Labels matter. Girls who are reminded of stereotypes about how girls perform in math do worse on math exams (in some circumstances). Publicly revealing the caste of students in India led to worse performance of students from castes that were traditionally lower in the caste hierarchy. In the U.S., posting a banner with vegetables in the form of cartoon characters increased schoolchildren’s consumption of vegetables by 90 percent. These are all forms of labeling. New research suggests that labeling matters in school scholarships – merit-based versus needs-based – as well.
Scholarships based on merit have a real allure. They create incentives for children to exert more effort to learn in school rather than just to attend. In Kenya, merit-based scholarships for primary school students led to learning gains even among children whose initial performance was so low that they had little chance of winning the scholarships. But merit-based scholarships can have adverse impacts as well. In the U.S. – where most scholarships are focused on higher education – merit-based scholarships tend to go to higher income students, and those students are more likely to go on in school anyway. Furthermore, when merit-based scholarships do boost school completion among better off students, they can increase the achievement gap between lower- and higher-income students. There’s nothing wrong with well-off students doing well in school, but is that really where scarce scholarship dollars (or shillings or pesos or francs or riels) should be focused? If not, then perhaps need-based scholarships make the most sense. Need-based scholarships for secondary students in Ghana improved learning, reduced fertility, and raised earnings once the students left school.
Scholarships affect student learning through at least three pathways. The first is the direct effect of the money. This comes after the scholarship is awarded, as in the Ghana example, and it should come whether the scholarship is based on merit or need. The second is the effect of merit scholarships on students’ incentives to work hard in school. This comes before the scholarship is awarded – and after, if renewal depends on continued performance, as in the Kenya example. We’d only expect to see this in a merit-based scholarship. The third effect is the labeling effect, which has been little tested and which we observe in a scholarship program from Cambodia.
Ten years ago, fourth-grade students in Cambodia received scholarships that would offset some of their schooling costs through primary school. In some schools, these scholarships were based on merit; in others, they were based on financial need. Three years into the program, Barrera-Osorio and Filmer show that students receiving either type of scholarship were more likely to stay enrolled in school and to show up. Scholarship recipients were between 23 and 32 percent more likely to complete primary school. But only students who received the merit scholarship did better on exams.
Of course, the first suspect in such a result is that different kinds of students received the scholarships. If higher performing students received the merit scholarships and lower performing students received the need scholarships, then we could easily imagine higher learning results among the merit scholarship recipients. But here’s the rub: “High baseline achievers who received the poverty-based scholarship performed no better in follow-up tests than the” control group. Alternatively, “poor individuals who received a merit-based scholarship did perform better on the follow-up test.” In other words, if you look at kids who would have qualified for either program – high-performing, low-income kids – learned more with the merit scholarship, not with the need scholarship. So it’s not just about the students selected: It’s about the labeling.
Nine years into the program – with the recipients now in their early 20s – new results from Barrera-Osorio, de Barros, and Filmer show that the two scholarships diverge in even more ways. Both types of scholarships still lead to more educational attainment. And again, only the merit scholarships led to increased learning – as measured by test scores. But the merit scholarship recipients also reported higher health and higher employment rates. (Notably, the benefits of the merit scholarships did not translate into higher socioemotional skills.) And as before, the results suggest that the same students do better with a merit scholarship than a need scholarship.
Why would this labelling matter? Students who receive merit scholarships are reminded of their academic potential by the scholarship itself, whereas students who receive need scholarships are reminded only of their financial needs. It isn’t hard to imagine which of those would encourage students to focusing on fulfilling their academic potential.
As the authors point out, the solution may be to use merit scholarships, but to ensure that high-need students are among the beneficiaries. This may happen through targeting schools or districts where the vast majority of students have high need, so merit scholarships will reach needy students. Or it could happen with a two-stage targeting process, where merit scholarships are rewarded among students that have been identified as high-need.
It’s great to target the most vulnerable students with financial help, but these results suggest there is value in doing it so that it emphasizes their ability over their needs.