Here’s my attempt at a characterization of the distinction (people should feel free to correct me if they think I’m wrong):
The GHDF made several grants to IDinsight and IPA for RCT research, which could help with the development of new proven interventions (for Covid prevention in particular). It also made grants to Instiglio for technical design of results-based financing, to J-PAL’s Innovation in Government Initiative to promote evidence-based policies in developing countries, and Fortify Health, a new potential GiveWell top charity.
These grants are all related to proven, evidence-based interventions, and help scale or promote that approach. The only exception is the grant to One For The World, which is more like an EA meta intervention. EDIT: There’s also the Centre for Pesticide Suicide Prevention grant, some info here.
In contrast, one could try to do things that are even more hits-based. An example of a past EA success in this area is helping reallocate £2.5 billion in DFID resources towards research funding for neglected tropical diseases (as I understand it, there were some specific reasons to believe that EAs actually had a large impact on that budget change that weren’t discussed publicly). Of course, some of that research was in the form of RCTs, but I guess a lot of it was more fundamental. The key motivating factor is more like “NTD research still is a very important, neglected, and tractable area” rather than “we want to scale proven interventions”. The indirectness of the policy advocacy route makes it more hits-based as well; there are no RCTs on whether that kind of policy advocacy works.
Or take the idea of developing-world public health regulation, e.g. tobacco taxation. This is not an RCT-backed intervention in a narrow sense, but nonetheless estimated to be extremely cost-effective based on some back-of-the-envelope calculations.
Another example might be ballot initiatives, though these are less scalable.
It’s not clear that any of this requires an additional fund. Perhaps the GHDF can simply do more of both. (edited)
Thanks for raising these questions! I work at GiveWell, and we’re planning to update the EA Global Health and Development Fund page to make the distinction between it and the Maximum Impact Fund clearer—we think we can do better to explain the difference.
Here’s a quick summary:
The Maximum Impact Fund is granted regularly to the highest-value funding opportunities we see among GiveWell’s recommended charities. This is a great option for donors who want to support GiveWell’s top charities and are open to their funding being used wherever it can do the most good among them.
You can read more about the Maximum Impact Fund, and see our past allocations, here.
The Effective Altruism Global Health and Development Fund has a broader remit, including programs that are new and about which little is known, as well as policy advocacy, public health regulation, technical assistance, and direct delivery programs, including GiveWell’s recommended charities.
There’s no minimum requirement for evidence for grantees’ work. The Fund Manager (Elie) takes an expected value approach to calculating the potential impact of grants from the Global Health and Development Fund. He’ll recommend higher-risk grants when they are more effective (in expectation) than GiveWell top charities. He’ll fund GiveWell top charities if no such higher-expected-value grant opportunity exists.
This is a great option for donors who are open to taking calculated risks when the expected value appears higher than GiveWell’s top charities, but are happy to support GiveWell’s top charities if not.
We don’t expect to change the overall portfolio of opportunities that we pursue in response to the allocation of donations between these two funds. In other words, as with most giving, there is fungibility. For example, if there are insufficient funds in the Global Health and Development Fund to support a high-leverage opportunity we want to fund, we would expect to seek funding for that opportunity from Open Philanthropy, with whom we work closely, or another donor.
Here’s my attempt at a characterization of the distinction (people should feel free to correct me if they think I’m wrong):
The GHDF made several grants to IDinsight and IPA for RCT research, which could help with the development of new proven interventions (for Covid prevention in particular). It also made grants to Instiglio for technical design of results-based financing, to J-PAL’s Innovation in Government Initiative to promote evidence-based policies in developing countries, and Fortify Health, a new potential GiveWell top charity.
These grants are all related to proven, evidence-based interventions, and help scale or promote that approach.
The onlyexception is the grant to One For The World, which is more like an EA meta intervention. EDIT: There’s also the Centre for Pesticide Suicide Prevention grant, some info here.In contrast, one could try to do things that are even more hits-based. An example of a past EA success in this area is helping reallocate £2.5 billion in DFID resources towards research funding for neglected tropical diseases (as I understand it, there were some specific reasons to believe that EAs actually had a large impact on that budget change that weren’t discussed publicly). Of course, some of that research was in the form of RCTs, but I guess a lot of it was more fundamental. The key motivating factor is more like “NTD research still is a very important, neglected, and tractable area” rather than “we want to scale proven interventions”. The indirectness of the policy advocacy route makes it more hits-based as well; there are no RCTs on whether that kind of policy advocacy works.
Or take the idea of developing-world public health regulation, e.g. tobacco taxation. This is not an RCT-backed intervention in a narrow sense, but nonetheless estimated to be extremely cost-effective based on some back-of-the-envelope calculations.
Another example might be ballot initiatives, though these are less scalable.
It’s not clear that any of this requires an additional fund. Perhaps the GHDF can simply do more of both. (edited)
Thanks for raising these questions! I work at GiveWell, and we’re planning to update the EA Global Health and Development Fund page to make the distinction between it and the Maximum Impact Fund clearer—we think we can do better to explain the difference.
Here’s a quick summary:
The Maximum Impact Fund is granted regularly to the highest-value funding opportunities we see among GiveWell’s recommended charities. This is a great option for donors who want to support GiveWell’s top charities and are open to their funding being used wherever it can do the most good among them.
You can read more about the Maximum Impact Fund, and see our past allocations, here.
The Effective Altruism Global Health and Development Fund has a broader remit, including programs that are new and about which little is known, as well as policy advocacy, public health regulation, technical assistance, and direct delivery programs, including GiveWell’s recommended charities.
There’s no minimum requirement for evidence for grantees’ work. The Fund Manager (Elie) takes an expected value approach to calculating the potential impact of grants from the Global Health and Development Fund. He’ll recommend higher-risk grants when they are more effective (in expectation) than GiveWell top charities. He’ll fund GiveWell top charities if no such higher-expected-value grant opportunity exists.
The list of past grants shows the breadth of programs that have received Global Health and Development Fund support. Here’s a grant for an RCT on the effects of mask-wearing on COVID-19, a grant for a policy advocacy organization, and a grant to GiveWell top charities.
This is a great option for donors who are open to taking calculated risks when the expected value appears higher than GiveWell’s top charities, but are happy to support GiveWell’s top charities if not.
We don’t expect to change the overall portfolio of opportunities that we pursue in response to the allocation of donations between these two funds. In other words, as with most giving, there is fungibility. For example, if there are insufficient funds in the Global Health and Development Fund to support a high-leverage opportunity we want to fund, we would expect to seek funding for that opportunity from Open Philanthropy, with whom we work closely, or another donor.
I hope that helps clarify!