Perhaps children learn better on a full belly? Or perhaps parents keep children away from school simply so that they can feed them? Or even send them to work so there’s enough money to feed that child? Any of those would suggest that feeding the child at school will lead to more learning and a better attendance record. Michael Kremer has evaluated just that, and finds that it’s true. Feeding children in absolute poverty (the $1.90 a day income kind, not something that exists anywhere in the United States) leads to much better educational outcomes. So if we subsidize school lunch in places too poor to afford it, we make the world a better place.
At which point we come back to one of the older lessons of economics: Just because something should be done doesn’t mean that getting government to do it works. A charity called Mary’s Meals does just this for $20 per child per year. The U.S. government has a program to do the same, which costs about $200 per child per year. Both are feeding millions a year — which is great, but think how much more could be done if it were all being done efficiently?
Banerjee, Duflo, and Kremer are well deserving of their Nobel Prize — not so much for making this vale of tears less dreadful, but for arming us with more of the tools we need to work out how to do that.