GiveDirectly

2016-12-22 · ~132 words

Thanks. I think I understand the argument, but it seems like it would require GD to only target people in a very narrow income band, which might be hard to measure. Eg. suppose GD targets people with thatched roofs, and everyone over 300 USD/month household income gets a metal roof. As the income distribution shifts right, more people get metal roofs and are moved out of the "extreme poverty" category, but the distribution also shifts within the "extreme poverty" category, as the in-category distribution has more people at close to 300/mo. and fewer at 100/mo. or 200/mo. If you only give to people in eg. the 120-140/mo. range, then the effectiveness remains pretty constant as long as there's more people in that range than you have money, but that seems logistically hard.