Note: You quote me as claiming that GiveWell adopts the deprivationist account. GiveWell’s 2015 cost-effectiveness estimate for AMF implies a deprivationist view, but the 2016 estimate explicitly calculates the QALY-equivalent value of saving a 5-year old’s life. This means there’s not a single “GiveWell view” because the reported cost-effectiveness estimate takes the median of about a dozen GiveWell employees’ individual estimates, but most employees appear to follow the time-relative interest account while a few adopt the deprivationist account.
Largely because of this change, GiveWell now claims that AMF is 4x as cost-effective as GiveDirectly, not 11x.
Thank for this! I’ll admit I’m not up to date on Givewell’s internal working. What is the difference between the deprivationist view and the QALY-equivalent of saving a 5-year old’s life? Isn’t the QALY-value of saving a life that the child would live until old age?
And yeah, I’m was just following you when you said there was a ‘GiveWell view’. I know in your post you explain how it’s a composition of staff views.
And thanks for 4x figure, I’ll edit the article. I’d mistakenly used last year’s figures.
What is the difference between the deprivationist view and the QALY-equivalent of saving a 5-year old’s life?
It sounds like you’re slightly misunderstanding me. GiveWell’s 2015 estimate said that the value of saving a 5-year old’s life was ~36 QALYs, which is a time-discounted estimate of the number of quality-adjusted years of life the 5-year old will now have. In the 2016 estimate, employees explicitly input how valuable they think it is to save a 5-year old in terms of QALYs—on the spreadsheet, look at the “Bed Nets” tab in the row “DALYs averted per death of an under-5 averted — AMF”. The median value is 8.25, and estimates range from 3 to 26. The highest estimate, 26, is still lower than last year’s estimate of 36, which suggests that none of the employees who filled this out adopt the deprivationist view.
And yeah, I’m was just following you when you said there was a ‘GiveWell view’. I know in your post you explain how it’s a composition of staff views.
Last year GiveWell’s cost-effectiveness estimate used 36 QALYs per life saved, which implies a deprivationist view. That’s not a composite of staff views, that’s the result implied by GiveWell’s reported cost-effectiveness numbers. It now appears that no GiveWell employees (or at least none who contributed to this cost-effectiveness analysis) actually hold a deprivationist view.
I’ve spent 2-3 hours going over GiveWell’s cost-effectiveness spreadsheet, so don’t expect to understand it immediately. GiveWell has a video explaining how the 2015 spreadsheet works. I haven’t much looked at the 2016 spreadsheet but it looks a lot better designed so it shouldn’t take as long to understand.
Note: You quote me as claiming that GiveWell adopts the deprivationist account. GiveWell’s 2015 cost-effectiveness estimate for AMF implies a deprivationist view, but the 2016 estimate explicitly calculates the QALY-equivalent value of saving a 5-year old’s life. This means there’s not a single “GiveWell view” because the reported cost-effectiveness estimate takes the median of about a dozen GiveWell employees’ individual estimates, but most employees appear to follow the time-relative interest account while a few adopt the deprivationist account.
Largely because of this change, GiveWell now claims that AMF is 4x as cost-effective as GiveDirectly, not 11x.
Thank for this! I’ll admit I’m not up to date on Givewell’s internal working. What is the difference between the deprivationist view and the QALY-equivalent of saving a 5-year old’s life? Isn’t the QALY-value of saving a life that the child would live until old age?
And yeah, I’m was just following you when you said there was a ‘GiveWell view’. I know in your post you explain how it’s a composition of staff views.
And thanks for 4x figure, I’ll edit the article. I’d mistakenly used last year’s figures.
It sounds like you’re slightly misunderstanding me. GiveWell’s 2015 estimate said that the value of saving a 5-year old’s life was ~36 QALYs, which is a time-discounted estimate of the number of quality-adjusted years of life the 5-year old will now have. In the 2016 estimate, employees explicitly input how valuable they think it is to save a 5-year old in terms of QALYs—on the spreadsheet, look at the “Bed Nets” tab in the row “DALYs averted per death of an under-5 averted — AMF”. The median value is 8.25, and estimates range from 3 to 26. The highest estimate, 26, is still lower than last year’s estimate of 36, which suggests that none of the employees who filled this out adopt the deprivationist view.
Last year GiveWell’s cost-effectiveness estimate used 36 QALYs per life saved, which implies a deprivationist view. That’s not a composite of staff views, that’s the result implied by GiveWell’s reported cost-effectiveness numbers. It now appears that no GiveWell employees (or at least none who contributed to this cost-effectiveness analysis) actually hold a deprivationist view.
Okay. I’ve gone back to GiveWell’s estimates, thanks. They look confusing and complicated, and it seems that quite a bit has changed in the past year.
I’ve spent 2-3 hours going over GiveWell’s cost-effectiveness spreadsheet, so don’t expect to understand it immediately. GiveWell has a video explaining how the 2015 spreadsheet works. I haven’t much looked at the 2016 spreadsheet but it looks a lot better designed so it shouldn’t take as long to understand.