Revise & Resubmit:

The Economic Value of Breaking Bad: Misbehavior, Schooling and the Labor Market. [pdf]

         -  joint with Nicholas Papageorge and Yu Zheng - R&R at Journal of Political Economy.

Press Coverage: [JHU Magazine][Brookings Institution][Frankfurter Allgemeine]

Abstract: Prevailing research argues that childhood misbehavior in the classroom is bad for schooling and, presumably, bad for labor market outcomes. In contrast, we argue that childhood misbehavior represents underlying non-cognitive skills that are valuable in the labor market. We follow work from psychology and categorize observed classroom misbehavior into two underlying latent factors. We then estimate a model of educational attainment and earnings outcomes, allowing the impact of each of the two factors to vary by outcome. We  nd one of the factors, labeled in the psychological literature as externalizing behavior (and linked, for example, to aggression), reduces educational attainment yet increases earnings. Unlike most models where non-cognitive skills that increase human capital through education also increase labor market skills, our ndings illustrate how some non-cognitive skills can be productive in some economic contexts and counter-productive in others. Policies designed to promote human capital accumulation could, therefore, have mixed effects or even negative economic consequences, especially for policies that target non-cognitive skill formation for children or adolescents which are aimed solely at improving educational outcomes.

Working Papers:

The Effect of Maternal Psychological Distress on Children's Cognitive Development. [pdf]

Abstract: This paper investigates how maternal mental health, measured by psychological distress, affects family investments and shapes children’s cognitive skills. I provide a model that incorporates ideas from both sociology and psychology into a fairly standard economic model of maternal investments. This model allows me to separate the different mechanisms that relate maternal mental health to children’s cognition. In order to estimate the causal effect of mental health, I control for the endogeneity of mental health as well as the inherent measurement error in mental health constructs. For the former, I use variation among U.S. states in mental health insurance coverage laws. For the latter, I use an item response theory approach. Using a longitudinal data set from the U.S., the Panel Study of Income Dynamics (PSID) and its Child Development supplement (PSID-CDS), I find that maternal psychological distress mainly affects children through a decrease in the productivity (quality) of maternal time investments. My findings support two policy interventions that mitigate this effect. My findings suggest that mental health treatment for at-risk mothers can have significant payoffs for children and is significantly more cost-effective than comparable income transfers. Moreover, my findings suggest that programs that improve maternal parenting can have large benefits for children of at-risk mothers.

The Multidimensionality of Teacher Quality: Teaching Skills and Students' Noncognitive Skills. [available upon request]‚Äč

          - joint with Alanna Bjorklund-Young.

Abstract: Prevailing research shows that noncognitive skills are important for students achievement in school and later life outcomes. Unfortunately, very little is known about how noncognitive skills can be taught and developed in school. We add to the literature by demonstrating that certain teaching skills, measured by trained observers using video recordings, are related to teacher quality measured by both students' cognitive and noncognitive skills. We first demonstrate that teacher quality is multidimensional, as teachers associated with higher student cognitive skills are not the same teachers associated with higher student noncognitive skills. We then show that different teaching skills are related to different dimensions of teacher quality. For example, in our most surprising result, we show that teacher sensitivity is positively related to students' effort and grit but negatively related to students' test scores. This highlights our main finding that teaching practices that help students develop cognitively are not the same teaching practices associated with students' noncognitive skills. Our results suggest that teachers should engage in different teaching practices depending on which student skill they are trying to develop. Moreover, our results suggest that school districts should be strategic when designing training programs, as different teaching skills are associated with different student outcomes and some skills thought to be productive can actually be harmful to students.

Family Disadvantage, Gender and the Returns to  Genetic Human Capital. - [available upon request]   

           - first author - joint with Esben Agerbo, Dorthe Bleses, Preben Bo Mortensen and Michael Rosholm. 

Abstract: This study relies on a large-scale sample of genotyped individuals linked with detailed register data in Denmark to study whether family disadvantage moderates the effects of genes on human capital formation. To measure the effect of genes, we rely on a polygenic score (PGS) derived from the most recent genome-wide association study (GWAS) of educational attainment. We show that the educational attainment polygenic score (EA PGS) explains a significant portion (up to 9%) of the variation in human capital in our sample. We then test whether the effect of the polygenic score differs across family disadvantage groups. We find that childhood disadvantage significantly reduces the relationship between the EA PGS and human capital outcomes. The effect of the EA-PGS on years of education is 30% smaller for individuals that experienced childhood disadvantage than for individuals that did not. We find similar moderation effects on post-secondary education, Danish and mathematics test scores and the probability of mental health diagnosis. The patterns remain after controlling for family fixed effects. Moreover, we find significant gender differences in the reported gene-environment interactions. In particular, we find significantly larger effects of childhood disadvantage on the returns of the EA PGS for males than for females. We also show that our findings extend to genes related to psychiatric diagnoses. That is, we show that childhood disadvantage also moderates the returns of mental health disorders polygenic scores on psychiatric diagnosis.