The Nonlinear Library EA - How we work, #2: We look at specific opportunities, not just general interventions by GiveWell
Nov 12, 2023
07:39
Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: How we work, #2: We look at specific opportunities, not just general interventions, published by GiveWell on November 12, 2023 on The Effective Altruism Forum.
This post is the second in a multi-part series, covering how GiveWell works and what we fund. The first post, on cost-effectiveness, is here. Through these posts, we hope to give a better understanding of our research and decision-making.
Author: Isabel Arjmand
Looking forward, not just backward
When we consider recommending funding, we don't just want to know whether a program has generally been cost-effective in the past - we want to know how additional funding would be used.
People sometimes think of GiveWell as recommending entire programs or organizations. This was more accurate in GiveWell's early days, but now we tend to narrow in on specific opportunities. Rather than asking whether it is cost-effective to deliver long-lasting insecticide-treated nets in general, we ask more specific questions, such as whether it is cost-effective to fund net distributions in 2023 in Benue, Plateau, and Zamfara states, Nigeria, given the local burden of malaria and the costs of delivering nets in those states.
Geographic factors affecting cost-effectiveness
The same program can vary widely in cost-effectiveness across locations. The burden of a disease in a particular place is often a key factor in determining overall cost-effectiveness. All else equal, it's much more impactful to deliver vitamin A supplements in areas with high rates of vitamin A deficiency than in areas where almost everyone consumes sufficient vitamin A as part of their diet.
As another example, we estimate it costs roughly the same amount for the Against Malaria Foundation to deliver an insecticide-treated net in Chad as it does in Guinea (about $4 in both locations). But, we estimate that malaria-attributable deaths of young children in the absence of nets would be roughly 5 times higher in Guinea than in Chad (roughly 8.8 deaths per 1,000 per year versus roughly 1.7 per 1,000), which leads AMF's program to be much more cost-effective in Guinea.
This map from Our World in Data gives a sense of how deaths from malaria vary worldwide.[3]
Because cost-effectiveness varies with geography, we ask questions specific to the countries or regions where a program would take place. When we were investigating an opportunity to fund water chlorination in Malawi, for example, we wanted to know:
How does baseline mortality from poor water quality in Malawi compare with that in the regions where the key studies on water chlorination took place?
What is the overall morbidity burden from diarrhea in Malawi?
Might people be more or less likely to use chlorinated water in this area than in the areas where the key studies took place?
What does it cost to serve one person with in-line chlorination for one year? We calculate this, in part, by estimating how many people are served by each device.
What proportion of the population is under the age of five? This is important to our calculations because we think young children are disproportionately susceptible to death from diarrhea.
What is the baseline level of water treatment in the absence of this program?
Where relevant, we also consider implementation challenges caused by security concerns or other contextual factors.
Why do cost-effective funding gaps sometimes go unfilled?
People are often surprised that some high-impact funding gaps, like the ones GiveWell aims to fund, aren't already filled. Of course, many high-impact opportunities are already supported by other funders, like Gavi or the Global Fund, to name just a couple examples. When we see remaining gaps, we think about how our grant might affect other funders' decisions, and whether another funder would step in to fill a particular gap if we didn't.[4]
The...
This post is the second in a multi-part series, covering how GiveWell works and what we fund. The first post, on cost-effectiveness, is here. Through these posts, we hope to give a better understanding of our research and decision-making.
Author: Isabel Arjmand
Looking forward, not just backward
When we consider recommending funding, we don't just want to know whether a program has generally been cost-effective in the past - we want to know how additional funding would be used.
People sometimes think of GiveWell as recommending entire programs or organizations. This was more accurate in GiveWell's early days, but now we tend to narrow in on specific opportunities. Rather than asking whether it is cost-effective to deliver long-lasting insecticide-treated nets in general, we ask more specific questions, such as whether it is cost-effective to fund net distributions in 2023 in Benue, Plateau, and Zamfara states, Nigeria, given the local burden of malaria and the costs of delivering nets in those states.
Geographic factors affecting cost-effectiveness
The same program can vary widely in cost-effectiveness across locations. The burden of a disease in a particular place is often a key factor in determining overall cost-effectiveness. All else equal, it's much more impactful to deliver vitamin A supplements in areas with high rates of vitamin A deficiency than in areas where almost everyone consumes sufficient vitamin A as part of their diet.
As another example, we estimate it costs roughly the same amount for the Against Malaria Foundation to deliver an insecticide-treated net in Chad as it does in Guinea (about $4 in both locations). But, we estimate that malaria-attributable deaths of young children in the absence of nets would be roughly 5 times higher in Guinea than in Chad (roughly 8.8 deaths per 1,000 per year versus roughly 1.7 per 1,000), which leads AMF's program to be much more cost-effective in Guinea.
This map from Our World in Data gives a sense of how deaths from malaria vary worldwide.[3]
Because cost-effectiveness varies with geography, we ask questions specific to the countries or regions where a program would take place. When we were investigating an opportunity to fund water chlorination in Malawi, for example, we wanted to know:
How does baseline mortality from poor water quality in Malawi compare with that in the regions where the key studies on water chlorination took place?
What is the overall morbidity burden from diarrhea in Malawi?
Might people be more or less likely to use chlorinated water in this area than in the areas where the key studies took place?
What does it cost to serve one person with in-line chlorination for one year? We calculate this, in part, by estimating how many people are served by each device.
What proportion of the population is under the age of five? This is important to our calculations because we think young children are disproportionately susceptible to death from diarrhea.
What is the baseline level of water treatment in the absence of this program?
Where relevant, we also consider implementation challenges caused by security concerns or other contextual factors.
Why do cost-effective funding gaps sometimes go unfilled?
People are often surprised that some high-impact funding gaps, like the ones GiveWell aims to fund, aren't already filled. Of course, many high-impact opportunities are already supported by other funders, like Gavi or the Global Fund, to name just a couple examples. When we see remaining gaps, we think about how our grant might affect other funders' decisions, and whether another funder would step in to fill a particular gap if we didn't.[4]
The...