determinants of adoption subsidies

20

Click here to load reader

Upload: david

Post on 16-Mar-2017

217 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Determinants of Adoption Subsidies

This article was downloaded by: [University Of Pittsburgh]On: 14 November 2014, At: 03:00Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

Adoption QuarterlyPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/wado20

Determinants of AdoptionSubsidiesDeborah A. Gibbs MSPH a , Barbara T. Dalberth MPH a

, Nancy D. Berkman PhD a & David Weitzenkamp PhDa

a RTI International (a trade name of ResearchTriangle Institute , USAPublished online: 11 Oct 2008.

To cite this article: Deborah A. Gibbs MSPH , Barbara T. Dalberth MPH , Nancy D.Berkman PhD & David Weitzenkamp PhD (2006) Determinants of Adoption Subsidies,Adoption Quarterly, 9:2-3, 63-80, DOI: 10.1300/J145v09n02_04

To link to this article: http://dx.doi.org/10.1300/J145v09n02_04

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all theinformation (the “Content”) contained in the publications on our platform.However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness,or suitability for any purpose of the Content. Any opinions and viewsexpressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of theContent should not be relied upon and should be independently verified withprimary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages,and other liabilities whatsoever or howsoever caused arising directly orindirectly in connection with, in relation to or arising out of the use of theContent.

Page 2: Determinants of Adoption Subsidies

This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan,sub-licensing, systematic supply, or distribution in any form to anyone isexpressly forbidden. Terms & Conditions of access and use can be found athttp://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 3: Determinants of Adoption Subsidies

Determinants of Adoption Subsidies

Deborah A. Gibbs, MSPHBarbara T. Dalberth, MPHNancy D. Berkman, PhD

David Weitzenkamp, PhD

ABSTRACT. Most children adopted from foster care in recent years re-ceived an adoption subsidy. However, little is known about factors asso-ciated with subsidy receipt and amount. Analyses of the Adoption andFoster Care Analysis and Reporting System (AFCARS) shows that na-tional data obscure substantial state-level variation in the proportion ofadopted children receiving subsidies and the median subsidy amount.Several characteristics of adopted children were associated with subsidyreceipt and amount, including age, race and ethnicity, and special needs

Deborah A. Gibbs is a Senior Health and Social Policy Analyst, Barbara T. Dalberthis a Social Research Associate, Nancy D. Berkman is a Research Analyst, and DavidWeitzenkamp is a Research Statistician. The authors are affiliated with RTI Interna-tional (a trade name of Research Triangle Institute).

Address correspondence to: Deborah A. Gibbs, RTI International, PO Box 12194,Research Triangle Park, NC 27709 (E-mail: [email protected]).

The authors thank Rick Barth and Laura Radel for their comments on earlier ver-sions of this article.

This project was funded by the U.S. Department of Health and Human Services(DHHS), Office of the Assistant Secretary for Planning and Evaluation (ASPE) undercontract number HHS-100-99-0006. RTI International conducted the research. LauraRadel was the ASPE project monitor. AFCARS data used in this publication were usedwith permission of and made available by the National Data Archive on Child Abuseand Neglect, Cornell University, Ithaca, NY. AFCARS is supported by the Children’sBureau, Administration on Children, Youth and Families, Administration for Childrenand Families, U.S. Department of Health and Human Services. AFCARS data wereoriginally collected by the Children’s Bureau.

The collector of the original data, the funding agency, the Archive, Cornell Univer-sity, and their agents or employees bear no responsibility for the analyses or interpreta-tions presented here.

Adoption Quarterly, Vol. 9(2/3) 2006Available online at http://www.haworthpress.com/web/AQ

© 2006 by The Haworth Press, Inc. All rights reserved.doi:10.1300/J145v09n2_04 63

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 4: Determinants of Adoption Subsidies

status. Subsidy receipt and amount also varied with adoptive families’characteristics, such as pre-adoptive relationship with the child, familystructure, and maternal race and ethnicity. [Article copies available for a feefrom The Haworth Document Delivery Service: 1-800-HAWORTH. E-mail ad-dress: <[email protected]> Website: <http://www.HaworthPress.com> © 2006 by The Haworth Press, Inc. All rights reserved.]

KEYWORDS. Adoption subsidies, AFCARS

INTRODUCTION

Adoption subsidies are perhaps the single-most powerful tool usedby the child welfare system to encourage adoption and support adoptivefamilies. Most children adopted from foster care in recent years re-ceived adoption subsidies, yet little is known about factors associatedwith their receipt and amount. This study uses data from the Adoptionand Foster Care Analysis and Reporting System (AFCARS) to analyzethe relationship between the characteristics of adopted children andadoptive families and subsidy receipt and amount.

BACKGROUND

In recent years, approximately 50,000 children have been adoptedannually from foster care (Administration for Children and Families,2005). Federal policy, as enacted in the Adoption Assistance and ChildWelfare Act (U.S. Public Law 96-272 [1980]) and the Adoption andSafe Families Act (U.S. Public Law 05-89 [1997]) recognizes that chil-dren who are unable to return to their birth families are better off in safeand permanent homes rather than remaining wards of the state. Manychildren in the foster care system have special needs–they are older, areminorities, are part of a sibling group, or have medical or emotionalneeds–that may make it difficult to place them with adoptive families.Adoption assistance, commonly known as adoption subsidies, is de-signed to remove financial barriers for families considering adoption ofthese special needs children. In recent years, adoption subsidy case-loads and expenditures have grown rapidly, in pace with increases infoster care adoptions. However, little information is available regardingthe factors that determine subsidy receipt and amount.

Federal funds support most adoption subsidies through Title IV-E ofthe Social Security Act. For fiscal year 2005, nearly 382,000 children

64 ADOPTION QUARTERLY

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 5: Determinants of Adoption Subsidies

are estimated to receive adoption assistance with Title IV-E funds, atenfold increase in less than 20 years (U.S. House of Representatives,2004). Federal costs for adoption subsidies, excluding administration andtraining, are estimated at $1.7 billion for fiscal year 2005, and are pro-jected to double by fiscal year 2013 (Congressional Budget Office[CBO], 2005).

To be eligible for federal adoption support, children must have beenremoved from families that would have met income criteria for the Aidfor Families with Dependent Children (AFDC) program, and have spe-cial needs that would preclude their adoption without subsidies (U.S.House of Representatives, 2004). Each state defines their criteria forspecial needs within broad federal guidelines. Title IV-E adoption assis-tance funds a part of the cost of eligible subsidies, with the remainingshare funded by state and/or county dollars. If the child does not meetTitle IV-E eligibility criteria, child welfare agencies may offer state-only subsidies using state and/or county funds. During fiscal year 2001,an estimated 78,000 children, representing 23 percent of the subsidypopulation, received state-only subsidies (Gibbs et al., in press).

States and counties vary widely in the assumptions that underlie thedesign of their subsidy programs. Some jurisdictions consider that sub-sidies should be set at a rate sufficient to provide general support forneeded services. Other jurisdictions set subsidy amounts at a level thatcan support only basic care for a child, unless there are time-limited re-quests for subsidy funds to address specific problems. A recent reportfrom the North American Council on Adoptable Children (NACAC)found that the subsidy rate in four states slightly exceeded the U.S. De-partment of Agriculture (USDA) rate needed to raise a child in a low-income family (Bower & Laws, 2002). In three states, however, thestandard state subsidy is just half the USDA estimate. States can requestfederal matching funds for subsidies up to the amount of the mainte-nance payment the child would have received if in foster care. Subsidiesin excess of this level can be funded only by state and/or county dollars.Standard adoption subsidy rates for 12 states are higher than basic fostercare rates (Bowers & Laws, 2002).

Preliminary analyses indicate that 88 percent of the children adoptedduring fiscal year 2003 received adoption subsidies (U.S. Departmentof Health and Human Services [U.S. DHHS], 2005). Research showsthat children who receive subsidies are adopted more quickly, and withbetter outcomes, than those who do not. A review of 2,200 adoptioncase records for children with special needs found clear evidence thatsubsidies facilitated adoption (Sedlak & Broadhurst, 1993). Among

Gibbs et al. 65

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 6: Determinants of Adoption Subsidies

adoptive parents interviewed as part of the study, 35 percent reportedthat the availability of subsidies influenced their decision to adopt, and29 percent said the adoption would not have been possible without asubsidy. Although time from termination of parental rights (TPR) toadoption was generally longer for older children, for children with se-vere disabilities, children who needed ongoing treatment, sibling groups,and for children with prior disrupted adoptions, these differences werereduced or eliminated when adoption assistance was provided. Simi-larly, interviews with case workers for 75 New York children who hadbeen waiting longest for adoptive placement (mean time since beingfreed for adoption was 8.6 years) indicated that increased subsidieswould facilitate adoption for 59 percent (Avery, 1999). An economicanalysis of adoptions reported in AFCARS between 1995 and 1997found that a one percent increase in adoption subsidies was associatedwith a 1.5 percent increase in the number of adoptions per 100,000 pop-ulation. In addition, the seven states in which adoption subsidies werelarger than foster care board rate had higher rates of adoption than otherstates (Hansen & Hansen, in press).

In addition to facilitating adoption, limited available evidence sug-gests that subsidies may improve their stability for children with thegreatest needs. Researchers have found lower rates of disruption amongsubsidy recipients when the adopted children were adolescents (Berry& Barth, 1990) or when they had special problems (Barth et al., 1988).No relationship between subsidies and disruption rates was foundwithin a larger sample of adopted children.

Although the estimated difference in savings varies, studies haveshown that there is an economic advantage for states to place childrenin permanent homes, even though a subsidy is provided to the family(Sedlak & Broadhurst, 1993; Barth 1993, 1997). A more recent article(Barth, in press) compares the costs of foster care with subsidized adop-tion, using a more sophisticated methodology than previous studies. In-corporating factors such as court costs, medical costs, and administrativecosts as well as foster care maintenance and adoption subsidies, the au-thors estimate subsidized adoptions result in federal savings of approxi-mately $1 billion dollars for the 50,000 children adopted each year.

Despite their influence on adoption decisions and adoption outcomes,little is known about factors associated with subsidy determination. Anearlier study (Avery & Mont, 1991) of New York’s state adoption sub-sidy records reports that higher subsidies were associated with childrenplaced in single-parent adoptive homes, children who were in fosterhomes, older children, and children adopted by families residing in a

66 ADOPTION QUARTERLY

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 7: Determinants of Adoption Subsidies

different county or state. Recent increases in the rate of adoptions fromfoster care, and the long-term financial commitment represented by adop-tion subsidies, underscore the need for better understanding of subsi-dies, particularly at the national level. This study addresses this need byusing national data to examine the extent to which state-level variablesand the characteristics of adopted children and adoptive families deter-mine subsidy receipt and subsidy amount.

METHODS

Data

All analyses described in this paper are based on AFCARS data,which compiles child-level information on all children for whom statechild welfare agencies have responsibility for placement, care, or super-vision. Two separate data files represent the population of childrenadopted from foster care during the year and the population of childrenresiding in out-of-home care during the year. Although some childrenare included in both files, no identifying information links these twosets of data, nor is there identifying information linking data from oneyear to another. All 50 states, as well as Puerto Rico and the District ofColumbia, submitted usable adoption and foster care data to AFCARSfor fiscal year 2001.

The analysis population for the adoption file comprised the 50,703children under the age of 18 years who were adopted during fiscal year2001, with some data from 1999 and 2000 used to examine trends. Vari-ables of interest in the adoption file include child’s characteristics (e.g.,age, sex, race, ethnicity, special needs classification, months from TPRto adoption); adoptive family’s characteristics (e.g., family structure;pre-adoptive parent-child relationship; mother’s age, race and ethnicity;father’s age, race, and ethnicity); and subsidy data (e.g., source of sub-sidy, subsidy amount). Because it is not possible to link records in theAFCARS adoption file to those in the foster care file, some analysesused state-level data from the AFCARS foster care file to represent me-dian foster care maintenance payments and time to adoption.

Characteristics of children and their adoptive parents are describedmore fully elsewhere (Dalberth et al., 2005). These children were evenlydivided by gender. Children were slightly more likely to be White andnon-Hispanic (38 percent) than African-American and non-Hispanic.An additional 16 percent were Hispanic, and 5 percent were American

Gibbs et al. 67

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 8: Determinants of Adoption Subsidies

Indians, Asians, Native Hawaiians or multiracial children. Nearly half(48 percent) were 5 years of age or less at the time of adoption (with 2percent less than one year of age), and 9 percent aged between 13 and 17years. Two-thirds (67 percent) of adoptive families were two-parentfamilies, with an additional 30 percent single females; single males andunmarried couples comprised less than 4 percent. Just over one-half ofchildren were adopted by their foster parents, 21 percent by a relativeother than a parent, and 15 percent by parents who were neither. Raceand ethnicity for adoptive parents are similar to those of adopted chil-dren, since 93 percent of adoptive parents were of the same race-ethnic-ity as their adopted children.

AFCARS data are reported to the Administration for Children andFamilies (ACF) by states. Although ACF continues to work with statesto improve the quality of AFCARS data, neither ACF nor RTI can ver-ify the validity or completeness of these data. Although the AFCARSdata elements are straightforward, the analysis took into account poten-tial concerns regarding the reliability of specific variables by identify-ing outliers for continuous variables that appear to be data errors andsetting them to missing. For example, there were several cases wherethe monthly subsidy amount was reported as greater than $10,000. Al-though this amount may be valid for a small number of cases, othercharacteristics of these cases suggest the likelihood of data reporting er-rors. Four states (Mississippi, Nevada, New York, and West Virginia),collectively representing 7 percent of adoptions in 2001, were excludedfrom multivariate analyses because subsidy data were missing or incom-plete for at least 30 percent of cases. These analyses are thus based onchildren in the remaining 46 states, the District of Columbia, and PuertoRico, for whom all model variables were consistently available. Althoughthe analysis population is thus not truly a national one, it represents the93 percent of 2001 adoptions nationally with the best available data.

Analyses

Preliminary descriptive analyses examined the characteristics of adop-ted children and adoptive parents, rates of subsidy receipt for subsidiessupported by federal Title IV-E and state-only subsidies, and subsidyamounts. Analyses were conducted at the national level and by state.

Multivariate analyses used AFCARS data to explore the individual-and state-level characteristics associated with whether a child receivesan adoption subsidy, and which characteristics are associated with sub-sidy amount among children receiving a subsidy. The first question was

68 ADOPTION QUARTERLY

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 9: Determinants of Adoption Subsidies

addressed with a logistic regression analysis, since the outcome is a 0/1binary variable indicating whether children adopted in FY 2001 re-ceived a subsidy. To answer the second question, analysts used multipleregression analysis, because the amount of the subsidy is a continuousvariable ranging from $1 to $8,000 (the highest subsidy value) monthly.If the factors that determine subsidy receipt are different from those that de-termine subsidy amount, then the estimation requires a two-stage structuralequation model. However, a two-stage model requires an explanatoryvariable in the first-stage equation that is not a predictor in the second stage.Since no such variable was identified, two separate equations were used.

Most statistical techniques require all observations to be independentof each other. In order to account for possible correlation among caseswithin a state, the analysis was constructed as a hierarchical model.Such correlation within states may come from unmeasured predictors orfrom predictors that, individually, play little role, but in aggregate havea stronger effect. State-specific random effects were represented by arandom intercept for unmeasured heterogeneity. This approach was ap-plied both to a logistic regression model and a linear regression model.In each model the random effects, those defining the hierarchical por-tion of the model, entered the model as part of the “linear predictor”–alinear equation whose parameters can be translated to odds ratios (forlogistic regression) or left as is, as chosen here. The final model for eachwas the most parsimonious set of random effects that provided meaning-ful improvement in model fit, measured by a likelihood ratio test, overthe next-simpler model. All fixed effects were included in all models.

Analyses used all records with non-missing data for all model vari-ables. The population used to answer the first research question was25,744 (51 percent of adopted children). The population used to answerthe second research question, including only those children who re-ceived a subsidy, was 22,150 (50 percent of subsidy recipients). Explan-atory variables, the fixed effects, are considered statistically significantlyat the p < 0.05 level of significance. For determining variables enteredinto the random effects portion of the model, statistical significance wasdefined as p < 0.01, to avoid undue model complexity.

FINDINGS

As noted earlier, the great majority of children adopted in FY 2001received subsidy assistance (88 percent). Among subsidy recipients, most

Gibbs et al. 69

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 10: Determinants of Adoption Subsidies

received subsidies with Title IV-E funds (84 percent), with the remain-ing 16 percent receiving state-only subsidies.

Nationally, the median subsidy amount for children adopted in 2001was $444 a month, a 10 percent increase from the $404 median for 1999.During the same period, the number of children adopted increased from33,655 children in 1999 to 39,135 children in 2001, a 16 percent in-crease, further increasing subsidy expenditures. Across the three yearsexamined, median subsidies increased in 26 states, with the amount ofincrease ranging from 3 to 67 percent. Eight states showed a reductionin median subsidies, with decreases ranging from 1 to 27 percent, andeight states reported the same median subsidy across the three years.Subsidy amounts vary substantially among states, as seen in Table 1.Median monthly subsidies were less than $300 in five states, and morethan $600 in six. Analysis of variance found no relationship between re-gion and either the percent of adopted children receiving subsidies orthe median subsidy amount (p = 0.578; p = 0.371).

Characteristics of Adopted Children

Descriptive analyses indicate that subsidy receipt varied somewhatby children’s characteristics, as seen in Table 2. Children six years ofage or older and non-Hispanic African American children were some-what more likely to receive adoption subsidies than other children, al-though the differences among groups are not large.

Nearly all (93 percent) children identified as having special needswith respect to adoption received subsidies. Of those children withoutidentified special needs, 59 percent received subsidies, of which one-third were state-only subsidies and the remainder federal. The presenceof special needs with respect to adoption is a criterion for federal subsi-dies. When these are not documented in AFCARS it is likely due to dataextraction errors (Maza, 2006). For fiscal year 2001, the proportion ofchildren classified as having special needs with respect to adoptionranged from 43 to 100 percent among states. These variations may re-flect differing criteria for special needs, as well as states’ generalpropensity to provide subsidies.

Children who were adopted within 6 months after TPR were lesslikely to receive subsidies, but little variation was seen among those withvarying lengths of time to adoption after 6 months. Most (63 percent) ofthese children were less than 6 years of age. State practice regarding thetiming of TPR varies widely, and some states avoid terminating parental

70 ADOPTION QUARTERLY

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 11: Determinants of Adoption Subsidies

Gibbs et al. 71

TABLE 1. Median Monthly Adoption Subsidy Amount by Age, by State,Fiscal Year 2001

Age at Adoption PercentReceiving

Subsidy(%)

0-5 6-12 13-17 Total

State ($) ($) ($) ($)

Alabama 241 241 241 241 46.8Alaska 653 603 689 650 97.5Arizona 479 479 479 479 94.3Arkansas 400 425 475 425 89.2California 405 471 569 441 91.5Colorado 401 601 786 510 92.1Connecticut 638 659 727 659 16.4Delaware 479 518 543 479 97.4District of Columbia 741 741 817 741 57.3Florida 295 304 364 300 64.2Georgia 388 411 433 411 48.3Hawaii 529 529 529 529 83.1Idaho 251 365 410 275 84.6Illinois 369 410 444 410 96.2Indiana* 169 162 194 171 51.8Iowa 587 904 971 856 76.5Kansas 304 400 400 400 72.8Kentucky 600 600 722 600 70.8Louisiana 273 392 395 353 91.1Maine 581 772 733 650 99.2Maryland 600 535 550 543 98.4Massachusetts 454 471 522 471 89.3Michigan 439 731 846 591 96.4Minnesota 397 552 612 427 99.3Mississippi* – – – – 81.4Missouri 225 275 304 275 96.4Montana 388 399 508 408 90.2Nebraska 421 641 627 527 68.8Nevada* – – – – 94.7New Hampshire 535 600 708 552 88.4New Jersey 412 457 516 437 90.1New Mexico* 487 520 545 503 89.4New York* – – – – 97.9North Carolina 315 365 415 365 94.2

North Dakota 396 388 558 402 62.8Ohio 471 517 575 500 96.0

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 12: Determinants of Adoption Subsidies

rights until adoption is imminent. Therefore, this measure should not beused as an indicator of time to adoption.

Median monthly subsidy amounts also varied with children’s charac-teristics. The largest variation was among children of different ages, withthose over age 13 receiving, on average, $116 more than those underage 6. Subsidy amount did not vary by race and ethnicity, except for aslight increase for those in the “other” category, which includes Ameri-can Indians, Asians, and multiracial children. The increased subsidyamounts received by children without identified special needs ($507 vs.$444) may be influenced by the wide variation in both special needs

72 ADOPTION QUARTERLY

TABLE 1 (continued)

Age at Adoption PercentReceiving

Subsidy(%)

0-5 6-12 13-17 Total

State ($) ($) ($) ($)

Oklahoma 300 360 420 360 99.5Oregon 369 500 540 400 99.0Pennsylvania 450 522 600 510 90.2Rhode Island 387 418 422 407 99.6South Carolina 332 359 425 359 100.0South Dakota 390 390 469 390 64.9Tennessee 362 313 422 402 81.2Texas 516 516 516 516 72.6Utah 270 348 411 300 77.3Vermont 478 567 789 549 85.3Virginia 294 344 436 344 95.5Washington 531 612 728 572 97.7West Virginia 400 400 456 400 91.7Wisconsin 518 796 1,002 639 99.2Wyoming 399 399 399 399 93.5Puerto Rico 207 174 174 174 13.2Total 406 471 522 444 88.1

Notes: 1. Adoption subsidy amounts exclude deferred payment amounts.2. Eight states (Arizona, Arkansas, Indiana, Mississippi, Nevada, New Mexico, New York, and West

Virginia) had missing or invalid subsidy amount data for at least 10 percent of their cases.3. Except where noted, data presented in these tables are reported to ACF by states. Although

ACF continues to work with states to improve the quality of AFCARS data, neither ACF nor RTI can verifythe validity or completeness of these data.*These states had missing or invalid subsidy amount data for more than 30 percent of their cases.Source: AFCARS 2001, adoption data.

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 13: Determinants of Adoption Subsidies

classification and subsidy amounts among states, rather than actual sub-sidy decisions favoring children without special needs. Children whowaited longer from TPR to adoption (more than 18 months) received ahigher median subsidy compared with children who were adopted morequickly after TPR.

Characteristics of Adoptive Families

Subsidy receipt and amount also varied according to characteristicsof adoptive families, as seen in Table 3. Children whose adoptive moth-ers were non-Hispanic African American were more likely to receive

Gibbs et al. 73

TABLE 2. Subsidy Receipt and Median Subsidy Amount by Characteristicsof Adopted Children, Fiscal Year 2001

Subsidy Recipients

AdoptedChildren1

(N)

AdoptedChildren2

(N)

Percent ofAdopted

Children (%)

MedianSubsidy

($)

Child’s age at adoption

0-5 years 24,350 20,861 86 406

6-12 years 21,514 19,460 90 471

13-17 years 4,701 4,232 90 522

Race/ethnicity

White, non-Hispanic 19,415 16,806 87 444

African- American, non-Hispanic 17,606 15,907 90 444

Hispanic 8,239 7,070 86 444

Other 2,673 2,342 88 469

Special needs with respect to adoption

Yes 43,115 40,060 93 444

No 6,069 3,583 59 507

Wait time (from TPR to adoption)

< 6 months 40,420 25,842 64 420

6 to 12 months 36,917 32,674 89 441

12 to 18 months 40,581 35,617 88 444

>18 months 34,793 30,407 87 475

Total 50,703 44,553 88 444

1Numbers in categories may not add to the total number of adopted children due to missing data.2Subsidy recipients include 398 children with subsidy amounts coded as $0 or $1, which may representdeferred subsidy agreements. These cases are excluded from calculation of median subsidy amount.Source: AFCARS 2001, adoption data.

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 14: Determinants of Adoption Subsidies

besidies, and received somewhat higher subsidies, than others. Childrenadopted by foster parents or relatives were more likely to receive subsi-dies than those adopted by non-relatives other than foster parents. Sin-gle females were more likely to receive subsidies than married couples.Although adoptions by stepparents, unmarried couples, and single maleswere relatively rare, they were associated with higher median subsidies.

Multivariate Models of Subsidy Receipt and Amount

Table 4 shows results of two multivariate models. The first data col-umn shows coefficients for the logistic regression analysis predictingwhether a child received a subsidy. Age was a significant factor: theolder the child, the more likely he or she was to receive an adoption sub-sidy. Boys and girls were equally likely to receive subsidies. Non-His-

74 ADOPTION QUARTERLY

TABLE 3. Subsidy Receipt and Median Subsidy Amount by Characteristics ofAdoptive Families, Fiscal Year 2001

Subsidy Recipients

AdoptedChildren1

(N)

AdoptedChildren

(N)

Percent ofAdopted

Children (%)

MedianSubsidy

($)

Adoptive mother’s race/ ethnicity

White, non-Hispanic 19,415 16,806 87 444

African-American, non-Hispanic 17,606 15,907 90 471

Hispanic 8,239 7,070 86 441

Other 2,673 2,342 88 454

Preadoptive parent-child relationship2

Foster parent 26,416 23,456 89 450

Stepparent 176 55 31 535

Other relative 10,707 9,672 90 441

Nonrelative 7,704 6,125 80 471

Adoptive family structure

Married couple 30,703 26,426 86 441

Unmarried couple 635 582 92 454

Single female 13,161 12,358 94 444

Single male 1,023 880 86 475

Total 50,703 39,135 77 444

1Numbers in categories may not add to the total number of adopted children due to missing data.2 Not all states report step-parent adoptions; therefore, these data are underreported.Source: AFCARS 2001, adoption data.

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 15: Determinants of Adoption Subsidies

panic African American children were 40 percent more likely to receivesubsidies than non-Hispanic White children. However, the differencesassociated with the race and ethnicity of adoptive mothers, seen in thedescriptive analysis, were not statistically significant. Children eligiblefor federal adoption assistance under Title IV-E, and those with specialneeds relative to adoption, were more likely to receive subsidies. Adop-tive family structure was not related to the likelihood of subsidy receipt.

The second data column shows coefficients for the regression modelpredicting subsidy amount for those children who received a subsidy.As with subsidy receipt, subsidy amount was positively related to theage of the child; the older the child, the larger the subsidy, with subsidyamount increasing by nearly $14 for each additional year of age. Subsi-

Gibbs et al. 75

TABLE 4. Multivariate Models of Subsidy Receipt and Amount

Model Parameter(reference category)

Model 1 Estimate(Logistic Regression

Model of SubsidyReceipt)

Model 2 Estimate(Multiple Regression

Model of SubsidyAmount)

Intercept �1.5157 160.97*Age

Age at adoption 0.07523* 13.382*Sex (female)

Male �0.01263 10.1299*Child’s race/ethnicity (non-Hispanic White)

Non-Hispanic African-American 0.3932* 49.858*Hispanic �0.01387 0.6433Non-Hispanic other �0.3671 �15.05

Title IV-E adoption subsidy (not eligible)Title IV-E eligible 7.1882* 213.3*

Special needs (not eligible)Special needs eligible 1.8986* 62.2079*

Adoptive mother’s race/ethnicity(non-Hispanic White)Non-Hispanic African-American �0.3866 �2.7268Hispanic 0.02951 �33.6417*Non-Hispanic other 0.2985 2.475

Pre-adoptive relationship (foster parent)Other relative �0.1199 �87.9039*Nonrelative �0.6675* �36.0257*

Adoptive family’s structure (married couple)Unmarried couple �0.08096 32.2784Single female 0.05311 36.1174*Single male �1.3586 �40.8746

* p < .011. Excluded are children in Mississippi, Nevada, New York, and West Virginia.2. Subsidy recipients include 398 children with subsidy amounts coded as $0 or $1, which may representdeferred subsidy agreements. These cases are excluded from calculation of median subsidy amount.Source: AFCARS 2001, adoption data.D

ownl

oade

d by

[U

nive

rsity

Of

Pitts

burg

h] a

t 03:

00 1

4 N

ovem

ber

2014

Page 16: Determinants of Adoption Subsidies

dies received by boys were, on average, $10 greater than those receivedby girls. Children classified as having special needs with respect toadoption received significantly higher subsidies than those not so clas-sified, as did those eligible for adoption assistance under Title IV-E.

Race and ethnicity of children and adoptive parents were related tosubsidy amount. Non-Hispanic African American children received sub-sidies that were, on average, nearly $50 higher than those received bynon-Hispanic White children. Being adopted by a Hispanic mother(compared with a non-Hispanic white mother) was significantly relatedto receiving a smaller subsidy.

Other characteristics of adoptive families influenced subsidy amounts.Children adopted by relatives, or by non-relatives other than foster par-ents, received smaller subsidies than those adopted by foster parents.Children adopted by single females, but not single males, received largersubsidies than those adopted by married couples.

Additional tests of the multivariate models confirm the importance ofstate-level practice variations in determining subsidy amount. The Intra-Class Coefficient (ICC) for Model 2 in Table 4 calculates the proportionof variation explained by the random intercepts representing state-spe-cific effects, (i.e., unobservable differences among states). ICC mea-sures cannot be appropriately calculated for the logistic regressionmodel of subsidy receipt due to the relationship between the mean andvariance present in logistic regression. In a test model including nochild-level fixed effects, state-specific effects accounted for 20 percentof unexplained variation. After adding child-specific variables and twoavailable variables representing state practice–median time to adoptionand the proportion of subsidy costs reimbursed with federal funds forTitle IV-E eligible subsidies–the ICC was reduced to 16 percent. Thereduction in the ICC achieved by adding child-specific fixed effects in-dicates that approximately one-quarter of the unexplained variationamong states is due to differences in individual characteristics such as ageor special needs status, or the specific state effects for which AFCARSprovides data. State-specific effects still accounted for 16 percent of un-explained variation within the models. The magnitude of this effect in-dicates that some unmeasured factor (or combination of factors) at thestate level plays a significant role in determining subsidy amount. Evenafter controlling for variations in the characteristics of adopted childrenand adoptive families, some unmeasured factor or combination of fac-tors at the state level still play a significant role.

76 ADOPTION QUARTERLY

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 17: Determinants of Adoption Subsidies

DISCUSSION AND CONCLUSIONS

These analyses demonstrate the far-reaching importance of subsi-dies. Among children adopted in 2001, nearly all (88 percent) receivedsubsidies. However, national data can obscure the very substantial vari-ations among states in subsidy practice. State-level subsidy receipt ratesranged from 13 percent to 100 percent. Similarly, while the median sub-sidy was $444 nationally, median subsidies reported by states rangedfrom $174 to $856. Although federally supported subsidies are limitedto the level of foster care maintenance payments, these levels are ulti-mately under the control of states.

Children’s age and special needs status influenced subsidy receiptand amount. Older children were more likely to receive subsidies, and toreceive larger subsidies, presumably reflecting their greater need for ser-vices. This finding is consistent with state policies that tend to have higherbasic subsidy rates for older children (U.S. House of Representatives,2004). After controlling for state-level variation, some associations werefound between children’s race and ethnicity, with non-Hispanic AfricanAmerican children more likely to receive subsidies, and to receivelarger subsidies. Although gender was not associated with subsidy re-ceipt, among children who received a subsidy, boys received slightlyhigher subsidies than did girls.

Pre-adoptive relationship and other characteristics of adoptive fami-lies influenced children’s subsidies. Children adopted by foster par-ents–who represent more than half of all adoptions–were more likely toreceive subsidies, and received higher subsidies than did other children.One possible reason foster parents receive higher rates is that they aremore adept with working with the foster care system and learning howto advocate for their child and ask for necessary support. Although noassociations were found between adoptive parents’ race and ethnicityand subsidy receipt, children adopted by Hispanic mothers receivedlower subsidies than those whose adoptive mothers were non-Hispanicwhites. Children adopted by single females received higher subsidiesthan those adopted by married couples. These findings suggest the in-fluence of both family needs and adoptive parents’ ability to advocateon subsidy decisions. However, even after controlling for child andfamily characteristics, state-level variations remained a significant de-terminant of subsidy receipt and amount.

Some limitations to these analyses should be noted. Although the qual-ity of AFCARS data improves each year, missing or invalid data limitedthe number of cases available for multivariate analyses and suggest the

Gibbs et al. 77

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 18: Determinants of Adoption Subsidies

need for caution when comparing data across states. In addition, some dataelements that would provide valuable insights into adoption subsidy pro-vision are not available within the AFCARS adoption data file, for exam-ple, sibling group membership, amount of foster care support payment andtime in foster care. Although AFCARS remains a rich and largely untappeddata resource even with these limitations, state administrative data maysupport more sophisticated subsidy analyses (see Barth et al., in press).

Using two separate regression models, rather than a two-stage model,reduces estimation efficiency. As a result, some explanatory variablesthat would have been statistically significant may not appear to be so. Inspite of this limitation, the multivariate models are robust, with consis-tent findings that suggest that the subsidy program is operating as in-tended. Children who have historically faced greater barriers to findingadoptive homes–older children, African American children and thosewith special needs–are more likely to receive subsidies, and to receivehigher subsidies. These findings indicate that subsidies are, in fact,being used to reduce financial barriers to adoption.

Dramatic variations across states in both subsidy receipt and amountsuggest that children in some states may face greater barriers to adoptionbecause of limited subsidy provision. In fact, economic models suggestthe elasticity of adoption rates in response to subsidy variations (Hansen& Hansen, in press). Adoptive parents may be attracted to high-subsidystates as a result of the increasing use of photolisting services such asAdoptUsKids, dissemination of subsidy data by the National Associationfor Adoptable Children (NACAC), and recent state efforts to reduce bar-riers to interstate placement (Dalberth et al., 2006). At the same time, thefact that nearly three-quarters of adoptions are by foster parents or rela-tives suggests that in many instances, existing relationships may out-weigh financial considerations. The demonstrated benefits of adoptionsubsidies, in terms of both child outcomes and public finances, suggesttheir potential usefulness for states that currently use them less extensively.

REFERENCES

Administration for Children and Families. (2005, September). Trends in foster care andadoption. http://www.acf.hhs.gov/programs/cb/stats_research/afcars/trends.htm. Asobtained on January 31, 2006.

Avery, R.J (1999). Identifying Obstacles to Adoption in New York State’s Out-of-HomeCare System. Child Welfare 78(5): 653-671.

78 ADOPTION QUARTERLY

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 19: Determinants of Adoption Subsidies

Avery, R.J & D. Mont (1991). Children Who Enter and Leave the New York StateAdoption System. Human Ecology 19(4):2-7.

Barth, R.P, C.K. Lee, J. Wildfire & S. Guo (in press). A Comparison of the Govern-mental Costs of Long-Term Foster Care and Adoption. Social Service Review

Barth, R.P. (1993). Fiscal Issues in Special Needs Adoption. Public Welfare 41(4):7-11.Barth, R.P., M. Berry, R. Yoshikami, R.K. Goodfield & M.L. Carson (1988). Predict-

ing Adoption Disruption. Social Work 33: 227-233.Berry, M. & R.P. Barth (1990). A Study of Disrupted Adoptive Placements of Adoles-

cents. Child Welfare 69(3):209-225.Bower, J.W., & R. Laws (2002). Forever Families: Support for Families of Children

with Special Needs: A Policy Analysis of Adoption Subsidy Programs in the UnitedStates. North American Council on Adoptable Children.

Congressional Budget Office (2005). CBO Baseline for Foster Care and Adoption As-sistance.

Child Welfare League of America (2004). Data from the National Data Analysis Sys-tem. <http://ndas.cwla.org>. As obtained on August 20, 2004.

Dalberth, B., D. Gibbs & N. Berkman (2005). Understanding Adoption Subsidies:An Analysis of AFCARS Data. Washington, DC: U.S. Department of Health andHuman Services, Office of the Assistant Secretary for Planning and Evaluation.

Dalberth, Barbara T., Jennifer Hardison, Deborah Gibbs, & Susan Smith (draft, 2006).Interjurisdictional Placement of Children in the Child Welfare System: Improvingthe Process. Report prepared by RTI International for U.S. Department of Healthand Human Services; Administration for Children and Families; Administration onChildren, Youth and Families; Children’s Bureau. Washington, DC: U.S. Depart-ment of Health and Human Services.

Gibbs, D., B. Dalberth, S. Hawkins, S. Harris, R. Barth, & J. Wildfire (2004). Termina-tion of Parental Rights for Older Foster Children: Exploring Practice and Policy Is-sues. Report prepared for the Office of Planning, Research and Evaluation, Children’sBureau, Administration for Children, Youth and Families, U.S. Department of Healthand Human Services. Research Triangle Park, NC: RTI International.

Gibbs, D., B. Dalberth, N. Berkman & D. Weitzenkamp (in press). Federal Support forAdoption Subsidies. Journal of Public Child Welfare

Hansen, M.E., & B.A. Hansen (in press). The Economics of Adoption of Children fromFoster Care. Child Welfare

Maza, P.L (2006). Personal communication from Penelope Maza, February 2, 2006.Sedlak, A. J, & D.D. Broadhurst (1993). Study of Adoption Assistance Impact and

Outcomes: Final Report. Rockville, MD: Westat, Inc.Spar, K., & C. Devere (2001). Child Welfare Financing: Issues and Options. Congres-

sional Research Service.U.S. Census Bureau, Population Division. June 18, 2003. Table NA-EST2002-ASRO-

03–National Population Estimates–Characteristics. <http://www.census.gov/popest/archives/2000s/vintage_2002 /files/NA- EST2002-ASRO-03.csv>.

U.S. Department of Health and Human Services (2000). Federal Financial Participa-tion in State Assistance Expenditures for October 1, 2000 through September 30,2001. < http://aspe.hhs.gov/health/fmap01.htm>. As obtained on January 15, 2004.

Gibbs et al. 79

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14

Page 20: Determinants of Adoption Subsidies

U.S. Department of Health and Human Services (2005). The AFCARS Report: Prelim-inary FY 2003 Estimates as of April 2005 (10). http://www.acf.hhs.gov/programs /cb/stats_research/afcars/tar/report10.htm. As obtained on January 31, 2006.

U.S. House of Representatives, Committee on Ways and Means (2004). 2004 GreenBook. < http://waysandmeans.house.gov/documents.asp>. As obtained on June 2,2004.

Wulczyn, F.H., & K.B. Hislop (2002). Issue Papers on Foster Care and Adoption:Growth in the Adoption Population. Chicago: Chapin Hall Center for Children atthe University of Chicago.

Received: 08/01/05Revised: 01/20/05

Accepted: 03/01/05

80 ADOPTION QUARTERLY

Dow

nloa

ded

by [

Uni

vers

ity O

f Pi

ttsbu

rgh]

at 0

3:00

14

Nov

embe

r 20

14