140 likes | 228 Views
The Incentive Impact of Merit Scholarships. Michael Kremer Harvard University and NBER Edward Miguel University of California, Berkeley and NBER Rebecca Thornton Harvard University. Motivation. Merit scholarships historically important, fell out of fashion, now back:
E N D
The Incentive Impact of Merit Scholarships Michael Kremer Harvard University and NBER Edward Miguel University of California, Berkeley and NBER Rebecca Thornton Harvard University
Motivation • Merit scholarships historically important, fell out of fashion, now back: • We examine impact on student learning • Kenyan context: younger ages • Role of student effort
This Project • The Girls Scholarship Program (GSP) • Randomized evaluation in 127 Kenyan primary schools • Survey data on attendance, study habits, attitudes • Preview of the main findings: • Test score gains among girls (0.17-0.29 s.d.) • Spillover gains for boys in program schools • Geographic variation in test score impacts • Attendance gains for students, teachers • Cost-effective way to boost test scores
Background • Poor farming area (Luhya, Teso) • Per capita income < $400 • Primary grades 1-8; national exam; secondary 9-12 • School fees, other costs substantial (recent free primary education) • Typical kid starts primary, but drops out, doesn’t go to secondary
Related Literature • Monetary incentives for passing exams in poor Israeli high schools (Angrist and Lavy 2002, WP) • Positive impact on exam performance • Small sample size, do not explore behavioral changes • Georgia HOPE Scholarship (Cornwell et al. 2002, 2003) boosted SAT scores but students completed fewer credits • U.K. Education Maintenance Allowance (Ashworth et al 2001, WP) boosted high school enrollment
Theoretical Perspectives • Merit awards increase short-term returns to studying • Possibility of spillover benefits of incentives • Effort of students, and effort of students and teachers may be complements in the classroom • Rationale for publicly-funded merit scholarships • Some psychology research stresses possible detrimental effects of “extrinsic motivation” • Negative demoralization impacts
The Girls Scholarship Program (GSP) • Non-profit organization International Child Support • Top 15% of grade 6 girls in program schools win: • 500 KSh (US$6.40) for school fees • 1000 KSh (US$12.80) for school expenses (uniforms, books, etc.) • Public recognition, certificate at an award ceremony • Randomized evaluation: 63 program, 64 comparison schools (Balanced groups) • Attrrition in Teso district; measurement issue
Program Impacts on Test Scores (1) • Overall impact 0.12 s.d. (s.e. 0.05) • All Girls 0.17 s.d. (s.e. 0.06)
Program Impacts on Test Scores (2) • Overall impact 0.12 s.d. (s.e. 0.05) • All Girls 0.17 s.d. (s.e. 0.06) • Overall Busia district impact 0.20 s.d., Teso -0.02 s.d. • Busia Cohort 1 girls in 2001: 0.29 Busia Cohort 2 girls in 2002: 0.21
Program Impacts on Test Scores (3) • Overall impact 0.12 s.d. (s.e. 0.05) • All Girls 0.17 s.d. (s.e. 0.06) • Overall Busia district impact 0.20 s.d., Teso -0.02 s.d. • Busia Cohort 1 girls in 2001: 0.29 • Busia Cohort 2 girls in 2002: 0.21 • Persistent effect for Cohort 1 girls in 2002: 0.28 • Spillover effect for boys: 0.13-0.21 s.d.
Evidence on Mechanisms • Both student and teacher effort increases: Attendance increased by 5%, reducing school absenteeism for students, teachers by one-third • Increased use of textbooks at home in last week • But no effect on tutoring, household textbook purchases, self-esteem, attitudes toward school, amount of chores at home
Impact of Scholarship Awards • Regression discontinuity method using sharp award threshold on 2001 test • For Busia district cohort 1 girls in 2002, large school participation gains among winners • No school participation gains for Teso winners • No difference between winners, non-winners in terms of 2002 self-esteem, attitude toward school, test scores in either Busia or Teso
Cost-effectiveness vs. Other Programs • Compare cost per pupil per 0.1 s.d. gain in average test scores to teacher incentives, textbooks, deworming, flipcharts in rural western Kenya • Merit scholarships ($1.41-4.92 per pupil per 0.1 s.d. gain) are more effective than textbooks ($5.61), deworming, flipcharts, and similar to teacher incentives ($1.36-4.77) • If consider Busia district alone, merit scholarship are far more cost-effective ($0.71-2.48 per pupil)
Summary and Implications • The merit award led to large and persistent test score gains for adolescents, and positive classroom externalities • Equity concerns with merit scholarships • Competitions restricted to particular geographic areas or populations may marry efficiency with equity • Secondary scholarships an attractive policy for Kenya • Future work: Future surveys to estimate long-run impacts (i.e., labor market returns) • U.S.- merit scholarships for tertiary for kids from poor, poorly performing areas?