Association for Public Policy Analysis & Management

Teacher performance pay programs have been primarily evaluated through student test scores. Test scores, however, have come under much criticism as a tool for evaluating student performance. They can be manipulated by “teaching to the test.” They do not capture important skills that are difficult to test but are important on the labor market. Test scores are measured on ordinal scales which makes cross-teacher comparisons in student tests scores difficult if not impossible. We propose to instead evaluate teacher pay for performance using labor market outcomes. We construct a comprehensive data set of schools which have implemented teacher performance pay programs across the United States since 1986. While it is impossible to know whether the data set is complete, we believe it is the most comprehensive such data set ever constructed. We use our data to calculate the fraction of students in each grade in each state who are affected by a teacher pay for performance program in a given year. We then match this with birth place data from the American Community Survey to create a data set of individuals with labor market outcomes and their exposure to teacher pay for performance programs during schooling. We find that cohorts with a larger share of individuals enrolled in schools with performance pay programs earn lower wages as adults. This effect holds even after controlling for the fact that performance pay programs are correlated with public school spending. This negative effect is concentrated on women and high school graduates with no significant effect for men. Our point estimate suggests that a 12 years of schooling in a district where teachers are paid by performance leads to a 15 percent decrease in adult wages. However, the effect is positive for those who do not graduate from high school.