Teacher Complementarities in Test Score Production: Evidence from Primary School

The existence of teacher complementarities across grades can have important implications for the efficient assignment of students to teachers and for the evaluation of teachers using standard value-added models (VAMs). In this paper, I extend the basic VAM to allow for interactive effects between teachers. Using data from North Carolina's primary schools, I find that teacher complementarities within schools are extremely small. This finding suggests that there is little to gain by assigning students to particular teacher sequences. Moreover, measures of teacher effectiveness generated from a typical VAM are not likely to be biased by interactions with past teacher inputs.