Working Paper: NBER ID: w26037
Authors: Francesco Agostinelli; Morteza Saharkhiz; Matthew J. Wiswall
Abstract: We develop a unified empirical framework for child development which nests the key features of two previously parallel research programs, the Child Development literature and the Education Production Function literature. Our framework allows for mis-measured cognitive and non-cognitive skills, classroom effects, parental influences, and complementarities/interactions. Although both are important, we estimate that differential parental investments are the more important source of end-of-kindergarten inequality than classroom quality. Higher quality classrooms and home investments have a larger effect on children entering kindergarten with skill deficits, a negative complementarity. Our estimated model replicates patterns by excluded race and family income variables and experimental results from the Tennessee STAR experiment. Finally, we find that the estimated classroom quality from standard value-added models would systematically be confounded with the student selection into classrooms based on their (omitted) non-cognitive skills and their home quality.
Keywords: Child Development; Parental Investments; Classroom Quality; Educational Inequality
JEL Codes: I21; J13
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| differential parental investments (J12) | end-of-kindergarten inequality (I24) |
| classroom quality (I24) | end-of-kindergarten mathematics skills (C02) |
| classroom quality (I24) | end-of-kindergarten reading skills (A21) |
| classroom quality (I24) | end-of-kindergarten noncognitive skills (D29) |
| home environment (D19) | end-of-kindergarten skills (A21) |
| household income (D19) | home environment (D19) |
| home environment (D19) | skill development (J24) |
| classroom investments (A21) | skill development for disadvantaged children (I24) |
| measurement error (C20) | bias in estimates (C51) |
| classroom quality assessments (I24) | confounding treatment effect (C32) |