A couple of years ago, I about an ultra-poor program in Bangladesh that was having lasting positive impacts. This week we got some good news out of Afghanistan where Guadalupe Bedoya, Aidan Coville, Johannes Haushofer, Mohammad Isaqzadeh and Jeremy Shapiro have a new looking at a similar program. And as a neat bonus, it has the most interesting results so far on women’s empowerment in the literature on these type of programs.
Bedoya and co. are looking at the same BRAC designed Targeting the Ultra Poor program when it’s implemented by government and a local NGO (and, in full disclosure, funded by the World Bank). For those of you just joining us, the program includes a spectrum of activities: transfer of a large asset (worth between $309 to $405), cash (maintenance) transfers for 12 months, training on livestock rearing and entrepreneurship, a health subsidy, and fortnightly check ins by organizers and vet services.
Bedoya and co. are working in 80 villages in the Balkh province of Afghanistan. They start with the poorest villages (subject to them being secure and accessible and having a minimum of financial and veterinary services). Within villages there is an extensive selection process, first there is a participatory wealth ranking to identify the ultra-poor. These households are then verified by the NGO and government and weighed against a set of criteria (e.g. the household is financial dependent on women’s domestic work or begging) to winnow them down further. They end up with 1,219 households which are randomized into treatment (491) and control (728).
Bedoya and co. do a baseline and then a follow up in 2018 – about one year after program completion and two years after the asset transfer. They collect a host of outcomes – especially going the extra mile on mental health and women’s empowerment. Here’s what they find (for those of you who might be worried, there are indices and multiple hypothesis correction, I am going to focus on the many significant results here):
Consumption. This goes up by a lot: 30% over the control group. It’s mostly driven by increases in food expenditure. And drops the share of households below the poverty line by 20 percentage points (context: 82 percent of the control group is below the poverty line).
Food security. The index here improves by 0.49 standard deviations. This includes effects such as the fact that the likelihood that all household members are eating at least two meals a day goes up by 11 percentage points.
Finance. Their index of financial inclusion goes up by 2.38 standard deviations – a heck of a lot. The biggest mover is the likelihood that folks have a bank account which jumps 28 percentage points, relative to the 1 percent of the control group that has an account.
Assets. This program gives folks assets, so of course this goes up. The value of livestock goes up by 315 percent relative to the control group, with folks moving into higher-return assets like cows. But Bedoya and co. are careful to note that “while large and significant the impact on livestock assets value is lower than the original value of the livestock transfer, suggesting that households may be reducing their asset base over time. This is consistent with findings by Banerjee et. al. (2015), who show a similar phenomenon across six countries initially, although with no further decline three years after the asset transfer.”
Time use. Women work more: 2.3 full time days per month, or 55 percent higher than the control group. And they also see a 22 percentage point increase in labor force participation. And men also work more, increasing their days working by 1.6 or 14 percent relative to the control group.
Income and revenue. This goes up by 22 percent relative to the control group, and the main driver is a 281 percent increase in revenue from livestock revenues. Part of this increase is coming from livestock sales – so Bedoya and co. note that this means that sustainability will have to be monitored (so someone please fund a follow up survey for them).
Psychological well being. This goes up meaningfully, with the index increasing by 0.58 SD for women and 026 SD for men. Bedoya and co. do a comprehensive job of measuring this, including life satisfaction, self esteem, depression, happiness and stress. And they throw in cortisol measures. For women all of these measures improve (and keep in mind that at baseline, 69 percent of ultra-poor women report major depression).
Women’s empowerment. When Bedoya and co. use the same index as that used by Banerjee and co. in their six country study they get similar results: nothing. Now this index focuses on household finances and expenditures. So Bedoya and co. go outside of their pre-analysis plan (hold on to your hats folks!) and give us a broader index. And this has a significant increase of 0.38 SD. The results are driven by three things: 1) increases in women’s decision making power over their own bodies and time, 2) an increase in their participation in income-generating activities (which follows from the program), and 3) an increase in political involvement and social capital (e.g. having an ID, reaching out to community leaders). This is cool – and argues that we need to look at a range of different dimensions, rather than narrow financial decision making, when looking at these types of programs.
The kids. Kids in the control group aren’t healthy: caregivers report that 51% of under five kids in the control group experienced diarrhea in the last two weeks. The program causes a 8 percentage point drop in this. For school enrollment, there are some gender differential effects: enrollment goes up by 7 percentage points for boys (against a control group mean of 56 percent) and by 5 percentage points (significant at the 10 percent level) for girls (where the mean in the control group is 53 percent). There is a significant reduction in absenteeism for boys, but nothing significant for girls. So the program seems to be widening the existing gender inequalities in education.
Bedoya and co. do a nice job at looking at likely heterogeneity of impacts and find nothing of note – this program seems to work equally well for most folks.
So the measurement, especially of psychological well being and women’s empowerment is one neat thing about this paper. They also do a nice job of showing all dollar values not only in nominal terms but also purchasing power parity – to give us a better sense of what this means for beneficiaries. They also do a serious cost benefit analysis – estimating the return against the increase in consumption alone – and find a return of 2.3 dollars for every dollar spent.
All in all these results brightened my day. Bedoya and co. show us that this program can work well, quite well actually, in a pretty fragile environment. And they shed some new insights into how it can improve not only people’s material well being, but their psychological health and women’s empowerment to boot. Now we just need to make sure the next generation benefits as well.