Estimating a Coordination Game in the Classroom: Evaluating the Role of Student and Teacher Performance Incentives

政策制定者和教育研究人员对奖励薪酬作为改善教育成果的一种手段的有效性越来越感兴趣。在过去的十年里，美国和其他国家实施了各种实地实验，以评估基于学生考试成绩向学生和/或教师提供的激励的影响。激励计划的效果通常被发现是温和的，无论学生或教师是否受到激励。贝尔曼等人。艾尔(2011)研究是个例外。这项研究分析了墨西哥教育部在88所联邦高中进行的为期三年的实地实验。实验发现，只对教师提供奖励没有显著的统计意义，但只对学生提供奖励的效果很大，对学生和教师都提供奖励的效果更大。实施任何激励制度的一个问题是，如何优化设计制度，奖励教师、学生或两者都是最优的，以及奖励多少。实地试验的一个局限性是，它们只提供有关所实施的激励方案的效果的信息，而不是关于替代设计的信息。例如，墨西哥实验提出的一个关键问题是，为什么教师激励只与学生激励相结合才有效，而学生激励本身也是有效的。本研究开发并实现了一个框架，用于研究实验效果背后的机制，并分析替代激励设计。为此，该项目开发了一个学生成就模型，在该模型中，学生和教师做出影响学习结果的努力决定。这些选择，以及其他教师和学生的特点，共同决定了考试成绩结果。该模型可以解释墨西哥项目中的激励效应模式。

There is a growing interest among policy makers and education researchers in the effectiveness of incentive payments as a means for improving educational outcomes. Over the last decade, various field experiments were implemented in the U.S. and in other countries to evaluate the impact of incentives provided to students and/or to teachers based on student test score performance.The effects of incentive programs have usually been found to be modest, regardless of whether the student or teacher receives the incentive. The Behrman et. al. (2011) study is an exception. That study analyzes a three-year field experiment conducted by the Mexican Ministry of Education in 88 Federal high schools. The experiment found no statistically significant effects of offering incentive payments to the teachers alone, but large and statistically significant effects of offering incentives only to students and a still larger effect of offering the payments to both the students and teachers.A question in implementing any incentive system is how to optimally design the system, whether it is optimal to reward teachers, students or both and by how much. A limitation of field experiments is that they are only informative about the effects of incentive schemes implemented and not about alternative designs. A key question raised by the Mexican experiment, for example, is why teacher incentives were only effective in combination with student incentives, whereas student incentives were also effective on their own.This research develops and implements a framework for studying the mechanisms underlying the experimental effects and for analyzing alternative incentive designs. To this end, the project develops a model of student achievement in which students and teachers make effort decisions that affect learning outcomes. These choices, along with other teacher and student characteristics, jointly determine test score outcomes. The model can explain the pattern of incentive effects found in the Mexican program.