Would it be a good idea to create a ‘GiveWell’ for U.S. charities?
In 2007, GiveWell was founded with the goal of “giving every donor convenient, usable access to the kind of intelligence and due diligence that is currently exclusive to large grantmakers.” Under its original plan, GiveWell would evaluate charities within seven distinct cause areas and recommend to donors the best charities within each area. Three of those cause areas were related to global health and development while four of them were related to U.S. educational/economic opportunity. GiveWell largely followed this plan for four years, during which it recommended a variety of U.S. charities. In 2011, GiveWell decided to stop recommending charities working on U.S. educational/economic opportunity and focus exclusively on global health and development charities. Its reasoning was that U.S. educational/economic opportunity was much less cost-effective than global health and development and that its recommendations were moving relatively little money to charities working on U.S. educational/economic opportunity.
In this post, I will consider whether it would be a good idea to start a GiveWell-style evaluator focused exclusively on U.S. charities.* The evaluator could find cause areas with high impact interventions that are appealing to donors and then find the best charities within each cause area.** Possible cause areas include:
1. Climate change mitigation. The goal would be to find the charity that is most effective at reducing or offsetting greenhouse gas emissions. (Potential recommendation(s): Cool Earth)
2. Criminal justice. The goal would be to find the charity that is most effective at minimizing the burden of the criminal justice system on people accused of minor crimes. (Potential recommendation(s): various bail funds)
3. Immigration relief. The goal would be to find the charity that is most effective at securing immigration relief for people in removal proceedings. (Immigrants who cannot afford an attorney have no right to an appointed attorney in removal proceedings. Those who are provided an attorney are much more likely to obtain relief, and providing immigration attorneys is relatively inexpensive on a per-case basis.) (Potential recommendation(s): unknown)
4. Educational opportunity. The goal would be to find the charity that is best at helping disadvantaged youths. (Potential recommendation(s): Nurse Family Partnership)
(I have purposefully left out public health and mental health because I think they would be less appealing to donors.)
Would It Be Successful?
In my view, for the evaluator to be successful, it would have to have an impact that is larger than the impact its employees would have if they each pursued a high earning career and donated a large fraction of their salary to GiveWell top charities. Since GiveWell top charities are about 100 times as effective (under certain assumptions) as Nurse Family Partnership and various bail funds, the evaluator would have to move 100 times as much money to its recommended charities as its employees would be donating to GiveWell top charities if they instead took high paying jobs.*** For example, if the evaluator eventually grows to the point where it has 20 employees (the current number of GiveWell employees) and each of those employees would otherwise be donating $10,000 to GiveWell top charities, then the evaluator would have to move $20 million to its recommended charities (about half of what GiveWell moved to its recommended charities in 2016 if you exclude Good Ventures).
If GiveWell was only able to move a relatively small amount of money to U.S. charities, what reason is there to think that a new evaluator would be able to move such a large amount of money?
1. GiveWell focused on several closely-related cause areas, while this evaluator would focus on several quite distinct cause areas. This would not only broaden the direct appeal of the evaluator, but also allow awareness of it to spread more easily (since someone who cares about one cause could learn about the evaluator through someone who cares about a different cause).
2. GiveWell promoted its recommended U.S. charities alongside more effective global health and development charities, and it took a more utility-oriented approach to marketing its recommended charities (under which giving to global health and development charities would make more sense). This evaluator could take a more justice-oriented approach, focusing mainly on the problem underlying each cause area and treating the high cost-effectiveness of the top charities as more of a bonus.
3. Due to the current political environment, there is a greater desire to support various U.S. causes (including among some EAs), which expands the pool of potential donors. Since the first few years of an evaluator are the most critical for its long-term success, now would probably be an ideal time to start an evaluator like this.
[There is also the possibility that the evaluator could influence institutional donors, such as foundations, corporations, and governments. Influencing foundations seems relatively unlikely since foundations tend to focus on supporting riskier endeavors. However, influencing corporations seems possible since they are often geographically limited and at least one has reached out to an EA group. Additionally, it seems possible that high quality research could influence governments (e.g. municipalities deciding how best to offset greenhouse gas emissions).]
Benefits (if successful)
1. Its recommendations could move large amounts of money to existing effective charities.
2. It could move money to new effective charities by incubating and then recommending them.
3. It could make it mentally easier for people to accept EA by allowing them to first accept prioritization within a cause area they like and later accept prioritization between cause areas.
4. It could help GiveWell make the case for overseas giving by providing a basis for comparison.
5. It could result in charities and donors placing greater value on evidence and transparency.
Costs (if successful)
1. There would be a large opportunity cost to the money and time invested in the evaluator.
2. If U.S. charities end up being somewhat close in cost-effectiveness to global health charities, some donors may decide to shift their donations from global health charities to U.S. charities.
3. It could shift money away from higher risk U.S. charities that have a higher expected impact.
4. It could reinforce the perception that EA is misleading since it gets people involved through a relatable cause and then tries to convince them that other causes are even more important.
5. It could undermine cause neutrality by promoting the idea that people should just choose their favorite cause area and then support the most cost effective charity working in that area.
6. If its recommendations rely on non-rigorous research or non-rigorous assumptions, then it could harm GiveWell’s reputation by undermining confidence in GiveWell-style evaluation.
Ways It Could Fail
The evaluator could fail if...
1. none of the most appealing interventions are sufficiently cost-effective and evidence-based.
2. all of the charities implementing the most promising interventions decline to be evaluated.
3. it cannot find donors who value cost-effectiveness but only want to give to U.S. charities.
4. it lacks a core community of dedicated supporters that will promote it and give money to it.
5. it is unable to find employees who are as competent and motivated as GiveWell employees.
Costs of Failure
1. There would be a large opportunity cost to the money and time invested in the evaluator.
2. It could harm the reputation of EAs by making people think EAs cannot execute their ideas.
3. It could result in people thinking that GiveWell-style evaluation only works in global health.
4. Charities would probably be less likely to cooperate if someone tried again after a failure.
*This is distinct from the question of whether GiveWell should have continued to evaluate U.S. charities since it’s possible for it to be a good idea to evaluate U.S. charities but a bad idea for GiveWell to do it.
**This proposal to identify the most effective charities within highly effective cause areas is narrower than previous proposals to identify the most effective charities within every cause area.
***This is somewhat of a simplification for a number of reasons. 1) If some of the money moved would otherwise have gone to other charities, then the evaluator would have to move more money to offset the reduced impact of those other charities. 2) If the evaluator has benefits other than moving money (see “Benefits (if successful)” above), then the evaluator would have to move less money to have the same overall impact. 3) If some of the employees of the evaluator donate to GiveWell, then the evaluator would only have to move 100 times the additional amount that its employees would have donated to GiveWell had they taken high paying jobs. However, it is quite possible that the employees of the evaluator will mostly donate to the evaluator’s recommended charities since GiveWell employees mostly donate to GiveWell recommended charities.
Disclaimer: I am not a current or former GiveWell employee.
I suspect that this would be a net good. We need to move more people towards cost effectiveness and this involves meetings people where they are. If this succeeded in creating a large group of people who were somewhat closer to the EA perspective, then EA might have a significantly easier time bringing in more members. And I suspect that this might be more than enough to counter-balance any people who change from donating to the most effective globally to the most effective locally.
(EDIT: The commenter below is right that this was addressed and I missed it): You missed one key consideration. It could also redirect money from more efficient overseas charities to less efficient US charities and even if this was a small amount, it may still be net negative if the utility difference was large enough.
I believe this is number 2 under “Costs (if successful).”
You’re right. Edited my comment to note that this was actually addressed.
Nice! I think it could be really valuable to create “GiveWell-style” charity evaluators for other areas. ACE started this off with animal charities, but I think some of the areas you listed could be good fits, as well as others e.g. biorisk/AI charities.
You mention this in the 5th benefit, but a major upside in my mind is incentivizing the space to place greater value in effectiveness and transparency. These effects could be far reaching and hard to quantify. You might see if ACE thinks this happened with animal advocacy because of their work.
There are potentially major benefits even if you fail:
You could gain valuable insights around starting such a venture, and around the focus area. This could be valuable both to you, and to the broader movement if you can distill it into a postmortem.
Starting an ambitious venture and failing can still be valuable career capital if you can show you hit certain milestones or can take away major lessons.
In general, I think you are overweighing the possible effects on EA / GiveWell. As the project grows you can decide how much explicit association you have with EA and GiveWell.
The biggest cost seems in my mind is the opportunity cost. Differences in cause areas can be pretty huge, so if you are working on a suboptimal cause, you might have a much lower impact.
Some people/organizations are doing similar things, but in general I agree with others that opportunity cost is the main downside. At the same time, I could see some of these, especially 2-3, actually offering cost-effective impacts because of low-probability systemic changes.
Impact Matters isn’t doing exactly this, but they do something similar to it: https://www.impactm.org/
Another resources is the Center for Evidence-Based Programs: evidencebasedprograms.org
Charity Navigator already exists. GuideStar too. If people would just use those, it would be better than something like 2⁄3 of Americans not looking into the causes they support. If an American is so set on rich country charities, just mentioning or encouraging those tools would be enough.
An American charity evaluator would have to compete with the charity ranking sites too. I don’t think it would get off the ground very easily.
I also don’t think it would be associated with Effective Altruism. Doesn’t make sense if it’s just based on location.
I support the spirit of this comment: use already existing resources, instead of creating new ones, and don’t make the solution more complicated than it needs to be. That said, neither Charity Navigator nor GuideStar currently make much of an attempt to calculate the cost-effectiveness of the charities in their database. They are both moving in the direction of encouraging charities to self-report impact data, but I’m not aware of any plans to use the kinds of standardized metrics or outcome definitions that would be necessary for a cost-effectiveness calculation. So I actually do think there would be a lot of value in an independent analysis of cost-effectiveness within a US framework, even a back-of-the-envelope one.