I haven’t read the whole report and I don’t know anything about development economics, so I might be misinterpreting it, but I was really surprised by:
If I read this table correctly, GiveWell estimates there’s only a 50% chance that they’ll make a >=40% adjustment to GiveDirectly’s main program estimated cost-effectiveness, right after making a 330% adjustment and with many uncertainties still unresolved
Looking at this table and this graph, it seems that GiveDirectly’s program has had increasing marginal cost-effectiveness, instead of diminishing returns, by expanding to Malawi, Rwanda and Mozambique. This is another update against https://www.givedirectly.org/dont-wait/
The 46% reduction in all-cause under 5 mortality seems absurdly high, even the 23% that GiveWell uses after discounting it is way higher than I would have ever thought, and has extremely depressing implications.
Oh yeah that’s super interesting that the mortality effect doesn’t change the cost-effectiveness estimate that much. I wonder why that is excactly? Might look into it later!
Cash transfers are not targeted (i.e. lots of households receive transfers that don’t have young children) and are very expensive relative to other ways to avert child deaths ($1000 vs a few dollars for a bednet). The latter varies over more orders of magnitude than child mortality effects, so it dominates the calculation.
I haven’t read the whole report and I don’t know anything about development economics, so I might be misinterpreting it, but I was really surprised by:
If I read this table correctly, GiveWell estimates there’s only a 50% chance that they’ll make a >=40% adjustment to GiveDirectly’s main program estimated cost-effectiveness, right after making a 330% adjustment and with many uncertainties still unresolved
Looking at this table and this graph, it seems that GiveDirectly’s program has had increasing marginal cost-effectiveness, instead of diminishing returns, by expanding to Malawi, Rwanda and Mozambique. This is another update against https://www.givedirectly.org/dont-wait/
The 46% reduction in all-cause under 5 mortality seems absurdly high, even the 23% that GiveWell uses after discounting it is way higher than I would have ever thought, and has extremely depressing implications.
Interesting one nice observations. What do you mean when you say that the 23% mortality reduction has “extremely depressing implications”
I think he’s referring to the paragraph lower down which says
Oh yeah that’s super interesting that the mortality effect doesn’t change the cost-effectiveness estimate that much. I wonder why that is excactly? Might look into it later!
Cash transfers are not targeted (i.e. lots of households receive transfers that don’t have young children) and are very expensive relative to other ways to avert child deaths ($1000 vs a few dollars for a bednet). The latter varies over more orders of magnitude than child mortality effects, so it dominates the calculation.