Abstract
The 2012 Deferred Action for Childhood Arrivals (DACA) program granted work authorization and protection from deportation to more than 800,000 young undocumented immigrants who arrived to the United States as minors. We estimate the association between this expansion of legal rights and birth outcomes among 72,613 singleton births to high school–educated Mexican immigrant women in the United States from June 2010 to May 2014, using birth records data from the National Center for Health Statistics. Exploiting the arbitrariness of the upper age cutoff for DACA eligibility and using a difference-in-differences design, we find that DACA was associated with improvements in the rates of low birth weight and very low birth weight, birth weight in grams, and gestational age among Mexican immigrant mothers.
Introduction
In 2012, the Obama administration created the Deferred Action for Childhood Arrivals (DACA) program, which granted some undocumented immigrants who came to the United States as minors work authorization and protection from deportation. By June 2020, more than 825,000 initial DACA applications had been approved (U.S. Department of Homeland Security 2020). DACA reduces two major sources of stress for undocumented immigrants: the threat of deportation and the inability to work legally (Abrego 2018; Gonzales et al. 2014; Patler and Pirtle 2018). Studies have found that DACA increased high school graduation, employment, and public service receipt; reduced poverty; and improved the mental and self-rated health of participants (Amuedo-Dorantes and Antman 2016, 2017; Bae 2020; Hamilton et al. 2020; Kuka et al. 2020; Patler et al. 2019; Pope 2016; Venkataramani et al. 2017, 2018).
By reducing the threat of deportation and providing new economic opportunities, DACA may also have improved the health of infants born to DACA participants. The threat of deportation through immigration enforcement is a known stressor (Ayón 2020) that affects the health of pregnant women and their newborns (Novak et al. 2017; Ro et al. 2020; Torche and Sirois 2019). Deportation threat also inhibits social service usage and undermines confidence in U.S. institutions, with direct and indirect consequences for health (Cruz Nichols et al. 2018; Toomey et al. 2014; Watson 2014).
An estimated 256,000 U.S.-born children have at least one parent who is a DACA participant (Svajlenka 2019). Two studies have found health improvements among children of DACA participants (Hainmueller et al. 2017; Patler et al. 2019). Mothers who participate in DACA may have a healthier pregnancy and birth through stress reduction and improvements in socioeconomic status resulting from the program. DACA may also affect birth planning and timing (Kuka et al. 2019). Births after DACA may be healthier because participants planned and achieved pregnancies within a broader set of opportunities.
In this study, we consider how DACA affected the health of participants' children at the start of their lives by examining two standard birth outcomes: birth weight and gestational age. We exploit the arbitrariness of the upper age cutoff for DACA eligibility to identify DACA's impact on birth outcomes among Mexican immigrant women in the United States.
Research Design
We analyze U.S. birth record data from June 2010 to May 2014 provided by the National Center for Health Statistics (2018), as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program. We conduct a difference-in-differences (DID) analysis, comparing the difference in birth outcomes to DACA-eligible mothers (the treatment group) before and after DACA (the treatment) was announced with the difference in birth outcomes to DACA-ineligible mothers (the control group) before and after DACA. Because DACA may affect both stress during pregnancy and pregnancy planning, we compare births across three periods: (1) pre-DACA, referring to births before the DACA announcement in June 2012; (2) post-DACA period 1, referencing births conceived before but born after DACA was announced; and (3) post-DACA period 2, encompassing births conceived after DACA was announced. DID analyses control for time trends common to all mothers and for group differences that are common across time. The model estimates the treatment effect with an interaction term between the treatment group and the post-policy periods under the assumption that the treatment group would have followed the same trend as the control group if the treatment had not occurred.
The birth records files include four characteristics of mothers that allow us to approximate DACA eligibility: mother's birthplace, mother's state of residence, mother's age, and mother's highest level of education.1 We limit the analytic sample to Mexican-born mothers who reside in the United States and have at least a high school diploma or GED, who are most likely to be eligible for DACA (Migration Policy Institute 2020). We further limit all analyses to singleton births between 22 and 44 weeks of gestation.
We use the DACA eligibility upper age criterion (<31 on June 15, 2012) to define the treatment and control groups.2 The treatment group is mothers who were just below the upper age cutoff for DACA eligibility, 29–30 years old at the time of the birth. Ideally, we would compare mothers ages 29–30 with those ages 31–32. As Table 1 shows, however, mother's age relates to eligibility by date of birth over the post-DACA period: eligible mothers could turn 31 as early as June 16, 2012; 32 in the second half of 2013; and 33 in the second half of 2014. Furthermore, we know mother's age at the time of the birth, but we do not know her birthdate. Therefore, we cannot determine the eligibility of 31-year-old mothers who gave birth in the second half of 2012 or the first half of 2013.
To balance the model assumption of similarity between the control and treatment groups with an interest in examining births conceived after the announcement of the program, we use 33- to 34-year-olds as the control group, which includes (full-term) births conceived through August 2013 and born through May 2014.3 Our analytic sample includes 72,613 births.
We examine low birth weight (LBW), defined as birth weights of less than 2,500 grams; very low birth weight (VLBW), defined as <1,500 grams; birth weight in grams; and birth weight in grams at the 6th, 50th, and 93rd percentiles, representing the distributional cut points for LBW, median birth weight, and macrosomia, respectively. We also examine preterm birth (<37 weeks gestation); gestational age in weeks; and small for gestational age (SGA), or birth weights below the 10th percentile of birth weight at each completed week of gestation from 22 to 44 weeks. For LBW, VLBW, preterm birth, and SGA, we employ linear probability regressions with robust standard errors to obtain unbiased coefficients under heteroscedasticity; logistic regression models (not shown but available upon request) produce similar results. For birth weight in grams and gestational age in weeks, we employ ordinary least squares regressions. For birth weight at the 6th, 50th, and 93rd percentiles, we use conditional quantile regression; unconditional quantile regression (not shown but available upon request) produces similar results.
The models presented here control for birth and mother characteristics that are unlikely to be mediators of DACA's impact on health: child sex, parity, whether the mother is married, and whether the father is Hispanic. In the online appendix, we show nested models, including one with no controls and a third, fully mediated model that includes controls for whether the mother has obtained a bachelor's degree or higher, whether the birth was paid for by Medicaid, and three characteristics of the mother's county of residence: the percentage of households below the federal poverty line, median household income, and the rate of immigration detainers issued in the 12-month period prior to the birth per 1,000 people.4 The nested models show no evidence of mediation by covariates.
Results
Table 2 describes the characteristics of singleton births in the United States to high school–educated Mexican-born mothers between June 2010 and May 2014, comparing DACA-eligible (29- to 30-year-old) mothers to ineligible (33- to 34-year-old) mothers. Group differences are small and correspond to age. Infants of eligible (i.e., younger) mothers have better birth outcomes, but eligible mothers are less likely tohave a college degree or to be married. There are few county-level differences. Mexican-born mothers in our sample live in counties that are 17.7% poor, have a median income of $54,000, and where local law enforcement issued 1.6 ICE detainers per 1,000 people in the year prior to the birth.
Figure 1 shows birth outcomes for eligible and ineligible mothers by three-month period from June 2010 to May 2014. The graphs show that the groups follow parallel time trends prior to the DACA announcement for all birth outcomes. For some outcomes—especially LBW, VLBW, and birth weight at the 6th percentile—a noticeable gap emerges between eligible and ineligible women after DACA is announced and particularly in the period that begins nine months after the announcement of DACA.
We estimate the summary impact of the policy with the DID analysis. Table 3 shows condensed results for the model with no potential mediators; Table A1 in the online appendix shows full regression coefficients across three nested models. Consistent with Figure 1, the results show that DACA was not associated with birth outcomes for births in utero at time of the announcement (post-DACA period 1), but DACA was associated with improvements to some outcomes for births conceived after DACA was announced (post-DACA period 2). For the latter, DACA was associated with a 1 percentage point decrease in the proportion LBW (Model 1) and a 0.04 percentage point decrease in VLBW (Model 2). These correspond to a 28.8-gram increase in average birth weight (Model 3). The conditional quantile regressions show that improvements were concentrated among births at the bottom of the birth weight distribution (Models 4a–4c): DACA was associated with an average increase of 79 grams at the 6th percentile, an increase of 25 grams at the 50th percentile, and no difference in birth weight at the 93rd percentile. DACA was not associated with preterm births (Model 5) or SGA (Model 7) but was associated with 0.09 weeks longer average gestational age (Model 6).
Figure 2 illustrates the conditional quantile regression estimates of DACA's impact in the second post-DACA period across the birth weight distribution, adding results from a regression using the standard birth weight cut points of the 10th, 25th, 75th, and 90th percentiles. The figure makes clear that DACA's association with birth weight is concentrated at the lower end of the distribution.
Conclusion
This study supports the theory that the expansion of legal rights can lead to important improvements in the health of beneficiaries and their children (Hainmueller et al. 2017; Patler et al. 2019). Our analysis of births to Mexican-born mothers in the United States between 2010 and 2014 shows that DACA was associated with substantially improved birth outcomes for births conceived in the nine-month period following the announcement of DACA. Births conceived to eligible mothers after the DACA announcement experienced a lower risk of LBW and VLBW, and larger average birth weight. These benefits were concentrated among births at the lower end of the birth weight distribution, meaning that births at greatest risk of poor outcomes benefited the most from the program. Given the association between birth outcomes and later-life health and development, we can assume that DACA also reduced those concomitant risks (Boardman et al. 2002; Hack and Borawski 2002). We did not find evidence that the policy impacted births in utero at the time of the announcement, perhaps because such pregnancies were exposed to DACA for shorter periods.
A causal interpretation of our results assumes that there was no other event occurring at or around the DACA announcement that differentially affected births to high school–educated Mexican immigrant women who were 29–30 years old versus 33–34 years old. Our results are specific to mothers who were just below and above the upper age criteria for DACA eligibility from June 2012 through May 2014; the results may not generalize to other DACA-eligible age groups. By virtue of their age, our sample disproportionately includes married mothers having their second or higher-order birth (cf. Martin et al. 2019). DACA may affect birth timing and pregnancy health differently for younger mothers, but different research designs are necessary to identify those effects.
We cannot directly measure DACA eligibility with the data we use. Because some women in our treatment group are documented or are undocumented but do not meet the other criteria for DACA, our estimate of DACA's impact is conservative. The true effect of DACA on maternal and infant health is likely larger among the eligible population than our estimates reflect.
In June 2020, the U.S. Supreme Court determined that the Trump administration had not provided sufficient legal justification to terminate DACA, and shortly thereafter the program began accepting new applications. However, legislation to provide permanent legal status for DACA recipients and other immigrants is needed in order to put an end to the stressful uncertainty that threats to the temporary executive action created (e.g., Patler et al. 2019). As of early 2021, the Biden Administration has proposed immigration reform legislation, including a route to citizenship for undocumented immigrants, but may face an uphill political battle to the law's passage. Congress should consider evidence of DACA's direct and intergenerational health benefits as they debate the Biden proposal. Millions of immigrants and their U.S.-born children would benefit from such a law.
Acknowledgments
This work was supported by the National Science Foundation (Award #1822787) and the UC Davis Global Migration Center. We thank Caprice A.R. Edwards at the National Association for Public Health Statistics and Information Systems (NAPHSIS), Susan B. Long at the Transactional Records Access Clearinghouse at Syracuse University, and Dewey LaRochelle at the Division of Vital Statistics at the National Center for Health Statistics for their help accessing the data we use in this study. We thank participants at the UC Davis Sociology Department Colloquium held in March 2020, Konrad Franco, and Ryan Finnigan for their feedback and help on the work in progress. Any errors are our own.
Notes
The Department of Homeland Security limited DACA eligibility to undocumented immigrants who (1) were resident in the United States from June 15, 2007, to June 15, 2012; (2) were between the ages of 15 and 30 on June 15, 2012; (3) arrived in the United States before June 15, 2007, at the age of 16 or younger; (4) had completed high school or a GED, were enrolled in school, or were active military or honorably discharged veterans; and (5) had not been convicted of a felony, significant misdemeanor, or three or more other misdemeanors, and did not otherwise pose a threat to national security or public safety. Birth records do not include information about mother’s legal status, year or age of arrival, current enrollment in school, military service, or criminal record.
We do not use the lower age cutoff because births to 14-year-olds are rare and are unlikely to follow patterns similar to those of births to older mothers (Martin et al. 2019).
We estimated the same models using alternate control groups: (1) births to 32- to 33-year-olds through May 2013, and (2) births to 34- to 35-year-olds through May 2015. The results using the first alternate control group showed no effects of DACA on births through May 2013. The results for the second alternate control group were consistent with those presented here. An alternate design would be to compare Mexican-born women with similarly aged U.S.-born women, but these groups violate the DID model’s assumption of similar time trends in the pre-DACA period.
County economic characteristics come from the Census Bureau’s Small Area Income and Poverty Estimates program (https://www.census.gov/programs-surveys/saipe/data/datasets.html). The county detainer data are from the Transactional Records Access Clearinghouse at Syracuse University (https://trac.syr.edu/). Detainers hold a person suspected of an immigrant violation, who would otherwise be released after apprehension by local law enforcement, until Immigration and Customs Enforcement (ICE) can assume custody. The detainer rate approximates local immigration enforcement.
