Previously we reported the size of impacts relative to average values in the control group (e.g. a 58% increase in assets); now we report the size of impacts relative to total transfer costs (e.g. a $430 increase per $1000 spent). The bottom lines are all the same, but we think this will make it easier for you to think about cost-effectiveness (which we think you should).
I’d love to see others follow suit. I spent 30 minutes or so on various prominent NGO websites and wasn’t able to find on any of them both (i) experimental estimates of the impacts of a program, and (ii) credibly documented unit costs of the program, so that you can calculate its cost-effectiveness. (A hat tip though to Against Malaria Foundation, where I found a link to equivalent analysis on GiveWell’s site.)
It’s easy to knock the NGOs. But remember this — by and large charities put the stuff on their web pages that donors like to see. GiveDirectly is different in that we mostly put up content we think is great. So if you want to see cost-effectiveness estimates on more charity websites, start making some noise.