I was recently at the GW conference on the economics & political economy of Africa where I saw an interesting paper by Richard Akresh, Emilie Bagby, Damien de Walque, and Harounan Kazianga on Burkina Faso. Akresh and co. make another compelling argument for focusing on early childhood (and indeed, in utero). Kids whose household has a shock during this critical period are less smart – and this leads to them going to school less.
Before turning to the results, it’s worth talking about Akresh and co’s data, which has some neat twists. They start with a household survey that is a baseline for an impact evaluation of a cash transfer program (bonus way to use your baseline: see Dave’s recent post on this, and more). And to get a measure of household shocks they need rainfall data. This took some multinational footwork – the area they are examining is best served by rainfall gauges in both Ghana and Burkina Faso. Once they get the rain gauge data, they are faced with the problem that there’s not a gauge in every village (the average distance is 9.6 km). So they use GIS data and inverse distance weighted interpolation to get a (nicely weighted) estimate for each village.
Now, in Burkina Faso, not every kid is enrolled in school. So Akresh and co. zero in on the households with more than one school age (5-15) kid. Of these households 17 percent send no kids to school, 24 percent send all and the other 59 percent send less than all of them. This last group is going to be the focus of most of the analysis, as Akresh and co. tackle the question of which siblings get to go to school and why.
Before we get to that, let’s go back in time. Using historical rainfall data, Akresh and co. can estimate the effect of shocks on child development. It turns out that low rainfall (more than one standard deviation below the village’s historical mean) in the village where a kid is in utero is correlated with a 0.23 standard deviation decrease in that kid’s cognitive ability z-score. There are similar effects for shocks when the kid is age zero and age 1 (although the last isn’t statistically significant). Shocks before or after this period don’t affect kids’ cognitive ability (as measured through a Raven’s test).
Now, fast forward 5 to 15 years and let’s look at whether that somewhat less smart kid is going to school, relative to siblings who didn’t suffer similarly early on in their life. Using an IV regression for ability, Akresh and co. show that a negative shock while in utero lowers the chance of being enrolled by 38 percent (relative to baseline levels), is 24 less likely to progress across grades on time, and is 113 percent more likely to start school late.
And these kids who experience a shock and have lower cognitive ability work more. They are 7 percentage points more likely to tend siblings or sick household members and 6.6 percentage points more likely to work on the family farm. During the two days preceding the survey, these children also spend more hours working: 0.65 more hours doing household chores, 0.24 hours tending siblings, 1.19 hours on the family farm.
So it’s a double disadvantage. Because their household has a shock when they’re young, these kids have lower cognitive ability. And then their parents (who are poor and resource constrained) choose them over their siblings to stay out of school and help with household labor and the family farm.
This paper makes a pretty compelling case for policy interventions so that parents aren’t faced with these kinds of decisions. And a clear priority here is making sure that the mom gets what she needs during the pregnancy, and then mom and baby get what they need to avoid these negative impacts during the first two years of life.
Before turning to the results, it’s worth talking about Akresh and co’s data, which has some neat twists. They start with a household survey that is a baseline for an impact evaluation of a cash transfer program (bonus way to use your baseline: see Dave’s recent post on this, and more). And to get a measure of household shocks they need rainfall data. This took some multinational footwork – the area they are examining is best served by rainfall gauges in both Ghana and Burkina Faso. Once they get the rain gauge data, they are faced with the problem that there’s not a gauge in every village (the average distance is 9.6 km). So they use GIS data and inverse distance weighted interpolation to get a (nicely weighted) estimate for each village.
Now, in Burkina Faso, not every kid is enrolled in school. So Akresh and co. zero in on the households with more than one school age (5-15) kid. Of these households 17 percent send no kids to school, 24 percent send all and the other 59 percent send less than all of them. This last group is going to be the focus of most of the analysis, as Akresh and co. tackle the question of which siblings get to go to school and why.
Before we get to that, let’s go back in time. Using historical rainfall data, Akresh and co. can estimate the effect of shocks on child development. It turns out that low rainfall (more than one standard deviation below the village’s historical mean) in the village where a kid is in utero is correlated with a 0.23 standard deviation decrease in that kid’s cognitive ability z-score. There are similar effects for shocks when the kid is age zero and age 1 (although the last isn’t statistically significant). Shocks before or after this period don’t affect kids’ cognitive ability (as measured through a Raven’s test).
Now, fast forward 5 to 15 years and let’s look at whether that somewhat less smart kid is going to school, relative to siblings who didn’t suffer similarly early on in their life. Using an IV regression for ability, Akresh and co. show that a negative shock while in utero lowers the chance of being enrolled by 38 percent (relative to baseline levels), is 24 less likely to progress across grades on time, and is 113 percent more likely to start school late.
And these kids who experience a shock and have lower cognitive ability work more. They are 7 percentage points more likely to tend siblings or sick household members and 6.6 percentage points more likely to work on the family farm. During the two days preceding the survey, these children also spend more hours working: 0.65 more hours doing household chores, 0.24 hours tending siblings, 1.19 hours on the family farm.
So it’s a double disadvantage. Because their household has a shock when they’re young, these kids have lower cognitive ability. And then their parents (who are poor and resource constrained) choose them over their siblings to stay out of school and help with household labor and the family farm.
This paper makes a pretty compelling case for policy interventions so that parents aren’t faced with these kinds of decisions. And a clear priority here is making sure that the mom gets what she needs during the pregnancy, and then mom and baby get what they need to avoid these negative impacts during the first two years of life.
Join the Conversation