title 03Bettinger_PEPG08-03.pdfProgram on Education Policy and Governance Working Papers Series Paying to Learn: The Effect of Financial Incentives on Elementary School Test Scores Eric Bettinger Case Western Reserve University PEPG 08-03 Preliminary draft Please do not cite without permission Prepared for the CESifo/PEPG joint conference “Economic Incentives: Do They Work in Education” Insights and Findings from Behavioral Research CESifo Conference Center Munich, Germany May 16-17, 2008Paying to Learn: The Effect of Financial Incentives on Elementary School Test Scores Eric P. Bettinger Case Western Reserve and NBER March 27, 2008 Abstract: In recent years, policymakers and academics have become increasingly interested in applying financial incentives to individual decision-making in education. This paper presents evidence from a pay for performance program which took place in Coshocton, Ohio. Since 2004, the Coshocton City Schools has provided cash payments to students for successful completion of their standardized testing. Students in third, fourth, fifth, and sixth grade who passed district and state-mandated standardized exams are eligible for these rewards. Coshocton determined eligibility for the program using randomization. By exploiting the randomization of Coshocton's incentive program, this paper attempts to identify the effects of student incentive programs on students' academic behavior. Additionally, the structure of Coshocton's program creates some "kinks" in student incentives. These kinks reveal places where students who were eligible for the incentive program no longer had any incentive for subsequent "high stakes" work. The results present evidence that young children respond to incentives. Math scores improved about 0.15 standard deviations higher for elementary school students who were eligible for the program relative to the control group. We find little evidence that reading, social science, and science test scores changed in response to the incentive program. However, students' behavior at the specific discontinuities in the cash incentive program suggests that students respond to incentives even in ways which may not be desirable to educators. The author thanks Bob Simpson and the Coshocton City Schools, especially Patty Cramer, Wade Lucas and David Hire, for help throughout the project. The author also thanks Jim Rebitzer, David Cooper, Michael Kremer, Bridget Long, Ted Miguel, Phil Oreopoulos, and seminar participants at Case Western Reserve University, University of British Columbia, University of Arizona, University of Pittsburgh, Stanford University, UCLA, Georgetown, and Ohio State University for helpful comments. All opinions and mistakes are my own.1I. Introduction In recent years, economists, policymakers, and education researchers have become increasingly interested in the role that incentives play in education. In the United States, for example, No Child Left Behind and other recent educational reforms have changed the incentives surrounding state-mandated test scores by creating penalties if students' academic performance is not improving or if it does not meet a certain level. In developing countries, parents and children can receive cash rewards for school attendance and regular check-ups (e.g. Progresa in Mexico), for student test scores (e.g. recent experiments in Kenya and Israel), or for college attendance (e.g. England's EMA program). In Fall 2004, Coshocton City Schools (Coshocton, Ohio) developed a financial incentive program focused on improving students' academic performance in primary school. With the financial support of a local foundation,1 Coshocton began making cash payments to students for successful completion of their standardized testing. Students in third, fourth, fifth, and sixth grade who passed district and state-mandated standardized exams became eligible for these rewards. The rationale for these cash payments is straightforward. While academic achievement in early grades can greatly improve students' long-run success – facilitating college attendance and greater job opportunities (e.g. Schweinhart and Weikart 2002), these long-run benefits are intangible to many young children. Few third, fourth, fifth, or sixth graders actively think about college attendance or employment. Additionally, children are inherently impatient, and many studies in education, psychology, and economics document how children are often more motivated by short-run rewards than less tangible long-run rewards (e.g. Chelonis, Flake, Baldwin, Blake and Paule 2004, Harbaugh and Krause 1998, Bettinger and Slonim 2007). The Coshocton experiment makes a unique contribution to previous research in economics, education, and psychology. Recent studies in economics (e.g. Angrist and Lavy 2007, Angrist, Lang, and Oreopoulos 2007, Kremer, Miguel and Thornton 2008, Leuven, Oosterbeek, and van der Klaauw 2003) have largely focused on the effects of external incentives on students' academic achievement among students in secondary and post-secondary schools. Coshocton is the first study in economics to focus on financial incentives for student achievement in primary schools. The Coshocton experiment also builds on prior literature in 1 The funding for these cash payments comes from a directed grant from the Simpson Family Foundation.2psychology. Psychologists have been particularly active in the development of theory and evidence on the role of financial incentive programs. Most of the early literature on “token economies” and incentives focused on the effects of external or extrinsic motivators on individual’s intrinsic motivation (e.g. Ayllon & Azrin 1968, Kazdin 1975a & 1975b, Kazdin & Bootzin 1972, Deci 1975, Lepper, Greene, & Nisbett 1973). In Coshocton, we were able to gather data on intrinsic motivation among students and can measure the impact of the program on intrinsic motivation. In our discussion of the results, we attempt to reconcile the recent findings in economics with previous literature from psychology. One of the most important features of the Coshocton incentive program is the fact that the district assigned eligibility for the program using randomization. This was a condition mandated by the sponsoring foundation. The unit of randomization is a grade-school level (i.e. grade i at school j), and Coshocton city schools conducted lotteries at the