3835 + .0346 Single (2.08) (1.97)

-.0018 Labor Force Participation (1.61)

-.0524 Unemployment + .0771 Education (2.43) (2.08)

-.0018 Aid Families Dependent Children (3.53)

+ .0118 Fundamentalism + .0016 Black (1.99) (.21)

The income variables were included in the estimation because of its theoretical implications.

16. See Kalmuss, Namerow, and Cushman |1991~ for a discussion of this issue.

17. Suggested by McLaughlin, Manninen, and Winges |1988~.

18. The complete empirical results are available upon request.

19. In 1989, the U.S. Supreme Court ruled that states could impose some restrictions on abortions performed in the first twenty-four weeks. Up until that time the Supreme Court had required all states to permit abortion on demand through the second trimester (less than one percent of all abortions occur in the third trimester). State laws mandating a waiting period, spousal involvement, parental consent, or information about other alternatives were not in effect during the 1982 period under study either because such laws were not enforced or enjoined because of court challenges or not yet enacted. Thus there did not exist any state variance in abortion regulation in 1982.

20. All the information pertaining to abortions was obtained from the Alan Guttmacher Institute which is the research institution affiliated with Planned Parenthood. The information provided by the Guttmacher Institute is from a yearly national survey, and its results are acknowledged by the U.S. Department of Commerce in the Statistical Abstract of the United States to be the most complete available.

21. See Judge et al. |1985~ chapter for a description of the multinomial logit model.

22. A complete description of the multinomial logit model is available in Judge et al. |1985~. The other estimated equation was log(Prob childrear/Prob abort) = f(X). The complete empirical results are available upon request.

23. The state regulation variables were also statistically insignificant when any two, three or all of them were added to equation (1). The state regulation variables were also found to be statistically insignificant when the dependent variable was ADOPT/PREG. The complete empirical results are available upon request.

REFERENCES

Ammerman, Nancy T. Bible Believers. New Brunswick, N.J.: Rutgers University Press, 1987.

Becker, Gary S. "A Theory of Marriage," in Economics of the Family, edited by T. W. Schultz. Chicago: University of Chicago Press, 1974, 299-344.

-------. "An Economic Analysis of Fertility," in Demographic and Economic Change in Developed Countries. Princeton, N.J.: Princeton University Press, 1960, 209-31.

Cain, Glen, and Adriana Weininger. "Economic Determinants of Fertility: Results Using Cross-Sectional Aggregate Data." Demography, May 1973, 205-33.

DeTray, Dennis N. "Child Quality and the Demand for Children." Journal of Political Economy, March/April 1973, S70-95.

Dooley, Martin D. "Labor Supply and Fertility of Married Women: An Analysis with Grouped and Individual Data from the 1970 U.S. Census." Journal of Human Resources, Fall 1982, 499-532.

Ellwood, David T., and Mary Jo Bane. "The Impact of AFDC on Family Structure and Living Arrangements," in Research In Labor Economics, edited by R. G. Ehrenberg. Greenwich, CT: JAI Press, 1985, 137-207.

Garfinkel, Irwin, and Sara S. McLanahan. Single Mothers and Their Children. Washington, D.C.: Urban Institute Press, 1986.

Goldscheider, Calvin, and William D. Mosher. "Patterns of Contraceptive Use in the United States: The Importance of Religious Factors." Studies in Family Planning, March/April 1991, 102-15.

Jones, Ethel B. "ERA Voting: Labor Force Attachment, Marriage and Religion." Journal of Legal Studies, January 1983, 157-68.

Judge, George C., W. E. Griffiths, R. C. Hill, Helmut Lutkepohl, and T-C Lee. The Theory and Practice of Econometrics, 2nd edition. New York: John Wiley, 1985.

Kalmuss, Debra, Pearila B. Namerow, and Linda F. Cushman. "Adoption Versus Parenting Among Young Pregnant Women." Family Planning Perspectives, January/February 1991, 17-23.

Leibowitz, Arleen, Marvin Eisen, and Winston K. Chow. "An Economic Model of Teenage Pregnancy Decision-Making." Demography, February 1986, 67-77.

Maddala, G.S. Econometrics. New York: McGraw-Hill, 1984.

McLaughlin, Steven D., Debra L. Manninen, and Linda D. Winges. "Do Adolescents Who Relinquish Their Children Fare Better or Worse Than Those Who Raise Them." Family Planning Perspectives, January/February 1988, 25-32.

Medoff, Marshall H. "An Economic Analysis of the Demand for Abortions." Economic Inquiry, April 1988, 353-59.

-------. "Constituencies, Ideology, and the Demand for Abortion Legislation." Public Choice, February 1989, 185-91.

Michael, Robert T. "Education and the Derived Demand for Children." Journal of Political Economy, March/April 1973, S128-64.

Mincer, Jacob. "Market Prices, Opportunity Costs, and Income Effects," in Measurement in Economics: Studies in Mathematical Economics and Econometrics in Memory of Yehuda Grunfeld. Stanford: Stanford University Press, 1963, 67-82.

National Committee For Adoption. Adoption Factbook. Washington, D.C., 1985.

Quinn, Bernard. Churches and Church Membership in the United States: 1980. Washington, D.C.: Glenmary Research Center, 1982.

Ryder, Norman B., and Charles F. Westoff. Reproduction in the United States, 1965. Princeton, N.J.: Princeton University Press, 1971.

Sloan, Irving J. The Law of Adoption and Surrogate Parenting. New York: Oceana Publications, 1988.

U.S. Bureau of the Census. State Reports, Detailed Characteristics: 1980. Washington, D.C.: Government Printing Office, 1983.

Ward, Michael P., and William P. Butz. "Completed Fertility and its Timing." Journal of Political Economy, October 1980, 917-40.

Westoff, Charles F., and Norman B. Ryder. The Contraceptive Revolution. Princeton, N.J.: Princeton University Press, 1977.

Willis, Robert J. "A New Approach to the Economic Theory of Fertility Behavior." Journal of Political Economy, March/April 1973, S14-64.