A large literature examines performance pay for managers in the private sector, but less is known about the use of performance pay for their public sector counterparts. Improving public service delivery may nonetheless depend heavily on aligning the incentives of public sector managers with social objectives. In this paper we study performance incentives for school administrators and how their responses to incentives vary with the amount of resources under their control. Our specific focus is the implementation of new, school-based programs to reduce childhood anemia in rural China. We randomly assigned 170 schools to three levels of performance pay for reductions in student anemia across which we orthogonally assign two levels of block grants. We emphasize three key findings. First, with a smaller block grant, large incentives were effective, but smaller incentives (10% of the size) were ineffective in reducing anemia. Second, absent explicit anemia-based incentives, increasing the size of block grants under the control of school administrators lead to sizeable reductions (but was nearly twice as costly as incentives alone). Third, we find that incentives crowd out the effect of additional resources (or vice-versa). Our evidence suggests that this crowding-out result is attributable, at least in part, to risk avoidance and the nature of existing ‘bureaucratic incentives’ facing school administrators.