Updated Cost-effectiveness Estimates

Givewell's cost-effectiveness estimates are of widespread interest. Aggregating their estimates for the 2013 giving season shows Schistosomiasis Control Initiative to be the best buy in public health, saving a life for $6714. Note that Givewell avoid relying excessively on individual cost-effectiveness estimates, and on a more global assessment, they rate SCI roughly equally effective as Deworm the World and Give Directly.

In the 2013 Giving Season, five Givewell employees made cost-effectiveness estimates for the top three charities. Elie apparently enjoyed filling out the spreadsheet so much that he did it twice - giving one estimate that was optimistic for deworming, and one that was optimistic about cash-transfers. I geometrically averaged these to provide a single estimate for Elie. Then, I geometrically averaged Elie's estimate with the other four.

This gave an aggregate estimate that Schistosomiasis Control Initiative stimulates 2.4 times more consumption than Give Directly and saves the equivalent of one life (36.5 disability-adjusted life years) for $6,714. This is lower than the previous estimate of $3,400 for the Against Malaria Foundation, which expired when Givewell came to believe that it was running out of room for more funding.

Givewell's estimated cost of saving a life has been trending up for a while now and I think there are at least two things occurring here:

First, as Givewell and the EA community grow, we become wiser, and our estimates regress to the mean.

Second, as Givewell grows, more good organisations like AMF will reach the limits of their room for more funding. In the last three years, the amount of money moved has continually increased: from $2m in 2010 to $5m in 2011 to $10m in 2012. In 2013, they reported continued strong growth. Each year, the number of low-hanging fruit decreases. Eliezer describes this model:

Isn't this what we would expect to see if GiveWell is working? As Earth's most urgent needs are addressed, marginal impact of additional donations should fall. I'm not saying this is what's happening, but it's worth noting that diminishing marginal effectiveness of the next dollar is exactly what you expect to see as deficiencies are corrected.

As Givewell-donors meet more organisations' funding needs, there will be a growing need for foundations to avoid scaling-up any one organisation too rapidly and it is good to see that Giving What We Can Trust and GoodVentures are stepping into this role.

At the same time, I think the increasing cost of doing good in the developing world should drive us to diversify our search for giving opportunities.