Not all pay-for-performance schemes are created equal.
While performance pay for teachers has gained quite a bit of attention lately, the reality is that fewer than four percent of schools have implemented incentive schemes. Tales of success are tempered by stories of failure (and teacher cheating), which has made policymakers and school administrators hesitant to embrace pay for performance.
Much research has been done to determine which performance pay schemes are successful. But systematic reviews of the structure of performance pay, and its its effective and ineffective components, are infrequent.
Derek Neal addresses this gap in research on performance pay in the 2011 NBER working paper “The Design of Performance Pay in Education.” Neal reviews dozens of studies of performance pay schemes from schools around the world. He identifies several systematic problems with these schemes and uses economic theory to formulate solutions.
Neal finds that school districts have difficulty setting optimal performance standards. Teachers aren’t motivated to pursue standards that they perceive to be unattainable. On the other hand, teachers can achieve low standards without increasing effort, potentially leading to a salary increase despite stagnant student performance.
Absolute performance standards are also a problem because teachers have an incentive to pressure regulators to lower standards or testing agencies to change the testing scale. Neal believes that relative performance standards, which evaluate teachers against one another, would alleviate both problems. Relative performance schemes’ reliance on competition precludes standard manipulation. A well-designed relative “performance contest” should naturally reveal the performance standard that elicits efficient teacher effort.
Still, studies show that while incentive schemes increase student performance on high-stakes tests they do not produce equal gains on comparable low-stakes tests. This suggests that teachers are coaching students to excel on specific high-stakes tests and decreasing time devoted to broader learning that might promote long-term student success. Consequently, high-stakes test results considerably overestimate actual student and teacher performance gains.
To circumvent this problem, Neal proposes using assessments with unpredictable and varied questions and a separate low-stakes test with year-to-year overlap in questions.
Finally, Neal argues that student achievement measures should take into account students’ backgrounds and baseline achievement to disrupt teachers’ incentive to avoid low-performing students and schools.