Please note that prior to September 2017, the Center on Global Poverty and Development was known as the Stanford Center for International Development (SCID).
By Emily Miller
Sub-Saharan Africa is home to the majority of the world’s HIV/AIDS cases. In Kenya, HIV/AIDS is the number one cause of death, accounting for nearly one-fifth of all deaths. Reducing transmission through risky sexual behavior is critical, but little is known about what policies are most successful. Teen pregnancy rates are also very high in Kenya, more than double the global average. For Kenya then, an important question is whether HIV/AIDS and early pregnancy can be addressed simultaneously by policies focused on reducing unprotected sex.
In “Education, HIV, and Early Fertility: Experimental Evidence from Kenya,” Pascaline Dupas — Associate Professor of Economics at Stanford and SCID Faculty Affiliate –– and co-authors Esther Duflo (Massachusetts Institute of Technology) and Michael Kremer (Harvard University) created an experiment to test how teen pregnancy and sexually transmitted infections (STIs) are affected by cheaper education and abstinence teaching. This study is unique by virtue of its size, length, successful follow-up of participants, randomization, and accuracy of its data on disease. The study followed about 9,500 girls and 9,800 boys enrolled in sixth grade in 2003 in 328 schools in Kenya’s Western Province. The students were followed for seven years, with data collected on school attendance, pregnancies, and marriage. A high percentage of students also agreed to be tested for the herpes simplex virus type 2 at the seven-year mark. A positive result indicated ever having had the disease, thus it is an indicator of risky sexual behavior in the past. The result was a rich set of data on educational attainment, fertility rates, and prevalence of STIs.
Dupas and co-authors evaluate three alternative policies: an education subsidy, the official HIV prevention curriculum for sixth through eighth grade students in Kenya, and a joint program combining the two stand-alone programs.
The education subsidy provided two free school uniforms over the last three years of primary school, removing the largest out-of-pocket education cost for families. Although school fees were abolished in Kenya in 2003, uniforms were still the norm until recently and students faced strong social pressure to wear them. The idea behind the education subsidy was that lowering the cost of education might be a means of giving girls an incentive to reduce unprotected sex since pregnancy typically forces girls to leave school. For girls who know they will have to drop out anyway due to the inability to afford a new uniform, avoiding pregnancy may not matter, but for those who know they can stay in school thanks to the subsidy, getting pregnant would mean missing out on the opportunity to be in school.
The second intervention, the HIV prevention education, provided teachers with training to help them deliver Kenya’s national HIV/AIDS curriculum. The curriculum stresses abstinence until marriage, similar to HIV/AIDS education programs in other African countries, and is a more direct way of increasing girls’ perceptions of how risky unprotected sex might be.
Dupas, Duflo, and Kremer found that lowering the cost of education by providing free uniforms succeeded in reducing early pregnancies. School dropouts were averted, and as a result, childbearing and marriage were delayed. Despite the fact that pregnancy fell when free uniforms were provided, the level of STIs did not drop. The HIV curriculum also failed to affect STI levels. Instead, abstinence education seemed to encourage girls to marry earlier, reducing the number of out-of-wedlock teen pregnancies, but not STIs.
In contrast, the joint program that combined the uniform subsidy and abstinence-focused HIV education did reduce STIs, while leaving the teenage pregnancy rate unchanged.
To explain this intriguing set of results where effects on pregnancy and STIs diverge, Dupas and co-authors suggest this had to do with whether girls were involved in committed relationships with single partners or casual relationships with multiple partners. STI risk is greater when a girl has a higher number of partners. However, girls may be better able to prevent pregnancy in casual relationships than in marriage because the frequency of intercourse is lower. Committed relationships on the other hand involve one partner, often a spouse, and carry greater risk of pregnancy due to higher levels of unprotected sex.
Under the education subsidy, girls had an incentive to choose casual relationships over marriage, since the lower risk of pregnancy may have allowed them to stay in school. Thus, early pregnancy levels dropped, but STI levels did not. The HIV program, which touted abstinence until marriage, encouraged girls to choose a committed relationship early on, bypassing the period of search through casual relationships. Under this program, girls tended to marry at an earlier age to lower their perceived STI risk. Yet STI levels remained unchanged. This is because girls still faced a risk of infection in marriage if the husband was already infected. The abstinence education only reinforced the notion that STI risk is higher in casual relationships, but did not encourage safer sex practices in marriage.
The joint program had the greatest impact on reducing STIs because it is the only one that increased abstinence: among girls who chose to delay marriage in order to stay in school with the free uniform, the HIV curriculum convinced some to abstain altogether in order to avoid the STI risk associated with casual relationships. Yet the joint program did not decrease pregnancy much because among those who did not abstain, the curriculum’s emphasis on abstinence until marriage may have persuaded some girls who would have delayed marriage thanks to the free uniforms to instead privilege committed relationships, where pregnancies are more likely.
This research challenges the notion that any program that is successful at reducing unprotected sex can reduce both early pregnancy and STIs. Dupas, Duflo, and Kremer show that, when policies affect the choice between a committed relationship with one partner and having multiple partners, pregnancy rates may fall, but STI levels may not. Therefore, analyzing these relationship choices is fundamental to understanding policy impacts on education, fertility, and STI transmission.
In addition, the study highlights how policies interact with each other. The effect of the joint program was not simply the sum of the two individual program effects. Policies should therefore not be viewed in isolation.
Dupas’ research provides evidence that lowering the cost of education – giving girls the opportunity to go to school if they want – is a powerful way of reducing school dropout and early pregnancy rates. Nevertheless, education subsidies alone may not be sufficient to reduce STIs.
The HIV curriculum was also unable to affect STI rates since many girls married an older, already infected partner. In addition, girls married at a younger age. Finding ways to reduce unsafe sex practices in casual relationships will probably be more effective in reducing STI, pregnancy, and school dropout rates than policies that encourage committed relationships early on, the authors conclude. In related work, Dupas found that informing teenagers that older men pose a greater STI risk than younger men encouraged girls to find partners their own age, reducing STI transmission without increasing pregnancy.