Theoretical Framework and Existing Literature

nature and nurture in this area appear to be large. For children with one variant of MAOA, I fi nd that increases in household income have the expected positive associa- tion with college enrollment, college graduation, and total years of schooling com- pleted. For children with another MAOA variant, who comprise over half of the popu- lation, I fi nd that there is a much weaker association between economic background and educational attainment. These results hold when the genetic component of the interactive effects are identifi ed using MAOA variation across full biological siblings, which is determined entirely by chance at the time of conception. The paper proceeds in fi ve additional sections. Section II lays out a simple theoreti- cal framework and briefl y reviews existing work in this area. Section III describes the data. Section IV presents the main sets of results. Section V considers several addi- tional issues and the mechanisms underlying the main results. Section VI concludes by discussing several of the study’s limitations and suggesting directions for future research in this area.

II. Theoretical Framework and Existing Literature

The basic concept underlying my empirical exercises can be concisely expressed as a conventional economic production function. Specifi cally, let y be a measure of educational attainment, and let g and e be measures of genetic and envi- ronmental factors that infl uence y. If these cross partials are nonzero, then the process by which academic achievement is produced includes gene- environment interactions. One familiar functional form that helps to illustrate several conceptually important cases is a CES production function. 1 Recall that the CES production function takes the form y = A[αg γ + 1 − αe γ ] 1γ where γ ≤ 1 and 0 ≤ α ≤ 1. In the special case where γ = 1, genetic and environmental inputs are perfect substitutes and this equation reduces to y = A[αg + 1 − αe]. In this case, the genetic and environmental infl uences on academic achievement are additively separable, ∂ 2 f ∂e∂g = 0, and there are no gene- environment interactions. The relative importance of genetic as opposed to environmental infl uences is given by the param- eter α. Much of the economic and psychometric literature on heritability has implicitly assumed a functional form of this kind, and the primary research exercise in that litera- ture has been to estimate the value of α. These estimates most commonly derived us- ing twin and adoptee research designs are then interpreted as the proportion of popula- tion variance in the trait under study that is due to genes, implicitly or explicitly ruling out gene- environment interactions of any importance. At the other extreme, when γ = −∞, genetic and environmental inputs are perfect compliments. This can be seen as an extreme case of gene- environment interactions, where changes in one factor are of no consequence when the other factor is held constant. Likely more realistic than either of these extreme cases is a CES production function where −∞ ≤ γ ≤ 1, so that the marginal effect of environmental improvements vary with genetic endowment and vice- versa. The empirical evidence presented below is consistent with such an intermediate case. 1. Cuhna, Heckman, and Schennach 2010, Cuhna and Heckman 2007, and Todd and Wolpin 2003 all use this same production function to study how various inputs occurring at different ages including genetic endowments affect child development. While direct measurement of human genotypes is now a reasonably straightforward process, studies of how genetic endowments modify the effects of environmental con- ditions have also been conducted with nonbiological data, typically using twin and adoptee research designs to identify genetic and environmental effects. A prominent example is Turkheimer et al. 2003, who found that the heritability of IQ is greater among high SES families than it is among low SES families, suggesting interactive complementarities between genetic and environmental advantages in infl uencing cog- nitive skills. 2 Another important contribution comes from Bjö rklund, Lindahl, and Plug 2006, who study the educational and earnings outcomes of Swedish adoptees. The authors fi nd signifi cant interactions between the characteristics of the adopted children’s biological parents and those of their adoptive parents, which they credibly argue refl ect genetic and environmental factors, respectively. 3 While these studies demonstrate that much can be learned without directly observ- ing genetic markers, relatively recent technological advances have made such direct observation practical in many cases, considerably increasing the level of scientifi c rigor possible in gene- environment interaction studies. One of the fi rst and most widely cited studies taking advantage of genetic markers was Caspi et al. 2002, which addressed the question of why some abused children themselves go on to de- velop aggressive and antisocial behaviors while others function relatively normally. Caspi and his collaborators obtained data indicating the functional presence of MAOA the same gene used in the present study for participants in a study of 1,037 children who were followed from ages three to 26. Each study participant was then classifi ed as having been severely maltreated, probably maltreated or not maltreated during child- hood, and the relationship between maltreatment status and an index of antisocial be- havior was estimated for children with and without MAOA gene presence. The authors found a striking gene- environment interaction: Among children with high MAOA activity, childhood maltreatment was associated with a relatively modest 0.24 standard deviation increase in antisocial behavior but among children with low MAOA activity childhood maltreatment was associated with a much larger increase of 0.68 standard deviations. This difference was highly statistically signifi cant and was robust to vari- ous changes in the specifi cation and in how antisocial behavior was measured. A large subsequent literature has explored gene- environment interactions with re- spect to a wide range of psychological traits, including child temperament Sheese et al. 2007, depression Wilhelm et al. 2006, alcoholism Hutchison et al. 2002, schizophrenia Caspi et al. 2005, and ADHD Retz et al. 2008, among others. 4 While the overwhelming majority of gene- environment interaction studies have focused on psychological traits and conditions like those noted above, researchers 2. In the more general literatures on intergenerational mobility and heritability, an important methodologi- cal distinction can be made between studies that regress child outcomes onto parent outcomes and studies that use different types of twins or adoptees to decompose the proportion of variation in a characteristic that is attributable to genetic versus environmental infl uences. The decomposition approach has tended to fi nd a larger overall role for genetic factors, while the regression approach is more amenable to incorporating genetic market data to estimate gene- environment interactions. Black and Devereux 2010 review both the regression and decomposition approaches, while Sacerdote 2011 provides a thoughtful discussion of issues related to the decomposition approach. 3. Other notable studies in this literature include Cadoret et al. 1996 and Kendler, Karkowski, and Prescott 1999. 4. An informative and accessible review of this literature is provided by Rutter 2006. have recently begun to explore the moderating effects of genetic background on a more general set of social outcomes. For example Guo, Roettger, and Cai 2008 study the determination of juvenile delinquency and fi nd signifi cant interactions between ge- netic markers and factors like eating regular family meals and repeating a grade, while Settle et al. 2010 fi nd a signifi cant interaction between a dopamine receptor gene and social networks in the formation of political ideologies. 5 However, I have been able to identify only two previous studies that examined the role of gene- environment interactions in determining any measure of academic achievement. First, Conley and Rauscher 2010 use within twin- pair birthweight differences to study how genetic traits may moderate the relationship between birth weight and several outcomes including high school grade point average. They fi nd only one sig- nifi cant gene- environment interaction, and its sign is the opposite of what had been suggested by prior research. Whereas the plausible exogeneity of within twin- pair birth weight differences make Conley and Rauscher’s research design attractive, its scope is inherently limited since birth weight is only one environmental factor that may impact educational outcomes. The present study analyzes a distinct and arguably broader environmental condition, economic background, and also expands the set of educational outcomes under consideration. Second, a study by Shanahan et al. 2008 analyzes the determination of educational continuation beyond high school, and fi nds signifi cant interactions between a variant of the dopamine receptor gene DRD2 and environmental factors such as having a par- ent that belongs to the PTA and how often parents discuss school related issues with the student. While their fi ndings are suggestive of important interactive effects, the qualitative methodological approach of Shanahan et al. makes their results diffi cult to interpret, and no attention is given to issues of selection bias. In the present study, I take advantage of family level clustering in my data to estimate sibling fi xed- effects models that identify the critical interaction terms using plausibly exogenous varia- tion in genetic status, and do so within a standard multiple regression methodologi- cal approach. Additionally, I analyze distinctly economic environmental factors and a broader set of educational outcomes.

III. Data