Teacher-Student Matching and the Assessment of Teacher Effectiveness

We use administrative data on North Carolina public schools to document the tendency for more highly qualified teachers to be matched with more advantaged students, and we measure the bias this pattern generates in estimates of the impacts of various teacher qualifications on student achievement. One of the strategies we use to minimize this bias is to restrict the analysis to schools that assign students to classrooms in a manner statistically indistinguishable from random assignment. Using data for 5th grade, we consistently find significant returns to teacher experience in both math and reading and to licensure test scores in math achievement. We also find that the returns in math are greater for socioeconomically advantaged students, a finding that may help explain why the observed form of teacher-student matching persists in equilibrium.