Incentives for Effort or Outputs? A Field Experiment to Improve Student Performance

One key choice in designing an incentive is whether to reward outcomes directly (outputs) or to reward the actions and behaviors that lead to those outcomes (effort/inputs). I conduct a novel direct test of an input incentive designed to increase student effort against both an output incentive and a control that does not receive an incentive. The interventions were implemented in a classroom-level randomized experiment with school children in India. A math software curriculum is implemented in all classrooms regardless of which activity is incentivized. It includes learning modules (the incentivized input) that are completed throughout a unit as well as a test at the end of the unit (the incentivized output). The two incentives are both piecerate and announced at the beginning of each unit. Students who receive an input incentive perform .57 standard deviations better than the control group on a non-incentivized outcome test. This performance is statistically significantly larger than the impact of the output incentive (which .24 standard deviations and not significant relative to the control). The input incentive is also almost twice as cost-effective as the output incentive. These results provide evidence that there can be large returns to directly inducing student effort in the classroom. The input incentive works better for present-biased students along an incentive-compatible measure of time preferences collected at baseline, which provides evidence to support the hypothesis that more frequent payments can address time inconsistency. This study also provides direct evidence that piecerate input incentives can be more effective than piecerate output incentives.