Adolescent girls, school, HIV, and pregnancy: evidence from Kenya Michael Kremer, Harvard University Esther Duflo, Pascaline Dupas, Samuel Sinei; Edward Miguel and Rebecca Thornton
Context Poor, rural region in Western Kenya Most children start primary school, but drop out before secondary school Free primary education Teen childbearing (DHS) 16% of 17-year old women have begun childbearing 46% of 19-year old women Two randomized evaluations
Reproductive health programs Teacher training on HIV/AIDS curriculum Government curriculum Limited discussion of condoms Debate on condoms and essay on protecting self against HIV/AIDS Information about HIV age profile (Dupas) Reducing cost of education
Teacher training in HIV/AIDS curriculum No significant/limited impact on knowledge, self-reported sexual behavior, childbearing Increases marriage conditional on childbearing by 6 %
Debate on condoms/ Essay on protecting self No increase in self-reported sexual activity Increases knowledge of and self- reported use of condoms Waiting for childbearing outcomes
Information about age-profile of infection Large gender differences in age profile imply cross-generational sex is key route of infection Girls not aware Information on age profile of infection by gender (delivered by NGO workers) Video
Outcomes Reduces girls’ childbearing by 32% Reduces childbearing with adult men by 65%
Reducing cost of education Primary education is free, but de facto costs (uniforms) School vs. childbearing tradeoff Reduces dropout by 13.5% Increases confidence in girls for saying “No” if partner wants to have sex by 5% Reduces probability of reporting ever having had sex by 13% Reduces teen childbearing by 10%
Targeted merit scholarships Policy challenge: secondary education – how allocate funds Prior to FPE; merit scholarships for girls scoring well in 6 th grade exams Girl Scholarship Program (GSP) in two districts in Western Kenya
Girls Scholarship Program Overall incentive effect: test-score gains of 0.14 sd (~6 percentage points) Teacher attendance up Girls with low pre-test scores gain No evidence of weakened intrinsic motivation/gaming Effects persist Heterogeneous program effect by district In successful district: test scores increase 0.25 s.d., gains for boys too, student attendance up In other district: cannot reject zero program impact
Conclusion Cost effective programs exist Informing girls about risks of relationships with older men Reducing cost of education Secondary school scholarships? Jury out? HIV/AIDS education training for new teachers Debates on condoms and essays on protecting self How to expand access to secondary education?
Outcomes of interest HIV education in schools Knowledge and attitudes Condom use Teenage childbearing Marriage Drop-out
HIV education: Improvement in students’ knowledge?
HIV education: Improvement in students’ attitudes?
HIV education: Self-reported sexual behavior
HIV education: Childbearing rates
HIV education: Dropout
HIV education: Relative risk information