Estimating Group Effects Using Averages of Observables to Control for Sorting on Unobservables: School and Neighborhood Effects

Abstract

We consider the classic problem of estimating group treatment effects when individuals sort based on observed and unobserved characteristics. Using a standard choice model, we show that controlling for group averages of observed individual characteristics potentially absorbs all the across-group variation in unobservable individual characteristics. We use this insight to bound the treatment effect variance of school systems and associated neighborhoods for various outcomes. Across multiple datasets, we find that a ninetieth versus tenth percentile school/neighborhood increases the high school graduation probability and college enrollment probability by at least 0.04 and 0.11 and permanent wages by 13.7 percent. (JEL C51, H75, I21, I26, J24, J31, R23)

Quantifying Family, School, and Location Effects in the Presence of Complementarities and Sorting

“…In this paper, we build on the rich literature on sorting, school and neighborhood effects, and multilevel modeling to assess the relative importance of neighborhood, school, and broader local area factors in shaping student's educational attainment and early career wages, and the degree to which this relative importance differs across students from different backgrounds. We extend the identification results of Altonji and Mansfield (2018) to models with multiple group levels and with interactions between observed individual factors and both observed and unobserved group level factors. The identification results are based on the idea that group averages of individual-level observables can fully control for sorting bias from group averages of individual-level unobservables.…”

Evidence of Neighborhood Effects from Moving to Opportunity: LATEs of Neighborhood Quality

“…Implementing Masten and Torgovitsky (2016)'s strategy for Average Partial Effects (APEs) is left for future work.19 See Clampet-Lundquist and Massey (2008) orQuigley and Raphael (2008) for related discussions. Our effects are from changes in neighborhoods that are comparable to those inPinto (2018), smaller than those considered inAltonji and Mansfield (2018), and in a different part of the distribution than inGalster et al (2016).…”

Do rural migrants benefit from labor market agglomeration economies? Evidence from Chinese cities

“…Occupation choice is endogenous in a wage model; but in our setting, the inclusion of occupation can help reduce the correlation between error term and the two agglomeration variables. A rule of thumb to test this method is to see whether and by how much the coefficients of education category variables become attenuated after the inclusion of occupation fixed effects since in general unobserved ability should be positively correlated with observed ability such as education attainment (Altonji and Mansfield, 2018;Fu and Ross, 2013;Oster, 2019).…”