This paper uses eleven years of administrative data from North Carolina public high schools to examine the extent to which the allocation of teachers within and across public high schools is contributing to inequality in student test score performance. The existence of nearly 3,500 teacher transfers allows separate identifica-tion of each teacher’s quality from other school-level factors. I find that teaching quality is fairly equitably distributed both within and across high schools: students among the bottom (top) decile of a student back-ground index are taught by teachers who are, on average, at the 41st (57th) percentile of the value-added distribution. I have greatly benefited from the input of Joseph Altonji, as well as Fabian Lange, Melissa Tartari, Costas Meghir, Lisa Kahn, Amanda Kowalski, and several anonymous referees. I also thank Mark Klee, Matthew Johnson, Priyanka

We extend Altonji and Mansfield’s control function approach to allow for multiple group levels and complementarities. Our analysis provides a foundation for a causal interpretation of multilevel mixed effects models in the presence of sorting. In our empirical application, we obtain lower-bound estimates of the importance of school and commuting zone inputs for education and wages. A school/location combination at the 90th versus 10th percentile of the school/location quality distribution increases high school graduation and college enrollment probability by at least .06 and .17, respectively. Treatment effects are heterogeneous across subgroups, primarily due to nonlinearity in the educational attainment model.