Association for Public Policy Analysis & Management

This paper examines the effect of the adoption of teacher value-added (VA) on student performance. Many states in the US have incorporated student outcomes into teacher evaluation systems. Policy makers understand the important role of teachers and recognize the need to measure their effectiveness more objectively. In the area of teacher quality, using student growth may offer a fairer and potentially more meaningful way to evaluate teacher’s effectiveness than previous methods schools used. As of 2017, 39 states require student growth data to be included in teacher evaluation.

Providing teacher VA information to the schools would be a useful tool for school districts because VA scores may exert social pressure to improve teacher performance which leads to the improvement of students’ learning. Despite the universal use of student outcomes in teacher evaluation, there is still little empirical evidence that examines the impact of this policy on students’ academic achievement. Rigorous empirical evidence of this important policy can go a long way in ensuring that all students have access to a high quality education. This work seeks to fill this gap by investigating the incorporation of teacher-level VA measures into teacher evaluations.

This paper exploits variation introduced by the Ohio Teacher Evaluation System, which piloted the incorporation of student growth measures as a part of teacher evaluations. The pilot involved 139 of 611 school districts and ran during the 2011-2012 school years before being incorporated statewide in 2013. I employ a differences-in-differences (DID) model using the school report card data from Ohio. The 4th- to 8th-grade teachers in pilot schools received the VA information while teachers in control schools did not. In addition, I use the DDD method, which includes 3rd graders within treatment districts as another comparison group. As third-grade teachers did not receive VA scores, it allows me to control for any unobserved district-level shocks that would coincide with adopting the VA policy.

I find no evidence of the policy increasing student achievement in various outcomes that demonstrate the student learning. Overall, I find a negative effect on school-level proficiency in mathematics, although it is not statistically significant, and a null result in reading. In order to gain more insight on the overall effect, I examine how the policy affected the proportion of students in each proficiency level category. I find a significant shift in the share of students at the more advanced levels towards the proficiency margin for both reading and mathematics. This result shows the evidence of student distribution is being shifted downward or compressed to the mean. With the 2001 NCLB reform, there was concern that schools would target students who are at the margin of passing, rather than students who were already proficient or those far from becoming proficient. My findings support this critique, even when teachers have less of an incentive to focus on the proficient threshold.