Bonus pay for teachers in U.S. is analyzed. We quantitatively argue that, because of the decentralized education finance system in U.S., this policy may lead to higher teacher and household sorting across school districts. This then may lead to higher variance of achievement and lower mean achievement. Formally, we use an equilibrium political economy model of education at which households, heterogeneous in exogenously set income, and teachers, heterogeneous in exogenously set quality, are endogenously allocated across two school districts. Public spending per pupil, income tax rate and student achievement in each district are also endogenously determined. Public education expenditures, which includes teachers' wage payment and non-teacher related education spending, are financed through local income tax revenue collected from all households residing in the district. Income tax rate in each district is determined via majority voting. Achievement depends on the efforts chosen by teachers and non-teacher related education spending. Teacher efficiency wage per unit of quality is determined at the national teacher labor market. We first calibrate our model so as to match certain statistics from 2000 U.S. data. Then in a counterfactual experiment, we introduce bonus pay for teachers which rises with average achievement. We find that for the current levels of bonus pay (2%-7% out of average base salary), variance of achievement is higher. For higher levels of bonus pay (13% and above), variance continuously falls. Also mean achievement always falls as bonus pay percentage rises.