Panel Paper: When Merit-Based Scholarship Does Not Work: Evidence from a Field Experiment in Malawi

Friday, November 3, 2017
Water Tower (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Jim Berry1, Hyuncheol Bryant Kim1 and Hyuk Son2, (1)Cornell University, (2)Yonsei University


We study the impacts of two types of merit-based scholarships on performance of primary school students through a field experiment in Malawi. One criticism of merit-based scholarship programs is that by providing rewards to only the very top performers, lower-performing students who are unlikely to receive the incentive may not respond to the programs. An incentive design that could address this concern follows that proposed by Barlevy and Neal (2012), in which students are grouped by baseline score, and incentives are awarded to the top performers in each group.

We study the impacts of this “relative” merit-based scholarship program alongside a more typical program that provided scholarships to the top students in the sample. The “standard” merit-based scholarship was given to the top 15% students in the final exam. In the relative merit-based scholarship scheme, students were divided into groups of 100 with similar baseline test scores, and within each group, the top 15 students in the final exam receive a scholarship. One hundred seventeen classrooms in 31 schools were randomly assigned to either the standard merit-based scholarship program, relative merit-based scholarship program, or to a control group.

We find no evidence that either scholarship program increased test scores. In fact, the standard merit-based scholarship decreased student test scores by 0.14SD compared to the control group. This corresponds to decreased motivation to study hard, and study time, as measured in follow-up surveys. Moreover, the decreases of motivation, study time, and test scores are larger in those with lower baseline test scores, who are less likely to receive the award. The relative merit-based scholarship program had no significant impacts on student test scores, although there is little evidence of negative effects, as in the standard merit-based scholarship scheme. This study shows that by themselves scholarship programs may not motivate students enough to increase performance in school, and the design of standard merit-based scholarships may in fact result in decreases in motivation and performance.