Children and Youth Services Review 78 (2017) 917

Contents lists available at ScienceDirect

Children and Youth Services Review

journal homepage: www.elsevier.com/locate/childyouth

Privatization and subsidization of adoption services from foster care: MARK

Empirical evidence
Joseph Deutscha, Simon Hakimb,, Uriel Spiegelc, Michael Sumkind
a
Department of Economics, Bar-Ilan University, Israel
b
Center for Competitive Government, Temple University, USA
c
Department of Management, Bar-Ilan University, Zefat Academic College, Israel
d
MA Student, Bar Ilan University, Israel

A R T I C L E I N F O A B S T R A C T

Keywords: This paper analyzes empirical dierences in adoption services of public and private agencies. The empirical
Public adoption investigation includes cross-sectional time series aggregated data for the 50 states within the United States from
Private adoption 1996 to 2010 with detailed statistical analysis of the period from 2000 through 2010 for which consistent and
Foster home comprehensive data exists. Under private agencies, only 11.6 months elapse from the time the courts terminate
Subsidies
the natural parents' custody until the child is adopted, while with public agencies the same process lasts for
JEL classications: 16 months. Furthermore, during the decade from 1996 to 2006 private agencies completed more adoptions than
H0 public agencies. However, the performance gap in favor of private agencies was eliminated in 2006 and in the
L33
following years. The results suggest that privatization of adoption of young and healthy children did not show an
J4
advantage for private services. However, transitioning adoption services to private agencies for older children or
children with complex special needs, improves the adoption services compared with those of public agencies.
Subsidization especially improves the adoption of older children and of all children with special needs while it
appears to be statistically insignicant or implied as unnecessary for healthy babies.

1. Introduction
Every year almost one million children in the United States are
victims of neglect or abuse. In a survey conducted at California state
prisons, 74% of male inmates and 75% of female inmates had been in
foster care during one year or more (California Senate, 2011; California
Oce of Research, 2011; Cusick et al., 2010) Spending extensive time
in foster care homes has shown devastating eects on children. Thirty-
three percent of male inmates and 40% of female inmates had spent ve
or more years in foster care. A University of Chicago study found that
one third of the former foster care children who were tracked in three of
the Midwestern states had a high level of involvement with the
criminal justice system. In a survey of prisoners under the age of 30,
20% of them spent part of their lives in foster care, and 25% of the
inmates that had convictions spent some of their childhood in foster
care (California Senate, 2011.)
Children growing at foster care homes fare worse than children
growing at normal homes with signicant educational achievements
and professional skills. Foster care children are less likely to gain
employment and their earnings are signicantly lower lacking sucient
earnings for decent living. Many of them have children that they cannot
support. Many of them struggle with mental and physical health
problems and receive government welfare support, and they are more
likely than their same age peers to be involved in criminal activities
(Courtney et al., 2005).
In 2014, the median stay in foster care was 13.5 months, an increase
from 11.9 months in 2004. Approximately 100,000 children awaited
adoption during an average of 37.3 months, including 13.7 months
after court termination of parental rights. In 2013, 11% or 43,083
children spent 5 years or more awaiting adoption. Nearly 30,000
children reached the age 18 without nding an adoptive home. The
average age of a child at the time of adoption was eight. Fifty-three
percent of the children were adopted by former foster parents, 32% by
relatives, and the remaining 15% were adopted by non-relatives
(USDHHS, 2015a; Tides Center, 2015). In the year beginning October
1, 2013, 50,634 children were adopted, while 38% of them were in
California, Texas, Florida, Arizona, and Michigan, in declining order
(USDHHS, 2015b).
In September 2010, among 408,425 children in foster care, 26%
were in homes of relatives, 48% were in foster homes of non-relatives,
and the remaining children were in group homes. Among those in foster
care, 25% had the goal of adoption while 21% of those who left foster

Corresponding author.
E-mail addresses: Joseph.Deutsch@biu.ac.il (J. Deutsch), hakim@temple.edu (S. Hakim), spiegeu@biu.ac.il (U. Spiegel), michasu@gmail.com (M. Sumkin).

http://dx.doi.org/10.1016/j.childyouth.2017.04.022
Received 28 December 2016; Received in revised form 30 April 2017; Accepted 30 April 2017
Available online 02 May 2017
0190-7409/ 2017 Elsevier Ltd. All rights reserved.
J. Deutsch et al.
care were adopted. Among the 254,114 children who left foster care,
the median amount of time spent in the adoption process was
13.5 months, an increase from 12.0 months in 2000. Thirteen percent
of them spent less than one year in foster care (Child Welfare
Information Gateway, 2012). The number of children waiting for
adoption after the termination of parental rights and as of the last
day of the year increased from 74,000 in 2004 to 82,000 in 2007. Since
2007, the number has steadily declined to 61,000 in 2011.
These gures showed signicant improvement in 2011 as compared
with the statistics of a decade earlier. The problem that was faced by
many states until 2011 was the heavy caseload imposed upon social
workers responsible for transferring eligible children from foster care to
adoption. This caseload resulted in insucient attention to attaining
either fast or reliable adoption, or both. Thus, several states decided to
contract out at least some of the adoption services to private companies.
The objectives were to transition more children from temporary foster
care to permanency and to improve the quality of the adoption. The
states that have in some form contracted their adoption services include
Michigan in 1992, Kansas in 1996, Illinois in 1997, Connecticut in
1999, Massachusetts in 2000, Colorado in 1997, Florida in 2005, Texas
in 2005, Oklahoma in 1992, and Tennessee in 1995.
In this paper, we analyze whether contracting out adoption services
increases the likelihood that children are adopted from foster care.
Simultaneously, we analyze whether contracting out adoption services
diminishes the time that children spend in the process. We also evaluate
whether contracting out leads to a higher and better quality of
adoption. While a fast adoption process may shorten the length of stay
in foster care, it may increase the probability of failed adoption.
In some sense, our results correspond with earlier ndings of
Hansen (2009) which suggest that the use of private agencies strength-
ens faster adoptions. Adoptions were mostly successful when the state
guarantees post-adoption nancial support to the adoptive parents. This
result coincides with the conclusions of Hansen and Hansen (2006) in
an earlier paper that cutting adoption subsidies may cut adoption
creation.
Section 2 discusses the eects on children of the two forms of foster
care and adoption. Section 3 presents the economic views and
experiences of market exposure and privatization of adoption services.
Section 4 describes our empirical model. We analyze a model that
explains the length of the adoption process in relation to the attributes
of children, privatization, and dierences among states. In Section 5 we
introduce a pooled database by year and state which shows that a
quantitative success of privatized adoption does not necessarily reect
qualitative success. Section 6 summarizes the paper and suggests some
policy implications.
2. Foster-care versus adoption
The eld of personality development shows that children that grow
up without permanent parents suer lasting and devastating psycholo-
gical eects. Such adverse eects exist even when the separation was
for a few months. Therefore, the research concludes that in the case of
foster care children, uniting the child with a permanent family needs to
be done as soon as possible to minimize psychological damages (Katz,
1990). It should be emphasized that the main objective of the welfare
authorities is to shorten as much as possible the duration of foster care
and to nd warm adoptive homes for children as quickly as possible.
We mention below several references to this approach to transition-
ing from a foster home to an adoptive home. However, the main
supporting argument can be found in the long-established and im-
portant Attachment Theory that is based on Bowlby. Its development
began in 1930 and continued until 1990 in a more recent paper by
Ainsworth and Bowlby (1991).
The basic concept derived from the Attachment Theory is that a
stable mother gure is important for children. Early family relation-
ships are important for the personality development of children.
10
Children and Youth Services Review 78 (2017) 917
Actual family experiences are often a very important if not the
crucial cause of emotional disturbances in children.
The presence of one's own attached and permanent mother is
important for a child.
This concept is introduced by Bretherton (1992) and the following
quotation is from page 770 of his paper regarding Efe parent-infant
attachment.
Young Efe infants receive more care (including nursing) from other
adult women than from their own mothers, except at night. Despite this
multiple mothering system, by 6 months infants begin to insist on a
more focalized relationship with their own mothers although other
female caregivers continue to play a signicant role.
The analogy may be made to the foster home in which multi-
motherhood may exist. Although one mother is involved, the distinc-
tion between the biological and the adoptive mothers is very clear to
the child.
The need for a stable and continuous relationship to enable healthy
social-emotional child development is a very convincing argument.
Comparing long-term foster care and adoption shows that those
growing up in adopted homes have higher levels of emotional security,
sense of belonging and general well-being. Long-term fostering does not
form a positive psychological base in the lives of children. Foster care is
temporary and can be removed at the initiative of local authorities, the
child, the foster caregivers or the birth parents. Even when foster care is
long-term, the child and the foster caregivers perceive it as temporary
and therefore prevent, in many cases, the formation of close relation-
ships. On the other hand, adoption conrms legal permanency with all
parental responsibilities transferred to the adoptive parents. Only
neglect or abuse by the adoptive parents can dissolve adoption and
this is not dierent when the child lives with the biological parents
(Triseliotis, 2002). In a similar study in the UK, Selwyn and Quinton
(2004) examined 130 children that all wished to be adopted but some
of whom had had to spend childhood in foster care. Foster caregivers
had authoritative diculties in making decisions in favor of the child
that are obvious and easily done by adoptive parents. Social workers
wanted to control what is happening in the lives of the children in foster
care and therefore maintained their full decision-making concerning
the child. Bureaucratic delays resulted, for example, in children missing
school trips. When a child is in foster care any desire that a child has,
even sleeping overnight at a friend's home, requires the social worker's
approval. Another diculty in foster care was that at the age of 16, the
young adults were referred to independent living while both they and
their foster caregivers wanted them to stay home. The concern over
living independently adversely aected the young adults in foster care
while their peers in adoptive homes could more easily live peacefully,
concentrate, and better perform in school. The legally insecure foster
placement adversely aected the emotional link in both directions for
the foster parents and the children while a positive link was evidently
stronger for the adoptive parents and their adopted children. However,
it is possible that part of this distrust can be attributed to the age of
children in foster care, who are usually older than their adopted peers.
The Adoption and Safe Families Act (ASFA) of 1997 addressed the
concern regarding the long wait in foster care for permanency with
families. The Act prioritized permanency by instituting reduced periods
for reunication eorts and deadlines for terminating parental rights.
The Act required that parental rights be terminated even when an
appropriate adoption is not available. It required the states to terminate
parental rights if a child has been in out-of-home care for 15 out of
22 months. High numbers of foster care placements contributed to
increases in both violent and nonviolent crime. Lengthy stay in foster
care linked to violent crime was particularly strong among African-
American youth. Foster youth were also much more likely to be
arrested as they made the transition to adulthood (Doyle, 2008).
Smith, 2003 analyzed a cohort of 1995 foster children whose parental
rights were terminated in the same month. The lower rate of exiting
foster care after Termination of Parental Rights (TPR) is attributed to
J. Deutsch et al.
being older, African-American, placed in kinship care, or having
multiple placement settings. Smith, 2003 suggests that shortening the
length of time after TPR could be harmful for older and minority
children. Eorts at reunication with their biological parents have been
shortened, and after being available for adoption, they remain in foster
care because of lack of adoption interest. Thus, Smith, 2003 suggests
that ASFA has expedited TPR but for some children caused a reduction
in their chances for reunication. At the same time, these children
remain in foster care, often in several such homes, while raising their
expectation for permanency to no avail.
The most signicant delaying factor in transitioning children from
foster care to adoption has been a usually unpredictable delay in the
court's termination of parental rights. Festinger and Pratt (2002)
investigated how the legal process can be expedited by following the
adoption service and control groups with dierent legal procedures.
The study concluded that the factors most important for expediting the
legal procedure are for the case to be kept in the same county, before
the same judge who is familiar with the case, and on the court calendar.
Maintaining these conditions shortened the time from leaving the
biological parents' home and entering the foster home until nal
adoption from one year to six months.
We reviewed many academic and professional articles that stressed
the benets to children of reducing the process from foster care to
adoption. It is imperative to reduce the duration of the case with the
often unpredictable delay of the judicial authority. On the other hand, it
appears that shortening the adoption process could prevent, at least in
some case, reaching the most desirable goal of reunication of the child
with biological parents or relatives. A set deadline for TPR reduces the
probability of reunication. Since adoption policies are set in laws,
there is little chance that each case is evaluated with respect to its
specic conditions. Therefore, a quantitative comparison is necessary
between the benets of reduced time in the process and the opportunity
costs of missed reunication. Only an empirical benet cost analysis
should be conducted before shortening the process. Such a database
analysis could possibly lead to the desired length of the process for
various attributes of children under specic environmental conditions.
3. Privatization of adoption services
The seminal paper by Landes and Posner (1978) and Posner (1987
and 1992) argued that introducing markets to child adoption increases
social welfare. They were the rst to explain the huge gap between the
availability of adoptive parents and babies for adoption by creating an
institutionally derived non-market price for adoption. Posner (1992)
concluded that the birth parents, the adoptive parents, and the children
would be better o if prices would uctuate. In a subsequent article,
Boudreaux (1995) claimed that expectant mothers should be allowed to
freely sell their parental rights to adoptive parents. His market solution
is limited to the transfer of rights to raise the child and is not the sale of
the child. The distinction is necessary to eliminate the right of adoptive
parents to abuse the child. Such a direct transaction will increase the
welfare of all three parties involved.
A more limited privatization of social services is contracting out or
outsourcing services that are provided by government agencies
(Freundlich, 2000). It can have two forms. Government agencies either
contract the delivery of single components of a large program or they
contract private agencies to deliver a comprehensive set of program
services in a single location (Winston et al., 2002: 910; Flaherty,
Collins-Camargo, & Lee, 2008). The delivery of adoption services re-
ects the rst option when private contractors provide social workers to
reduce the load of their public peers. New York City began contracting
with non-prot agencies more than a century ago to provide services for
children under its custody. Indeed, the private providers usually
compete among themselves and with the public and nonprot entities.
The greater the number of such vendors, the greater the competition
and the more ecient is the service. Experience shows that delivery of
11
Children and Youth Services Review 78 (2017) 917
private services also raises the quality and level of government service.
Blackstone and Hakim (2003) analyzed the eorts to privatize
adoption and foster care in Kansas and Michigan. In a later eort that
included Illinois, Blackstone, Buck, and Hakim (2004) showed that
when adoption services are partially privatized, the changes improved
the speed and number of adoptions, and reduced the costs of the
services.
A unique problem in privatizing adoption services is that unlike
adults, children are often unable to complain in the event of a mismatch
or in the extreme case of abuse. Thus, continuous, frequent, ethical, and
professional control and oversight are needed before, during and
following an adoption. Hansen and Hansen (2002 and 2006) stress
the need of the workers to be familiar with the child's wellbeing. On one
hand, it is desirable as stated above, to achieve permanency as soon as
possible. However, unsuccessful placement and repeated trials could
have long-term devastating eects on the child, and be of signicant
cost to the public welfare agency. Again, Hansen and Hansen (2006)
assert that privatization of adoption institutions requires long prepara-
tion and the development of control and oversight by the public
agencies. High pay oered for private adoption services frequently
brings about adoption with no regard to permanency. Hansen (2009),
estimated that each dollar currently spent on the adoption of a child
from foster care yields between two and three dollars in future benets
to society.
Flaherty et al. (2008), and Snell (2000) claim that economic
motives, competition, and long-term business considerations will make
the use of private entities a reliable and secure method for delivering
child welfare services. Both papers suggest that American experience
has shown that privatization is preferable in the long term.
Hollingsworth, Bybee, Johnson, and Swick (2010) express an
opposite view to that of Flaherty et al. (2008) and Snell (2000) by
asserting that the experience of social workers is crucial to the success
of the adoption process. Private social workers are paid less than public
social workers, and thus are less experienced. In addition, in private
companies the social workers often change, leading to less familiarity
with specic needs of the children. Most of the children and their
biological families are non-white, as are the social workers in the public
sector. Their eorts to keep the child with biological relatives are more
successful. However, the fact that social workers in private agencies are
often white may make the communication more dicult among the
child, the biological parents and the private social worker. This can lead
to an unsuccessful outcome in the eort to keep the child with
biological relatives. Economic considerations may lead private social
workers to persuade biological relatives not to maintain parental rights
and to place the child in the unknown environment of adoption.
Craig et al. (1998) oer the following few suggestions for contract-
ing out foster care and adoption services successfully: The contract with
private agencies should quantitatively specify outcomes (or goals) that
reect the children's needs. States that initiate contracting out must
always rely on competitive bidding among several private agencies to
avoid monopoly power. Rewards should be provided for good perfor-
mance of measurable goals. They also recommend that the states that
do contract out their foster care and adoption services should hire
external contract monitors and evaluators to assess contractor perfor-
mance.
4. Methodology
The methodology used in the current paper is based on multivariate
regression analysis to examine the adoption process. We consider the
process that is performed by agencies funded by the public sector as
compared to the process managed by private for-prot agencies. In
addition, we try to statistically estimate the eects of variables on the
adoption rate, including among them subsidies granted to public or
private adoption agencies and several other socioeconomic variables.
The empirical analysis is based on aggregate panel data sets for fty
J. Deutsch et al.
states within the United States during eleven years from 2000 to 2010
(The National Data Archive on NDACAN Child Abuse and Neglect,
2013, and Child Welfare Information Gateway, 2012), generating ve
hundred and fty observations.
4.1. The target variable and the eect of privatization
In recent years, several states within the United States have
transitioned from public service for adoption to privatization of the
adoption rate provided by private for-prot agencies. We expect that
this transition should lead to an increased adoption rate i.e. target
variable. According to our presumption, prot motivation may encou-
rage agencies to promote adoption and to act quickly and eectively, as
in many other areas in which privatization has proved itself. A priori
our hypothesis is that a private agency is more ecient and motivated
for successful adoption as compared to a public agency that is funded
by the public sector.
4.2. The target variable and the eect of other factors
Our research examines the impacts of several socioeconomic factors
and other characteristics of the state that may inuence in the
aggregate level the adoption rate (target variable) of adoption. The
following are several corollaries regarding these expectations:
a. Government involvement using a scal policy of subsidies for the
adoption agency should lead to a more ecient adoption process.
b. Income per capita of the state, that most likely represents better
health and education and as a result a higher standard of living as
well as greater individual happiness, may encourage people to
voluntarily adopt children, especially those children who are less
healthy or more neglected by society.
c. In the category of states that are more liberal or democratic as
compared to those that are more conservative or religious, the
adoption process may be more successful.
d. The duration of foster home care and of education provided by
foster mentors may negatively aect the chance of a successful
adoption process and the probability of a peaceful stay with the
adoptive family later on. It may postpone the chance on one hand of
nding an adoptive family and on the other hand of succeeding with
the adoption itself.
4.3. Statistical methods
We analyze the eects of privatization, subsidies for adoption
agencies, sociodemographic factors, and other variables on the aggre-
gate adoption rates of the states.
The hypothesis that the introduction of a private adoption service
aects the adoption rate is tested using a dummy variable equal to one
for those states in which the privatization process was initiated starting
with the relevant year. The dummy is entered only from the year in
which the private adoption began so we can compare not only between
periods but also between states.
Similarly, we introduce a xed eect statistical model (FE) to
capture the state specic sociodemographic attributes and other state
specic characteristics. In order to correct possible bias resulting from
endogeneity of the subsidy variable, we estimate the model using an
instrumental variable method (IV).
The estimation uses Fixed Eects as representing the special
characteristics of each state. In addition we apply the instrumental
variables system in order to resolve the endogenous problem due to the
use of several explanatory variables.
5. The empirical model
As discussed above in the methodological section, the empirical
12
Children and Youth Services Review 78 (2017) 917
analysis is based on aggregate panel data set 50 states within the United
States during 11 years from 2000 to 2010 generating 550 observations.
The estimated macro model statistically explains the rate of adop-
tion ( ) as a linear function of the following group of variables:
Adopt
Pop
(1) Characteristics of the Child
(2) Characteristics of Adoptive Parents
(3) Characteristics of the State in which the Adoption Takes Place
(4) Characteristics of the Adoption System
(5) Incentives for Adoption in the State: Subsidy
(6) Privatization of public adoption agencies
The sources for the aggregate data are the Survey of the National
Council for Adoption (NFCA) and the State and Metropolitan Area Data
Book. The following tables describe the dierent groups of explanatory
variables:
(1) Characteristics of the child demographic factors.
Name of Description of variable
variable
agegrp Age group of child. The variable of the child's age is
divided into 3 age groups: 05, 612, 1318. Each of
the age variables includes information regarding the
rate of child adoption in certain age groups.
gender Gender of child. This is a dummy variable that is
given the value of 1 if the child is female and 0 if the
child is male.
race Race of child. This is a dummy variable that is given
the value of 1 if the child is in the white racial group
and 0 if the child is in the black racial group.
Characteristics of the child mental and physical health of the
adopted child.
There are four fundamental kinds of impairments among the
population of children awaiting adoption: mental disturbance, mental
retardation, physical disability that impacts daily activity of the child,
and a visual or hearing impairment that signicantly impacts the child's
development.
Name of Description of variable
variable
mental Adoption rate of children with mental disturbance
motor Adoption rate of children with mental and motor
impairments
physical Adoption rate of children with a physical disability
that impacts daily activity
visualhear Adoption rate of children with a visual or hearing
impairment that signicantly impacts development
(2) Characteristics of adoptive parents family structure.
Name of variable Description of variable
traditionalfam Traditional family. Adoption rate of children by
parents who are a couple
nontraditionalfam Nontraditional family. Adoption rate of children
by a male or female single foster parent or by a
couple who is not married
Characteristics of adoptive parents The type of relationship with
the child awaiting adoption.
J. Deutsch et al.
Name of variable Description of variable
foster Adoption rate of children by a foster family
that is not related to the child
relative Adoption rate of children by a foster family
that is related to the adopted child
otherrelative Adoption rate of children by family that is not
included in any of the categories
whiteadoptivemother Rate of white mothers that want to adopt a
child
adoptivemotheragegrp Rate of mothers that want to adopt a child
according to 3 age groups: a. rate of women
from ages 2535; b. rate of women from ages
3645; c. rate of women from ages 4660.
(3) Characteristics of the state in which the adoption takes place.
Name of Description of variable
variable
incpercapi Income per capita. These are a dummy variables
receiving the value of 1 if the state belongs to quartile
i (i = 1,2,3) of income distribution and 0 otherwise.
aborti Abortion rate. These are a dummy variables receiving
the value of 1 if the state belongs to quartile i
(i = 1,2,3) of abortion rates distribution and 0
otherwise.
region This is a dummy variable that is given the value of 1 if
the state belongs to one of 3 geographic regions:
South, Northeast, Midwest and 0 otherwise
republican This is a dummy variable that is given the value of 1 if
in the last four elections within a state, a Republican
candidate was elected at least three times.
conservative The rate of people in a state that have a conservative
worldview.
liberal The rate of people in a state that have a liberal
worldview.
religious The rate of religious people in a state.
nonreligious The rate of nonreligious people in a state.
(4) Characteristics of the adoption system average duration of the
adoption process:
Name of Description of variable
variable
stayleveli These are a dummy variables receiving the value of 1
if the state belongs to quartile i (i = 1,2,3) in the
distribution of waiting time in foster homes and 0
otherwise.
Characteristics of the adoption system average duration of time
from the termination of parental rights by a court until the adoption.
Name of Description of variable
variable
waitingleveli These are dummy variables receiving the value of 1 if
the state belongs to quartile i (i = 1,2,3) in the
distribution of waiting time for the child in foster
homes from termination of parental rights until the
adoption and 0 otherwise.
13
Children and Youth Services Review 78 (2017) 917
(5) Incentives for adoption in the state: subsidy.
Name of variable Description of variable
subsidy Rate of adopted children who received subsidies
(6) Privatization of public adoption agencies
Name of Description of variable
variable
private This is a dummy variable that is given the value of 1
if the state completed a privatization process
The panel data include aggregate annual data for 50 states during
all years from 2000 to 2010 (The National Data Archive on NDACAN
Child Abuse and Neglect, 2013, and Child Welfare Information
Gateway, 2012.) Since 1995, all states are required to provide detailed
information concerning every child that is under the auspices of the
state Department of Health and Human Services (DHHS). However,
only 14 states provided the data in 1995, and the problem continued
until 1998 when federal sanctions of reduced child subsidies went into
eect, raising the number of reporting states to 41. The database
became comprehensive in 2000 when the individual states started to
report full information on child adoption.
We have used three methods for the statistical estimation: the basic
model we used is the OLS where there is no control over the specic
eects of the year and the state. The Fixed Eect (FE) method considers
the heterogeneous eects among the states and the year however is
consistent within them. Each state has specic demographic, popula-
tion attributes, level of subsidy and the extent of adoption which the FE
method identies. Each state may have characteristics that dier
among states but are relevant in the explanation of the dependent
variable. Exclusion of one such independent variable may distort the
statistical results. The FE method corrects for such omission of a
relevant explanatory variable. Introduction of the state eects may
not correct the endogenous problem in our model. For example, a
positive correlation may exist between the level of the subsidy and the
error term that is lowered by the introduced dummy variable for the
state but is not eliminated. The IV method corrects for such a problem
by including assisting variables like the rate of adopting mothers who
are white and the age groups of such adopting mothers. These two
variables do not directly aect the dependent variable of the rate of
adopted children however are highly collinear with the level of subsidy.
White women concentrate on their careers may choose not to bear a
child. However, availability of subsidy may motivate them to adopt.
6. Evaluation of the aggregate model
The main purpose of the aggregate analysis is to test empirically the
eect of state privatization and subsidization provided for successful
adoption on the extent of child adoption. At rst glance, the adoption
rates under a private service are higher, in the short run, than under the
public service.
Fig. 1 illustrates the transitory eect of privatizing adoption
agencies, a process that began in some states approximately 20 years
ago. The nancial incentive of private agencies encourages them to be
more ecient and to make better matches between the children and the
adopting parents. Such matching is more likely achieved in the
adoption of young and healthy white girls who are generally more
easily adopted. During the years 19972005, the states that include
both private and public agencies showed 30% favorable gap in
adoptions in comparison to states that relied merely on public agencies.
However, during the years 20062010 this gap closed.
J. Deutsch et al. Children and Youth Services Review 78 (2017) 917

Number of Children adopted by 100,000 Table 2

30 The eect of private vs. public adoption on the rate of adoption among children with
diagnostic special need.
25 The dependent variable is the rate of adopted children by category of special need.

20 Independent Dependent variable

variables Mental and motor Special mental Physical special
15 special needs rate needs rate needs rate
(2) (4) (6)
10
FE FE FE
5
Private 0.01*** 0.01* 0.01***
0 (0.005) (0.006) (0.004)
1996 1998 2000 2002 2004 2006 2008 2010 2012 Subsidy 0.04*** 0.03** 0.01
(0.010) (0.012) (0.008)
Privatized Not Privatized Race 0.00 0.01 0.01*
(0.004) (0.004) (0.003)
Fig. 1. Number of adopted children by year.
Northeast 0.02*** 0.02** 0.00
(0.005) (0.006) (0.004)
Table 1 South 0.02*** 0.02*** 0.02***
The eect of private vs. public adoption on the rate of adoption. (0.004) (0.005) (0.003)
Midwest 0.01** 0.04*** 0.01***
Dependent variable (0.005) (0.005) (0.004)
Adoption rate Traditionalfam 0.04*** 0.04*** 0.05***
(2) (3) (0.011) (0.013) (0.009)
FE IV Relative 0.00 0.01 0.03***
(0.007) (0.008) (0.006)
Private 7.68** 8.67** Republican 0.03*** 0.04*** 0.00
(3.198) (3.473) (0.005) (0.006) (0.004)
Subsidy 5.64*** 27.05* Conservative 0.29*** 0.33*** 0.06
(1.944) (14.208) (0.047) (0.055) (0.039)
Incpercap1 4.21*** 2.29 Religious 0.04 0.00 0.11***
(1.604) (2.467) (0.027) (0.032) (0.023)
Incpercap2 4.66*** 3.12** Constant 0.08*** 0.09*** 0.03*
(1.362) (2.164) (0.019) (0.022) (0.015)
Incpercap3 3.73*** 2.36 Year FE Yes Yes Yes
(1.176) (2.014) State FE No No No
Staylevel1 15.61*** 14.12*** Observations 561 561 561
(4.091) (4.657) R-squared 0.04 0.11 0.12
Staylevel2 9.63*** 13.60*** Adj. R-squared 0.03 0.10 0.10
(1.826) (4.662)
Staylevel3 15.44*** 12.89* ***, **, * - Signicant at 1%, 5% and 10% condence level, respectively.
(2.301) (6.767) Robust standard errors below estimates.
Waitinglevel1 1.65** 1.33*
(0.726) (0.814)
Waitinglevel2 2.77*** 1.77*
The tables also display the results of instrumental variable estima-
(0.844) (1.132) tion (IV) to deal with possible simultaneous eect of the privatization
Waitinglevel3 3.65*** 3.38*** variable substituted with a subsidy variable (the instrumental variable).
(0.963) (1.131) In the FE model, we observe that the sign of the coecient for the
Abort1 10.11*** 11.31
privatization variable is negative and the sign for the subsiding
(2.147) (9.985)
Abort2 5.40*** 17.53*** adoption is positive.
(1.720) (5.271) When considering the possibility of an endogenous relationship
Abort3 10.66*** 7.74 between the privatization variable and the subsidy variable, we
(1.857) (8.374)
introduce an independent variable as IV that replaces the subsidy and
Female 9.00 4.12
(7.314) (8.643)
thereby enhances the negative statistical eect of privatization and
Agegrp2 22.37*** 8.22 increases the positive eect of the subsidy on the rate of adoption.
(6.509) (12.268) In both the FE and IV models, the coecients for the control
Agegrp3 0.48 8.99 variables are stable and robust. We will concentrate on the results of the
(10.000) (13.874)
FE model which explicitly keeps constant the eects of both the state
Nontraditionalfam 10.08* 3.15
(5.140) (8.677) and the year. The FE statistic for the entire equation also appears
Relative 2.02 2.24 robust. When the eects of the state and the year are kept constant (FE),
(1.241) (1.498) the negative and signicant eect of income is revealed for all four
Foster 2.98*** 3.76**
income ranges. As the state income rises, the rate of adoption
(1.148) (1.590)
Republican 13.83*** 4.39
diminishes. The wealthier the state, the lower is the rate of children
(3.455) (4.022) available for adoption and the lower is the rate of adoptions.
CONSTANT 5.64 15.94** The average foster care stay for all children reects the pressure on
the adoption services and the greater involvement of social workers. In
***, **, * - Signicant at 1%, 5% and 10% condence level, respectively.
general, the longer a child stays in foster care, given a constant budget
Robust standard errors below estimates.
for social services, fewer resources are required for adoption services.
Indeed, the states in the 25 percentile lowest average stay in foster care,
A multivariate regression analysis that incorporates all hypothe-
showed a positive and signicant statistical eect on the adoption rate.
sized aecting independent variables is expected to provide robust
Thus, such states can divert resources for adoption services. In the states
statistical results. The aggregated model, Tables 14, includes xed
in which the average foster care stay is longer, the positive statistical
eects dummy variables (FE) for each state and each year to capture the
eect still exists. However, when we observe the size of the coecient,
eect of specic state characteristics and the year trend.

14
J. Deutsch et al. Children and Youth Services Review 78 (2017) 917

Table 3 Table 4
The eect of private vs. public adoption on the rate of adoption among children with The eect of private vs. public adoption on the rate of adoption by age group.
diagnostic special need. The dependent variable is the rate of adopted children at age group 0 to 5.
The dependent variable is the rate of adopted children by category of special need.
Independent variables Dependent variable: adoption rate at age group
Independent variables Dependent variable Age 05 Age 612 Age 1318
Visually or hearing impaired rate Mentally disturbed
(8) (10) Private 0.02** 0.00 0.02***
FE FE (0.010) (0.008) (0.005)
Subsidy 0.08*** 0.06*** 0.01
Private 0.01*** 0.04*** (0.022) (0.018) (0.012)
(0.004) (0.014) Race 0.01 0.00 0.01***
Subsidy 0.03*** 0.10*** (0.008) (0.007) (0.005)
(0.008) (0.030) Female 0.14*** 0.14*** 0.01
Race 0.01** 0.00 (0.042) (0.035) (0.024)
(0.003) (0.011) Northeast 0.00 0.01 0.01
Northeast 0.00 0.06*** (0.011) (0.009) (0.006)
(0.004) (0.015) South 0.03*** 0.01* 0.02***
South 0.02*** 0.06*** (0.009) (0.007) (0.005)
(0.003) (0.012) Midwest 0.03*** 0.01 0.02***
Midwest 0.00 0.02 (0.010) (0.008) (0.006)
(0.004) (0.014) nontraditionalfam 0.03 0.01 0.02
Traditionalfam 0.01 0.09*** (0.026) (0.022) (0.015)
(0.009) (0.034) Relative 0.03** 0.01 0.01*
Relative 0.02*** 0.01 (0.014) (0.012) (0.008)
(0.005) (0.020) Republican 0.01 0.00 0.00
Republican 0.03*** 0.11*** (0.010) (0.008) (0.005)
(0.004) (0.015) Liberal 0.14 0.03 0.16**
Conservative 0.01 0.91*** (0.112) (0.094) (0.063)
(0.039) (0.138) nonreligious 0.11** 0.05 0.15***
Religious 0.10*** 0.19** (0.049) (0.041) (0.027)
(0.022) (0.080) Constant 0.58*** 0.33*** 0.09***
Constant 0.01 0.22*** (0.041) (0.034) (0.023)
(0.015) (0.055) Year FE Yes Yes Yes
Year FE Yes Yes State FE No No No
State FE No No Observations 561 561 561
Observations 561 561 R-squared 0.14 0.18 0.10
R-squared 0.13 0.14 Adj. R-squared 0.12 0.16 0.08
Adj. R-squared 0.11 0.12
***, **, * - Signicant at 1%, 5% and 10% condence level, respectively.
***, **, * - Signicant at 1%, 5%, and 10% condence level, respectively. Robust Robust standard errors below estimates.
standard errors below estimate.
Tables 2 and 3 provide equations that address children with
the states with the longer foster care stay exhibit smaller coecient or aggregate and specic diagnosed special needs.
less of a statistical eect. Thus, greater eciency in foster care services The specic special needs include emotional needs, mental impair-
enables increased success of adoption services. ment, special physical needs, and visual or hearing impairments.
We turn our attention now to the length of the average transition Our major nding shows that when all children have medical
period by state, from the time that the biological parents lose the right impairments, the sign of the privatization variable is negative; suggest-
to raise the child until the actual adoption takes place. We hypothesize ing that privatizing the service reduces the rate of child adoption. The
rst that ecient adoption service is reected in a shorter such time subsidy variable is positive, suggesting that as the subsidy for the
period. An alternative hypothesis is that a detailed and diligent children increases, the adoption rate rises for children with special
matching between the adoptive parents and the child requires a longer needs. Even when we separately analyze the dependent variable by type
period of time. The results support the rst hypothesis since as the of special need, the negative sign for privatization persists for all
average time the child spends in transition increases, the rate of categories except for children with special physical needs. Privatization
adoption declines. Adoptions are successful in those states in which a appears to reduce statistically the rate of adoption both in the aggregate
child spends less time in transition prior to adoption. and for specic children diagnosed with special needs, except for
States in which the abortion rate is low have more children who are children with physical impairments. In addition, as the subsidy per
not raised by their biological parents and thus the rate of adoption is child increases, the adoption rate rises in the four categories of children
high. Further, the extent of the adoption eect, or the size of the with special needs.
coecient, increases with the decline in the rate of abortions. We will estimate the same equations to explain adoption of children
On the demand side, the rate of adoption is higher for non- in three age groups (Table 4).
traditional families including a female or male foster care parent and We conducted the two estimation methods of OLS and FE which
non-married couples. The sex of the child is insignicant in explaining showed similar results. The equation for the youngest age group of
the adoption rate. 05 years old shows a negative statistical eect of privatization and a
Republican majority is a dummy variable that equals 1 if in three negative sign for the subsidy. Interestingly, race seems not to aect
out of the four most recent elections, a Republican governor was adoption for this youngest group of children.
elected. Until the middle of 2015, Republicans objected to public When we shift our attention to the age group of adopted children
funding for private adoption agencies and did not allow public agencies from 6 to 12 years old, privatization shows no statistical eect while
to support adoption by homosexual couples. Bills were presented in subsidies do show a positive eect.
some states (e.g. Florida) which allowed faith-based agencies to follow Privatization, however, reveals a positive eect on the older age
their religious beliefs and prevent homosexuals from adopting children. group of 1318, while subsidies show a positive but not signicant
As such, a signicant positive coecient was hypothesized and was statistical eect.
indeed supported by the FE estimated model (Table 1). We see an interesting trend in privatization and the public subsidy

15
J. Deutsch et al.
for the adopted children. For the youngest group of children, 05 year
old, government services are better and there is no need for the
subsidies. In the age group of 612 years old, private and public
providers show no signicant dierent eect, but subsidies do have a
strong positive eect. When we turn to the oldest group of 1318 year
olds, which is the most challenging age group for adoption, privatiza-
tion has a positive eect while the subsidy is statistically insignicant.
This could suggest an important policy implication that still requires
further analysis. Demand for very young children is high so that using
private rms or subsidizing the children is unnecessary. When we
observe the middle group of 6 to 12 year old, privatization is still
unnecessary but subsidies seem to improve adoption. Privatization
appears to be of signicant positive eect in increasing adoptions in the
oldest group of children of 13 to18 years old who are the least
demanded by adopting parents. Hence, at this initial stage of analysis,
the use of private adoption companies may be more eective for the
oldest group of children.
7. Summary and conclusions
This paper analyzes the eects of public and private intervention in
child adoption services, and of government subsidies provided for such
children. The analysis also considers the relevant environmental
variables including the attributes of the adopted children, the adoptive
parents, the state, and the adoption process. The statistical analysis
includes two parts. This aggregate analysis uses the attributes of both
the adopted children and the adoptive parents to statistically explain
the extent of adoption in the 50 states for the years 1995 to 2010.
The empirical results show that a child with no special medical
needs is adopted within 14 months, while a child with such special
needs is adopted within 17.7 months. For older healthy children, the
process lasts on the average for 18.8 months. The adoption process
takes longer for children with serious medical or other special needs
than for children with no special needs or with mild vision and hearing
impairments. When observing the special medical needs in detail we
realize that vision and hearing impairments delay adoption the least,
with the average adoption period of 16.8 months. In contrast, for
adopted children with emotional special needs and mental impairment
the process lasts almost 20 months. There is no signicant dierence
between public and private adoption agencies with respect to the
duration of the process for healthy children and for children with mild
special needs.
The ndings further show that when all cross-sectional and time
series intervening variables are kept constant, privatizing adoption
services shortens the adoption process. Another nding that may be
relevant for public policy is that the level of government subsidies for
adoption services signicantly increases the number of adoptions per
100,000 people.
The states which are most ecient in shortening the length of the
adoption process are Idaho, West Virginia, and Wyoming, while the
least ecient states in this respect are Louisiana, Washington, and New
Mexico.
The study leads to several interesting and promising recommenda-
tions. Several other studies in privatization of state and local services,
including the privatization of adoption services in Kansas, suggest that
privatization or contracting out is appropriate only when sucient
competition exists among agencies. Government needs to extend
monitoring eorts when transitioning adoption services from a mono-
polistic government to a monopolistic private agency has not proved to
be eective. As the competition becomes stronger and prevails, the
need for government monitoring decreases. As the competition inten-
sies, contractors have greater incentive to excel.
Contracting out requires that the expected outcomes are known and
can be quantied to enable easier and eective monitoring. In addition,
other public and private providers, whether for prot or nonprot,
should be allowed to compete. Adoption providers and marketing
16
Children and Youth Services Review 78 (2017) 917
eorts should be extended from the existing state residence of the child
to the national level. With respect to foster care, the child needs to
remain close to the biological parents to encourage and enable reunion
with them. Thus a search for foster care within the state is required.
However, when adoption is concerned, the ties with the biological
parents are severed, and extending the market raises the probability of
the extent of the process while it diminishes its duration.
This study shows that privatization of adoption services was slightly
less ecient for young and healthy children. For older children and
children with special needs, contracting out to private agencies
improves the performance of adoption services as compared with
public agencies. Public adoption services are more ecient and
successful for children with mild special, emotional, motoric, auditory
and visual needs that do not signicantly interfere with daily activities.
However, privatization seems to have a positive and signicant eect
for children who have mild physical special needs and experience
diculties in daily activities. Subsidization appears to improve adop-
tion of older children and all children with special needs, while it
appears to be insignicant for babies. Thus, government could subsidize
private agencies for the adoption of children with mild special needs,
while still providing public adoption services for younger children.
Young and healthy children are most often sought out for adoption
rather than those who are older or are not as healthy. Thus, government
support in the form of subsidies could be limited to the challenging
cases. Contracting out the services together with limited subsidies could
improve the welfare of adopted children.
We analyze and demonstrate in Fig. 1 the eect of privatization of
child adoption on successful adoption measured by the number of
adopted children multiplied by 10,000. As the gure illustrates,
privatization has accelerated the volume of adoption by the public
agency. However, it has more signicantly increased adoption by the
private agencies. Competition led to a signicant increase of adoption
in both the public and the private agencies during the years 19972000.
However, from the beginning of the millennium, private adoptions
decreased for several years. From 2005 on we nd that private and
public adoptions show a similar trend. We may suspect that the reason
is that the private agencies worked more eciently and quickly in the
years that many unsuccessful adoptions occurred. After several years,
when the surpluses in adoption disappeared, the gap was closed
between the public and private agencies in providing adoption services.
Based on our ndings, we recommend that to avoid stagnation in
adoption procedures, government should accelerate adoption using
economic tools such as subsidies rather than by raising competition
between private and public adoption agencies. However, it should also
enable privatization in the form of contracting out and competition in
the more dicult adoption of older children or children with special
needs. With respect to the adoption of these children, the private sector
appears to perform better than the public sector.
The panel data analysis of this paper provides general trends
derived from panel data for 50 US states within the United States
during the years 20002010. Verication of the ndings in future
research requires case studies of individual communities with more
specic data. If the results are veried then the suggested policies
should be implemented in some communities before widespread
implementation.
References
Ainsworth, M. D. S., & Bowlby, J. (1991). An ethological approach to personality
development. American Psychologist, 46, 331334.
Blackstone, E. A., Buck, A. J., & Hakim, S. (2004). Privatizing adoption and foster care:
Applying auction and market solutions. Children and Youth Services Review, 26(11),
10331049 (http://d3iovmfe1okdrz.cloudfront.net/cms/wp-content/uploads/2013/
01/childyouth.pdf Last visited June 19, 2015).
Blackstone, E. A., & Hakim, S. (2003). A market alternative to child adoption and foster
care. Cato Journal, 22(3), 485494 Winter. http://www.fox.temple.edu/cms_
research/institutes-and-centers/center-for-competitive-government-2/
publications/#articles Last visited June 19, 2015.
J. Deutsch et al.
Boudreaux, D. J. (1995). A modest proposal to deregulate infant adoption. The Cato
Journal, 15(1), 117135.
Bretherton, I. (1992). The origins of attachment Theory: John Bowlby and Mary
Ainsworth. Developmental psychology, 28(5), 759775.
California Oce of Research (2011). State survey of California prisoners: What
percentage of the state's polled prison inmates were once foster care children. Policy
mattersCalifornia State Senate (December. See, http://www.sor.govoce3.com/
vertical/Sites/%7B3BDD1595-792B-4D20-8D44-626EF05648C7%7D/uploads/
Foster_Care_PDF_12-8-11.pdf last visited December 30, 2012).
California Senate (2011). Policy matters. Oce of Research (Dec.). See, http://sor.senate.
ca.gov/sites/sor.senate.ca.gov/les/State%20Survey%20of%20California
%20Prisoners.pdf Last visited June 23, 2015.
Child Welfare Information Gateway (2012). Foster care statistics, 2010. 1250 Maryland
Avenue, SW, May: Children's Bureau/ACYF (See, www.childwelfare.gov/pubs/
factsheets/foster.pdf. Last visited, December 30, 2012).
Courtney, M. E., et al. (2005). Midwest Evaluation of the adult functioning of former foster
youth: Outcomes at age 19. Chapin Hall Center for Children at the University of
Chicago (May).
Craig, C., et al. (1998). Blueprint for the privatization of child welfare. Policy Study No.
248Institute for Children, Inc. Boston, Ma (Dec.). See, http://www.nfpcar.org/
Archive/Susan_Orr_Privatization.pdf Last visited June 21, 2015.
Cusick, G. R., et al. (2010). Crime during the transition to adulthood: How youth fare as they
leave out-of-home care. Washington D.C.: US Department of Justice (Document No.:
229666. https://www.ncjrs.gov/pdles1/nij/grants/229666.pdf Last visited 11/
20/11).
Doyle, J. J., Jr. (2008). Child protection and adult crime: Using investigator assignment to
estimate causal eects of foster care. MIT Sloan School of Management & NBER
(December).
Festinger, T., & Pratt, R. (2002). Speeding adoption: An evaluation of the eects of
judicial continuity. Social Work Research, 26(4), 217224 (December).
Flaherty, C., Collins-Camargo, C., & Lee, E. (2008). Privatization of child welfare services:
Lessons learned from experienced states regarding site readiness assessment and
planning. Children and Youth Services Review, 30(7), 809820.
Freundlich, M. (2000). The market forces in adoption. The child welfare league of America
for the Evan B. Washington D.C: Donaldson Adoption Institute.
Hansen, M. E. (2009). Using private contracts to create adoptions from foster care. In S. J.
Quintero (Vol. Ed.), Child Welfare Issues and Perspectives. chapter 7. Child Welfare Issues
and Perspectives (pp. 113126). Inc. New York: Nova Science Publishers.
Hansen, M. E., & Hansen, B. A. (2002). An overview of federal adoption policy legislation: An
economic prospective. Shrewsbury, Ma: Center for Adoption Research, University of
17
Children and Youth Services Review 78 (2017) 917
Massachusetts.
Hansen, M. E., & Hansen, B. A. (2006). The economics of adoption of children from foster
care. Child Welfare, 85(3), 559583.
Hollingsworth, L. D., Bybee, D., Johnson, E. I., & Swick, D. C. (2010). A comparison of
caseworker characteristics in public and private foster care agencies. Children and
Youth Services Review, 32, 578584.
Katz, L. (1990). Eective permanency planning for children in foster care. Social Work,
35(3), 220226 (May.).
Landes, E. M., & Posner, R. A. (1978). The economics of the baby shortage. The Journal of
Legal Studies, 7(June), 323348.
National Data Archive on NDACAN (2013). Child abuse and neglect, adoption and foster care
analysis and reporting system (AFCARS). (Adoption http://www.ndacan.cornell.edu/
datasets/datasets-list-afcars.cfm and User Guide, 2000Present, see Http://www.
ndacan.cornell.edu/datasets/pdfs_user_guides/AFCARS_Guide_2000-Present.pdf Last
visited 5/25/2015).
Posner, R. (1987). The regulation of the market in adoption. Chicago Unbound: University of
Chicago Law School http://chicagounbound.uchicago.edu/cgi/viewcontent.cgi?
article=2903&context=journal_articles.
Posner, R. (1992). Sex and reason. Cambridge: Harvard University Press.
Selwyn, J., & Quinton, D. (2004). Stability, permanence, outcomes and support. Adoption
and Fostering, 28(4), 615.
Smith, B. D. (2003). After parental rights are terminated: Factors associated with exiting
foster care. Children and Youth Services Review, 25(12), 965985.
Snell, L. (2000). Child welfare reform and the role of privatization. Policy study no. 271.
Los Angeles, Ca: Reason Public Policy Institute.
Tides Center (2015). Foster care resource center. (See, http://www.
childrensactionnetwork.org/resources.html# Last visited 6/24/2015).
Triseliotis, J. (2002). Long-term foster care or adoption? The evidence examined.
Child & Family Social Work, 7(1), 2333.
U.S. Department of Health and Human Services (USDHHS) (2015a). Foster care statistics,
2013. Numbers and trends (April). https://www.childwelfare.gov/pubPDFs/foster.pdf
Last visited June 24, 2015.
U.S. Department of Health and Human Services (USDHHS) (2015b). Administration for
children and families, 2015. Age at Adoption Finalization https://www.acf.hhs.gov/
programs/cb.
Winston, P., et al. (2002). Privatization of Welfare Services: A Review of the Literature.
Washington D.C.: Mathematica Policy Research Inc (May. See, le:///C:/Users/
Simon%20Hakim/AppData/Local/Microsoft/Windows/INetCache/IE/PG084803/
privatization.pdf Last visited 4/23/17).