Introduction

The safe and permanent placement of the children committed to its care is one of the chief objectives of the substitute child care system. Ideally, institutional child care facilities (orphanages) and foster homes provide temporary living arrangements for parentless children as a prelude to placing them into stable, traditional family units. However, the adoptability of children varies considerably with factors like age, race, and physical and behavioral problems. As a result, children can remain within the substitute child-care system for very long periods. Some are never adopted. The longer a child is left in the system, the more likely it is that the system itself will influence his or her adoption prospects.

The substitute child-care system in the United States has undergone revolutionary changes during the 20th century. The most obvious change, of course, has been the decline of the traditional American orphanage and its subsequent replacement by a network of foster families. In 1933, for instance, 144,000 children were cared for in orphanages in the United States. The number of children in institutional care declined steadily throughout the following decades. By 1977, only 43,000 children were living in orphanages. And by 1980, the orphanage had for all practical purposes ceased to exist.

Less obvious, but no less important, has been the influence of major changes in the federal welfare system instituted during the 1960s. Specifically, the expansion of a program of government transfer payments to single parents, Aid to Families with Dependent Children (AFDC), has essentially meant that children are no longer removed from the biological mother's care simply because of poverty due to the breadwinner's absence. Abuse, neglect, alcoholism, or other dysfunctional conditions that characterize an unacceptable environment for raising a child are now necessary before the state will intervene. As a result, fewer children enter the substitute child-care system these days and proportionately more of them who do come from backgrounds that make successful adoption less likely.

This essay explores the determinants of adoption rates in the United States. Of specific interest is the effect of moving from an orphanage-based child care system to a foster family-based system. The linkage between adoptions and the institutional characteristics of the substitute child care system has perhaps seemed irrelevant because many of the children who now enter the system come from backgrounds that promote behavioral problems inimical to successful adoption. Just the opposite is true, however: It is precisely because most parentless children are not adopted quickly that there is a pressing need to understand how the characteristics of the substitute care system impact the adoptability of its wards.

In what follows, we consider whether successful adoption is better facilitated by an institutional-based child care system or a foster family care-based system. This is accomplished by estimating a pooled cross-section time-series adoption equation in reduced form. Because of data limitations, the analysis focuses on the years 1955 through 1959, when both orphanages and foster homes were important components of the substitute child-care system in the United States. Observations are collected across states. The model's dependent variable is an adoption rate measured relative to the number of displaced children in a state (unrelated adoptions divided by the number of children in group homes and foster care). Key explanatory variables are the per capita numbers of children in orphanages and in foster care. Other explanatory variables include government welfare payments and state demographic characteristics.

Welfare, birth control, abortion, and a desire on the part of social workers to "deinstitutionalize" parentless children have all played roles in the disappearance of the traditional American orphanage. At the same time, however, these forces may have created a permanent underclass of unadoptable boys and girls, forced to spend their childhood years being shifted from one foster-care home to another.

The next section provides some historical background on the causes and consequences of the decline of the American orphanage. Section 3 contains a description of the data. Our empirical model is specified in Section 4 and our empirical results are reported in Section 5. The last section offers some concluding remarks.

Where Have All the Orphans Gone?

There are no longer any orphans born in the United States. Indeed, the only place the word "orphan" appears in the entire Statistical Abstract of the United States nowadays is as an entry in a table headed "Immigrants Admitted, by Class of Admission", where orphaned children of U.S. citizens are identified as comprising an immigration category exempt from numerical limitations. Slightly more than 9,000 such children were admitted to the United States in 1991 (U. S. Department of Commerce 1993, p. 10). Apparently, America must import its orphans.²

This state of affairs contrasts sharply with circumstances only a few decades ago: There were about 144,000 children in orphanages nationally during the depths of the Great Depression; beginning with the outbreak of World War II, the number fell steadily "to 95,000 in 1951 to 77,000 in 1963 to 43,000 in 1977". Orphanage care had essentially disappeared by 1980 (Jones 1993, 459).

Three reasons are commonly given to explain the postwar decline of the American orphanage (Jones 1993). One factor relates to an anti-institutional bias that gradually developed among social workers and public welfare agencies. As early as the first White House conference on children in 1909, the conclusion was reached "that, wherever possible, children should be placed in foster families rather than institutions." The belief that "home life is the highest and finest product of civilization", and that children should not be deprived of it "except for urgent and compelling reasons" reinforced, wittingly or not, the growing movement toward professionalization of social work.

The origins of this second key factor in the decline of the American orphanage can be traced to the mid 1950s, when the Child Welfare League of America (CWLA), the leading national organization for social workers, issued a series of reports stressing the importance of professionalization and tightening the standards orphanages were required to meet in order to qualify for CWLA accreditation. These accreditation standards, which called for

more staff in proportion to children; more professional social workers, that is, with M.S.W. degrees; more social workers with specialty training (usually psychiatric social work); professional social workers as administrators; higher salaries; secretarial help for caseworkers; psychiatric and other consultants; policy manuals; up-to-date personnel policies (provision for sick leave and vacations); and regular case conferences (Jones 1993, 468),

raised the average cost of care considerably and, because the rate of public compensation for children placed in their care did not keep pace, exerted increasingly heavy financial pressure on traditional orphanages. The cost spiral was accelerated by the growing preference for foster-care placements, which reduced institutional populations and raised fixed costs per child further.³ By the late 1960s, "a representative board rate for foster families . . . would have been $120-$150 per month, less than half of the . . . $310 it would have cost to maintain that same child in an institution" (Jones 1993, pp. 472-73).

Aid to Families with Dependent Children (AFDC), which began following passage of the Social Security Act of 1935, was the third factor leading to the decline of the American orphanage. The availability of AFDC payments meant that single mothers who might otherwise abandon their children were more likely to continue to care for them at home. That this welfare program had a major impact on the population potentially served by traditional orphanages is evidenced by the fact that from 1950 on, the number of children receiving AFDC payments increased at an average rate of 100,000 per year (Jones 1993, 464).

The Adoption Assistance and Child Welfare Act of 1980 spelled the end of the traditional orphanage "not because it had anything in particular to say about institutional care, but because it did not." To achieve the twin goals of reducing the number of children placed in foster care, which had ballooned from 165,000 to 394,000 between 1961 and 1977, and reducing "the replacement rate to unity, that is, to the point that a child who had to be placed in a foster family either remained with that family for a short while before returning to his or her parents or remained there permanently", the act mandated that children be placed in "the least restrictive (most family-like) setting available" (Jones 1993, 476).

The 1980 law attempted to deal with a crisis in substitute child care which had been growing since the 1950s and which in many respects still plagues the system. Because the children entering foster care are disproportionately characterized by behavioral problems associated with abuse and neglect,⁴ and because the number of foster children needing placement greatly exceeds the number of families willing and able to care for them, most children in foster care are shifted repeatedly from one foster home to another. This "replacement rate" rises progressively with the age of the child, adding yet another source of insecurity to crucial formative years.

An early study by Maas and Engler (1959) found that three-fourths of the children in foster care never return home. A few years later, Jeter (1963) reported evidence that return home was expected for only about 12 percent of the 62,000 foster children she studied, while adoptive placement accounted for another 13 percent. The typical foster child in both studies "had been placed at least twice, and as many as 16 percent had experienced four or more placements. One child in the Jeter study had been placed 22 times." When Walter Ambinder (1965) later documented the experiences of 410 boys in foster care homes, he found that "the average number of placements increased regularly from 2.6 at age 8 to 5.7 at age 15" (Jones 1993, p. 475).⁵

Coupled with the fact that by 1980 "most foster families were caring for two or more unrelated foster children" (Jones 1993, 476), the emotional toll levied by such instability is surely quite heavy.⁶ Some evidence of this is reflected in available adoption statistics.⁷ For example, whereas there were in excess of 12,000 adoptions annually in California during the late 1960s, less than half that number of children were being adopted each year by the middle 1970s. Although adoptions of all types (except intercountry adoptions) trend sharply downward over this period, the decline in the number of children adopted under the auspices of public agencies, which we have seen tend to rely more heavily on foster care than institutional care (Wolins and Piliavin 1964, 44-46), is particularly noteworthy (Wingard 1987, 306). These changes, which are reasonably reflective of national trends, coincide with a change in public policy toward abortion (following passage of the California Therapeutic Abortion Act in 1967, for instance, the number of legal abortions there increased from 5,031 in 1968 to 116,749 in 1971), and with increases both in the number of births to unmarried women and the number of single mothers of all ages choosing to keep their children (Wingard 1987, 310).

Adoption has not been a placement priority for social workers for some time. Of the nearly 90 percent of the children in substitute care for which written "permanency plans" existed in 1983, adoption was considered to be the best option for less than 15 percent of them. The plans for placing children permanently reported by the respondents to a 1985 study by the American Public Welfare Association were, in descending order of frequency, "return to parents, relatives, or other caretakers (46.6 percent), long-term foster care (18.4 percent), adoption (14.7 percent), and independent living or emancipation (10.2 percent)" (Stein 1987, 642).⁸ Precedence is given to the goal of returning children to their parents' or relatives' homes despite the fact that "the reintegration of children to their biological homes is a difficult process, and reentry to [substitute] care occurs at a fairly high rate -- 2 percent reentry for children returned to biological families compared to 2 percent for children adopted" (Stein 1987, 648).

While sociological factors are undoubtedly important, our task to isolate the impact of the decline of the American orphanage on adoptions from the background noise generated by changes in welfare policy, the legalization of abortion, the introduction of effective contraceptive techniques, and other such influences. We conjecture that, other things being the same, the substitution of foster family care for institutional care plays a significant role in explaining the decrease in adoptions (especially those not associated with the remarriage of a biological parent) by lowering the quality of potentially adoptable children and increasing the cost of adoptions. The reduction in quality is due to the aforementioned forces of self-selection which bias the characteristics of the children entering the substitute child care system away from social norms. This bias is particularly remarkable in adoptions taking place under the auspices of public agencies whose "children are older at the time of placement (median age 48 months vs. less than three months), and have more health problems (34% vs. 11% of private agency adoptions)" (Wingard 1987, 307). The increase in cost is due to the dispersion of potentially adoptable children over a relatively large number of foster families each of whom cares for only two or three children,⁹ rather than having them housed in a smaller number of larger institutions where the cost of searching for an adoptable child with desired characteristics is arguably lower.

Previous research on the determinants of adoption has focused on the supply side. Medoff (1993) estimates a regression model designed to explain cross-state variations in "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)" during 1982. He finds that the decision to place a child for adoption is negatively related to the female labor force participation rate, the amount of a state's AFDC payment, and the state unemployment rate. These results are consistent with an economic interpretation of the adoption decision in which working women are less likely to have unwanted births than nonworking women and in which lower opportunity costs of child rearing (higher unemployment rates) and larger welfare payments induce more women to keep their children. Medoff also finds that the decision to place a child for adoption is positively related to a woman's marital status, her education, and her religious affiliation. Other things being the same, single women, women with higher levels of educational attainment, and fundamentalist women are significantly more likely to place their children for adoption. Medoff's evidence suggests that the prices and availabilities of abortions do not have statistically significant effects on adoptions.¹⁰ Moreover, the adoption rate is not influenced by the existence of state regulations designed to encourage them.

In the remainder of this essay, we report evidence suggesting that the decline of the orphanage and the growth of foster care have adversely affected the number of children who get adopted in the United States. That is, in addition to the economic and sociological factors identified in prior research as influencing the decision to place a child for adoption, we find that institutions matter: Adoption rates are lower in those states where the shift from orphanages to foster care took place earlier. Far from encouraging the permanent placement of parentless or abandoned children in stable family-like environments, the foster care system seems to have accomplished just the reverse.

The Data

To explore the relationship between adoption rates and the institutional features of the substitute child care system empirically, observations are needed on three key variables: adoptions, children in orphanages, and children in foster care. Unfortunately, constraints on the availability of data for each of these variables severely limit the time period that can be studied.

Surprisingly, data on adoptions in the United States are virtually nonexistent prior to the early 1950s. Then, as now, there was no uniform data collection system in place for keeping track of the number of adoptions each year. The first generally accessible statistics on adoptions were collated from states voluntarily reporting to the Children's Bureau of the Social Security Administration. Slightly more than half the states supplied data of sufficient accuracy and comprehensiveness to be deemed useful. Thus, adoption data in this country become available for the first time in the mid 1950s -- and then for only about half the states.

Moreover, any analysis of the substitute child-care system must take into account the evolving nature of these institutions over time. The children placed in "group care" facilities nowadays almost all tend to be either emotionally disturbed, delinquent, or physically challenged. The functions of the traditional orphanage have for all practical purposes been taken over by foster family care.

Hence, we need to go back far enough in time to identify a period in which children who are simply dependent on substitute care (rather than being emotionally, behaviorally, or physically challenged) could be reared either in an institution or in a foster family home. Evidence that such an era existed at least up through the early 1960s is provided in Table 1.

As reported in Table 1, the shift of the U.S. substitute child care system from one in which the majority of children were placed in orphanages to one dominated by foster family care occurred over a time span of some 50 years. In 1910, 65 percent of parentless children were cared for in group homes. This figure dropped modestly over the next 23 years, falling to 58 percent by 1933. Taken together, the data in Table 1 tell a story of a generally steady decline in the percentage of children living in group-care homes.

Yet, importantly for our purposes, the orphanage system was still a major institution for caring for parentless children through the 1950s and early 1960s. Thirty-four percent of the children needing substitute care were still being reared in orphanages in 1960 (again, these are not special-needs or delinquent children). Moreover, the data in Table 1 are based on all children in the U.S. substitute care system and therefore do not fully capture the important role still being played by traditional orphanages in many states as late as the early 1960s. In 1960, for example, 79 percent of Oklahoma's parentless children were living in group homes; the corresponding figures were 66 percent in Tennessee and 48 percent in Illinois (Wolins and Piliavan 1964, 39).

* Group home data exclude children who were in temporary detention facilities, delinquents in correctional institutions, and children in institutions for the mentally or physically handicapped.

Source: Wolins and Piliavin (1964).

The foregoing discussion suggests that only a relatively narrow time frame exists for which adequate data on adoptions are available and during which orphanages were still performing their traditional function of caring for children without special needs. Because of limited adoption data, this period begins in the 1950s; because of the changing nature of group homes, the period ends somewhere in the mid- to late 1960s. Based on these considerations, this study uses observations across a sample of states for the years 1955 through 1959.¹¹

Over this period, observations on all of the variables included in the adoption model described in the next section are available for 24 states plus the District of Columbia. A list of these states along with the corresponding number of unrelated adoptions per capita in 1955 is provided in Table 2. With 45 unrelated adoptions for every 100,000 persons in the state, Oregon and Vermont are the most adoption-intensive states on this list. The fewest number of adoptions per capita occurred in Rhode Island, which had 15 unrelated adoptions per 100,000 persons.

State	Adoption Rate
Arkansas	17.867
Connecticut	28.610
Delaware	25.000
District of Columbia	32.690
Florida	38.227
Georgia	21.076
Indiana	33.364
Iowa	30.236
Maine	43.961
Minnesota	26.945
Missouri	20.438
Nevada	35.802
New Hampshire	33.454
New Mexico	42.139
North Dakota	18.936
Oregon	45.159
Rhode Island	15.236
South Dakota	27.647
Tennessee	19.606
Texas	21.600
Utah	34.375
Vermont	45.082
Virginia	28.319
Washington	44.270
Wisconsin	23.955

The Empirical Model

Adoption rates across states vary according to variations in factors that affect the supply of children who are adoptable and the demand for these children. Below, these factors are included in a single-equation linear regression model of the determinants of adoption rates. A simultaneous-equations model specifying separate demand and supply equations is not feasible because there is no empirically practical proxy for the "price" of an adoption. Because of legal and ethical restrictions in the market for adoptions, and because many of the costs of adoption are borne in-kind -- arrangements made for the birth mother's medical expenses, resources expended by adoptive parents to establish their financial and emotional qualifications, and so on -- there is no observable price for adopted children.

The absence of a market-clearing price means that excess demand or excess supply will tend to characterize the market for adoptable children. Anecdotal evidence suggests that while the market for white infants is usually characterized by excess demand, other potentially adoptable children are in excess supply. Because some adoption markets are characterized by excess demand and others by excess supply, we generally avoid referring to the model developed here in terms of either of these concepts.

This study focuses on the influence of group care versus foster family care on adoption rates. As such, we limit our analysis to what are called "unrelated adoptions". A related adoption occurs when a child is adopted by a blood relative other than a biological parent or, more commonly, when a biological parent remarries and the child is adopted by the new spouse. All other cases are considered to be unrelated adoptions.

Our empirical model of adoption rates across states is specified as follows:

ADOPT_it = f (SUBSTITUTE_it, BIRTH_it, DIVORCE_it, MARRIAGE_it, AGENCY_it, OUTWED_it, INCOME_it, UNEMP_it, AFDC_it, WHITE_it, EDUC_it, URBAN_it, AGE_it).

The i subscript indexes states (i = 1, . . ., 25) and t denotes the year of observation (t = 1955, 1957, 1958, 1959). In what follows, the definition of each of these variables is given and the possible a priori relationships between the dependent variable and the independent variables are discussed.

ADOPT. The dependent variable is defined as the number of unrelated adoptions per displaced child (the sum of children in foster care and group homes) for each of the 25 jurisdictions in the sample during each of the four years, 1955, 1957, 1958, and 1959.

SUBSTITUTE. This variable represents three different measures of the number of children who require care that substitutes for the care normally provided by a biological parent or parents. There are two principal kinds of substitute care available in the time period covered by this study-orphanage care and foster family care. FOSTER is defined as the number of children in foster care per capita, ORPHAN is defined as the number of children in orphanages per capita, and ORPHAN/FOSTER is defined as ratio of the number of children in orphanages divided by the number of children in foster homes.

Two different specifications of the adoption equation are estimated. The first specification includes the variables FOSTER and ORPHAN. A larger number of children in foster care and orphanages implies a greater supply of adoptable children. Although for supply-side reasons the variables FOSTER and ORPHAN are both expected to have positive effects on adoption rates, of key importance is the relative size of these effects. If the estimated coefficient on FOSTER is larger than the coefficient on ORPHAN, then it can be argued that foster care promotes the adoptability of displaced children better than a group home-based system. On the other hand, if the coefficient of ORPHAN is greater than the coefficient of FOSTER then the opposite interpretation is indicated-adoptions are better facilitated by an orphanage-based system than by a foster family-based system.

As an alternative to entering each of the two categories of displaced children separately, we have created a third variable defined as the ratio of these variables. The variable ORPHAN/FOSTER is the number of children in orphanages divided by the number of children in the foster care system. If an orphanage-based system of caring for displaced children is more amenable to adoptions then we expect this variable to be positively related to the dependent variable. On the other hand, a negative sign would indicate that foster care is more adoption-friendly than group homes.

BIRTH, DIVORCE, MARRIAGE. These variables are defined, respectively, as birth rates, divorce rates, and marriage rates across states. If these life events primarily affect the supply of adoptable children, then BIRTH and DIVORCE should be positively related to the dependent variable, while MARRIAGE is expected to be negatively related to ADOPT.

AGENCY. This variable controls for variations in the mix of institutions through which adoptions are arranged. In particular, AGENCY is the proportion of adoptions mediated by (public and private) child welfare agencies.¹² If agency adoptions are more restrictive than independent adoptions, or if the children served by agencies are more difficult to place, then it is expected that this variable will be negatively related to adoption rates across states.

OUTWED. OUTWED is the proportion of children born out of wedlock in a state. To the extent that out-of-wedlock births are an important source of supply of adoptable babies, it is expected to have a positive relationship with the dependent variable. The estimated coefficient on OUTWED will be negative if illegitimate children are more difficult to place adoptively.¹³

INCOME, UNEMP, AFDC. These three variables control for the effects of general economic conditions on the market for adoptions. INCOME is state income per capita, UNEMP is the number of unemployment insurance beneficiaries per capita, and AFDC is the amount of state aid received by families with dependent children, per recipient. If these economic variables primarily affect the supply of adoptable children, then INCOME and AFDC would be positively associated with adoptions across states, while UNEMP would be negatively associated with adoptions across states. The opposite relationships are expected if these variables are proxies for birth mothers' opportunity costs of rearing their children at home. In this case we expect adoption rates to be lower in states with lower per capita incomes, higher AFDC payments per recipient, and more unemployment insurance beneficiaries per capita.

WHITE, EDUC, URBAN, AGE. Adoption rates are linked to any number of general characteristics of the populace, including race and education. WHITE is the proportion of a state's population that is white, EDUC is the number of high school graduates per capita, URBAN is the fraction of a state's population that lives in an urban area, and AGE is the proportion of a state's population that is between the ages of 15 and 44.

From the time the first statistics on adoption were collected in the 1950s, it has been clear that adoptions of white children overwhelmingly dominate nonwhite adoptions. In 1951, for instance, only 6 percent of total adoptions involved nonwhite children. The variable WHITE is expected to be positively related to state adoption rates for this reason. The effect of education on adoptions is somewhat ambiguous. If education is associated with more "liberal" attitudes toward adoption, then it could be positively associated with adoptions across states. It could be negatively related if more education leads to fewer unwanted births or if more highly educated people are more willing or better able to care for their own children.

Adoptions may be positively related to urbanization for two reasons. First, there may be scale economies in the sense that larger populations are needed to support the institutions (lawyers, child welfare agencies, and so on) that facilitate the adoption process. Second, because the market is thicker, unwed mothers who wish to place their children for adoption may tend to temporarily relocate to large cities for this purpose.

AGE, the proportion of a state's population between the ages of 15 and 44, may be positively or negatively related to cross-sectional adoption rates. It will be positively related if this variable reflects the effect of having relatively more women of childbearing age supplying adoptable children; it will also have a positive sign if the variable reflects the effect of having relatively more couples in the age range that tend to demand adoptable children. On the other hand, as the proportion of the population between 15 and 44 increases, the proportions of the population younger than 15 or older than 44 must necessarily get smaller. If the size of the 15-to-44 age range is inversely related to the size of the less-than-15 age range, then AGE may be inversely related to ADOPT insofar as there will tend to be fewer adoptable children.

Recall that our data set includes observations collected from 24 states and the District of Columbia for the years 1955, 1957, 1958, and 1959 (to reiterate, adoption data are not available for 1956). The means of the dependent and independent variables comprising the data set are reported in Table 3.

Variable	Definition	Mean
ADOPT	Adoptions per displaced child	.432
ORPHAN	Children in orphanages per capita	.267
FOSTER	Children in foster care per capita	.980
ORPHAN/FOSTER	Ratio of children in orphanages to those in foster care	.289
BIRTH	Birth rate	25.39
DIVORCE	Divorce rate	3.494
MARRIAGE	Marriage rate	17.22
AGENCY	Fraction of adoptions arranged by public and private adoption agencies	.622
OUTWED	Fraction of children born out of wedlock	.423
INCOME	State income per capita	$1,875.00
UNEMP	Unemployment insurance beneficiaries per capita	.032
AFDC	Monthly AFDC payment	$29.23
WHITE	Fraction of the population that is white	.89
EDUC	Fraction of the population with a high school education	8.06
URBAN	Fraction of the population living in urban areas	61.76
AGE	Fraction of the population between the ages of 15 and 44	40.64

* Population is measured in thousands. All per capita variables should therefore be interpreted as "per thousand population."

Source: Statistical Abstract of the United States (various issues).

Empirical Results

The empirical results are presented in Table 4. Of particular interest are the estimated coefficients on the first three variables listed there-ORPHAN, FOSTER, and ORPHAN/ FOSTER. As predicted, ORPHAN is positively and significantly related to adoptions per displaced child across states. Somewhat surprisingly, FOSTER is a negative and significant determinant of adoptions rates. Our expectation was that both ORPHAN and FOSTER would be positive determinants of adoption rates and that a comparison of their estimated coefficients would indicate which system provided a more favorable setting for adoption. The results are much stronger than this. Adoption rates are positively related to orphans per capita and negatively related to foster children per capita. (The difference in the coefficients is statistically significant on the basis of the relevant Chow test.) These results suggest that traditional orphanages provided an atmosphere that instilled characteristics in children that made them more adoptable than the atmosphere provided by foster family care. Alternatively, orphanages may better facilitate adoptions by lowering the costs of search for prospective adoptive parents.

The remainder of the results in Table 4 indicate that only five of the remaining twelve explanatory variables are statistically significant in our sample. The percentage of the population that is white is positive and significant in both equations; the proportion of the population that lives in urban areas is positive and significant in one of them. Three explanatory variables have negative and significant coefficients in both regressions-the index of children born out of wedlock, the state unemployment rate, and the generosity of a state's AFDC benefits. The first of these results suggests that illegitimate children are more difficult to place adoptively; the latter two indicate that adoption rates are lower in states where the opportunity costs of rearing children at home are lower. Overall, the regressions explain about 60 percent of the cross-state variation in adoptions per displaced child.

Explanatory Variable	(1)	(2)
Intercept	-.533 (.520)	-.934 (.938)
ORPHAN	.707 (2.477)‡	.
FOSTER	-.525 (4.206)†	.
ORPHAN/FOSTER	.	.980 (4.416)†
BIRTH	.011 (.577)	.004 (.216)
DIVORCE	.033 (.709)	.050 (1.102)
MARRIAGE	-.008 (1.084)	-.009 (1.247)
AGENCY	-.109 (.338)	-.084 (.263)
OUTWED	-1.870 (3.627)†	-2.138 (4.505)†
INCOME	.0003 (1.808)	.0001 (.738)
UNEMP	-11.531 (3.081)†	-13.200 (3.751)†
AFDC	-.020 (2.474)‡	-.027 (3.459)†
WHITE	2.534 (3.753)†	2.348 (4.209)†
EDUC	-.046 (.988)	.013 (.308)
URBAN	.003 (.647)	.014 (3.157)†
AGE	.196 (.102)	-.593 (.320)
R2	.593	.603

*Absolute values of t-statistics in parentheses.

†Significant at the 1 percent level.

‡Significant at the 5 percent level.

Concluding Remarks

Orphanages had disappeared from the American scene by 1980 in large part due to social workers' expressed preference for placing children needing substitute care in foster families rather than in large, impersonal institutions. Dickensian scenes of drab regimentation could not possibly compete with pictures of the bucolic freedom and healthy lifestyles supposedly experienced by the streetwise New York City kids Charles Loring Brace sent to work on Midwestern farms in return for room, board, schooling, and the loving attention of caring foster mothers and fathers. But professional preferences aside, there has been little or no systematic empirical work hitherto of the relative merits of institution-based versus foster-family-based substitute care. Are the children placed in foster care nowadays in fact better off than those formerly placed in old-fashioned orphanages?

At least in terms of their potential adoptability, the answer given in this essay is no. Based on data from the late 1950s, we find evidence that, other things being the same, children entering the substitute care system in states which relied more heavily on foster family placements than institutional placements were less likely to be adopted by someone other than a blood relative. The lower rate of unrelated adoptions amongst children placed in foster care can be explained either by the heavy emotional toll levied by the system's relatively high replacement rate, which causes foster children to experience rejection and to develop behavioral problems that reduce their adoption prospects, or by the relatively high costs borne by prospective adoptive parents in searching over more than 100,000 foster families for children with desired characteristics.

In any case, the empirical results presented here call for rethinking the substitute care paradigm universally accepted by the social work profession. If the goal is to place parentless children as quickly as possible into stable, traditional family environments, then orphanages rather than foster families seem to be the institution of choice.

Ambinder, Walter J. 1965. "The Extent of Successive Placements among Boys in Foster Family Homes." Child Welfare (July).

Barbour, Charles A. and William F. Shughart II. 1997. "Abortion Rights and Wrongs: The Impact of Waiting Periods and Parental Consent on Abortions in Mississippi." Unpublished manuscript.

Brown, Robert W. and R. Todd Jewell. 1996. "The Impact of Provider Availability on Abortion Demand." Contemporary Economic Policy 14 (April).

Cole, Elizabeth S. 1987. "Adoption." In Anne Minahan et al., eds., Encyclopedia of Social Work, 18th edition, vol. 1. Silver Spring, MD: National Association of Social Workers.

Flango, Victor Eugene. 1990. "Agency and Private Adoptions, by State." Child Welfare 69 (May-June).

Jeter, Helen R. 1963. "Children, Problems, and Services." In Child Welfare. Publication No. 403. Washington, DC: U.S. Children's Bureau.

Jones, Marshall B. 1993. "Decline of the American Orphanage, 1941-1980." Social Service Review 67 (September).

Maas, Henry S. and Richard E. Engler, Jr. 1959. Children in Need of Parents. New York: Columbia University Press.

Medoff, Marshall H. 1993. "An Empirical Analysis of Adoption." Economic Inquiry 31 (January).

Stein, Theodore J. 1987. "Foster Care for Children." In Anne Minahan et al., eds., Encyclopedia of Social Work, 18th ed. 1. Silver Spring, MD: National Association of Social Workers.

Whittaker, James K. 1987. "Group Care for Children." In Anne Minahan et al., eds. Encyclopedia of Social Work, 18th ed., vol. 1. Silver Spring, MD: National Association of Social Workers.

Wingard, Deborah. 1987. "Trends and Characteristics of California Adoptions: 1964-1982." Child Welfare 66 (July-August).

Wolins, Martin and Irving Piliavin. 1964. Institution or Foster Family: A Century of Debate. New York: Child Welfare League of America.

U. S. Department of Commerce, Bureau of the Census. 1993. Statistical Abstract of the United States 1993. Washington, DC: U.S. Government Printing Office.