The past decades have seen large increase in primary enrollment, and a closing (for most, but not all, countries) of the gender gap in enrolment. The next step is to look at secondary school. A nice new paper by Esther Duflo, Pascaline Dupas and Michael Kremer looks at what happens when you make it accessible to more young people.
Duflo and co. are looking at the medium run impacts of a secondary school scholarship program for secondary school in Ghana. Working with the Ghana Education Service and IPA, Duflo and co. randomly assign scholarships to a group of students who took the entrance exam and got into high school, but aren’t planning to enroll, mostly because of the cost. High school in Ghana isn’t cheap – tuition costs about 500 Ghana cedis per year and there are additional fees on top of this. To put this in perspective, GDP per capita in 2011 was about 2400 cedis.
For their study, Duflo and co. make sure that the beneficiaries are equally spread across girls and boys and this will let them look at gender differences in the impacts. It’s also important to note that when students in Ghana enter the four years of senior high school (SHS) they are applying to a track (vocational versus academic) and a major. As one might expect, the results are different for those entering the vocational versus the academic tracks.
Before turning to the results, it’s worth discussing the data. At the start of the program, Duflo and co. give all respondents a cell phone and, twice yearly, airtime credit. In a country with a fair amount of switching across phone providers, this increases the chance that their respondents will keep the number active. This allows them to complement the 2008 baseline with phone survey follow ups (they crank out two 30 minute follow ups in 2015 and 2016), as well as in-person, longer survey (including math and reading tests) in 2013. Given the phones, as well as in-person tracking for those who don’t answer, they keep attrition to a remarkably low 2 percent in 2016.
On to the results. Immediately after getting the scholarship, 75 percent of the lottery winners enroll – a rate which is about 4 times higher than the comparison group. By the endline, a significant portion of the control group had managed to find the money for school: 74 percent of the scholarship kids completed SHS against 47 percent of the control. Breaking the treatment effect out by gender, Duflo and co. find that the scholarship increased the SHS completion rate for girls from 42 to 68 percent, and from 53 to 79 percent among boys. The lower completion rate among girls is largely in part due to the fact that, in order to get the size of the female they wanted, they had to take a second cohort of girls who had finished junior high school one year before they recruited the other girls and boys. And this one year out of school makes a difference in their trajectory.
While a lot of recent work has shown quality issues with education in sub-Saharan Africa (see the World Bank’s service delivery indicators work for example), these folks are still getting something out of their education. Scholarship winners scored 0.14 standard deviations higher on reading tests and 0.12 on math tests, with no significant differences across the vocational and academic streams. There is a big difference by gender, however, with females increasing by 0.25 standard deviations (overall) and males increasing by 0.05 standard deviations. The female effect size is relative to a control group where males strongly outperform females – female scholarship winners manage to catch up with male non-winners but still lag behind male scholarship winners.
Outside of test scores, scholarship winners also seem to have more general knowledge and are more likely to know how to use the internet. The female differences here are striking: those who win the scholarship are more likely to have a bank account, a Facebook account and an email account (the last significant at 10 percent).
In terms of going on to tertiary education, the gains are, not surprisingly, concentrated among students pursuing the academic track who are 5.3 percent more likely to go on in school (this is a 48% increase relative to the control). And this effect is super strong for females, who see a 9.3 percentage point jump (the result for men is neither large nor significant). This effect is big enough to close the female-male gap in tertiary enrollment that Duflo and co. see in their sample.
On the flip side, the gains in terms of employment are concentrated among those who pursue the vocational track. For these folks, scholarships lead to a 24 percent increase in monthly earnings (significant at the 10 percent level, with their other measure significant at better). This is driven by a sharp uptick in the probability of working rather than wages or hours worked. Interestingly, there are no big gender differences across these Using income outcomes, as well as other parameters, Duflo and co. estimate that there is a 13 percent return to SHS for those in the vocational stream. In talking about the employment effects, Duflo and co. do a good job of setting the macroeconomic scene noting that in 2016, things weren’t particularly great in Ghana and youth unemployment was quite high. They also add a nice discussion of how this might matter for external validity.
The scholarships also have strong fertility and marriage effects for females. By age 25, scholarship females were 9.1 percentage points (26%) less likely to have lived with a partner, 10.7 percentage points (18%) less likely to have ever been pregnant and have 0.217 fewer children regardless of whether they are in the academic or vocational stream. Duflo and co. show how these effects are consistent with a) increased opportunity costs of children, b) being better able to decode information to make informed choices, and c) education shaping fertility preferences. Scholarship winners also engage in healthier behaviors, with safer sex (significant for men) and increased hand-washing, bed net use, and use of mosquito repellent.
Taken together, these results show some strong medium-run effects of making secondary education more accessible to poorer youth. It’s also interesting to see what this access can do to the life trajectory of young men and women. Duflo and co. indicate that they plan to follow these folks in the years to come. I look forward to seeing what they find.
Duflo and co. are looking at the medium run impacts of a secondary school scholarship program for secondary school in Ghana. Working with the Ghana Education Service and IPA, Duflo and co. randomly assign scholarships to a group of students who took the entrance exam and got into high school, but aren’t planning to enroll, mostly because of the cost. High school in Ghana isn’t cheap – tuition costs about 500 Ghana cedis per year and there are additional fees on top of this. To put this in perspective, GDP per capita in 2011 was about 2400 cedis.
For their study, Duflo and co. make sure that the beneficiaries are equally spread across girls and boys and this will let them look at gender differences in the impacts. It’s also important to note that when students in Ghana enter the four years of senior high school (SHS) they are applying to a track (vocational versus academic) and a major. As one might expect, the results are different for those entering the vocational versus the academic tracks.
Before turning to the results, it’s worth discussing the data. At the start of the program, Duflo and co. give all respondents a cell phone and, twice yearly, airtime credit. In a country with a fair amount of switching across phone providers, this increases the chance that their respondents will keep the number active. This allows them to complement the 2008 baseline with phone survey follow ups (they crank out two 30 minute follow ups in 2015 and 2016), as well as in-person, longer survey (including math and reading tests) in 2013. Given the phones, as well as in-person tracking for those who don’t answer, they keep attrition to a remarkably low 2 percent in 2016.
On to the results. Immediately after getting the scholarship, 75 percent of the lottery winners enroll – a rate which is about 4 times higher than the comparison group. By the endline, a significant portion of the control group had managed to find the money for school: 74 percent of the scholarship kids completed SHS against 47 percent of the control. Breaking the treatment effect out by gender, Duflo and co. find that the scholarship increased the SHS completion rate for girls from 42 to 68 percent, and from 53 to 79 percent among boys. The lower completion rate among girls is largely in part due to the fact that, in order to get the size of the female they wanted, they had to take a second cohort of girls who had finished junior high school one year before they recruited the other girls and boys. And this one year out of school makes a difference in their trajectory.
While a lot of recent work has shown quality issues with education in sub-Saharan Africa (see the World Bank’s service delivery indicators work for example), these folks are still getting something out of their education. Scholarship winners scored 0.14 standard deviations higher on reading tests and 0.12 on math tests, with no significant differences across the vocational and academic streams. There is a big difference by gender, however, with females increasing by 0.25 standard deviations (overall) and males increasing by 0.05 standard deviations. The female effect size is relative to a control group where males strongly outperform females – female scholarship winners manage to catch up with male non-winners but still lag behind male scholarship winners.
Outside of test scores, scholarship winners also seem to have more general knowledge and are more likely to know how to use the internet. The female differences here are striking: those who win the scholarship are more likely to have a bank account, a Facebook account and an email account (the last significant at 10 percent).
In terms of going on to tertiary education, the gains are, not surprisingly, concentrated among students pursuing the academic track who are 5.3 percent more likely to go on in school (this is a 48% increase relative to the control). And this effect is super strong for females, who see a 9.3 percentage point jump (the result for men is neither large nor significant). This effect is big enough to close the female-male gap in tertiary enrollment that Duflo and co. see in their sample.
On the flip side, the gains in terms of employment are concentrated among those who pursue the vocational track. For these folks, scholarships lead to a 24 percent increase in monthly earnings (significant at the 10 percent level, with their other measure significant at better). This is driven by a sharp uptick in the probability of working rather than wages or hours worked. Interestingly, there are no big gender differences across these Using income outcomes, as well as other parameters, Duflo and co. estimate that there is a 13 percent return to SHS for those in the vocational stream. In talking about the employment effects, Duflo and co. do a good job of setting the macroeconomic scene noting that in 2016, things weren’t particularly great in Ghana and youth unemployment was quite high. They also add a nice discussion of how this might matter for external validity.
The scholarships also have strong fertility and marriage effects for females. By age 25, scholarship females were 9.1 percentage points (26%) less likely to have lived with a partner, 10.7 percentage points (18%) less likely to have ever been pregnant and have 0.217 fewer children regardless of whether they are in the academic or vocational stream. Duflo and co. show how these effects are consistent with a) increased opportunity costs of children, b) being better able to decode information to make informed choices, and c) education shaping fertility preferences. Scholarship winners also engage in healthier behaviors, with safer sex (significant for men) and increased hand-washing, bed net use, and use of mosquito repellent.
Taken together, these results show some strong medium-run effects of making secondary education more accessible to poorer youth. It’s also interesting to see what this access can do to the life trajectory of young men and women. Duflo and co. indicate that they plan to follow these folks in the years to come. I look forward to seeing what they find.
Join the Conversation