Academic Carrots

A carefully designed conditional cash transfer program that pays students to attend school can be an effective way to improve educational outcomes.

Education is a critical component of future economic growth. Developing countries in particular can significantly improve their fortunes by increasing school attendance and graduation rates. However, getting children to go to school and keeping them in school is a challenge in many of these countries. The net enrollment rate for primary education in Sub-Saharan Africa in 2004, for instance, was only 64 percent. This rate was lower among girls and children from poor families.

The precise factors that determine the decision to go to school are still unclear. Families may compare the future returns to schooling with the costs of enrollment, which includes foregone earnings if the child goes to school instead of working. Improving the quality of education might make schooling more attractive, especially in countries where the average quality is poor, but previous empirical research finds that it does not significantly tilt the balance in favor of increased attendance. On the other hand, interventions that change the cost of schooling either through subsidies, scholarships, reduced user fees, or cash incentives seem to work.

Cash incentives in conditional transfer programs are used to encourage positive behaviors in health, nutrition, and education. These programs are especially popular in emerging markets in Latin America. Low-income families receive cash in exchange for meeting such conditions as getting their children immunized, getting regular health check-ups, and enrolling and keeping their children in school.

Many of these initiatives in developing countries are patterned after Mexico's Conditional Cash Transfer (CCT) program. A key feature of CCT is to pay students once or twice a month for meeting a specified attendance or enrollment target. While this model has been largely successful, the city of Bogota, Colombia—which established a similar program in 2005 called Conditional Subsidies for School Attendance—wanted to find out whether there are other ways to design such a program that would lead to even better outcomes than the standard model.

With the help of experts—Chicago Booth professor Marianne Bertrand, Felipe Barrera-Osorio of the World Bank, Leigh L. Linden of Columbia University, and Francisco Perez-Calle of Corpovisionares—the city of Bogota ran a pilot project to compare the effectiveness of three different conditional cash transfer models. The results are analyzed in the expert group's study titled "Improving the Design of Conditional Transfer Programs: Evidence from a Randomized Education Experiment in Colombia," which will be used to design the city's final program.

The first model that the study looked at is similar to Mexico's CCT. The second model modifies this basic version by changing the timing of cash payments such that a third of the bi-monthly payments for good attendance is transferred to the families as a lump sum at the beginning of the next school year. This design acknowledges that families are sometimes unable to save money, either because they cannot commit to saving or because of imperfect savings institutions. Families typically have to pay a larger amount for fees and supplies at the beginning of the school year, which may prevent them from enrolling their children if they have not saved enough. The third model encourages good performance by using cash incentives not only to encourage children to attend school but also to graduate and eventually go on to college or vocational school.

The study finds that simply changing the timing of the transfer significantly increases enrollment over the basic model, especially for students from the poorest families and those who have the lowest school participation rates. Moreover, many more students graduate and enroll in higher education if they are offered a specific monetary incentive to do so. "The main message is that it is possible to improve the program's design in an easy and cost-neutral way to get better results," says Bertrand.

However, the way eligibility rules are set up means that some siblings may participate in the program while others do not, which may have adverse effects on an ineligible sibling. Girls are particularly vulnerable. The study finds that families that receive a subsidy seem to reallocate and concentrate their resources on the child that participates in the program.

Tweaking the Basic Model

The pilot project was carried out in Bogota, Colombia, where, like most urban areas in middle-income countries, the enrollment rate is close to 100 percent for children between the ages of 5 and 13, but begins to decline afterwards. The drop is faster among poorer children. About 74 percent of those who were not attending school between the ages of 5 and 18 in 2003 belonged to the bottom two categories of Colombia's poverty index.

More than 13,000 children registered to participate in the project, but only about 10,000 subsidies were available. The subsidies were randomly allocated by a lottery. The cash transfers were disbursed to participants through one of the three models or "treatments" designed by the authors.

In the basic model based on Mexico's popular CCT program, families receive the equivalent of $15 a month as long as the child attends school at least 80 percent of the time. The total annual value of this transfer is three times more than what students say they earn on average if they work and slightly more than what families report spending on educational expenses. The payments would be made twice a month through a bank debit card. Students would be removed from the program if they fail to matriculate to the next grade twice, fail to reach the attendance target in two successive payment periods, or are expelled from school.

The two additional treatments introduce some modifications to the basic model. The aim is to find a design that can better reach the goals of the program while keeping the costs of each intervention roughly the same.

The savings treatment is based on research that suggests some households may face difficulties saving money for their childrens' education. In this model, a student who meets the attendance target would receive only two-thirds of $15 a month, while the remaining third would be held in a bank account. The accumulated funds are given to families in time for students to enroll in the following school year.

Because of the lower monthly transfer, it is possible that children in households with very immediate cash needs would be forced to work instead of going to school. Thus, school attendance might suffer relative to the basic model. However, if a family's long-term savings constraint is more important for children's academic participation, then the savings treatment could lead to both higher attendance and higher re-enrollment rates when compared to the basic model.

The third model provides an incentive for students to perform well in school rather than just attend classes. Upon graduation, students receive a transfer of $300, about 75 percent of the average cost of the first year of vocational school, if they enroll in a higher educational institution. Like the savings treatment, the lower monthly subsidy could reduce school attendance if families struggle to make ends meet. But the anticipation of a larger reward could also motivate students to graduate and move on to college or vocational school.

Targeting At-Risk Students

The study finds that conditional cash transfer programs significantly increased school attendance by 3 to 5 percentage points. The basic and savings treatments were about equally effective in getting children to go to school every day despite the lower monthly transfers under the savings model. The same is true when students are given an incentive to graduate; there is no evidence that a lower monthly reward causes school attendance rates to fall.

The difference in impact between models becomes more apparent when it comes to re-enrolling children the following school year. The study finds that the savings treatment increased the enrollment rate by about 4 percentage points, while there is no significant change in enrollment under the basic model. Although families who wish to send their children to school every year know that they ought to put aside part of the monthly subsidy they receive, it seems that forcing families to save through the savings treatment works precisely because they are unable to save. Enrollment likewise increased by about 3.7 percentage points if students are given an incentive to graduate.

Putting the attendance and enrollment results together suggests that changing the structure of these incentives can generate significant increases in re-enrollment without weakening students' incentives to attend school. Moreover, the study finds that the savings treatment is especially effective at raising enrollment of students from families with the lowest incomes as well as students with the lowest probability of enrolling in the absence of a subsidy. The basic treatment has little effect on the enrollment rate of this group of students. Thus, tweaking the basic model can also help target those students who are most likely to drop out of school.

The two alternative models likewise succeed at getting students to go to college or vocational school. The study finds that giving a direct incentive to graduate boosted matriculation at institutions of higher education. The savings treatment likewise increased enrollment in higher education, but the basic treatment did not have a significant impact.

Students who participate in a cash incentive program will likely work fewer hours outside the home as they focus more on school. Indeed, the study finds that the basic and savings treatments reduced the number of hours worked by about one-third.

Family Dynamics

In theory, even if a child's siblings are not eligible to participate in the program, they might still benefit from being a brother or sister of someone who does. The additional resources that the program brings to a family, for instance, can be spread equally across children so that everybody benefits.

On the other hand, parents may decide to use the transfer only on their eligible child. Worse, they may redirect some of their prior educational investments away from the children who did not qualify to the child who participates in the program. In this case, siblings would be worse off than when their brother or sister had not received the subsidy. Studying these indirect effects is important because many developing countries face funding constraints that would force them to limit eligibility to certain groups.

The results suggest that receiving a subsidy changed the allocation of educational opportunities within a household. Siblings of children who participated in the program had lower attendance and enrollment rates compared to the siblings of children who did not receive a subsidy. The attendance rate fell by 3 percentage points while the enrollment rate dropped by 7.3 percentage points. Both brothers and sisters of those children who participated in the program were affected, but the magnitude of the effect was much stronger for girls.

Thus, another important consideration in designing these programs is deciding who gets the cash incentive and whether those rules make all or only a few children in a family eligible to participate. The study's intriguing result suggests that rules restricting eligibility to certain age groups rather than household income groups, for instance, may have unintended consequences. Nonetheless, these programs do increase the overall resources available to families and the authors do not rule out the possibility that future transfers from educated to non-educated siblings could more equally distribute welfare gains.