Choosing a “Paperwork Pregnancy”:  Determinants of International
Child Adoptions Completed by U.S. Parents 

Stephanie M. Brewer
Tiffany M. Hicks
James J. Jozefowicz*
Indiana University of Pennsylvania

Abstract
     We analyze international child adoptions completed by U.S. parents and fill a gap in the literature; because, previous studies have focused primarily on domestic adoptions. We estimate the impact of social, economic, and demographic variables on the number of international adoptions using multivariate regression analysis with a U.S. state-level data set for 2000. Data sources include the Adoption and Foster Care Reporting and Analysis System and the Census Bureau.  The analysis indicates that international adoptions are positively related to the following populations within a state: the $50,000-100,000 income range, the 25-44 year old age range, those holding a professional/graduate degree, and the marriage rate. In addition, international adoptions are negatively related to the population aged 45-65, the labor force participation rate of women, and the birth rate within a state.
1.   Introduction
     One of the most interesting points of study within the industrialized world has been the evolution of the family.  This is an area of study that can be better understood using an interdisciplinary approach combining economics and sociology. Many characteristics of the American household have changed drastically over the last hundred years.  Women entering the workforce, the shift from an agricultural to a service sector economy, and changing societal values are just a few of the reasons for a drastic change in the makeup of the American nuclear family.  Developed countries are characterized by easy availability of contraceptive options and legalized abortion, which give more control over reproduction and greater independence for women.  When combined with an observed increase in the labor force participation rate of women, it should not be surprising to observe lower birth rates in the developed world, where the average child per household ranges between 1.6 and two according to the Central Intelligence Agency (CIA).  Furthermore, it is no longer a widespread societal expectation for individuals to marry young and immediately begin having many children. Couples, as well as individuals, are waiting until later in life to begin families, and that choice may complicate conception.  One option open to those facing difficulties conceiving a child historically has been adoption.  
     However, the pursuit of adoption is not just for the infertile.  Adoption has become a major point of interest in many facets of our society.  Some prospective adoptive parents seek adoption from the beginning as their first-best choice to build a family, and others use it as a means of completing their family by adding to biological children. “Preferential adoption” is a term coined by Feigelman and Silverman (1979) for couples who are able to biologically reproduce but choose adoption. Additionally, the gay and lesbian community calls further attention to adoption as one approach to parenthood for these couples.
     Increasingly our world is getting smaller. In recent years attention has focused on recognizing the desperate life circumstances of children and adults living in other countries. One vein of humanitarian/social justice movements has been calling attention to children in critical need of a home and family of their own.   Additionally, some families embark on adopting a child from a foreign nation partly from a religious perspective; these families respond to faith-based calls to care for orphans in distress and may hope to introduce their own beliefs to a child who otherwise may never have been taught them in his/her own nation of birth. Furthermore, celebrities such as actress Angelina Jolie and contemporary Christian singer Steven Curtis Chapman have spotlighted international adoption through their own personal family stories.
     While the international adoption community goes out of its way to avoid language such as a “market” for children, one of the most basic ideas of economics, namely supply and demand, can be used as a framework to describe the phenomenon of international adoption.  Those who live in nations where there is a shortage of “healthy” (perceived at both physical and emotional levels) children to adopt, turn to other countries which open their borders in an attempt to prioritize placement of orphaned children in permanent, loving families above alternate outcomes of having these children grow up in orphanages, foster care, or on the streets.
     Relative to the degree to which adoption affects our world, its study is minimal, and few quantitative studies have been conducted. While Ishizawa, Kenney, Kubo and Stevens (2006) state that the literature on intercountry adoption is “rapidly growing” (p. 1208), most empirical work on the issue mainly focuses on the impact of adoptions on the physical and psychological well-being of the adopted child and adoptive family.   Examples of work done on outcomes include: Tieman, van der Ende and Verhulst  (2006), who focus on the social functioning of intercountry adoptees; Kim (1995), who studies the mental development of adolescent children who have been adopted and compares that to children who grew up with their biological families; and Glidden (2000), who investigates the adjustment of adopted children who have developmental disabilities.  While it is important to understand how placement affects a child’s well-being, this study does not attempt to explain outcomes.  Instead, the contribution of the present research is to focus on the social, economic, and demographic determinants of international adoption. Regressions are estimated to understand the impact of income, education, age, and substitutes on the number of foreign children adopted by Americans.  This research represents an attempt to better understand the characteristics of the population of people who successfully navigate the arduous task of becoming parents of an internationally adopted child.
     This paper is presented in seven sections. Section 2 introduces a brief history of adoption.  Section 3 presents relevant literature previously published on the topic.  Section 4 presents the data used, and Section 5 will explain the econometric model employed to estimate the relationships and econometric issues.  Section 6 contains the results, and Section 7 offers conclusions and possible extensions for future research.
2.  Brief Historical Background of Adoption
     Adoption is the legal severing of ties between birth parents and children. Historically, there has been a societal dilemma as to what should or can be done with children who need to be placed with alternate caregivers other than their biological parents.  In the early years of U.S. history, there were far more orphaned children in need of care in New York City and other cities in the Northeast than adequate means to care for them. From the mid-1800s until about 1930, children from the East were loaded into “Orphan Trains” bound for the Midwest, Canada, and Mexico, as a means of moving these children out of the overcrowded, crime-prone cities.  Interested families would show up at the train stations and choose children, based on little more than physical appearance, to take home with them.  No established, systematic mechanism of trying to match these children with the best available families existed.  Underlying motivation for the adoptive families included “Americanizing” poor Catholic and Jewish orphaned children to the Protestant way of life and a desire for an extra set of hands from a labor perspective (Herman, 2005).
     In the year 1851, Massachusetts passed what is considered the first modern law concerning adoption.  While its intent was to make child welfare the primary concern in adoptions, the law placed final adoption decisions in the hands of judges. In 1868, Massachusetts again stepped to the forefront of child welfare reform by initiating a deinstitutionalization of children and working to place them into foster care.  This initiated a period characterized by the term “placing out,” which referred to situations of “non-institutional arrangements to care for dependent children” (Herman, 2005).  This term is synonymous with baby-farming.  Agencies would pay families to take care of children, and children were essentially indentured servants forced to work to stay out of the institution. 
     The beginning of the 1900s saw the opening of the first specialized adoption agencies and social work organizations charged with the protection of displaced and under cared for children.  This was a time when adoption was viewed as the best option for illegitimate children, for unmarried mothers to place their children, and for infertile couples.  Adoption policies reflected societal norms at the time, and adoption could not be pursued by unmarried adults.  In 1912, Congress established the U.S. Children’s Bureau (USCB) within the Department of Labor with an emphasis on reducing instances of infant mortality, eliminating child labor, and as a force to reform adoption laws and set in motion research in the field.  The USCB worked to produce regulation for adoptions including minimum standards for safeguarding children for placement in adoptive homes.  Currently, the USCB is located within the Department of Health and Human Services arm of the U.S. government.
     Even with the introduction of the USCB, the foster care system has been plagued with issues such as a relative overabundance of special needs children who are harder to place, difficulties in the matching process, and a lack of standardization across states.    The 1970s brought about the beginnings of the so-called “Adoption Movement” in the United States (Herman, 2005). During this time, the buzz about adoption related to long waiting periods and an “undersupply of healthy white infants relative to the number of couples seeking to adopt” (Bachrach et al., 1991, p. 715). Additionally the National Association of Black Social Workers was urging adoption agencies to shy away from interracial adoptions that were popular during the 1950s and 1960s.  They felt this was killing their race and culture (Ishizawa et al., 2006). 
     The sexual revolution came concurrently. This time of social change brought increased use of contraceptives and abortions, as well as, increased numbers of women both leaving the home to join the labor force and waiting longer to marry and/or attempting to conceive.  These new social norms resulted in decreased numbers of “healthy” domestic children available for adoption through the foster care system and private agencies.  These trends continue into the present and partially can explain the growing desire among potential adoptive families to find alternative avenues for adoption.  
     Bartholet (1993, p. 92) describes that the world breaks down into two “camps” of international adoption—those that “send” and those that “receive.”  Receiving countries are typically those considered developed.  They are characterized by societies with low birth rates and a small number of children needing homes. These are nations where contraception, sexual education, and abortion are readily available, and single parents typically keep children.  The result is less children available for adoption. The opposite characteristics hold true in nations labeled as senders.
     The pursuit of international adoption by Americans began after World War II.  “Americans were eager to bring orphans of war-torn European nations to the United States for humanitarian reasons, as well as to fill the needs of childless couples” (Kim, 1995, p. 141).  This trend also continued following both the Korean and Vietnam wars. The passage of the Refugee Relief Act of 1953 in the U.S. facilitated the adoption of South Korean children after the war and began the flow of Asian adopted children to the U.S. At one time South Korean adoptees represented half of the children who came to the U.S. through international adoption. Nevertheless during the late 1980s and early 1990s, the perspective of some sending countries shifted.  This was a period marked by nationalism and outcries to deal with the problems internally in lieu of allowing children to be adopted by foreigners.  Opponents of transracial adoption, including international adoption, argued that children needed to stay in their “communities of origin” and close to their “roots” (Bartholet, 1993, p. 97). 
     In particular, the South Korean government bowed to this pressure and decreased the number of children available for international adoption. Thus in 1993, the number of children adopted from South Korea was only 1,765, a third of what it had been in 1988 (Kim, 1995).  That same year, 1993, marked a low point in international adoption with only 6,500 recorded adoptions. However, even as South Korea was becoming more restrictive, Russia and China were opening their borders to the international adoption process.  In recent years, these two countries have allowed the largest number (in absolute terms) of children to be adopted by foreign parents.
     The United Nations has sought to address the need to protect against the trafficking and selling of children for profit. In 1993, The Hague Convention on the Protection of Children and Co-operation in Respect of Intercountry Adoption was signed.  It was designed to establish a framework for legal adoption placements internationally. While The Hague Convention set out to originally streamline and standardize the process, it left individual nations to develop their own laws and policies regarding foreign adoptions.  The Hague Convention outlines that only orphans can be adopted, and strict guidelines for who is an “orphan” dictate that birth parents’ rights are already terminated, usually by death or abandonment, and do not hinder the process. The passage of the Intercountry Adoption Act of 2000 appropriates the Hague Convention standards for the U.S.
     One valuable piece of The Hague Convention framework concisely outlines what makes a child legally “adoptable.” This can be one of the most attractive characteristics of international adoption to an American.  Foster care and private adoptions in the U.S. are sometimes delayed or reversed due to legal complications from the birth parents.  In many respects, current U.S. laws favor the rights of birth parents over prospective adoptive parents and can hinder the process of domestic adoptions.  Given the inefficiencies and heartbreak that characterize U.S. adoption history and the current foster care system, combined with the movement to improve the international adoption process and greater multicultural awareness, it is of little wonder that more and more Americans are turning abroad to complete their families.
3.  Relevant Literature
     The vast majority of empirical research related to the topic of international adoption examines behavioral and psychological outcomes of the racial and ethnic issues that face transracial adoptions. Since approximately 85% of all international adoptions are considered transracial, this is a major focus of this research area. Lee (2003) outlines the topic of transracial adoption and its social impact.  International adoption remains a controversial topic (Tizard, 1991).  Data from The National Adoption Attitudes survey conducted by the Dave Thomas Institute (2002) indicate that 47% of the population surveyed believes “international adoptees have more medical and behavior problems than domestically adopted children.” With international adoptions becoming more and more common, this is one reason why research on the topic is mainly focused on the impact and outcomes of these adoptions.  Tienman et al. (2006) finds intercountry adoptees “are less likely to have intimate relationships” (p. 68) than nonadoptees; however both groups have similar levels of educational and professional attainment.  While Lee (2003) finds that internationally adopted children, as well as domestically adopted ones, are well-adjusted and follow normal patterns of development, the author also states that more research needs to be conducted to further understand psychological adjustment and socialization. 
     There is a missing gap in the previously published literature regarding the determinants of intercountry adoption. This study attempts to gain an understanding of the determinants of international adoption by adapting a model for state-level analysis similar to the one utilized in Hansen and Hansen (2006).  That study is a cross-sectional quantitative analysis of the economics of adoption from the American foster care system.   Their analysis uses the number of adoptions from public foster care as the dependent variable and uses independent variables to understand the population who choose public adoption in the United States.  The authors state that positive externalities of the foster care system help society as a whole and that by decreasing the price of adoption, benefits to society accrue.  They find a strong positive correlation between the number of adoptions from foster care and increases in subsidies given to households who adopt.   This outcome can be viewed as an application of the law of demand from consumer theory.
     Lawmakers recognized this relationship and through passage of The Adoption and Safe Families Act (ASFA) of 1997, they dedicated $20 million in federal money to assist families in the adoption process through reimbursements and monthly assistance, especially for children with special needs.  Statistics indicate an overall increase in adoptions across almost all 50 states following the enactment of this law (Kroll, 1999). As states decrease waiting periods and streamline the process for families and social workers, there is an increase in adoptions from foster care.  Hansen and Hansen (2006) recommend using resources to help the matching process and assist case workers to become more effective and efficient in their abilities to match adoptees to appropriate families. 
     Price is not the only aspect of adoption investigated in the literature.  Socio-demographic indicators are used in understanding the demand for adoptions.  Hansen and Hansen (2006) find a positive relationship exists between the number of children adopted and median household income. A population who has a large number of people aged 25 to 45 and/or a higher percentage of married couples also is correlated positively with the number of children adopted from foster care.
     While Medoff (1993) seeks to understand the supply of adoptions, and results cannot be compared to this study; the rationale for incorporating some of the components of that study is appropriate to the present investigation. Logically, Medoff (1993) finds unemployment and labor force participation rates of women have a significant, negative correlation with the adoption rate. These variables can be integrated into the study of the total “market” for international adoptions.
     Income is a proxy for socioeconomic status.  Another proxy is educational attainment. Medoff (1993) incorporates educational attainment into the analysis as does Bachrach, London, and Maza (1991), who find that increased levels of educational attainment lead to an increase in the adoption rate, or demand, for adoptions.  
      Another economic thread investigated in Medoff (1993) and Hansen and Hansen (2006) is the concept of substitutes. Childbirth, abortion, and international adoption are investigated as possible substitutes for adoption from foster care. While Bacharach, London, and Maza (1991) present the argument that these cannot be viewed as alternatives due to inherit differences in the experiences of each case, the authors do indicate that adoption is usually a result of a rational decision making process and the costs and benefits of alternatives must be weighed.  Their results indicate negative relationships exist between domestic adoptions and the birth rate. Also, populations who conduct more international adoptions are less likely to complete domestic ones. 
4.  Data
     The data for this study is a cross section collected from all 50 states in the U.S., plus the District of Columbia, for the year 2000.  The adoption data is collected from The Adoption and Foster Care Reporting and Analysis System (AFCARS) maintained by the U.S. Department of Health and Human Services.  The remaining demographic data is obtained from the U.S. Census Bureau.
4.1 Dependent and Independent Variables
     The transformation of the dependent variable and the rationale for the independent variables integrated into the analysis are modeled after Hansen and Hansen (2006).  The dependent variable is the number of international adoptions that occurred in the year 2000.  Since we are gathering this data from across all 50 states, there is a need to control for the varying population sizes.  Hence, the final form of the variable is the total number of international adoptions per 100,000 people. 
     This study expands upon the Hansen and Hansen (2006) analysis and attempts to improve the measurement of independent variables to better capture the characteristics of the population.  While Hansen and Hansen (2006) use single variables to proxy for income, age, and education, this study breaks out each category into three separate variables.   
     The age variables are broken down into three categories:  percents of the state population who are 18-24 years old, 25-44 years old, and 45-65 years old. Hansen and Hansen (2006) argue that the age bracket of 18-45 captures the population most likely to be building families, but with the delineation of three individual age variables, a greater degree of understanding can be ascertained and can better control for any omitted variable bias.          
     The educational variables represent the percent of individuals in a given state who have completed one of three different levels of education as their highest level of educational attainment.  PCTHS captures the percent of the population of a state who received only a high school diploma or GED equivalent.  PCTBACH is the percent of the state population holding a bachelor’s degree.  PCTGRAD represents those individuals within a state who have earned a graduate or professional degree. 
     The third expanded variable captures annual income for the population.  These categories are the percents of the state population earning less than $50,000 (LOGINCOME50), earning $50,000-100,000 (LOGINCOME50100), and earning more than $100,000 (LOGINCOME100).  These variables are transformed into the natural logs.
     By definition, substitutes are expected to be negatively related to the dependent variable, since they represent alternate pathways to building a family.  Three substitutes¹ for international adoption are introduced in this model. For various reasons, certain populations may prefer to pursue domestic adoption. Hence, PCTPUB controls for the percent of all adoptions from public foster care in 2000.  A second variable controlling for substitutes to international adoption is the birthrate (BIRTH), which is the number of live births per 1,000 people within a state.  The population’s desire to build a family through biological reproduction is proxied by the BIRTH variable.  The expected sign on the coefficient for both of these variables is negative. There are also those individuals in a population who may be averse to building their families at a given point in time and who therefore may instead choose abortion.  ABORTION captures the total number of abortions per 1,000 women in a given state.  The expected sign of this variable therefore would be negative.  However, higher abortion rates could result in a situation in which fewer children are available for domestic adoption if children who otherwise would be placed for adoption are aborted.  If this is the dominant explanation of the relationship between international adoption and abortion, then the ABORTION variable would be expected to have a positive sign.
     The last two demographic variables presented are the marriage rate and the labor force participation rate of women.  MARRIED captures the percent of the population of a state who is married.  The LFPRATE variable measures the fraction of the women in each state who are active participants in the labor force.  MARRIED is expected to be positively related to international adoptions, since a number of foreign countries do not allow single adults to adopt.   LFPRATE is expected to have a negative relationship.  This expectation is based on the premise that higher labor force participation rates may proxy for women who place a greater emphasis on their careers and may wish to have either no children or a smaller family and may be less likely to add children to their family through international adoption. 
4.2  Descriptive Statistics
     After controlling for population differences, Minnesota is the state with the most per capita international adoptions in 2000, with 15.06 per 100,000 people.  Conversely, Nevada has the fewest with only 1.15 per 100,000 people.  The mean for this variable is 7.19 international adoptions per 100,000 people.  
     West Virginia has the highest percent of people with only a high school diploma at 39.4%.  California has the lowest with 20.1%.  The states with the highest and lowest percentages of their population obtaining a bachelor’s degree as the highest educational attainment level are Colorado with 21.6% and West Virginia with 8.9%, respectively.  The District of Columbia is statistically the most educated region of the U.S. with 21.0% of its population holding a graduate or professional degree.  The population of North Dakota has the lowest percent of graduate or professional degrees with 5.5%.
     As for the substitutes for international adoptions, 82% of all adoptions that occurred in Illinois in 2000 were from public foster care, while Alabama and Idaho tie for the lowest with 10% being public. The nationwide average is 33.5%.  The average birth rate across all states for 2000 is 14.3 per 1,000 people.  Utah has the highest with a birth rate of 21.9, while Maine and Vermont’s rates of 10.8 and 10.9, respectively, are the lowest.  Wyoming reported only 0.9 abortions per 1,000 women ages 15-44.  The state that reports the next lowest number is Kentucky with 5.3.  On the other hand, 68.1 is the reported abortion rate for the District of Columbia for 2000.  The next highest is New York at 39.1.  The average abortion rate across all states is 17.1. 
     Across the United States, the average percent of the population who is married in 2000 is 54.8%. Utah reports the highest number of 60%, and New York reflects the minimum number of 50%; excluding an extreme outlier of 29.9% reported for Washington, D.C.  The last piece of demographic data collected is the labor force participation rate of women.  The average across all states is 94.4%.  Minnesota has the highest percentage, with 96.6% of women who are in the labor force.  The District of Columbia has the lowest rate at 89%.  Table 1 gives a detailed chart of descriptive statistics.
5.  Model
5.1 Econometric Model
     The model presented in this study is roughly adapted from Hansen and Hansen (2006).  We assume that the determinants of international adoption can be estimated from the following regression model:

INTERCOUNTRYi = βXi + Єi

where INTERCOUNTRY is the number of international adoptions per 100,000 people, X is a vector of the previously discussed explanatory variables controlling for various social, economic, and demographic factors, and Є is the stochastic error term.  Ordinary least squares (OLS) is used to estimate the regression model.
5.2 Econometric Issues
     Due to the cross-sectional nature of the data set, heteroskedasticity is a concern.  The White Test results in a test statistic of 22.24.  That number is less than even the 10% chi-square critical value of 37.9 and suggests heteroskedasticity is not present in this model.
     There is also a potential concern regarding several groups of variables and their combined effect on the dependent variable.  F-tests are performed for the following groups individually: the three age variables, the three income variables, the three education variables, and the three substitute variables.  The model is restricted four times to test the null hypothesis that each group has no jointly significant effect on the dependent variable.  Table 2 displays the calculated F-statistics.
     Based on these results, we can reject the null hypothesis in each case and conclude that all four of these groups of variables should be included in the analysis.
6.  Results
6.1 Model 1
     The results for the regressions can be found in Table 3.2  Model 1 includes all of the explanatory variables and, generally, yields coefficient signs that are consistent with expectations.  The coefficients on the variables for the percent of the state population earning $50,000-100,000 (LOGINCOME50100), the percent of the state population who is aged 25-44 (POP25-44), MARRIED, and LFPRATE are all significant at the 5% level. Three additional variables are significant at the 1% level: POP45-65, PCTGRAD, and BIRTH. The R-squared is 0.805, and the adjusted R-squared is 0.727.
     When income is measured as median household income in Hansen and Hansen (2006), a positive but statistically insignificant relationship to domestic adoptions is reported.  The results of the present paper are dissimilar.  The coefficients for LOGINCOME50 and LOGINCOME100 indicate a negative and statistically insignificant relationship to the dependent variable.  However, LOGINCOME50100 is statistically significant at the 5% level.  While Hansen and Hansen (2006) found a positive relationship between median household income and domestic adoptions, these results indicate the percent of state populations who make $50,000-100,000 annually are more likely to adopt internationally.  The negative sign on the lowest income level tested, LOGINCOME50, possibly could be a result of minimum income requirements imposed by state adoption regulations, adoption requirements of sending countries, and international adoption agency policies. 
     Adoption agency requirements may serve as another explanation for the negative relationship between international adoptions and POP18-24.  Agencies, as well as individual foreign countries, generally impose minimum and maximum age requirements for individuals whom they consider eligible to adopt a child.  Hansen and Hansen (2006) include only the population aged 25-45 as an independent variable, resulting in a positive but statistically insignificant finding. However, a positive relationship that is statistically significant at the 5% level is found for our POP25-44 variable.  Furthermore, the coefficient on our POP45-65 variable is negative and significant at the 1% level. 
     It is expected that higher levels of educational attainment within a state population will be positively correlated with international adoptions.  PCTHS and PCTBACH both have a positive, but statistically insignificant correlation to the international adoption dependent variable.  This finding is the same as Bachrach, London, and Maza (1991), who incorporate variables for high school, some college, and undergraduate degree recipients.  PCTGRAD perhaps returned the most interesting results. The coefficient is positive and statistically significant at the 1% level.  
     The last group of independent variables to consider is the substitutes for international adoption.  PCTPUB returns an estimated coefficient which is positive but not significant.  That relationship is similar to the result reported by Hansen and Hansen (2006). This positive sign may indicate that there is a competitive aspect to adoption and a shortage of children perceived as healthy to adopt from public foster care.  In other words, as the number of foster care adoptions rises, other families are forced to turn to international adoption in an attempt to find healthy children. While the ABORTION variable proved to not be statistically significant, it is negatively related to international adoptions.³ One explanation for a negative relationship could be that abortions are a proxy for gauging the level of aversion the population has for building families.   The t-statistic for the coefficient on the BIRTH variable is -3.57, which indicates significance at the 1% level.  Hansen and Hansen (2006) find the same negative relationship; however, it proved to be insignificant when looking at domestic adoptions.   
6.2 Model 2
     While the F-test indicates that the variables LOGINCOME50, POP18-24, and PCTHS are jointly significant as a part of their variable groups listed above, each proved to be statistically insignificant in the results for Model 1.  Theoretically, given that age and income restrictions exist for households wishing to pursue international adoption, those households in each of these groups are more likely to be unable to complete the process. Furthermore, we are concerned that multicollinearity may be a problem within these variable groups. Therefore, in Model 2, we drop out LOGINCOME50, POP18-24, and PCTHS to check the robustness of the findings from Model 1.  All of the signs remain the same and are as expected.  None of the variables that were statistically significant in Model 1 lose significance; however ABORTION becomes significant at the 10% level.  Also with a t-statistic of 3.055, LOGINCOME50100 increases to a level of 1% significance, up from 5% in Model 1.4 The R-squared and adjusted R-squared increase slightly to 0.79 and 0.74, respectively.
7.  Conclusions
     This model does a relatively good job of representing the overall relationship of international adoptions to the included independent variables.   Taking into account the requirements that states, sending countries, and agencies all place on applicants, the statistical significance of LOGINCOME50100, POP25-44, MARRIED, PCTGRAD and LFPRATE variable coefficients are consistent with theoretical assumptions. It is reasonable to find that the mid-range of both the age and income variables have the strongest explanatory relationship to international adoptions, particularly since state, agency, and/or country requirements often include minimum age and income levels for potential adoptive parents.            
     The negative correlation between international adoptions and the abortion rate that is reported by both models, in addition to the statistical significance indicated in Model 2, provides contrary evidence to Bacharach, London, and Maza’s (1991) proposition that abortion is not a substitute for adoption. However, this remains an interesting area of inquiry, and future research may shed light on the decision making process that is involved in family planning.
       It is also important to mention the limitations of the model presented and ideal directions for future research. The model would have more explanatory power if it was possible to control for infertility rates of a population, as well as, the number of private adoptions.  Furthermore, individual level data rather than state level data may generate better insights.
     Despite these caveats, our study fills an important gap in the child adoption literature and contributes significantly to the understanding of international child adoptions.  We have identified the key social, economic, and demographic characteristics of the state population most likely to pursue international adoption. In addition, we have revealed what avenues represent substitutes for intercountry adoption as approaches to family building.

Footnotes
*Corresponding author.
¹We recognize there is a clear fourth substitute of private domestic adoptions (not from foster care), but a lack of data availability precludes inclusion of this variable in our analysis. Furthermore, we ideally would like to investigate infertility rates in place of birthrates but a lack of data availability also precludes such analysis.
²The observations for the District of Columbia and California contained several outliers in the data set. Thus, regressions were run including and omitting these two observations. The results indicated that these data points have no appreciable effect on the findings.  All of the signs and significance levels for the estimated coefficients remained the same whether these observations were included or not.
³A regression was run dropping the ABORTION variable from Model 1 due to its statistical insignificance to check for robustness.  All signs remain the same, and no variables which had statistical significance lost it.  Statistical significance for a few variables increased: MARRIED to 5% significance, POP25-45 to 5%; while LOGINCOME100 gained significance at the 10% level.
⁴A regression was run additionally dropping the LOGINCOME100 and PCTBACH variables from Model 2 due to their statistical insignificance.  All signs and levels of statistical significance remained the same except the level of statistical significance for ABORTION increased to the 5% level.

References
Adoption and Foster Care Analysis and Reporting System.  2002.  1995-1999:  User’s guide and codebook.  Ithaca:  Cornell
       University Family Life Development Center.
Bachrach, Christine A., Kathryn A. London, and Penelope L. Maza.  1991.  “On the Path to Adoption: Adoption Seeking in
       the United States, 1988.”  Journal of Marriage and the Family 53:  705-718.
Bartholet, Elizabeth. 1993. “International Adoption:  Current Status and Future Prospects.”  The Future of Children 3:  89-103.
Dave Thomas Foundation for Adoption.  2002.  National adoption attitudes survey.  Available at
       <http://www.davethomasfoundationforadoption.org/ASSETS/...>.
Feigelman, William and Arnold R. Silverman.  1979.  “Preferential Adoption:  A New Model of Family Formation.”  Social
       Casework:  The Journal of Contemporary Social Work  60:  296-305.
Freundlich, Madelyn.  2000.  Adoption and Ethics, Volume 2:  The Market Forces in Adoption. Washington, DC:  The Child
       Welfare League of America.
Glidden, Laraine Masters.  2000. “Adopting Children with Development Disabilities:  A Long-term Perspective.” Family
       Relations 49:  397-405.
Hansen, Mary Eschelback and Bradley A. Hansen.  2006. “The Economics of Adoption of Children from Foster Care.” 
       Child Welfare 85:  559-583.
Herman, Ellen. 2005. Adoption history project. Available at  http://darkwing.uoregon.edu/~adoption/>.
United States Department of Health and Human Services. 2002. HHS adoption awards, bonuses, and grants.  Available at
       <http://www.hhs.gov/news/press/2000pres/200000920.html>.
Ishizawa, Hiromi, Catherine T. Kenney, Kazuyo Kubo, and Gillian Stevens.  2006.“Constructing Interracial Families through
       International Adoption.”  Social Science Quarterly 87:  1207-1224.
Kim, Wun Jung.  1995.  “International Adoption:  A Case Review of Korean Children.”Child Psychiatry and Human
       Development 25:  141-154.
Kroll, Joe.  1999.  “1998 U.S. Adoptions from Foster Care Projected to Exceed 36,000.”AdopTalk, Winter:  1-2.
Lee, Richard M.  2003.  “The Transracial Adoption Paradox: History, Research, and Counseling Implications of Cultural
       Socialization.”  The Counseling Psychologist  31:  711-744.
Medoff, Marshall H.  1993.  “An Empirical Analysis of Adoption.” Economic Inquiry 31:  59-70.
National Adoption Information Clearinghouse.  n.d., Multiple entries.  Available at  http://www.calib.com/naic>.
National Committee for Adoption.  1989.   Adoption factbook.  Washington, DC:      National Committee for Adoption.
Plug, Erik and Wim Vijverberg.  2000.  “Schooling, Family Background, and Adoption:  Does Family Income Matter?”
        Journal of Political Economy 111:  611-641.
United States Census Bureau.   Statistical abstract of the United States. Multiple years. Available at
      <http://www.census.gov/compendia/statab/past_years.html>.
Tieman, Wendy, Jan van der Ende, and Frank C. Verhulst.  2006.  “Social Functioning of Young Adult Intercountry Adoptees
       Compared to Nonadoptees.”  Social Psychiatry and Psychiatric Epidemiology 41:  68-74.
Tizard, Barbara.  1991.  “Intercountry Adoption:  A Review of the Evidence.”  Journal of Child Psychology and Psychiatry 32:
         743-56.
United States Department of Health and Human Services.  n.d.   Multiple entries. Available at
      <http://www.acf.hhs.gov>.
United States Census Bureau.  2000.  United States Census 2000.  Available at 
     <http://www.census.gov/main/www/cen2000.html>.

Table 1:  Descriptive Statistics

Table 2:  F-Statistics

Table 3:  Results