Tulane Economics Working Paper Series A Pint for a Pound? Reevaluating the Relationship Between Minimum Drinking Age Laws and Birth Outcomes Alan I. Barreca Department of Economics Tulane University New Orleans, Louisiana [email protected] Marianne E. Page Department of Economics University of California, Davis Davis, California [email protected] Working Paper 1220 October 2012 Abstract Previous research documents a substantive, positive, correlation between the minimum legal drinking age (MLDA) and birth outcomes. Using an improved empirical framework, we reach a different conclusion: there is little or no relationship between the minimum legal drinking age and the health of infants born to young mothers. We do, however, find that MLDA policies are associated with the sex ratio at birth. Our estimates suggest that raising the MLDA may reduce fetal losses. Keywords: alcohol, minimum drinking age, infant health, birthweight, fetal death JEL: I18, J13 A Pint for a Pound? Reevaluating the Relationship Between Minimum Drinking Age Laws and Birth Outcomes Alan I. Barreca Department of Economics Tulane University 206 Tilton Hall New Orleans, LA [email protected] Marianne E. Page Department of Economics University of California Davis One Shields Avenue Davis, CA [email protected] October 2012 Abstract Previous research documents a substantive, positive, correlation between the minimum legal drinking age (MLDA) and birth outcomes. Using an improved empirical framework, we reach a different conclusion: there is little or no relationship between the minimum legal drinking age and the health of infants born to young mothers. We do, however, find that MLDA policies are associated with the sex ratio at birth. Our estimates suggest that raising the MLDA may reduce fetal losses. Keywords: alcohol; minimum drinking age; infant health; birthweight; fetal death. We thank seminar participants at University of California-Merced, University of Colorado- Denver, and University of California-Irvine for their comments. We gratefully acknowledge funding support from National Institutes of Health (R01AA017990-01). We are solely responsible for the views expressed in the article. Barreca is an Assistant Professor at Tulane University and can be contacted at [email protected]. Page is a Professor at University of California, Davis and can be contacted at [email protected]. I. Introduction Do restrictive alcohol policies affect birth outcomes? Several studies have found evidence that minimum legal drinking ages (MLDA) affect young women’s alcohol consumption and the probability of engaging in risky sexual activity.1 Taken together with evidence on the negative correlation between prenatal alcohol consumption and birth defects2, it would seem that the answer to the question must be Yes. Yet, surprisingly little is known about whether, and through which mechanisms, the MLDA affects birth outcomes. The goal of our research is to further explore this question using cross-state changes in MLDA laws that occurred during the 1970s and 1980s. One previous study has used the same policy variation and documented a substantive, positive, correlation between MLDAs and birth outcomes. Fertig and Watson (2009) (hereafter FW) correlate changes in the MLDA with measures of health among infants born to women ages 14-20. To further control for policy endogeneity, FW use births to women ages 21-24 as an additional control group. FW find that, among young mothers, a lower MLDA is associated with a higher incidence of low birthweight and premature birth and they note that this result is most likely due to selection. In particular, they find that a lower drinking age is associated with lower levels of education among white mothers and the absence of paternal information on black infants’ birth certificates. This selection effect is further supported by Dee (2001), who finds that increasing the MLDA to age 21 reduced childbearing rates among black teens by roughly 6 percent, but had no statistically significant effect on whites. Using a similar identification strategy, we reach a different conclusion. Specifically, we find little or no correlation between an MLDA of 18 and birth outcomes once age specific time trends and state-by-age fixed effects, are included. FW acknowledge that their estimates are sensitive to the inclusion of age-specific trends; however, throughout their manuscript they continue to reject the null hypothesis that increases in the MLDA have no impact on birth outcomes.3 We present evidence that our specification better addresses possible omitted variables 1 See, for example, Carpenter (2005); Cook and Moore (2002); Cooper (2002); Grossman and Markowitz (2005); Kaestner (2001); Kaestner and Joyce (2000); Markowitz et al. (2005); Rashad and Kaestner (2004); Rees et al. (2001); and Sen (2002). 2 A non-exhaustive list of recent studies includes Albertsen et al. (2004); Berkowitz et al. (1982); Jaddoe et al (2007); Kesmodel et al. (2000); McDonald, et al. (1992); Mills et al. (1984); Shu et al. (1995); Whitehead and Lipscomb (2003); Windham et al. (1995). 3 Neither FW’s abstract, introduction, nor conclusion sections discuss this important robustness check, nor are the results of this check presented in the tables of the published manuscript. 2 and document the importance of our preferred control variables by examining their economic and statistical significance in 12 “placebo” states, which consistently maintained a MLDA of 21 during our sample period.4 The control variables turn out to have strong predictive power in both treatment and placebo states, which suggests that previous evidence on the relationship between MLDA laws and birth outcomes is likely driven by omitted variables. At the same time, we present novel evidence that a low MLDA increases the probability of fetal loss. Assuming that the surviving fetuses are the healthiest, our finding implies that selection may be biasing the estimated relationship between access to alcohol and infant health towards a positive relationship. In other words, the implications of our study are considerably different from FW. Our results suggest future research should explore the link between alcohol access and fetal loss, while FW’s results highlight the role of maternal selection into pregnancy. Further research is needed to establish the mechanisms through which alcohol policies affect birth outcomes. II. Background on Minimum Drinking Age Laws Empirical analyses of the relationship between restrictive alcohol policies and birth outcomes have largely used variation in minimum drinking age laws across states and over time as their source of identifying variation. Prior to the 1970s, most states maintained a legal drinking age (MLDA) of 21, but during the early 1970s a number of states reduced their MLDA, and in many states the new drinking age was as low as 18. A subsequent rise in the number of alcohol related fatalities induced the majority of these states to raise their minimum drinking age to 21 again during the 1970s and 1980s. With the passage of the Uniform Drinking Act, all states were compelled to adopt an MLDA of 21 by 1988.5 Figure 1 illustrates the changes in the MLDA between 1977 and 1989. For example, the number of states with a MLDA of 21 increased from 12 in 1977 to 23 in 1985, and 51 (including the District of Columbia) by 1988. We will exploit this variation to identify the effect of the laws on infant health. 4 These 12 states include: Arkansas, California, Indiana, Kentucky, Missouri, Nevada, New Mexico, North Dakota, Oregon, Pennsylvania, Utah, and Washington. Note that many of these states are geographically clustered. For example, California, Oregon, Nevada, and Washington, and are thus less likely to be affected by MLDA law changes in treated states. 5 Although the law mandating a MLDA of 21 was passed in 1984, the law was not enforced until late in 1986. The law withheld a portion of a state's Federal highway funds if the state did not enact a MLDA of 21 by October 1, 1986. (Distilled Spirit Council of the United States 1996) 3 III. Data Data on MLDA laws come from the Distilled Spirits Council of the United States.6 We have information on the month and year that the MLDAs changed for each state, between 1978 and 1988. We match the MLDA data with birth outcome data by mother's estimated age at conception, the month the child was conceived, and the mother’s state of residence at delivery. As we discuss below, treatment is assigned based on the mother's age at the time of conception, and whether the mother was residing in a “low MLDA” state (i.e. a state with a MLDA of 18). The birth-outcome data used in our state-year MLDA analyses come from the National Center for Health Statistics (NCHS) public-use Natality Files. The NCHS data are derived from information reported on birth certificates and include the near universe of all U.S. births that occurred in the 1970s and 1980s.7 The NCHS data have information on the child's gender, birth weight, length of gestation, presence of congenital anomalies, and five-minute Apgar score. Birth weight and Apgar score are commonly used indicators of infant health at birth and have been linked to a number of long-term outcomes (e.g. Black et al., 2007; Oreopoulos et al., 2008).8 In addition, we construct dichotomous indicator variables for whether the child's birth weight is below 2500 grams ("low birth weight") and whether the length of gestation was under 37 weeks ("premature birth"). We include the fraction of births that are female as a dependent variable because evolutionary theory suggests that males are more sensitive than females to in utero health shocks (Trivers and Willard, 1973) and because several recent studies have found evidence that stressors during pregnancy affect the probability of bearing a male child (Cagnacci et al., 2004; Almond et. al., 2007; Nilsson, 2008; Sanders and Stoecker, 2011). Note that FW do not present results for Apgar score or fraction female.9 To discern the mother’s age during her pregnancy, we rely on information in the NCHS data on the mother’s age at delivery, the mother's state of residence at delivery, and the month of the child's birth. A limitation of the data is that we observe the mother's age (in whole years) at the time of child's birth but not at the time of conception so we cannot tell whether the mother 6 We are grateful to Thomas Dee for providing us with these data. The vast majority of states reported 100-percent samples. The remaining states report 50-percent samples. 8 Apgar scores are on a 10-point scale based on five categories of infant health. 9 FW state that “Apgar scores are not available in the Vital Statistics data during the relevant time period” (footnote 17 p. 743). However, there were actually 38 reporting states as of 1978. The number of states reporting Apgar score increased gradually to 47 states by 1989. 7 4 was the same age during the majority of her pregnancy. To be consistent with FW, we assume the mother's age at conception is one year less than her age at the child's birth.10 Our NCHS analyses are restricted to mothers who are between 14 and 24 years of age at the time of conception. Women over the age of 21 are legally able to drink throughout the entire period under study, so changes in the MLDA should not causally affect these women (assuming there are no intertemporal effects or spillovers.) As such, changes in the health outcomes of infants born to these women can be used to control for within-state changes in birth outcomes that may be spuriously correlated with MLDA changes, or for any effects of the MLDA that are common to all age groups. For example, the MLDA policy changes may be endogenous to the changes in health conditions that affect infants born to women of all ages. Summary statistics for our sample of births are presented in Table 1. IV. Estimation Approach Our identification strategy, which builds on FW’s approach, exploits variation in the MLDA laws across states and over time. Because the MLDA laws vary across states, years and cohorts we conduct our analyses at the state-year-age cohort level. We begin by estimating the following regression via ordinary least squares: (1) Yast = !1 MLDA18st + !2 MLDA18st * age14-17ast + !3 MLDA18st * age18-20ast + "s + #t + $a + %s * t + &cst where Yast is an average outcome for infants born to mothers of age a residing in state s at the time of conception t;11 MLDA18st is an indicator for whether the drinking age in the mother’s state of residence s was 18 at time t, MLDA18st is interacted with age14-17ast and age18-20ast to allow the effects of the MLDA to vary for mothers who were 14 to 17 and 18 to 20 years old, respectively12; "s are state fixed effects; #t are year-month fixed effects; $a are age fixed effects; and %s * t are state-specific linear time trends. The inclusion of state-specific time trends allows us to account for the possibility that variation in the MLDA is correlated with unobserved factors 10 Following FW, we assume that the mother’s age at conception was one year less than her age at birth for all gestational lengths over 26 weeks. For gestational lengths less than 26 weeks, we assume the mother’s age at delivery and age at conception are equal. 11 The month the mother conceived the child can be inferred by using publicly available information on length of gestation; in the cases where gestation is not reported, we assume gestation began 40 weeks prior to the date of the child's birth. 12 We focus on whether the state had an MLDA of 18 in order to be consistent with FW. We have run similar models where we replace the MLDA18 dummy and its interactions with indicators for whether the mother could legally drink at the time of conception. These models generally produce qualitatively similar estimated relationships. 5 that vary by state and year that might affect infant health. States that initially had a low MLDA and were required to raise the MLDA to 21 when the Uniform Drinking Act was passed in 1984, are arguably different from states that maintained an MLDA of 21 throughout the 1970s and 1980s. The speed with which states complied with changes in the federal MLDA during the late 1980s may also be nonrandom. We address the possibility that the error term & is correlated within states by clustering our standard error estimates at the state level. Equation (1) is qualitatively similar to the model that FW use and produces very similar estimates.13 The two control groups in this model include: (i) young mothers in states where the MLDA was consistently above 18, and (ii) mothers who were too old to be affected by the MLDA changes. The MLDA was never higher than 21 during the sample period, therefore, infants born to mothers between the ages of 21 and 24 can effectively serve as the second control group. After estimating equation (1) we systematically add two important sets of control variables. First, we add age-specific linear trends. The inclusion of these trends allows us to account for the possibility that the estimated ! coefficients are capturing a convergence in birth outcomes among mothers of different ages over time. Second, we also include state-by-age fixed effects. This allows us to control for the possibility that within state differences in birth outcomes between mothers of different ages are correlated with states’ MLDA policies. For example, increases in the MLDA may have coincided with other initiatives designed to improve birth outcomes among young mothers. Alternatively, the MLDA could be a response to recent declines in birth outcomes to young mothers. We show that including these statistically significant controls substantively affects the estimated impact of alcohol access on birth outcomes. FW acknowledge that adding age trends reduces the estimated impact of the MLDA laws.14 They offer two hypotheses: First, there were secular trends in birth outcomes that impacted mothers differentially by age. Second, controlling for age-specific time trends absorbs “useful variation”, making identification problematic. Although we cannot rule out the second hypothesis, we present empirical evidence (below) to support the first and more worrisome 13 FW rely on individual level data and control for infant sex and plurality. We disagree with controlling for these characteristics since they are potentially endogenous to changes in the MLDA For example, we find that the probability of a female birth increases with an MLDA of 18. 14 FW show how their estimates change with the inclusion of the age-specific trends in the NBER working paper version of their manuscript, but this robustness check is not presented in the published manuscript. 6 hypothesis.15 FW do not thoroughly test the robustness of their results to the inclusion of stateby-age fixed effects.16 To help demonstrate the economic importance of our preferred control variables, we present two sets of figures. First, Figure 2 plots differences in the fraction of low birthweight and premature births to 14-17 old mothers and to 21-24 year old mothers. The differences are plotted separately for states that did (“treatment states”), and did not (“placebo states”), experience changes in their MLDA laws between 1978 and 1988. Panel A.1 illustrates that, relative to infants born to 21-24 year old mothers, there was a notable decrease in the fraction of infants born to 14-17 year old mothers who were classified as low birthweight. Panel B.1 illustrates that 18-20 year old mothers also experienced improved outcomes over time, relative to 21-24 year olds. Differences in pre-term delivery rates (Panels A.2 and B.2.) show similar patterns. Importantly, trends are similar in both the “treatment” and “placebo” states, which strongly suggests that the patterns are driven by omitted factors and secular trends that are unrelated to the changes in the MLDA. Thus, comparing birth outcomes to younger vs. older women, and failing to control for age-specific trends, likely results in coefficient estimates that overstate the benefits of MLDA increases. These relationships are similar for white mothers (Figure A1) and black mothers (Figure A2). Second, Figure 3 illustrates the age-outcome profiles for mothers in treatment and placebo states. State-by-age fixed effects help control for the possibility that maternal ageoutcome profiles differ between treated and untreated states. One can see immediately that such differences exist: Panels A and B show that, compared to young mothers in placebo states, young mothers in treated states give birth to less healthy infants. In 1978, for example, the probability that a 14-year-old mother gave birth to a low birthweight baby was approximately 0.12 in treated states, and approximately 0.11 in placebo states. Treatment and placebo differences are less stark among the older mothers in our sample. In 1978, for example, the 15 Testing the second hypothesis is difficult. However, we note that including age-specific time trends does not absorb useful “statistical variation”. In general, we find that the estimated standard errors increase with the addition of the age-specific time trends, although only slightly. For example, when the dependent variable is low birthweight, the standard error on the interaction term MLDA is 18 x mother is 14-17 increases from 0.157 to 0.183 (columns 1 and 2 of Table 2, respectively). However, when both age-specific trends and state-by-age fixed effects are included, the standard error estimates are actually smaller. In the previous example, the standard error is now 0.080 (column 3). Thus, we can rule out the possibility that our model is absorbing a substantial portion of the statistical variation. 16 FW mention testing the robustness of their results to the inclusion age-by-year interactions, state-by-age interactions and state–by–year interactions together. However, they note that they “do not have the power to identify any significant effects using this less restrictive model.” (p. 744) 7 probability that a 24-year-old mother gave birth to a low birthweight baby was close to 0.06 across both treatment and placebo states. Thus, the maternal age-outcome profile is both higher and steeper in treatment states, albeit slightly. The fact that treatment states had a steeper age-outcome profile in 1978 is consistent with a negative relationship between early access to alcohol and young mothers’ birth outcomes. If MLDA laws were truly behind the observed pattern, however, then we would expect differences between treatment and placebo states to narrow over the period of study. As can be seen in Panel B, the differences across states persist in 1988. Given that babies born to young mothers living in treated states were less healthy than those born to young mothers in placebo states throughout the sample period, failing to control for state-by-age fixed effects may lead to an overestimate of the effects of an MLDA of 18 on birth outcomes. The age-outcome profiles for white women and black women are presented in Figure A3 and Figure A4, respectively; the relationships follow a similar, although noisier, pattern across races. Figure 2 and Figure 3 provide evidence that age-specific trends and state-by-age fixed effects are likely to be economically important controls. In the regressions below, we also test the statistical significance of these controls. Our statistical analyses reject the null hypothesis that the controls do not belong in the model. V. Results Table 2 presents our estimates of the relationship between MLDAs and probability of being classified as a low birthweight or premature birth. We present estimates produced by four specifications: the first specification includes dummy variables that control for state of birth, month and year of birth, and mother’s age, along with state-specific time trends. This specification is qualitatively similar to that used by FW and produces estimates that are nearly identical.17 The second specification adds an age-specific linear trend to the set of control variables. Our third specification is similar to the first, but includes state-by-age fixed effects. Our fourth specification includes all of the controls. When age-specific trends and state-by-age fixed effects are omitted from the regression (columns 1 and 5), we find that an MLDA of 18 is associated with worse birth outcomes for both 17 For example, the three key coefficient estimates in Panel A column (1) are -0.18, 0.49, and 0.24. FW’s analogous coefficient estimates are -0.17, 0.50, and 0.26, respectively. 8 the affected group (infants born to 18-20 year olds), and infants born to women who are younger than the MLDA. The coefficient estimates suggest that relative to 21-24 year old mothers, and MLDA of 18 increases 18-20 year old mothers’ probability of giving birth to a low birthweight baby by 0.24 percentage points. The probability of a premature birth increases by 0.16 percentage points. To put these estimates into perspective, about 8 percent of infants born to 1820 year old mothers are born weighing less than 2500 grams, and about 12 percent are born before 37 weeks. The estimates, therefore, appear to be economically as well as statistically significant. The estimated impact of the MLDA on the health of infants born to older mothers is also substantive (and in the opposite direction). Among women 21-24 years old, who are too old to be materially affected by a change in the minimum legal drinking age,18 an MLDA of 18 is associated with a 0.18 percentage point lower probability of having a low birthweight baby, and a 0.26 percentage point lower probability of having a premature birth. This tells us that births to older mothers are helping to control for omitted factors that are associated with the MLDA changes. We do not know what these omitted factors are, but the fact that the estimated coefficient on MLDA is 18 is large (nearly half the magnitude) and of the opposite sign, is a concern. One potential explanation is that the estimated effects are biased by differential trends across age groups and/or differential age-outcome profiles in treated states. We investigate this possibility by adding age-specific trends to the regression in columns (2) and (6). Including age trends reduces the estimated MLDA coefficients dramatically. The estimated coefficient on the variable MLDA is 18 x Mother is 14-17 falls from 0.49 (column 1) to 0.17 (column 2) and is no longer statistically significant. For premature births, the MLDA is 18 x Mother is 14-17 estimate drops from 0.80 (column 5) to 0.48 (column 6) for 14-17 year olds. Similar to Figure 2, the diminished coefficient is consistent with the increase in MLDAs being correlated with relative improvements in the health of young mothers’ infants. In addition to their economic significance, an F-test of joint significance indicates that the control variables are statistically significant at the one percent level. The inclusion of age-by-state fixed effects without the age-specific trends (columns 3 and 7) has a modest impact on the estimates for premature births, but little impact on the estimates for low birthweight. For premature birth, the coefficient on MLDA is 18 x Mother is 14-17 drops 18 It is possible that 21-24 year olds may have been affected by the MLDA law change earlier in life. 9 from 0.80 (column 5) to 0.51 (column 7) once state-by-age fixed effects are included. For low birthweight, the coefficient falls only slightly from 0.49 (column 1) to 0.44 (column 3). As with the age-specific trends, the state-by-age fixed effects are jointly statistically significant at the one percent level, consistent with the notion that the age-outcome profiles vary substantively across states.19 Once the age-specific trends and state-by-age fixed effects are included (column 4 and 8), nearly all the coefficients are statistically insignificant. For example, among mothers ages 14-17, there is now a negative but statistically insignificant relationship between an MLDA of 18 and the probability that an infant is low birthweight. However, there is a positive and statistically significant relationship between an MLDA of 18 and low birthweight among mothers’ ages 1820, although the coefficient is less than half the magnitude (0.11 versus 0.24). The estimates on preterm birth are both statistically and economically insignificant. Importantly, the F-statistics on the added controls are statistically significant. This evidence, coupled with Figures 2 and 3, strongly suggests that both age-specific trends and state-by-age fixed effects belong in the regression model. Table 3 and Table 4 are the analog of Table 2, except the samples are restricted to native white mothers and native black mothers, respectively. There are a few estimates to highlight: first, the age-specific trends affect the estimates much more than the state-by-age fixed effects. Second, among white mothers 14-17, there is a statistically significant negative relationship between an MLDA of 18 and low birthweight once the preferred controls are included (Table 3 column 4). As in Table 2, the coefficient on MLDA is 18 x Mother is 18-20 is positive and statistically significant. Third, when we include our preferred controls and focus on pre-term births to black women, there is a statistically significant negative coefficient on MLDA is 18 x Mother is 18-20 (Table 4 column 8). This implies that an MLDA of 18 is correlated with better birth outcomes. Table 5 shows the relationship between alcohol policies and several other birth outcomes, by race of mother, using our preferred set of control variables. Nearly all of the coefficient estimates on the MLDA interactions are small and statistically insignificant. The patterns are similar for both white and black mothers. There are a few notable exceptions: First, an MLDA of 19 For this test of joint significance on state-by-age terms, we assume iid errors. With standard errors clustered at the state of residence, there are insufficient degrees of freedom to test the joint significance of the approximately 500 stage-by-age terms. Thus, this F-test should be interpreted as an upper bound of statistical significance. 10 18 is associated with a statistically significant increase in birthweight for both white and black mothers who are between the ages of 14-17. Second, among 14-17 year old black mothers, an MLDA of 18 is associated with a statistically significant reduction in the probability that the Apgar score is less than or equal to 5. Third, there is a statistically significant decrease in the probability of low Apgar scores (i.e. less than or equal to 5), especially among black women. Fourth, among 18-20 year old mothers, an MLDA of 18 is associated with a higher fraction of births that are female, especially among the sample of black mothers (Panel C). The magnitude of the relationship between an MLDA of 18 and percent female is meaningful: for black women 18-20, an MLDA of 18 increases the probability of giving birth to a female by 0.5 percentage points.20 Taken together with the fact that we see a statistically significant decrease in the probability of a low Apgar score, these results may be indicative of higher rates of fetal deaths (Trivers and Willard, 1973), and positive selection.21 Previous studies have found evidence that MLDA laws affect a different type of selection other than fetal loss. In particular, Dee (2001) finds that MLDA laws affect selection into motherhood. Table 6 explores this possibility in the presence of age specific trends and state-byage fixed effects. The coefficient estimates on the interaction term MLDA is 18 x Mother is 1417 suggest that an MLDA of 18 decreases the fraction of women without a high-school degree by 0.2 percentage points, and increases the fraction of births where the father’s information is missing by 1.1 percentage points, but most of these estimates are not statistically different from zero. The estimated relationships among mothers 18-20 are qualitatively similar, but smaller in magnitude. On the whole these estimates provide inconclusive evidence that an MLDA of 18 increases the fraction of births to more economically disadvantaged groups. In sum, when the full spectra of outcomes are considered in light of the full set of controls, we cannot reject the hypothesis that an MLDA of 18 has no impact on birth outcomes. The estimated relationship between MLDAs and the sex ratio at birth hints at potentially important selection effects, however. Such effects likely bias the estimated impact of infant health towards zero. Thus, while we believe that previous studies have overstated the magnitude 20 Assuming all the fetal deaths were males (i.e. a lower bound), our estimates imply an MLDA of 18 is associated with approximately 1,090 additional fetal deaths per 100,000 births for black women 18-20. 21 We cannot rule out the possibility that the MLDA selects mothers who are more likely to give birth to a female into the sample. 11 of the correlation between MLDAs and birth outcomes, our improved regression model points to an alternative link through which MLDA laws may exert important effects. VI. Conclusion Existing research suggests that raising the MLDA might lead to better birth outcomes via changes in the selection of women who give birth. The analyses presented in this paper suggest that such a conclusion may be premature. We show that after controlling for age-specific time trends and cross state variation in the maternal age gradient, there is little correlation between higher MLDAs and birth outcomes. Our results do not necessarily imply that prenatal exposure to maternal alcohol consumption has no effect on birth outcomes because our design focuses on the effects of MLDA policies, which may have little effect on the drinking behavior of pregnant women. Previous studies have shown that the MLDA does substantively affect drinking among young women (e.g. Carpenter and Dobkin, 2009) but it is less clear how such barriers affect the drinking behavior of pregnant women. Sample selection also complicates the interpretation of the estimates. Dee (2001) notes that higher MLDAs are associated with higher birth rates among black mothers. We also see an increase, although statistically insignificant, in the fraction of births for which paternal information is missing from the birth certificate. This selection may bias the estimates towards finding a negative relationship between a low MLDA and birth outcomes. Conversely, we find evidence that a low MLDA is associated with increases in fetal loss, and the associated positive selection of surviving fetuses may make it appear as though a low MLDA improves birth outcomes. It is unclear whether the maternal composition effects are larger than the fetal loss effects, so the net direction of the bias is difficult to determine. If we wish to obtain a clear understanding of how these policies affect infant health then engaging in further research that disentangles these countervailing forces is imperative. Another explanation for our inability to reject the null hypothesis is that MLDA laws affect mothers’ drinking behavior, but the effects of in utero exposure to alcohol surface later in childhood and/or adulthood. The infant health information that is available on birth certificates is collected within hours (or, more usually, minutes) of a child’s birth and represents only a tiny fraction of the child outcomes that might be affected. Using a natural experiment in Sweden, 12 Nilsson (2008) finds large long-term impacts as a result of changes in alcohol policies. Further research is needed to assess the full impact of MLDA laws on children’s long-term development. The main conclusion that we draw from our analyses is that, contrary to previous findings, the impact of MLDA policies on birth outcomes is still largely unknown. Our results resonate with Armstrong (2003) who argues that, given the quality of existing research, Americans may have jumped too quickly to the conclusion that consumption of alcohol during pregnancy in any amount has devastating consequences. Although the relationship between alcohol access and infant outcomes is intuitive, our research indicates that the causal link is far from established. 13 References Albertsen, K., A.N. Anderson, J. Olsen, and M. Gronbaek (2004). “Alcohol Consumption during Pregnancy and the Risk of Preterm Delivery.” American Journal of Epidemiology, 159(2): 155-161. Almond, Douglas, Lena Edlund and Marten Palme (2007). “Chernobyl’s Subclinical Legacy: Prenatal Exposure to Radioactive Fallout and School Outcomes in Sweden.” NBER Working Paper #13347. Armstrong, Elizabeth M. (2003). Conceiving Risk, Bearing Responsibility: Fetal Alcohol Syndrome and the Diagnosis of Moral Disorder. Johns Hopkins University Press, Baltimore, MD. Berkowitz, G.S., T.R. Holford, and R.L. Berkowitz (1982). “Effects of Cigarette Smoking, Alcohol, Coffee and Tea Consumption on Pre-term Delivery.” Early Human Development, 7: 239-50. Black, Sandra E., and Paul J. Devereux, and Kjell G. Salvanes (2007). “From the Cradle to the Labor Market? The Effect of Birth Weight on Adult Outcomes.” The Quarterly Journal of Economics, 122(1): 409-439. Cagnacci, A., A. Renzi, S. Arangino, C. Alessandrini and A. Volpe (2004). “Influences of Maternal Weight on the Secondary Sex Ratio of Human Offspring.” Human Reproduction, 19(2): 442. Carpenter, Christopher (2005). “Youth Alcohol Use and Risky Sexual Behavior: Evidence from Underage Drunk Driving Laws.” Journal of Health Economics, 24(3): 613-28. Carpenter, Christopher and Carolos Dobkin (2009). “The Effect of Alcohol Consumption on Mortality: Regression Discontinuity Evidence from the Minimum Drinking Age,” American Economic Journal - Applied Economics 1(1): 164-182. Cook, Philip and Michael Moore (2002). “The Economics of Alcohol Abuse and AlcoholControl Policies.” Health Affairs, 21(2): 120-133. Cooper, M.L. (2002). “Alcohol Use and Risky Sexual Behavior Among College Students and Youth: Evaluating the Evidence.” Journal of Studies on Alcohol, 14(14): 101-117. Dee, Thomas (2001). “The Effects of Minimum Legal Drinking Ages on Teen Chidlbearing.” Journal of Human Resources, 36: 823-838. Distilled Spirit Council of the United States (1996). “Minimum Purchase Age by State and Beverage, 1933-Present.” Washington, DC: Distilled Spirits Council of the United States. 14 Fertig, Angela, and Tara Watson, (2009). “Minimum drinking age laws and infant health outcomes.” Journal of Health Economics, 28: 737-747. Grossman, Michael and Sarah Markowitz (2005). “I Did What Last Night? Adolescent Risky Sexual Behaviors and Substance Use.” Eastern Economic Journal, 31(3): 383-405. Jaddoe V., R. Bakker, A. Hofman, J. Machenback, H. Moll, E. Steegers, and J. Witteman (2007). “Moderate Alcohol Consumption During Pregnancy and the Risk of Low Birthweight and Preterm Birth: The Generation R Study.” Annals of Epidemiology, 17(10): 834-40. Kaestner, Robert (2000). “A Note on the Effect of Minimum Drinking Age Laws on Youth Alcohol Consumption.” Contemporary Economic Policy, 18(3): 315-325. Kaestner, Robert and T. Joyce (2001). “Alcohol and Drug Use: Risk Factors for Unintended Pregnancy.” in M. Grossman and C. Hsieh (eds.) The Economic Analysis of Substance Use and Abuse: The Experience of Developed Countries and Lessons for Developing Countries, Edward Elgar Limited, United Kingdom. Kesmodel, U., Olsen, S.F., Secher, N.J. (2000). “Does Alcohol Increase the Risk of Preterm Delivery?” Epidemiology, 11: 512-18. Markowitz, Sara, Robert Kaestner, and Michael Grossman (2005). “An Investigation of the Effects of Alcohol Consumption and Alcohol Policies on Youth Risky Sexual Behaviors.” American Economic Review, 95(2): 263-266. McDonald, A.D., B.G. Armstrong, and M. Sloan, (1992). “Cigarette, Alcohol, and Coffee Consumption and Prematurity,” American Journal of Public Health, 82: 87-90. Mills, J.L., B.I. Graubard, E.E. Harley, G.G. Rhoads, and H.W. Berendes (1984). “Maternal alcohol Consumption During Pregnancy: How Much Drinking During Pregnancy is Safe?” Journal of the American Medical Association, 252(14): 1875-1879. Nilsson, J. Peter (2008). "Does a pint a day affect your child's pay? The effect of prenatal alcohol exposure on adult outcomes." cemmap working paper CWP22/08. Oreopoulos, Philip, Mark Stabile, Randy Walld, and Leslie L. Roos (2008). “Short-, Medium-, and Long-Term Consequences of Poor Infant Health: An Analysis Using Siblings and Twins.” Journal of Human Resources, 43:88-138. Rashad, I., and R. Kaestner (2004). “Teenage Sex, Drugs, and Alcohol Use: Problems Identifying the Cause of Risky Behaviors.” Journal of Health Economics, 23: 493-503. Rees, D.I., Laura Argys, and Susan Averett (2001). “New Evidence on the Relationship between Substance Use and Adolescent Sexual Behavior.” Journal of Health Economics, 20(5): 835845. 15 Sanders, Nicholas J. and Charles F. Stoecker (2011) “Where Have All the Young Men Gone? Using Gender Ratios to Measure Fetal Death Rates.” NBER Working Paper 17434. Sen, B (2002). “Does Alcohol-Use Increase the Risk of Sexual Intercourse Among Adolescents? Evidence from the NLSY97.” Journal of Health Economics, 21: 1085-1093. Shu, X. O, M.C. Hatch, J. Mills, J. Clemsn, M. Susser, (1995). “Maternal Smoking, Alcohol Drinking, Caffeine Consumption, and Fetal Growth: Results from a Prospective Study.” Epidemiology, 6(2): 115-120. Trivers, Robert L. and Dan E. Willard (1973). “Natural Selection of Parental Ability to Vary the Sex Ratio of Offspring.” Science, 179 (4068): 90. Windham, G.C., L. Fenster, B. Hopkins and S.H. Swan (1995). “The Association of Moderate Maternal and Paternal Alcohol Consumption with Birthweight and Gestational Age,” Epidemiology, 6(6): 591-597. Whitehead, N. and L. Lipscomb (2003). “Patterns of Alcohol Use before and During Pregnancy, and the Risk of Small-for-Gestational-Age Birth.” American Journal of Epidemiology, 158(7): 654-652. 16 Table 1: Summary of means, by mother’s race National Natality Data, 1978-1989 Women aged 14 and 24 at time of conception Birthweight Birthweight < 2500g Gestation Gestation < 37 weeks Apgar score (5 min) Apgar <= 5 Congenital anomaly Female HS only Some college Dad’s age missing Birthweight missing Gestation missing Apgar score missing Congenital anomaly missing Mom’s education missing Observations (1,000s) All 3,274 .08 39 .12 8.9 .014 .08 .49 .4 .15 .3 .002 .097 .12 .11 .079 19,780 Race of mother White Black 3,326 3,080 .068 .13 40 39 .096 .18 9 8.8 .012 .021 .08 .079 .49 .49 .41 .41 .16 .16 .22 .56 .0015 .002 .09 .12 .12 .13 .11 .11 .076 .089 13,859 3,743 Other 3,274 .073 39 .12 8.9 .015 .082 .49 .35 .16 .27 .0064 .12 .12 .11 .12 2,178 Notes: Age at conception is not reported in the National Natality data. We estimate age at conception by subtracting one from mother’s age at delivery for all gestational lengths greater than 26 weeks; we assume that age of delivery is the same as age at conception for gestational lengths less than 26 weeks. Our sample is restricted to births that were conceived between 1978 and 1988 inclusive. Date of conception is determined by subtracting gestational length. When gestational length is missing, we assume a 40 week gestational length. Also, delivery day-of-month is not available in the 1989 file, so we assume the delivery took place on the 15th of the month. White and black categorization of race includes natives only. The white and black categorization of race includes hispanic and non-hispanic women. 17 18 73,973 19749259 8.24 0.00 Yes Yes Yes Yes Yes No -0.096 (0.056)* 0.173 (0.183) 0.133 (0.091) 73,973 19749259 3.42 0.00 Yes Yes Yes Yes No Yes -0.178 (0.032)*** 0.437 (0.079)*** 0.246 (0.046)*** Low birthweight (x100) (2) (3) 73,973 19749259 13.50 0.00 3.37 0.00 Yes Yes Yes Yes Yes Yes -0.050 (0.032) -0.060 (0.080) 0.111 (0.051)** (4) 73,771 18071451 Yes Yes Yes Yes No No 73,771 18071451 17.41 0.00 Yes Yes Yes Yes Yes No -0.221 (0.096)** 0.478 (0.285)* 0.183 (0.124) 73,771 18071451 5.30 0.00 Yes Yes Yes Yes No Yes -0.165 (0.066)** 0.506 (0.130)*** -0.005 (0.075) Gestation < 37 weeks (x100) (6) (7) -0.269 (0.092)*** 0.800 (0.264)*** 0.163 (0.113) (5) 73,771 18071451 27.70 0.00 5.33 0.00 Yes Yes Yes Yes Yes Yes -0.076 (0.063) -0.052 (0.111) 0.008 (0.079) (8) Notes: * p<0.10, ** p<0.05, *** p<0.01. Standard errors are clustered on the state of residence. All regressions are weighted by the number of births, where the outcome variable in question is non-missing, in each state-age-year-month cell. The sample is restricted to women who conceived between 1978 and 1988 inclusive. More information on the data can be found in Table 1. Number of cells Observations 73,973 19749259 State f.e. Year-by-month f.e. Age f.e. State-specific trends Age-specific trends State-by-age f.e. Age-specific trends F-stat P-value State-by-age f.e. F-stat P-value Yes Yes Yes Yes No No MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 (1) -0.184 (0.052)*** 0.494 (0.158)*** 0.240 (0.085)*** MLDA is 18 Outcome: Column: Table 2: MLDA analysis National Natality Data, 1978-1989 Women aged 14-24 at conception All races 19 Notes: See notes to Table 2. Number of cells Observations Age-specific trends F-stat P-value State-by-age f.e. F-stat P-value State f.e. Year-by-month f.e. Age f.e. State-specific trends Age-specific trends State-by-age f.e. MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 MLDA is 18 Outcome: Column: 73,195 13842703 Yes Yes Yes Yes No No -0.096 (0.036)** 0.243 (0.122)* 0.116 (0.057)** (1) 73,195 13842703 5.31 0.00 Yes Yes Yes Yes Yes No -0.055 (0.035) 0.006 (0.122) 0.093 (0.057) 73,195 13842703 1.84 0.00 Yes Yes Yes Yes No Yes -0.091 (0.030)*** 0.203 (0.083)** 0.120 (0.040)*** Low birthweight (x100) (2) (3) 73,195 13842703 5.92 0.00 1.82 0.00 Yes Yes Yes Yes Yes Yes -0.036 (0.036) -0.167 (0.093)* 0.115 (0.054)** (4) (5) 72,894 12720724 Yes Yes Yes Yes No No 72,894 12720724 19.72 0.00 Yes Yes Yes Yes Yes No -0.113 (0.069) 0.101 (0.195) 0.060 (0.098) 72,894 12720724 2.36 0.00 Yes Yes Yes Yes No Yes -0.114 (0.071) 0.389 (0.075)*** -0.070 (0.077) 26.15 0.00 2.33 0.00 Yes Yes Yes Yes Yes Yes -0.064 (0.070) -0.132 (0.081) 0.008 (0.086) (8) 72,894 12720724 Gestation < 37 weeks (x100) (6) (7) -0.148 (0.068)** 0.439 (0.180)** 0.016 (0.080) Table 3: MLDA analysis National Natality Data, 1978-1989 Women aged 14-24 at conception Whites only 20 Notes: See notes to Table 2. Number of cells Observations Age-specific trends F-stat P-value State-by-age f.e. F-stat P-value State f.e. Year-by-month f.e. Age f.e. State-specific trends Age-specific trends State-by-age f.e. MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 MLDA is 18 Outcome: Column: 62,374 3,736,912 Yes Yes Yes Yes No No -0.581 (0.135)*** 1.162 (0.241)*** 0.717 (0.124)*** (1) 62,374 3,736,912 7.83 0.00 Yes Yes Yes Yes Yes No -0.249 (0.159) 0.620 (0.374) 0.162 (0.188) 62,374 3,736,912 2.04 0.00 Yes Yes Yes Yes No Yes -0.569 (0.143)*** 0.980 (0.118)*** 0.816 (0.114)*** Low birthweight (x100) (2) (3) 62,374 3,736,912 14.78 0.00 2.07 0.00 Yes Yes Yes Yes Yes Yes 0.005 (0.155) -0.046 (0.141) -0.059 (0.149) (4) (5) 60,968 3,382,284 Yes Yes Yes Yes No No 60,968 3,382,284 9.14 0.00 Yes Yes Yes Yes Yes No -0.387 (0.228)* 0.643 (0.370)* 0.158 (0.186) 60,968 3,382,284 1.86 0.00 Yes Yes Yes Yes No Yes -0.627 (0.133)*** 1.060 (0.211)*** 0.531 (0.163)*** Gestation < 37 weeks (x100) (6) (7) -0.736 (0.179)*** 1.345 (0.260)*** 0.644 (0.137)*** Table 4: MLDA analysis National Natality Data, 1978-1989 Women aged 14-24 at conception Blacks only 60,968 3,382,284 20.25 0.00 1.92 0.00 Yes Yes Yes Yes Yes Yes 0.023 (0.124) -0.284 (0.198) -0.335 (0.164)** (8) Table 5: Additional outcomes National Natality Data, 1978-1989 Women aged 14-24 at conception Outcome: Column: MLDA is 18 MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 MLDA is 18 MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 MLDA is 18 MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 State f.e. Year-by-month f.e. Age f.e. State-specific trends Age-specific trends State-by-age f.e. Birthweight Gestation (1) (2) Five Minute Apgar Score (x100) (3) Apgar <= 5 (x100) (4) Congenital Percent anomaly female (x100) (x100) (5) (6) -0.452 (1.075) 4.378 (2.162)** -1.258 (1.460) Panel A: All races 0.013 0.201 0.001 (0.007)* (0.441) (0.022) 0.014 0.696 -0.064 (0.013) (0.497) (0.030)** -0.012 0.309 -0.038 (0.007)* (0.257) (0.029) -0.912 (3.403) 0.580 (0.436) 0.358 (0.248) 0.020 (0.084) 0.023 (0.108) 0.157 (0.073)** -1.246 (1.240) 6.153 (2.648)** -0.442 (1.979) Panel 0.007 (0.006) 0.021 (0.010)** -0.009 (0.009) -0.841 (2.886) 0.466 (0.381) 0.323 (0.234) 0.093 (0.090) -0.090 (0.133) 0.027 (0.097) -0.568 (3.761) 7.836 (3.343)** 0.250 (3.563) Panel C: Blacks only 0.022 1.127 0.005 (0.011)* (0.538)** (0.053) 0.020 -0.104 -0.122 (0.019) (0.629) (0.058)** 0.003 0.360 -0.120 (0.015) (0.373) (0.066)* 1.756 (4.573) -0.170 (0.211) -0.104 (0.121) -0.042 (0.162) 0.229 (0.183) 0.542 (0.204)** Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes B: Whites only -0.083 0.007 (0.487) (0.021) 0.523 -0.034 (0.413) (0.036) 0.242 -0.028 (0.225) (0.025) Yes Yes Yes Yes Yes Yes Notes: See notes to Table 2. 21 Yes Yes Yes Yes Yes Yes Table 6: MLDA analysis Compositional effects National Natality Data, 1978-1988 Outcome: Less than HS (x100) HS only (x100) Some college (x100) Column: (1) (2) (3) MLDA is 18 MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 MLDA is 18 MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 MLDA is 18 MLDA is 18 x mother is 14-17 MLDA is 18 x mother is 18-20 State f.e. Year-by-month f.e. Age f.e. State-specific trends Age-specific trends State-by-age f.e. Dad’s info missing (x100) (4) 0.230 (0.217) -0.237 (0.393) -0.213 (0.247) Panel A: All races 0.041 -0.271 (0.238) (0.183) 0.171 0.066 (0.448) (0.332) -0.006 0.218 (0.273) (0.217) -0.510 (0.326) 1.064 (0.860) 0.483 (0.464) 0.396 (0.230)* 0.038 (0.331) -0.096 (0.274) Panel B: Whites only -0.161 -0.235 (0.204) (0.180) -0.088 0.051 (0.399) (0.361) 0.007 0.089 (0.286) (0.200) -0.348 (0.313) -0.136 (0.939) 0.351 (0.449) 0.057 (0.446) -0.043 (0.632) -0.523 (0.419) Panel C: Blacks only 0.410 -0.466 (0.547) (0.299) -0.463 0.506 (0.616) (0.435) 0.109 0.414 (0.447) (0.326) -0.243 (1.032) 1.323 (1.219) 0.227 (0.651) Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Notes: See notes to Table 2. 22 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0 10 Number of states 20 30 40 50 Figure 1: MLDA Law Changes 1977 1979 1981 1983 MLDA = 18 MLDA = 20 1985 1987 1989 MLDA = 19 MLDA = 21 Notes: Although the law mandating a MLDA of 21 was passed in 1984, the law was not enforced until late in 1986. The law withheld a portion of a state’s Federal highway funds if the state did not enact a MLDA of 21 by October 1, 1986. Source: Distilled Spirit Council of the United States (1996). 23 .04 1978 1980 1982 1984 Year of conception 1986 Panel A.1: Birthweight < 2500 g 1988 Placebo states 1982 1984 Year of conception Treated states 1986 Panel B.1: Birthweight < 2500 g 1980 Placebo states 1982 1984 Year of conception 1988 1978 1980 Placebo states 1982 1984 Year of conception Treated states 1986 Panel B.2: Gestation < 37 weeks Treated states 1986 1988 1988 Notes: The placebo states include Arkansas, California, Indiana, Kentucky, Missouri, Nevada, New Mexico, North Dakota, Oregon, Pennsylvania, Utah, and Washington. Y axes scales vary across panels. 1978 Treated states 1980 Panel B: Differences between 18-20 year olds and 21-24 year olds over time Placebo states 1978 Panel A.2: Gestation < 37 weeks Panel A: Differences between 14-17 year olds and 21-24 year olds over time Figure 2: Differences in birth outcomes across age groups over time, by treated and placebo states Birthweight < 2500g .025 .03 .035 .008 .01 Birthweight < 2500g .012 .014 .016 .018 .02 .07 Gestation < 37 weeks .05 .06 .04 .025 Gestation < 37 weeks .02 .015 24 Notes: See notes to Figure 2. 14 14 Placebo states 22 16 Placebo states 22 Treated states 18 20 Estimated age at conception 24 14 16 24 14 Treated states 22 16 Placebo states 22 Treated states 18 20 Estimated age at conception Panel B.2: Gestation < 37 weeks Placebo states 18 20 Estimated age at conception Panel A.2: Gestation < 37 weeks Panel B: Conception year is 1988 Treated states 18 20 Estimated age at conception Panel B.1: Birthweight < 2500 g 16 Panel A.1: Birthweight < 2500 g Panel A: Conception year is 1978 Figure 3: Mean birth outcomes by mothers’ estimated age at conception, by treated and placebo states .12 Birthweight < 2500g .08 .1 .06 .12 Birthweight < 2500g .08 .1 .06 .2 Gestation < 37 weeks .1 .15 .05 .2 Gestation < 37 weeks .1 .15 .05 25 24 24 .03 Birthweight < 2500g .022 .024 .026 .028 .02 .012 Birthweight < 2500g .009 .01 .011 .008 1978 1978 Treated states 1986 1980 1988 1978 1980 Placebo states 1982 1984 Year of conception Placebo states 1982 1984 Year of conception Treated states 1986 1988 1978 1980 Placebo states 1982 1984 Year of conception Treated states 1986 Panel B.2: Gestation < 37 weeks Treated states 1986 Panel A.2: Gestation < 37 weeks Panel B: Differences between 18-20 year olds and 21-24 year olds over time Placebo states 1982 1984 Year of conception Panel B.1: Birthweight < 2500 g 1980 Panel A.1: Birthweight < 2500 g Panel A: Differences between 14-17 year olds and 21-24 year olds over time Figure A1: Differences in birth outcomes across age groups over time, by treated and placebo states White mothers Notes: See notes to Figure 2. .007 .05 Gestation < 37 weeks .035 .04 .045 .03 .018 Gestation < 37 weeks .014 .016 .012 26 1988 1988 .02 Birthweight < 2500g .005 .01 .015 0 −.005 .015 Birthweight < 2500g 0 .005 .01 −.005 1978 1978 Treated states 1986 1980 1988 1978 1980 Placebo states 1982 1984 Year of conception Placebo states 1982 1984 Year of conception Treated states 1986 1988 1978 1980 Placebo states 1982 1984 Year of conception Treated states 1986 Panel B.2: Gestation < 37 weeks Treated states 1986 Panel A.2: Gestation < 37 weeks Panel B: Differences between 18-20 year olds and 21-24 year olds over time Placebo states 1982 1984 Year of conception Panel B.1: Birthweight < 2500 g 1980 Panel A.1: Birthweight < 2500 g Panel A: Differences between 14-17 year olds and 21-24 year olds over time Figure A2: Differences in birth outcomes across age groups over time, by treated and placebo states Black mothers Notes: See notes to Figure 2. −.01 .06 Gestation < 37 weeks .03 .04 .05 .02 .025 Gestation < 37 weeks .005 .01 .015 .02 0 27 1988 1988 Notes: See notes to Figure 2. 14 14 Placebo states 22 16 Placebo states 22 Treated states 18 20 Estimated age at conception 24 14 16 24 14 Treated states 22 16 Placebo states 22 Treated states 18 20 Estimated age at conception Panel B.2: Gestation < 37 weeks Placebo states 18 20 Estimated age at conception Panel A.2: Gestation < 37 weeks Panel B: Conception year is 1988 Treated states 18 20 Estimated age at conception Panel B.1: Birthweight < 2500 g 16 Panel A.1: Birthweight < 2500 g Panel A: Conception year is 1978 Figure A3: Mean birth outcomes by mothers’ estimated age at conception, by treated and placebo states White mothers .1 Birthweight < 2500g .07 .08 .09 .06 .05 .1 Birthweight < 2500g .07 .08 .09 .06 .05 .16 Gestation < 37 weeks .1 .12 .14 .08 .06 .16 Gestation < 37 weeks .1 .12 .14 .08 .06 28 24 24 14 16 18 20 Estimated age at conception 22 Panel A.1: Birthweight < 2500 g 24 14 16 18 20 Estimated age at conception Notes: See notes to Figure 2. Placebo states 16 22 24 Treated states 14 Placebo states 22 Treated states 18 20 Estimated age at conception 16 Placebo states 22 Treated states 18 20 Estimated age at conception Panel B.2: Gestation < 37 weeks Panel B: Conception year is 1988 Treated states Panel B.1: Birthweight < 2500 g Placebo states 14 Panel A.2: Gestation < 37 weeks Panel A: Conception year is 1978 Figure A4: Mean birth outcomes by mothers’ estimated age at conception, by treated and placebo states Black mothers .15 Birthweight < 2500g .12 .13 .14 Birthweight < 2500g .12 .13 .14 .11 .15 .11 .25 Gestation < 37 weeks .15 .2 .1 .25 Gestation < 37 weeks .15 .2 .1 29 24 24
© Copyright 2026 Paperzz