Pay for performance is a popular public education reform, and millions of dollars are currently being targeted for pay for performance programs. These reforms are popular because economic and management theories suggest that well-designed incentive pay programs could improve teacher effectiveness. There is little evidence about the characteristics of a well-designed incentive pay plan for teachers, however. This study takes advantage of a recent natural experiment in Texas to explore the optimal design of teacher incentive plans. We find that when given the opportunity, teachers design relatively weak incentive pay plans. In turn, those relatively weak incentives do not appear to have induced any significant changes in teacher productivity, although they did have a significant impact on teacher turnover. As such, our evidence suggests that the types of incentives that appeal to teachers (and their unions) may not be optimal from the employer perspective. (Contains 7 tables, 2 figures and 20 footnotes.)