State-specific licensing policies and pension plans create mobility costs for educators who cross state lines. We empirically test whether these costs affect production in schools--a hypothesis that follows directly from economic theory on labor frictions--using geocoded data on school locations and state boundaries. We find that achievement is lower in mathematics, and to a lesser extent in reading, at schools that are more exposed to state boundaries. A detailed investigation of the selection of schools into boundary regions yields no indication of systematic differences between boundary and non-boundary schools along other measured dimensions. Moreover, we show that cross-district labor frictions do not explain state boundary effects. Our findings are consistent with the hypothesis that mobility frictions in educator labor markets near state boundaries lower student achievement. [This paper was published in "Economics of Education Review" v60 p54-67 2017.]