Teacher compensation schemes are often criticized for lacking a performance-based component. Proponents of merit pay argue that linking teacher salaries to student achievement will incentivize teachers to focus on raising student achievement and stimulate innovation across the school system as a whole. In this paper, we utilize a policy experiment conducted in the New York City public school system to explore the effects of one performance-based bonus scheme. We investigate potential impacts of group-based incentive pay over two academic years (2007-2008 and 2008-2009) on a range of outcomes including: teacher effort, student performance in math and reading, and classroom activities, measured through environmental surveys of teachers and students. We also explore impacts on the market for teachers by examining teacher turnover and the qualifications of newly hired teachers. Overall, we find the bonus program had little impact on any of these outcomes. We argue that the lack of bonus program impacts can be explained by the structure of the bonus program. Group bonuses led to free-riding, which significantly reduced the program's incentives. Once we account for free-riding, we find evidence that the program led teachers to increase their effort through a significant reduction in absenteeism. When considering the effectiveness of performance-based teacher pay, the structure of incentives matter. (Contains 11 tables, 3 figures and 24 footnotes.)