State Contexts and Foster Care Adoption Rates

Abstract

Prior research has examined associations between state adoption policies and assisted reproductive technology insurance coverage and foster care adoption rates, but knowledge of the relationships between state policies and contexts and foster care adoption is still limited. In this study, we test adoption subsidy policy, alternative means of family formation, and the demographic characteristics of potential adoptive parents and children as predictors of foster care adoption rates at the state-year level using data from the Adoption and Foster Care Analysis and Reporting System and other sources for 2005 to 2016. We use between-within models to obtain random effects estimates between states and fixed effects estimates within states. We find that states with higher average adoption subsidies have lower foster care adoption rates. Foster parent rates, international adoption rates, and mandated in vitro fertilization insurance coverage are positively associated with foster care adoption. States with higher median household incomes and more same-sex couples have higher foster care adoption rates, but states with more women of childbearing age have lower rates of older child adoption. These findings suggest some ways states may seek to increase adoptions from foster care.

Introduction

Each year in the United States, it is estimated that over 100,000 children are adopted (Jones & Placek, 2017) and about 2.3% of all children have adoptive parents (US Census Bureau, 2010). There are multiple types of adoptions, however, that arise from different child and parent circumstances. Adoptions from the foster care system account for more than half of those adoptions, with over 63,000 children adopted from foster care in 2018 (US Department of Health and Human Services, 2019). Yet, each year, there are more than 120,000 foster children waiting for adoptive homes and approximately 10,000 youth each year reach the age of majority without having obtained a permanent and legally recognized family (U.S. Department of Health and Human Services, 2017). The majority of adoptions from foster care occur within state, and it is the responsibility of states to identify and screen prospective adoptive parents. Thus, an examination of how state policy and demographic factors are associated with foster care adoption rates may help to understand adoption patterns and suggest possible opportunities for better identification of permanent homes for eligible children in foster care.

In this study, we investigate the associations of state policies and population characteristics with the rate of adoption from foster care. Prior work in the field of economics has produced important causal evidence on the role of adoption tax credits (Hansen, 2007) and other economic incentives for adoption (Brehm, 2018; Buckles, 2013). We examine a wider variety of state characteristics which are potentially associated adoption rates, although we are unable to test causal relationships. We focus on foster care adoption because the supply of adoptive families remains far below the number of children in foster care with a goal of adoption (Hansen, 2007) and there are very few established evidence-based programs for recruiting adoptive families for children in foster care or improving rates of adoption for older children (for an exception, see Vandivere, Malm, Zinn, Allen, & Mcklindon, 2015). In this study, we examine how state-level patterns of adoption subsidies, alternative means of family formation, demographic composition, and the composition of adoption-eligible foster children are associated with state foster care adoption rates over a 12-year period. In doing so, we contribute new insight into the ways that state factors are associated with adoption rates.

Background

Under federal and state laws, adoption is the preferred outcome for children in foster care who cannot safely reunify with their birth parents. The child welfare system is required to make efforts to achieve family reunification, except in egregious or extreme circumstances (Congress, 1980; US Department of Health and Human Services, 1997). However, in many cases, reunification cannot be achieved or sustained. In such cases, adoption is considered the optimal means of providing children with a permanent, legally- and socially-recognized family. However, for children in a foster care placement with a relative, guardianship is also an approved form of permanency.

There are two primary avenues for adoption from foster care. Foster parents, including kinship foster parents (relatives of the child in foster care), may pursue adoption after it is determined that the foster child in their care cannot be safely reunified. Alternatively, individuals seeking to form or expand a family may be less interested in the often-temporary nature of foster care placements, and instead search for adoption-eligible children whose current foster parents are not pursuing adoption. Identification of families for adoption-eligible children has been aided by the advent of state and national websites (e.g., adoptUSkids.org and similar state-specific versions) that provide profiles of eligible children and contact information for the child’s caseworker. Typically, children will reside with a prospective adoptive family in what is referred to as a “pre-adoptive placement” for a period of at least 6 months before the family seeks to finalize the adoption. The vast majority of children adopted from foster care are eligible for adoption subsidies that offset the care of the child, as well as ongoing medical coverage through Medicaid (US Department of Health and Human Services, 2011).

The Decision to Adopt

In addition to childbearing, international, private, and foster care adoption are all potential avenues for creating or growing a family. The perceived benefits of childbearing and the different forms of adoption are likely to vary. Individuals or couples may prefer to produce biological offspring for reasons including a desire to have a child who shares their traits, to experience pregnancy and childbirth, and to be part of the child’s earliest life experiences. To understand the role of state policies and demographics in foster care adoptions, it is helpful to first consider the circumstances under which individuals pursue adoption.

For some, altruism and connection to or sympathy for a child affect the adoption choice (Bethmann & Kvasnicka, 2012); this may be particularly important for foster parents deciding whether to adopt a child who is already in their care. However, the decision to adopt a child may be a rational choice undertaken after prospective parents weigh the costs and benefits of foster care adoption against available alternative means of family formation, including childbearing, foster care, and other adoption avenues. Of course, individuals face different constraints on their options, such as infertility, physical health constraints on biological parenthood, or limited financial means that inhibit private adoption. However, the circumstances that raise the costs of one avenue indirectly subsidize the others, and vice versa (Bethmann & Kvasnicka, 2012). State policies have immense power to raise or lower the costs of foster care adoption – in absolute terms and relative to childbearing and adoption alternatives.

State Policy and Context

Adoption subsidies are provided in order to avoid creating a financial disincentive for adoption among foster families. A foster family typically receives what is referred to as a reimbursement or maintenance payment that covers a portion of a child’s expenses. Thus, a low-income foster family interested in adoption could face financial hardship in the absence of subsidies. Subsidies may be especially relevant for helping families seeking to adopt children with serious health needs, or large sibling groups. Subsidies are typically limited to the amount of the foster care reimbursement payment for that child had they not been adopted. States (or sometimes counties) set minimum amounts, but amounts may vary by child age and disability status. Minimum amounts range from less than $200 per month in Utah to $1,000 per month in the District of Columbia (North American Council on Adoptable Children, 2016). A state’s average foster care subsidy rate will be affected by its base rate, as well as the proportion of children who have special needs. Thus, states with a higher average subsidy rate may have fewer adoptions, because it suggests that more of their children have special needs.

The availability of assisted reproductive technology (ART) is also likely relevant to adoption rates as it may allow more individuals to have biological children than would otherwise be able. Over 60,000 babies were conceived using in-vitro fertilization (IVF) in 2012, a nearly 50 percent increase from 2003 (Doucleff, 2014). However, IVF may be cost-prohibitive, at approximately $12,400 a cycle with a 33 percent chance of success (Doucleff, 2014). Over 15 percent of women who wish to become pregnant struggle with infertility (Pal, 2018) and some evidence suggests that infertility is positively related to adoption (Cohen & Chen, 2010; Gumus & Lee, 2012). Gumus & Lee (2012) found that the use of ART is negatively associated with adoption rates, but do not examine policies influencing ART access. If individuals pursue adoption due to infertility, then policies that subsidize ARTs may reduce adoptions, particularly among middle or lower-income women who would otherwise find ART cost-prohibitive. However, Cohen and Chen (2010) found that state-level changes in IVF cycles were positively associated with rates of foster care adoption. Regardless, ART availability merits consideration as it may alter family expansion strategies.

Finally, policies which impact the costs of different adoption avenues influence the relative costs of adoption from foster care. There are far fewer constraints to foster care adoption than domestic private or international adoption. First, it is far less expensive to adopt from foster care than to pursue private domestic or international adoption, because states typically cover or reimburse the costs (Moriguchi, 2012) in addition to providing adoption subsidies (Malm, Vandivere, & McKlindon, 2011). In contrast, the estimated average costs range from $5,000 to $40,000 for a private domestic adoption, and between $15,000 and $30,000 for an international adoption (Moriguchi, 2012), though these costs are partially offset by federal adoption tax credits. Thus, higher-income families that are interested in adoption may select among multiple options, whereas families of limited financial means may find foster care adoption to be the only affordable option.

Aside from financial cost, families may face a variety of other barriers in private and international adoption. In private domestic adoption, the biological parent may select the adoptive family; in international adoption, the family must meet the criteria laid out by the source country. In both cases, prospective adoptive parents who are unmarried, older, or have health issues may be unable to adopt privately. In contrast, the criteria for adopting from foster care focus largely on safety and commitment to the child, though some health-related criteria apply there as well (Child Welfare Information Gateway, 2016). Moreover, because there is no supply constraint of children awaiting adoption from foster care, state systems cannot afford to erect additional barriers for prospective parents.

In contrast to foster care adoption, the access to private domestic and international adoption has become increasingly constrained. Domestically, as teen birth rates have declined and contraception has become readily available (Kearney & Levine, 2015), fewer of the babies who would otherwise be placed for adoption are born. Additionally, as single motherhood is more socially accepted, unwed mothers face less social pressure to voluntarily relinquish their babies. In the late 1960s, nearly nine percent of children born to never-married mothers were placed for adoption, but less than one percent of these children were relinquished by the mid-1990s (Chandra, Abma, Maza, & Bachrach, 1999). Thus, prospective adoptive parents may wait years for an adoptive placement domestically – especially for a healthy child. Shorter waiting periods affect the decision to pursue international adoption over private domestic adoptions (L. Hollingsworth & Ruffm, 2002; Ishizawa & Kubo, 2014). Although there remains a large supply of children in need of adoptive homes worldwide due to war, humanitarian crises, abandonment, and other factors, source and destination countries have made the process increasingly difficult. In part due to concerns about corruption, several source countries—countries with an oversupply of children without families—have restricted or completely banned adoptions from foreign parents (Voight & Brown, 2013) and international standards intended to improve standards and reduce corruption have been difficult for low-income countries to meet (Montgomery & Powell, 2018). Other countries have been reasonably effective in improving domestic adoptions. On the whole, constraints in access have drastically reduced international adoptions from 23,000 in 2004 to less than 5,000 in 2017 (U.S. Department of State, 2017). Contrasting the decline in children eligible for private domestic or international adoption, the number of children waiting for an adoptive home through the U.S. foster care system has declined at a much lower rate. In 2000, there were 131,000 children waiting for adoption, and in 2017, there were 123,000 (US Department of Health and Human Services, 2006, 2018).

Although foster care adoption is more widely accessible and far less costly than private or international adoption, it poses a variety of unique challenges. The characteristics of adoptable foster children are different – many are school age and have disabilities. They may also have siblings that need to be placed with them or want to remain in contact with birth parents or other biological relatives; adoptions from foster care are increasingly “open adoptions,” meaning there is shared information and potentially regular contact with the child’s birth parents. Although open adoptions may benefit children, they also may deter prospective adoptive parents who want a more traditional family arrangement. In sum, prospective adoptive parents weigh numerous factors when considering foster care adoption, and state policies have immense potential to influence these factors.

State Demographics

In addition to official policies, the demographic characteristics of potential parents and of adoption-eligible children may have implications for foster care adoption rates. First, income patterns determine the number of households with the resources to adopt and influence the type of adoption parents choose to pursue (Ishizawa & Kubo, 2014). Additionally, the age and sex structure of the state likely inform both the number of parents looking to form or expand families and the number of women facing infertility. Although studies indicate that one-third of adult women are interested in adoption, approximately 10% of those who express interest take steps toward adoption (such as obtaining a home study), and an even smaller proportion ultimately adopt a child (Macomber, 2005). Indeed, few heterosexual couples pursue adoption unless they encounter fertility problems; even then, a large majority pursue medical fertility treatment rather than adoption (L. D. Hollingsworth, 2000). Moreover, many who would consider adoption in general are not interested in adopting older children or children with severe disabilities, two populations overrepresented in the foster care system (Brooks, James, & Barth, 2002; Macomber, 2005; Zhang & Lee, 2011).

Same-sex couples adopt at much higher rates than heterosexual couples and are more likely to adopt across racial and ethnic groups and to adopt children with special needs (Gates, Badgett, Macomber, & Chambers, 2007). However, same-sex couples also face greater barriers to adoption, both through explicit policies under which religious nonprofits can decline to approve otherwise suitable same-sex couples for adoption, as well as subtler forms of institutional discrimination (Gates et al., 2007; Messina & D’Amore, 2018; Perrin, Hurley, Mattern, Flavin, & Pinderhughes, 2019).

Prospective parents who place fewer conditions on the characteristics of children they would consider for adoption are particularly important for states to recruit. The odds that a foster child will be adopted depend on a variety of factors, including the age, race, and sex of the child, degree of disability, and behavioral and mental health (Akin, 2011; Connell, Katz, Saunders, & Tebes, 2006; Skidmore, Anderson, & Eiswerth, 2016; Snowden, Leon, & Sieracki, 2008; US Department of Health and Human Services, 2011). Younger children are adopted more frequently, and some research has found adoption advantages for girls as well as White, Asian, and Hawaiian/Pacific Islander children (Akin, 2011; Snowden et al., 2008; US Department of Health and Human Services, 2011). The research on the relevance of physical disabilities and mental health problems is mixed, with some research finding both higher and lower rates of adoption for children with these characteristics (Akin, 2011; Connell et al., 2006; US Department of Health and Human Services, 2011). States have very different foster care populations, reflecting differences in willingness to remove children, evidentiary standards for removal, rates of reunification, use of informal placements, and other factors. States with higher proportions of children who are older, of a racial minority group, or male in their foster care systems may have lower adoption rates, as these children face additional barriers to adoption.

The Current Study

The current study will explore how state adoption subsidies, alternative means of family formation, demographic composition of potential parents, and demographic composition of the foster care population are associated with foster care adoption. The results will illuminate the characteristics of states that are positively and negatively related to foster care adoptions, and therefore the likelihood of children in the foster care system finding permanent homes.

Data and Methods

Data

In order to broadly examine state policies and populations, we utilized data from a wide variety of sources. Many key measures were taken from the Adoption and Foster Care Analysis and Reporting System (AFCARS) (Children’s Bureau, 2017). AFCARS collects case information on all children who have been placed in foster care or adopted with the involvement of a state agency. Little information is collected on children, but the dataset contains information on children’s history with the child welfare system, availability for adoption, and basic demographic information. Since reports to AFCARS are federally mandated, the data can also be used to assess child welfare populations at the state level. All remaining measures were taken from publicly available data sources, which will be discussed in greater detail in the Measures section.

The unit of analysis was the state-year, and the sample contained 51 states, including the District of Columbia, for 12 years from 2005 to 2016. The total sample comprised 612 state-years.

Measures

Dependent variables

Our dependent variable was the total number of non-relative foster care adoptions per state-year. We excluded kinship adoptions as the decision to adopt a relative is likely based on the specific child, and therefore related adoptive parents do not represent a state-level resource. Non-relative adoptions include adoptions from foster care. We included only adoptions in which adoptive children and parents resided in the same state. We excluded adoptions of Native American or Alaskan Native children, as data on these adoptions are unreliable, as well as adoptions in which children were placed by private agencies (1.2 percent of cases). We also used measures of the number of children three or younger who were adopted by non-relatives, as well as the number of older children. All dependent variables were measured using AFCARS data.

To account for differences in the number of children potentially eligible for adoption (which varies across state and year), we calculated foster care adoption rates as a function of both total children in the state and total children who were eligible to be adopted from foster care. The total child population was taken from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program data, and the number of adoption-eligible children in non-relative foster care was calculated using AFCARS data. Both denominators also excluded Native American and Alaskan Native children.

Independent variables

We included a number of measures of policies and demographic characteristics predicting demand for foster care adoptions. First, in order to evaluate adoption subsides, we measured the percent of adoptive families receiving adoption subsidies, ranging from 5 to 100, and the average adoption subsidy in hundreds of dollars, ranging from 1 to 30, using AFCARS adoption data. We adjusted the average adoption subsidy for inflation, standardizing to 2016 dollars.

To assess alternative means of family formation, we included a binary measure of mandated IVF coverage using information from the National Conference of State Legislatures (NCSL) (2014), which tracks state laws on numerous issues. IVF coverage is equal to 1 if states require health insurance plans to cover in vitro fertilization treatments for infertile women and 0 otherwise. We also calculated the number of foster families per 1,000 households using data from the American Community Survey (ACS) (U.S. Census Bureau, 2017), which collects annual information on the United States population. Lastly, we included both the international (per 100,000 households) and private (per 1,000 households) adoption rate as alternative family formation methods. Data on international adoptions was taken from the U.S. Department of State, Bureau of Consular Affairs (2017) and data on unrelated private adoptions came from the National Council for Adoption (NCA) (Jones & Placek, 2017). Data on unrelated private adoptions were not available for every year, so we included the rate of private adoptions only in 2007.

We analyzed the demographic composition of potential adoptive parents using three variables. We measured the state’s median household income, standardized to 2016 dollars, using data from the Current Population Survey (CPS) (United States. Bureau of the Census & United States. Bureau of Labor Statistics, 2018). We also included a measure of women aged 15–44 as a percentage of the total population using data from the ACS. Last, we calculated the number of same-sex couples per 1,000 households using ACS data.

We assessed the final group of independent variables, demographic composition of adoption-eligible foster children, using the percentage of adoption-eligible children in non-relative foster care who were Black, the mean age of children entering non-relative foster care, the percent of children in foster care who were placed with relatives, and the number of children entering foster care per 1,000 children, all calculated using AFCARS data. We also included year fixed effects as well as a control variable for the regional price parity, taken from the Bureau of Economic Analysis (BEA), in all models.

Supplementary Variables

In order to test the robustness of our results, we utilized several additional variables in supplementary analyses. We used a variable measuring the minimum foster care subsidy, adjusted for inflation, in each state-year, as well a measure of the average adoption subsidy as a percentage of the median household income, both calculated using AFCARS data. In addition, we tested a measure of whether a state has an explicit policy allowing adoption discrimination against same-sex couples, using data taken from familyequality.org and the Movement Advancement Project. We utilize an alternative exposure variable of the number of households per state-year as well as an indicator of the percent of children in foster care who were reunified with their birth families in the prior year.

Methods

To estimate the impact of state characteristics on foster care adoption rates, we used between-within models (Allison, 2009), a hybrid approach which combines fixed effects and random effects models and allows for the inclusion of both static and time-varying variables. These models allowed us to examine how both differences between states and differences within states over time predicted adoption rates, as well as include IVF coverage mandates and 2007 private adoption rates as predictors, as these variables did not vary within states over the study period. We calculated the mean and deviation across states for each time-varying variable for each state-year and used both as predictors in all models. We used negative binomial regression models because the dependent variable, number of adoptions from foster care, is an overdispersed count measure. Random effects models were used to obtain correct standard errors. In the between-within models, mean coefficients reveal how differences between states are associated with the outcome. Deviation coefficients are fixed effects estimates which control for stable state characteristics, revealing the ways changes within states are associated with the outcome. Including both measures allows us to obtain greater understanding of the relationships between state characteristics and foster care adoption rates.

We estimated six main models with variations in the dependent variable and the exposure population. Model 1 estimates the rate of FC (foster care) adoptions per child in the population (total FC adoption rate), while Model 2 estimates the rate of adoptions per adoption-eligible child who is not in relative care (eligible-child FC adoption rate). Models 3 through 6 estimate these models by age group. Models 3 and 4 estimate adoption rates for children ages 0–3 and 4–18, respectively, per child in the population. Models 5 and 6 estimate adoption rates for children ages 0–3 and 4–18 per adoption-eligible child.

We also tested several supplementary models. The first uses all independent variables as predictors of the rate of adoption per household, in order to match household-level independent variables. The second and third models test the minimum foster care subsidy and the average adoption subsidy as a percentage of median household income in place of the average subsidy amount, respectively. The fourth supplementary model includes the measure of state policy allowing same-sex couple adoption discrimination. The final supplementary model includes the percentage of children in foster care reunified with their birth families in the previous year. Supplementary models 2 through 5 predict total foster care adoptions per child in the population.

Data were unweighted as the sample includes all states and the District of Columbia for the years 2005–2016. We used multiple imputation with chained equations to handle missing data.

Results

Sample Description

Descriptive statistics for all variables are displayed in Table 1. Total foster care adoptions ranged widely from one to 4,467 by state-year, with a mean of 576. The rate of foster care adoptions per 1,000 children ranged from zero to 2.32 with a mean of 0.52. Adoptions of very young children (aged 0–3) were more common, with a mean of 252 and an average rate of 0.91 adoptions per 1,000 children aged 0–3.

Table 1.

Descriptive Statistics and Sources

	Mean	SD	Source
Dependent Variables
Non-Relative Public Adoptions (Total)	621.23	691.50	AFCARS
Rate per 1000 Children	0.52	0.35	AFCARS
Non-Relative Public Adoptions (0–3)	251.52	318.72	AFCARS
Rate per 1000 Children	0.91	0.55	AFCARS
Non-Relative Public Adoptions (4+)	370.04	397.77	AFCARS
Rate per 1000 Children	0.41	0.29	AFCARS
Independent Variables
Percent of Adoptive Families Receiving Subsidies	90.01	13.83	AFCARS
Average Adoption Subsidy in Hundreds	7.02	2.77	AFCARS
Mandated Insurance Coverage of IVF	0.22	0.41	NCSL
Foster Families per 1000 Households	2.37	0.98	ACS
International Adoptions per 100,000 Households	11.14	7.15	Dept. of State
Private Adoptions per 1000 Households	0.32	0.23	NCA
Median Household Income in Thousands	64.18	11.01	CPS
Women Aged 15–44 as a Percent of the Population	20.01	1.26	ACS
Same-Sex Couples per 1000 Households	6.06	2.60	ACS
Percent of Non-Relative Adoption-Eligible Children who are Black	30.97	19.52	AFCARS
Mean age at Entry into Non-Relative Foster Care	8.78	1.68	AFCARS
Percent of Foster Children who are in Relative Care	28.10	11.22	AFCARS
Children Entering Foster Care per 1000 Children	40.67	17.07	AFCARS
Regional Price Parity	97.44	8.73	BEA
Exposure Variables
Child Population (Total)	1,421,168	1,643,901	SEER
Child Population (0–3)	307,854	360,628	SEER
Child Population (4+)	1,113,314	1,283,621	SEER
Adoption Eligible Children (Not in Relative Care) (Total)	2,669.81	3,234.60	AFCARS
Adoption Eligible Children (Not in Relative Care) (0–3)	870.97	1,208.62	AFCARS
Adoption Eligible Children (Not in Relative Care) (4+)	1,798.93	2,093.11	AFCARS
Supplementary Variables
Minimum Foster Care Subsidy in Hundreds	1.53	1.06	AFCARS
Average Adoption Subsidy Percent of Median Income	13.26	4.97	AFCARS
Same-Sex Adoption Discrimination Policy	0.13	0.33	Various
Number of Households	2,251,554	2,374,497	ACS
Percent of Children Reunified in Prior Year	19.37	6.65	AFCARS

Most adoptive families received subsidies, with a mean of 90 percent, and the average subsidy was 702 dollars. IVF insurance coverage was mandated in 22 percent of state-years, including all years for Arkansas, Connecticut, Hawaii, Illinois, Maryland, Massachusetts, Montana, New Jersey, Ohio, Rhode Island, and West Virginia. IVF coverage mandates were not correlated with any variable above 0.25. The average state-year had few foster families, around 2.4 per 1,000 households. International and private adoptions were also both fairly rare, averaging 11.1 adoptions per 100,000 households and 0.32 adoptions per 1,000 households, respectively. Median household income averaged 64,180 a year. Women of childbearing age made up 20 percent of the population, on average, and there were 6.06 same-sex couples per 1,000 households. On average, 31 percent of adoption-eligible children were Black. The average child entered non-relative foster care at age 8.8 and 28 percent of foster children were placed with relatives.

The mean child population was around 1.42 million, although this figure varied widely from about 100,000 to 9.2 million. On average, about 2,670 children not in relative care were available for adoption in a state-year, about a third of whom were three years old or younger.

Model Results

Primary model results are displayed in Table 2. Coefficients are reported in incidence-rate ratios (IRRs). IRRs greater than one indicate that a one unit increase in the independent variable is predictive of a ([β – 1]*100) increase in the FC adoption rate. IRRs less than one indicate that a one unit increase in the independent variable is predictive of a corresponding decrease in the FC adoption rate.

Table 2.

Negative Binomial Regression Models Predicting Non-Relative Foster Care Adoptions

	Rate per Child in Population (Model 1)		Rate per Child Eligible for Adoption (Model 2)
	IRR	95% CI	IRR	95% CI
% of Adoptive Families Receiving Subsidies
Mean	0.986	[0.967, 1.005]	0.985	[0.965, 1.005]
Deviation	1.000	[0.998, 1.002]	1.000	[0.999, 1.002]
Average Adoption Subsidy in Hundreds
Mean	0.862***	[0.790, 0.940]	0.890*	[0.811, 0.978]
Deviation	0.979**	[0.966, 0.993]	1.000	[0.987, 1.013]
Mandated Insurance Coverage of IVF	1.759**	[1.218, 2.541]	2.009***	[1.372, 2.941]
Foster Families per 1000 Households
Mean	1.179	[0.864, 1.609]	1.247	[0.887, 1.751]
Deviation	1.044***	[1.018, 1.071]	1.039**	[1.015, 1.063]
International Adoptions per 100,00 Households
Mean	0.972	[0.917, 1.031]	1.023	[0.953, 1.098]
Deviation	1.011***	[1.005, 1.017]	1.002	[0.996, 1.007]
Private Adoptions per 1000 Households (2007)	1.492	[0.754, 2.951]	1.369	[0.642, 2.919]
Median Household Income in Thousands
Mean	1.015	[0.983, 1.047]	0.998	[0.962, 1.035]
Deviation	1.009**	[1.003, 1.015]	1.000	[0.994, 1.006]
Women Aged 15–44 as a % of the Population
Mean	0.820	[0.644, 1.044]	0.868	[0.684, 1.101]
Deviation	0.965	[0.924, 1.009]	0.999	[0.957, 1.042]
Same-Sex Couples per 1000 Households
Mean	1.514***	[1.327, 1.728]	1.401***	[1.226, 1.600]
Deviation	0.992	[0.968, 1.017]	1.002	[0.979, 1.026]
% of Eligible Children who are Black
Mean	0.994	[0.982, 1.006]	0.981**	[0.969, 0.993]
Deviation	0.997	[0.991, 1.004]	0.996	[0.991, 1.002]
Mean Age at Entry into Non-Relative Foster Care
Mean	1.119	[0.962, 1.301]	1.225*	[1.045, 1.436]
Deviation	0.958*	[0.926, 0.991]	0.963*	[0.933, 0.993]
% of Foster Children who are in Relative Care
Mean	1.030**	[1.012, 1.048]	1.051***	[1.031, 1.071]
Deviation	0.995*	[0.990, 0.999]	0.994**	[0.990, 0.998]
Children Entering Foster Care per 1000 Children
Mean	1.025***	[1.011, 1.038]	1.008	[0.995, 1.023]
Deviation	1.004***	[1.002, 1.006]	0.994***	[0.992, 0.996]
Regional Price Parity
Mean	0.924***	[0.884, 0.965]	0.933**	[0.888, 0.980]
Deviation	0.976**	[0.958, 0.994]	0.965***	[0.949, 0.981]
ln_r	1.478	[0.975, 2.240]	1.252	[0.834, 1.880]
ln_s	17.649***	[10.319, 30.187]	13.002***	[7.517, 22.490]
Observations	612		612

Exponentiated coefficients;

^*p<0.05

^**p<0.01

^***p<0.001;

Year fixed effects included in models

Adoption subsidies.

The proportion of adoptive families receiving adoption subsidies was not significantly related to FC adoption rates. However, the mean amount of the adoption subsidy was negatively associated with FC adoptions, such that a state with a $100 larger average adoption subsidy had a 14 percent lower total FC adoption rate and an 11 percent lower eligible-child FC adoption rate. When states increased their average adoption subsidy by $100, their total FC adoption rates were estimated to decline by 2.1 percent.

Alternative means of family formation.

IVF insurance coverage mandates were associated with significant increases in adoption rates: 76 percent higher for total FC and 101 percent higher for eligible child FC adoption rates. Additionally, net of other variables, states with an increase of 1 foster family per 1,000 households between years were estimated to have four percent higher total FC and eligible FC adoption rates. States which experienced an increase of one international adoption per 100,000 households had 1.1 percent higher total FC adoption rates. Conversely, private adoption rates in 2007 were not related to adoption rates.

Demographic composition of potential adoptive parents.

States with an increase in median household income had higher total FC adoption rates; a $1,000 increase in state median household income was associated with a one percent increase. Lastly, states with more identified same-sex couples had 51 percent higher total FC adoption rates and 40 percent higher eligible-child FC adoption rates. However, a within-state increase of one same-sex couple per 1,000 households was not associated with FC adoption rates.

Demographic composition of adoption-eligible foster children.

Finally, the demographic composition of children available for adoption influenced foster care adoption rates. The proportion of adoption eligible children who were Black was not significantly associated with total FC adoption rates, but states with a one PP higher proportion of Black adoption-eligible children had two percent lower eligible-child FC adoption rates. States with an older mean age at non-relative foster care entry had 23 percent higher eligible-child FC adoption rates, but a one year increase in the average age entry predicted a four percent decrease in total and eligible-child FC adoption rates. States with one PP more children placed in relative care had seven to eight percent lower FC adoption rates, while a one PP increase in relative placement predicted approximately 0.5 percent lower rates. The rate of children entering foster care was positively associated with total FC adoption rates both between and within states, while an increase in foster entry rates was negatively associated with eligible-child FC adoption rates. Finally, higher regional price parities predicted lower total and eligible FC adoption rates both between and within states.

Age-specific models.

Table 3 presents models (3–6) of total and eligible-child FC adoption rates for children ages 0–3 and 4+. Patterns of coefficients and statistical significance were quite similar to the models for all ages, though it is worth noting that states with one PP more women between the age of 15 and 44 had a 20.8 percent lower total FC adoption rate of children ages four and older.

Table 3.

Negative Binomial Regression Models Predicting Non-Relative Foster Care Adoptions by Age Group.

	Rate per Child in Population				Rate per Child Eligible for Adoption
	Ages 0–3 (Model 3)		Ages 4+ (Model 4)		Ages 0–3 (Model 5)		Ages 4+ (Model 6)
	IRR	95% CI	IRR	95% CI	IRR	95% CI	IRR	95% CI
% of Adoptive Families Receiving Subsidies
Mean	0.988	[0.969, 1.007]	0.982	[0.964, 1.000]	0.983	[0.963, 1.002]	0.984	[0.964, 1.004]
Deviation	1.000	[0.998, 1.002]	1.000	[0.999, 1.002]	1.000	[0.998, 1.002]	1.001	[0.999, 1.003]
Average Adoption Subsidy in Hundreds
Mean	0.908*	[0.835, 0.987]	0.870***	[0.802, 0.945]	0.936	[0.857, 1.022]	0.883**	[0.807, 0.966]
Deviation	0.969***	[0.954, 0.985]	0.981*	[0.967, 0.996]	0.992	[0.978, 1.006]	1.000	[0.987, 1.015]
Mandated Insurance Coverage of IVF	1.468*	[1.006, 2.143]	1.860**	[1.278, 2.708]	1.477	[0.995, 2.193]	2.124***	[1.414, 3.192]
Foster Families per 1000 Households
Mean	1.138	[0.828, 1.564]	1.246	[0.910, 1.706]	1.124	[0.800, 1.580]	1.248	[0.881, 1.768]
Deviation	1.054***	[1.024, 1.084]	1.037**	[1.010, 1.066]	1.040**	[1.014, 1.067]	1.036**	[1.010, 1.062]
International Adoptions per 100,000 Households
Mean	1.006	[0.948, 1.068]	0.949	[0.896, 1.006]	1.048	[0.977, 1.124]	0.981	[0.914, 1.052]
Deviation	1.014***	[1.007, 1.021]	1.010***	[1.004, 1.017]	1.004	[0.997, 1.010]	1.001	[0.995, 1.007]
Private Adoptions per 1000 Households (2007)	1.483	[0.728, 3.022]	1.456	[0.734, 2.890]	0.889	[0.411, 1.924]	1.524	[0.706, 3.289]
Median Household Income in Thousands
Mean	1.000	[0.967, 1.034]	1.012	[0.982, 1.044]	1.004	[0.969, 1.040]	0.998	[0.963, 1.035]
Deviation	1.010**	[1.003, 1.018]	1.011**	[1.004, 1.017]	1.000	[0.994, 1.006]	1.002	[0.995, 1.008]
Women Aged 15–44 as a % of the Population
Mean	0.862	[0.687, 1.083]	0.792*	[0.630, 0.996]	0.969	[0.779, 1.206]	0.836	[0.658, 1.062]
Deviation	0.961	[0.915, 1.010]	0.965	[0.921, 1.012]	1.016	[0.970, 1.064]	0.994	[0.949, 1.041]
Same-Sex Couples per 1000 Households
Mean	1.384***	[1.213, 1.578]	1.523***	[1.339, 1.732]	1.290***	[1.127, 1.478]	1.378***	[1.208, 1.572]
Deviation	0.990	[0.963, 1.019]	0.991	[0.966, 1.017]	1.010	[0.984, 1.037]	0.999	[0.974, 1.025]
% of Eligible Children who are Black
Mean	0.996	[0.984, 1.008]	0.995	[0.984, 1.007]	0.984**	[0.972, 0.995]	0.984**	[0.972, 0.996]
Deviation	0.995	[0.988, 1.002]	1.001	[0.995, 1.007]	0.999	[0.993, 1.006]	0.997	[0.991, 1.003]
Mean Age at Entry into Non-Relative Foster Care
Mean	1.029	[0.886, 1.194]	1.126	[0.967, 1.312]	1.122	[0.963, 1.308]	1.227*	[1.044, 1.444]
Deviation	0.933***	[0.899, 0.969]	0.977	[0.944, 1.012]	0.964*	[0.932, 0.996]	0.970	[0.938, 1.003]
% of Foster Children who are in Relative Care
Mean	1.023*	[1.005, 1.042]	1.024**	[1.006, 1.043]	1.049***	[1.030, 1.068]	1.043***	[1.024, 1.063]
Deviation	0.995*	[0.990, 1.000]	0.994*	[0.990, 0.999]	0.998	[0.993, 1.002]	0.993**	[0.989, 0.998]
Children Entering Foster Care per 1000 Children
Mean	1.030***	[1.016, 1.044]	1.021**	[1.008, 1.034]	1.019*	[1.004, 1.034]	1.007	[0.993, 1.022]
Deviation	1.005***	[1.002, 1.007]	1.003**	[1.001, 1.005]	0.995***	[0.993, 0.997]	0.994***	[0.992, 0.996]
Regional Price Parity
Mean	0.953*	[0.910, 0.999]	0.925***	[0.886, 0.965]	0.938*	[0.893, 0.985]	0.947*	[0.902, 0.993]
Deviation	0.976*	[0.956, 0.997]	0.974**	[0.956, 0.992]	0.959***	[0.942, 0.976]	0.969***	[0.953, 0.986]
ln_r	1.926**	[1.229, 3.019]	1.739**	[1.145, 2.644]	1.692*	[1.103, 2.594]	1.448	[0.952, 2.203]
ln_s	10.877***	[6.249, 18.934]	13.876***	[8.198, 23.484]	8.107***	[4.770, 13.778]	10.254***	[5.879, 17.884]
Observations	612		612		612		612

Exponentiated coefficients;

^*p<0.05

^**p<0.01

^***p<0.001;

Year fixed effects included in models

Supplementary models.

Our supplementary models are displayed in Table 4. The model predicting adoptions per household was substantively very similar to Model 1. In fact, none of our main results were significantly altered in the supplementary models. We did find that a $100 higher minimum foster care subsidy between states predicted a 67 percent increase in the total FC adoption rate. States with an average adoption subsidy which was a higher percentage of the median household income had 6.7 percent lower total FC adoption rates and an increase in the subsidy to income ratio also predicted a one percent decrease in total FC adoptions. Although a greater number of same-sex couples between states remained positively associated with total FC adoption rates, changes to policies allowing adoption discrimination against same-sex couples predicted a 16.5 percent increase in total FC adoption rates. Finally, states with one PP more foster children reunified with their birth families had 5.3 percent higher total FC adoption rates.

Table 4.

Supplementary Models

	Adoptions per Household		Minimum Foster Care Subsidy		Adoption Subsidy % of Income		Same-Sex Adoption Discrimination		% of Children Reunified in Prior Year
	IRR	95% CI	IRR	95% CI	IRR	95% CI	IRR	95% CI	IRR	95% CI
% of Adoptive Families Receiving Subsidies
Mean	0.984	[0.965, 1.003]	0.995	[0.974, 1.015]	0.984	[0.965, 1.003]	0.984	[0.965, 1.003]	0.984	[0.966, 1.003]
Deviation	1.000	[0.999, 1.002]	1.001	[0.999, 1.002]	1.000	[0.998, 1.002]	1.000	[0.999, 1.002]	1.000	[0.999, 1.002]
Average Adoption Subsidy in Hundreds
Mean	0.850***	[0.777, 0.928]	0.842***	[0.771, 0.919]	---	--- ---	0.876**	[0.802, 0.957]	0.911*	[0.834, 0.996]
Deviation	0.977**	[0.963, 0.992]	0.981**	[0.968, 0.995]	---	--- ---	0.985*	[0.972, 0.999]	0.981**	[0.967, 0.994]
Minimum Foster Care Subsidy in Hundreds
Mean	---	--- ---	1.668***	[1.418, 1.961]	---	--- ---	---	--- ---	---	--- ---
Deviation	---	--- ---	0.987	[0.962, 1.012]	---	--- ---	---	--- ---	---	--- ---
Avg Adoption Subsidy % of Median Income
Mean	---	--- ---	---	--- ---	0.933**	[0.891, 0.977]	---	--- ---	---	--- ---
Deviation	---	--- ---	---	--- ---	0.991*	[0.983, 0.998]	---	--- ---	---	--- ---
Same-Sex Couple Adoption Discrimination
Mean	---	--- ---	---	--- ---	---	--- ---	0.874	[0.456, 1.673]	---	--- ---
Deviation	---	--- ---	---	--- ---	---	--- ---	1.165***	[1.097, 1.238]	---	--- ---
% of Children Reunified in Prior Year
Mean	---	--- ---	---	--- ---	---	--- ---	---	--- ---	1.053***	[1.025, 1.081]
Deviation	---	--- ---	---	--- ---	---	--- ---	---	--- ---	1.001	[0.995, 1.007]
Mandated Insurance Coverage of IVF	1.628**	[1.126, 2.356]	1.719**	[1.177, 2.512]	1.842**	[1.277, 2.659]	1.533*	[1.046, 2.246]	1.888***	[1.306, 2.729]
Foster Families per 1000 Households
Mean	1.095	[0.801, 1.497]	1.145	[0.835, 1.570]	1.197	[0.877, 1.632]	1.125	[0.817, 1.550]	1.277	[0.941, 1.733]
Deviation	1.042**	[1.015, 1.069]	1.048***	[1.023, 1.073]	1.044***	[1.018, 1.071]	1.044***	[1.019, 1.070]	1.043**	[1.017, 1.070]
International Adoptions per 100,000 Households
Mean	0.968	[0.913, 1.027]	1.002	[0.944, 1.065]	0.974	[0.918, 1.033]	0.966	[0.909, 1.027]	0.958	[0.904, 1.016]
Deviation	1.013***	[1.007, 1.019]	1.011***	[1.006, 1.017]	1.011***	[1.005, 1.017]	1.012***	[1.006, 1.018]	1.010***	[1.005, 1.016]
Private Adoptions per 1000 Households (2007)	1.359	[0.685, 2.698]	1.235	[0.609, 2.505]	1.519	[0.770, 2.996]	1.507	[0.757, 3.001]	1.717	[0.873, 3.377]
Median Household Income in Thousands
Mean	1.015	[0.984, 1.047]	1.006	[0.974, 1.039]	1.000	[0.967, 1.034]	1.011	[0.979, 1.044]	1.019	[0.987, 1.052]
Deviation	1.009**	[1.002, 1.015]	1.007*	[1.000, 1.013]	1.007*	[1.001, 1.014]	1.008*	[1.001, 1.014]	1.009**	[1.003, 1.016]
Women Aged 15–44 as a % of the Population
Mean	0.862	[0.675, 1.101]	0.850	[0.658, 1.099]	0.832	[0.652, 1.060]	0.768*	[0.603, 0.979]	0.845	[0.664, 1.074]
Deviation	0.967	[0.924, 1.011]	0.962	[0.920, 1.006]	0.965	[0.923, 1.009]	0.969	[0.929, 1.011]	0.965	[0.923, 1.008]
Same-Sex Couples per 1000 Households
Mean	1.420***	[1.245, 1.619]	1.412***	[1.232, 1.617]	1.515***	[1.327, 1.729]	1.534***	[1.340, 1.755]	1.517***	[1.331, 1.731]
Deviation	0.991	[0.966, 1.016]	0.987	[0.966, 1.010]	0.993	[0.968, 1.018]	0.996	[0.972, 1.020]	0.993	[0.969, 1.018]
% of Eligible Children who are Black
Mean	0.993	[0.981, 1.005]	0.996	[0.983, 1.008]	0.994	[0.982, 1.006]	0.997	[0.986, 1.009]	1.000	[0.988, 1.012]
Deviation	0.997	[0.990, 1.003]	0.996	[0.990, 1.002]	0.998	[0.992, 1.004]	0.996	[0.990, 1.002]	0.998	[0.992, 1.004]
Mean Age at Entry into Non-Relative Foster Care
Mean	1.106	[0.951, 1.287]	1.046	[0.893, 1.225]	1.123	[0.964, 1.307]	1.115	[0.958, 1.297]	1.058	[0.909, 1.231]
Deviation	0.955*	[0.923, 0.989]	0.958*	[0.927, 0.990]	0.957*	[0.925, 0.990]	0.958*	[0.928, 0.990]	0.950**	[0.918, 0.984]
% of Foster Children who are in Relative Care
Mean	1.031***	[1.013, 1.049]	1.023*	[1.004, 1.041]	1.031***	[1.013, 1.050]	1.040***	[1.021, 1.059]	1.022*	[1.004, 1.041]
Deviation	0.994*	[0.990, 0.999]	0.994**	[0.989, 0.998]	0.994*	[0.990, 0.999]	0.993**	[0.988, 0.997]	0.993**	[0.988, 0.997]
Children Entering Foster Care per 1000 Children
Mean	1.025***	[1.012, 1.039]	1.031***	[1.017, 1.046]	1.024***	[1.011, 1.038]	1.028***	[1.014, 1.041]	1.021**	[1.008, 1.034]
Deviation	1.003**	[1.001, 1.006]	1.003**	[1.001, 1.005]	1.004***	[1.002, 1.006]	1.003**	[1.001, 1.005]	1.004***	[1.002, 1.006]
Regional Price Parity
Mean	0.936**	[0.897, 0.978]	0.942**	[0.902, 0.985]	0.921***	[0.881, 0.963]	0.922***	[0.882, 0.963]	0.922***	[0.883, 0.962]
Deviation	0.981*	[0.962, 1.000]	0.984	[0.966, 1.002]	0.974**	[0.956, 0.993]	0.981*	[0.965, 0.998]	0.981	[0.962, 1.000]
ln_r	1.450	[0.956, 2.199]	1.277	[0.844, 1.932]	1.494	[0.986, 2.264]	1.392	[0.927, 2.092]	1.569*	[1.032, 2.387]
ln_s	17.529***	[10.201, 30.120]	13.111***	[7.525, 22.843]	18.189***	[10.649, 31.066]	15.412***	[9.078, 26.164]	18.974***	[11.094, 32.451]
Observations	612		612		612		612		612

Exponentiated coefficients;

^*p<0.05

^**p<0.01

^***p<0.001;

Year fixed effects included in models

Discussion

This study documents wide variability in foster care adoption rates, both overall and per eligible child in foster care, across states and to a lesser extent over time. Differences across states in their system characteristics and population demographics explain some differences in state foster care adoption rates. We highlight key findings here and discuss their implications.

First, we found that the average adoption subsidy amount was negatively associated with foster care adoption rates both between and within states. Previous research has indicated that increased subsidy eligibility hastens time to adoption (Buckles, 2013) and that increased subsidy amount is associated with higher number and rates of adoption (Hansen, 2007). However, higher average adoption subsidies rates and amounts likely reflect an adoption-eligible population that is older, with more challenges and special needs – these attributes tend to lower the probability of adoption but, conditional on adoption, are associated with near-universal subsidy eligibility and higher subsidy amounts. In addition, subsidy eligibility may be expanded, or amounts raised by states seeking to increase their rates of foster care adoptions. Supporting this explanation, we found that the minimum foster care subsidy was positively related to adoption rates, suggesting that anticipated aid may entice potential parents to pursue adoption generally, while they may be discouraged from adopting children with special needs. In sum, while adoption subsidies incentivize families to pursue foster care adoptions, these incentives may not outweigh potential parents’ desires for young and healthy children (Brooks et al., 2002; Macomber, 2005; Skidmore et al., 2016; Zhang & Lee, 2011).

Our results also provide insight into the relationship between foster care adoption and alternative family formation strategies. Interestingly, foster care families, international adoptions, and IVF were all positively associated with foster care adoption rates, suggesting that populations with high desire for children may be willing to pursue a variety of avenues. Indeed, there may be cultural factors which encourage adoption, regardless of the type. This finding complements prior research that finds a positive relationship between use of assisted reproductive technology and foster care adoption (Cohen & Chen, 2010) In contrast, private adoption rates were not significantly associated with foster care adoptions. The contrast between private and international adoption coefficients may also reflect different preferences and resources among state pools of potential adoptive parents.

Finally, we found that some state demographics predicted foster care adoption rates. States with higher median household income had higher foster care adoption rates, consistent with prior findings that wealthier families adopt at higher rates (Ishizawa & Kubo, 2014), although parents who adopt from foster care tend to have lower income than other types of adoptive parents (Moriguchi, 2012). We also found that states with more same-sex couples per capita had higher adoption rates, which fits with existing knowledge that same-sex couples are more likely to both adopt generally and to adopt children with special needs (Gates et al., 2007). Lastly, states with a higher percentage of the population that is female and of childbearing age had lower total adoption rates of older children. This finding may imply that foster care adoption is more common among older women, who may already have biological children. This is compatible with previous research (Bernal, Hu, Moriguchi, & Nagypal, 2007; Ishizawa & Kubo, 2014; Jones, 2008) as well as the theory that some parents may choose to adopt out of altruism rather than the desire to form a family (Bethmann & Kvasnicka, 2012). Limited prior research has connected demographic variables to adoption rates, and so these findings in particular make an important contribution.

Demographics of adoption-eligible children were also relevant. States with more Black children waiting for adoption had lower rates of foster care adoption, supporting research that Black children are disadvantaged in the adoption process (Akin, 2011). Somewhat surprisingly, the percentage of foster children placed in relative care and the mean age at entry into non-relative foster care were positively associated with non-relative foster care adoption rates between states. However, both variables were negatively associated with adoption rates within states, which fits with prior research that child age is negatively associated with adoption at the individual level (Snowden et al., 2008).

This study has several important limitations. Although policies at the state level have important relationships with foster care adoption rates, there are likely important characteristics of smaller geographies that are not captured in this study. We are also unable to measure all state characteristics which are likely to be relevant. For example, policies influencing family leave and child tax credits may impact adoption rates, as may the way states handle sibling groups. Unreliable data in AFCARS also prevented us from including measures of adoption-eligible child disability and frequency of child removals due to parental drug use. Additionally, we could not measure private adoption by unrelated parents in every year, but were limited to 2007 data. Finally, while our analytic strategy reveals factors which are correlated with foster care adoptions, it does not allow for a causal interpretation of the results, particularly as many of our independent variables are likely endogenous to state adoption rates. Future research should seek to address these limitations.

Despite these limitations, the results of this study have strong implications for state foster care adoption rates. Numerous characteristics of states are related individual family formation choices, with important consequences for vulnerable children in need of permanent homes. Since our analyses are descriptive rather than causal, further research is necessary in order to apply our findings to state policy. Nevertheless, these findings expand current understanding of the ways state policies and populations are related to adoption rates.