Fair ranking of teachers

Economic theory suggests that it is optimal to reward teachers depending on the relative performance of their students. We develop an econometric approach, based on stochastic frontier analysis, to construct a fair ranking that accounts for the socio-economic background of students and schools and the imprecision inherent in achievement data. Using German PIRLS (IGLU) data, we exploit the hierarchical structure of the data to estimate the efficiency of each teacher. A parsimonious set of control variables suffices to get a “fair” estimate of unobserved teacher quality. A Hausman–Taylor type estimator is the preferred estimator because teacher efficiency and some exogenous variables may be correlated.