Scientific evidence to guide policy on MDGs

Governments need rigorous evidence on program effectiveness to make informed choices on where to invest scarce resources in the fight against poverty. Randomized evaluations help identify promising strategies by providing rigorous evidence on the effectiveness of an intervention in a specific context. When the same program is evaluated across multiple contexts, or when similar program mechanisms are tested and compared across different settings, it helps us better understand what is likely to work, and why. Like successful interventions, unsuccessful ones also add important knowledge to our understanding of what drives the persistence of poverty, and what we may be able to do to solve some of the problems facing the poor today.

Over the past 10 years, J-PAL researchers and others have conducted hundreds of randomized evaluations on a range of topics that span the different MDGs. Our seven best buys emerge from this wider evidence base and share common criteria: they cost very little to implement and are easy to administer; they have been shown to be effective in at least one randomized evaluation, but in most cases in more than one context; and, their effects are consistent with broad patterns of individual and collective behavior found in a variety of settings.

While we don’t know everything about what works in development—and we never will—our best buys provide examples of five more general lessons about development effectiveness:

First, the MDG goals are interconnected: Hence, children’s poor health may be an important reason for their low school attendance. This is illustrated with the example of in-school deworming programs in Kenya and India. In other contexts, fighting anemia or malaria, or addressing pregnant women’s nutritional deficiencies, have been shown to be effective in achieving broader development goals.

Second, improved access to schooling has limited benefits if children are not given a chance to learn. Heterogeneity in students’ level of school preparedness is an important challenge for teachers who are often grappling with curricula that are poorly adapted to the majority of learners. Children who fall behind early on find it difficult to catch up. Focusing on teaching basic skills can have great payoffs. We illustrate this with an intervention that uses volunteers or quasi-volunteers for remedial education in India. In other contexts, splitting classes into “tracks” by initial performance has been shown to improve learning for both stronger and weaker students, as has learning from educational computer games that adjust to the learner’s level of comprehension. In contrast, simply increasing resources to schools without making changes in pedagogy or school management has had small and generally disappointing effects.  

Third, asking people to pay—even small fees—strongly discourages them from doing things today that could benefit them in the future. This behavior is not specific to the poor: research in psychology and economics has shown this to be an important feature of how people behave. It might, however, be particularly damaging for people in developing countries who face stricter credit constraints and sometimes have to work hard to access services that are taken for granted in developed countries. Eliminating these costs, or providing small incentives, can have very large positive effects on people’s behavior. We illustrate this with four different, but related interventions: free bednet distribution, free uniforms, small, time-limited subsidies for fertilizer purchases, and incentives for immunization. 

Fourth, providing simple, actionable information can be highly effective in changing behavior but only if the content of the information is easily understood and gives clear direction to guide behavior change. We illustrate this with two HIV prevention programs from Kenya. When young girls in Kenya were provided with information about the relative HIV infection rates of men in their 20s and older, compared to younger men, the number of teen pregnancies with older men was dramatically reduced (by 65 percent), with no corresponding increase in pregnancies with younger men. In comparison, training teachers in the standard national HIV/ AIDS curriculum led to no reduction in teen pregnancies with older men. These findings are consistent with evidence that parents and pupils respond to simple, actionable information about returns to schooling to increase (and, in some cases, decrease) school attendance, and a rich literature in psychology.

Finally, persistent voter bias against women leaders can be changed. All over the world, women are underrepresented in political positions. Research in political science, economics and psychology has shown that this is in part due to bias among voters regarding women’s ability to be effective leaders. The good news is that if voters are shown that women can make good leaders—through exposure to female leaders—this can help voters lower their bias. We illustrate this with the impact of female quota in politics in India.

For more information about the J-PAL best buys:

1. For as little as 50 cents per child per year, deworming of children through mass school-based programs can cut school absenteeism by a quarter. By the end of their primary school education, children who have been regularly treated for worms will have an entire additional year of education.	read policy briefcase
2. It costs no more than $2.25 per child per year to provide remedial education to children who lack basic reading skills. Within three months, Pratham’s “Read India” remedial education program boosted letter recognition by 60 percentage points, and ability to read and comprehend a short story by 35 percentage points.	read policy briefcase read academic paper
3. Doing away with small user fees on bednets to make them available for free to pregnant women and mothers in health clinics costs less than $5 per net and can increase uptake by 75 percent. Women who receive free bednets are as likely to use them as those who pay for them. Free provision does not increase wastage. It does however make the benefits of bednets accessible to many more mothers and young children.	read academic paper
4. Quotas for women in politics costs practically nothing. Yet, it increases women’s political participation and shift spending towards women’s priorities, such as clean water. Voters systematically underrate women politicians but exposure to women leaders can eliminate biases against women.	read policy briefcase
5. It costs $4 per girl per year to provide free primary school uniforms that help keep girls in school and reduce teen pregnancies by 9 percent. Informing young girls about the added risk of unprotected sex with men in their 20s (or older) costs just $1 per student and can reduce the teen pregnancies with older men by 65 percent.	read policy briefcase
6. Smart subsidies to farmers boost technology adoption, farm productivity and income. Time-limited offers to purchase fertilizers in the harvesting season, with free delivery in the planting season, can massively increase uptake and usage of fertilizers.	read academic paper
7. Small incentives can be a minor additional price to pay to get children immunized. Vaccines are highly cost effective and provided for free in many countries.  Yet, there are areas where coverage is low. Small incentives – such as a bag of lentils per shot - if offered to households can massively boost uptake in low take-up regions.	(paper in progress)

Not every study finds a highly effective program. Some will find that potentially good new ideas did not work. This is useful and important knowledge too. Other studies find that things work but the cost is high given the magnitude of the benefits. Even so, all of these studies are critical as they allow us to assess comparative cost effectiveness and help achieve buy-in for scale-up of the most successful programs. Our Fighting Poverty: What Works? bulletins spell out general policy lessons from across different studies.

If you have any comments or questions or want to know more about the Abdul Latif Jameel Poverty Action Lab, you can contact us on povertylab@mit.edu.