Research

​I estimate a dynamic model of schooling on two longitudinal datasets (the national education longitudinal study of 1988 and the high school longitudinal study of 2009) and find that the effect of relative family income on education has decreased between the early 1990’s and the early 2010’s. After taking into account a cognitive ability measure (a math test score), family background variables and unobserved heterogeneity, the marginal effects of relative income on grade progression in college have become smaller for the younger cohort. After substituting the math test score quartiles as measures of cognitive ability by an estimated cognitive factor, the effect of relative family income in the two cohorts become smaller and the differences between the two cohorts are not statistically significant. Meanwhile in the same period, the effect of unobserved heterogeneity in explaining the variation in educational attainment has increased significantly. 


Students’ participation in the labor force is the norm for a large proportion of high school students in many countries including the US. For instance, data from the National Education Longitudinal Study of 1988 (NELS:88) show that in 1990, around 56 percent of 10th grade students in the US were working during school days while this number for the same cohort had increased to almost 70 percent in 1992 when most of the students were in 12th grade. Around 20 years later, data from the High School Longitudinal Study of 2009 (HSLS:09) show that although the number of students working while in school has decreased, the proportion is still high. Between 46 to 50 percent of students in 9th and 12th grades work for pay during school days. Using data from these two American surveys (NELS:88 and HSLS:09), this paper examines the effect of school year employment and participation in extracurricular activities on academic performance, specifically the grade point average. Since the decision to work or participate in extracurricular activities are endogenous, several methods were used in this paper to overcome this issue. The results show that in 1990, working while in school had no significant impact on students’ GPAs but in 1992, when most students in the cohort were in 12th grade, working decreased their GPAs. In 2009, working had a negative impact on GPAs among grade 9 students. Working in 2012, when most students were in 12th grade, also had a negative impact on their GPAs. Contrary to the effect of working, participating in extracurricular activities significantly improves academic performance. 


This paper estimates models of the evolution of cognitive and noncognitive skills and studies the role of family environment in shaping these skills in teenage years. I estimate a dynamic factor model to solve the issue of endogeneity of inputs and multiplicity of inputs relative to instruments. The model is estimated once for the 1988 sample between the ages of 14 to 16, and a second time, for high school graduates only, in both 1988 and 2009 cohorts between the ages of 14 and 18. This paper differs from others in the literature by looking at the evolution of skill production technology in teenagers and young adults over a span of 20 years instead of estimating the production function only for one cohort and in early childhood. The results show that parental investments (both in terms of time and money) improved noncognitive skills but not cognitive skills for the 1988 cohort. The effect on cognitive skills has increased in the later cohort.