Does improved human capital empower girls? An interesting paper  by Willa Friedman, Michael Kremer, Ted Miguel, and Rebecca Thornton give us some insight into the answers.
First, some theory on how more education might affect girls is in order. A first school of thought that Friedman and co. label early modernization argues that education is going to break down traditional attachments which are based on things like gender, ethnicity, religion and the like in favor of achievement and merit. A second school of thought is more grim: education is basically a tool of cultural indoctrination and social control. Third, we have education as politically and socially empowering. And this doesn’t have to be benign – the build of expectations can lead to frustration and violence.
Now the existing empirical literature doesn’t help us much in distinguishing among these – the cross country work and existing observational studies show a correlation between social and political empowerment of girls and education, but causality is elusive.
Enter Friedman and co. They are looking at a girls scholarship program in Kenya. Previous work  has shown that this program significantly boosted test scores for the scholarship girls, with externalities in terms of teacher attendance and the performance of non-scholarship students. Here they go longer term – looking at the impacts on these girls a couple of years later.
First, the set-up. This is a study in rural, western Kenya (as they put it: “residents of a rural backwater”) among girls from politically weak ethnic groups. In 2001, an NGO randomly selected 34 primary schools for a girls scholarship program and 35 control schools. Girls in the top 15% in terms of exam scores got an award of $19.20 – some for school fees, some for school expenses, a year for the last two years of primary school.
An earlier survey gives them the results discussed above, but they went back from 2005-2007 to look at more medium term effects (the girls were now between 17 and 21 years old). Tracking was pretty effective – they have a follow up rate of around 80%.
The first set of impacts they look at is human capital. Girls from scholarship schools are still scoring significantly higher after the program had ended. One interesting slice of heterogeneity that they look at is comparing girls above +2sd in terms of initial test scores (who had a good shot at getting the scholarships) versus those below +2sd (who were much less likely to get the scholarships). As with the earlier results, both sets of girls from scholarship schools showed similar benefits. Girls from treatment schools were also more likely to have attended at least some secondary school and were also more likely to still be in school.
They then turn to measures of autonomy within the household. Treatment school girls were significantly less likely to view domestic violence as acceptable – to the tune of 25% less than the control group. They were also significantly (at 90% confidence) less likely to have had arranged marriages (and here the alternative is elopement). They do not find significant effects on a range of other measures – attitudes on overall gender equality, fertility, contraception, and age of marriage – probably due to power issues.
They then move on to political and social attitudes. The first interesting result here looks at heterogeneity by survey timing. Looking at the surveys which happened before the run-up to the 2007 election, they find that the treatment appears to have heightened feelings that ethnic affiliation was “very important.” They then look at a range of attitudes towards democracy and governance, including, for example, agreement with statements such as “we should choose our leaders in this country through regular, open, and honest elections.” The treatment had no impact on these. So it looks like the increased education didn’t weaken ethnic attachments or promote more belief in democracy.
However, intellectual engagement with the political world does seem to have been boosted by the program. Girls in treatment schools were less likely to listen to the radio (not so informative on matters political) and more likely to read the newspaper (+14%) – which has much more coverage of politics. Knowledge goes up – girls from treatment schools were more likely to be able to name the health minister and the president of Uganda, for example. And the treatment girls don’t seem to like what they are reading about. They are less likely to agree with the statement “we should show more respect for authority” and more likely to agree with the statement “as citizens, we should be more active in questioning the actions of our leaders.”
Now treatment school girls aren’t participating more in political and social affairs – there are no significant differences here across a range of measures for this. But they are more significantly likely (by around 6%) to think that the use of violence is sometimes justified in politics (keep in mind the post-election violence of 2007, as well as Kenya’s history). So Friedman and co. sum this up as increased awareness and less satisfaction with government, but not much faith in changing this through democratic means.
So what do we have in the end? From these results, it looks like education did in fact empower girls – they show more autonomous attitudes within the household, and in society. So we can scratch the simple education as social control argument. But they’re not toeing the line in line with modernization either – they aren’t keen supporters of democratic means, at least not how it is practiced in Kenya.