You are here

Grants and Funding Recommendations in January and February 2021

Published: June 2021

Please note that this page describes our understanding of our top charities' funding needs and plans as of January and February 2021, which is when we made the recommendations and allocations discussed below.


This page explains the opportunities that GiveWell directed funding to in January and February 2021. These include (a) opportunities that we funded using donors' contributions to the Maximum Impact Fund and Effective Altruism Global Health and Development Fund, and (b) recommendations that we made to large donors, including Open Philanthropy, about where to direct their donations to maximize impact.

In January and February 2021, we granted or recommended grants totaling $80.2 million from:

  • Open Philanthropy, a philanthropic organization with which we work closely, provides substantial support to our recommended charities. In November 2020, Open Philanthropy committed to granting $100 million to GiveWell's recommended charities during the 2020 giving season. At our recommendation, Open Philanthropy granted $70 million of this funding in November, and we committed to recommending an allocation of the remaining $30 million by January 2021, when we expected to have more information about some promising funding opportunities. We have now recommended that Open Philanthropy grant this funding to the following opportunities:
  • At our recommendation, an anonymous donor gave $20 million to the Against Malaria Foundation (AMF) to support its work in the Democratic Republic of the Congo (DRC) in 2022-23.
  • In Q4 2020, donors gave $25.9 million to the Maximum Impact Fund. Effective Altruism Funds, Effective Altruism Australia, Effektiv Spenden, and RC Forward also offer donors the option to donate to versions of a "maximum impact fund." They have periodically asked GiveWell to recommend which organization to allocate these funds to. In early 2021, they asked GiveWell for recommendations for how to allocate the funds donors had given since they last granted out these funds. These funds totaled $0.7 million from Effective Altruism Funds, $0.3 million from Effective Altruism Australia, $0.4 million from Effektiv Spenden, and $0.1 million from RC Forward. (Note that the $0.7 million from Effective Altruism Funds is separate from the Effective Altruism Fund for Global Health and Development; see the next bullet point.) This funding went to AMF to support its work in Nigeria in 2021-22 ($21.4 million) and in DRC in 2022-23 ($5.9 million).
  • In February 2021, we recommended a grant of $3 million from the Effective Altruism Global Health and Development Fund to be split evenly between Malaria Consortium's SMC program and AMF.

Principles we followed

The principles we followed in arriving at this recommended allocation are the same as those we followed and described in detail in 2019:

  • Principle 1: Put significant weight on our cost-effectiveness estimates.
  • Principle 2: Consider additional information that we have not explicitly modeled about an organization.
  • Principle 3: Consider additional information that we have not explicitly modeled about a funding gap.
  • Principle 4: Assess charities’ funding gaps at the margin, i.e., how they would spend additional funding, where possible.
  • Principle 5: Default to not imposing restrictions on charities' spending.
  • Principle 6: Default to funding on a three-year horizon, modifying to preserve our options for the future where doing so is low-cost.
  • Principle 7: Ensure charities are incentivized to engage with our process.

A note on how we discuss cost-effectiveness on this page

We often use GiveDirectly's unconditional cash transfers as a benchmark for comparing the cost-effectiveness of different programs. When discussing cost-effectiveness, we generally refer to the cost-effectiveness of a program in multiples of "cash." Thus, if a program is estimated to be "10x cash," this means it is estimated to be ten times as cost-effective as unconditional cash transfers.

Maximizing cost-effectiveness over time

GiveWell allocates funding on a quarterly basis. For each allocation decision, we try to maximize the cost-effectiveness of GiveWell-directed giving over time. We do this by:

  • Funding opportunities above the cost-effectiveness threshold we expect to apply over the next couple of years. As of these grant decisions in early 2021, this threshold was roughly 10x cash, but it is liable to change over time depending on the size and cost-effectiveness of new opportunities we identify and how much total funding we are able to direct to our recommendations.
  • Not missing out on time-sensitive opportunities, i.e., opportunities that will not be available in the future or for which a delay in funding would cause delays in program implementation.
  • Among opportunities that are not yet time-sensitive, favoring opportunities that are more certain, i.e., that we're less likely to change our mind about before the next time we make a funding decision.

Excluding the three GiveWell Incubation Grants mentioned above (which we discuss separately1), in January and February 2021, we allocated $72.2 million to the following opportunities:

  • $21.4 million to AMF to fully fund a time-sensitive opportunity in Nigeria in 2021. The intention of this grant is to reduce the interval between long-lasting insecticide-treated net (LLIN) distributions in Akwa Ibom and Bauchi States from 4 years to 3 years (we roughly estimate that LLINs last 2-3 years). This is to supplement funding from the President's Malaria Initiative (PMI), which funds LLIN distributions in these states.2 We estimate the cost-effectiveness of this grant as 15x cash.3
  • $27.4 million to AMF to fill the remainder of its funding gap for DRC in 2022 and partially fund its work in DRC in 2023. We estimate the cost-effectiveness of this grant as 12x cash overall:4 13x for the first $20 million and 10x for the next $7.4 million.5
  • $23.4 million to Malaria Consortium's SMC program to partially fund its work in 2023. We estimate the overall cost-effectiveness of this grant as 13x cash; funding gaps for specific countries that we expect to be funded with this grant range from 9-15x cash.6

We prioritized fully filling the gap for AMF in Nigeria 2021 because it was time-sensitive and above our bar for cost-effectiveness.

After filling that gap, we roughly split the remaining funding between Malaria Consortium's SMC program for 2023 and AMF's work in DRC in 2022-23. Both opportunities are likely above our cost-effectiveness bar, and other considerations pull about equally in both directions:

  • AMF's gap in DRC is plausibly time-sensitive, but because we are modeling its cost-effectiveness in a new way (more detail below), we are less confident in our cost-effectiveness estimate, and it could increase or decrease as we learn more.
  • Malaria Consortium's gap for 2023 is not yet time-sensitive (though it will become time-sensitive by the end of 2021), but we are more confident that it meets our cost-effectiveness bar.

Against Malaria Foundation

Modeling impact in DRC and Nigeria

Evaluating grants using the counterfactual

In evaluating grants that we make or recommend, we routinely ask: what would happen to this program if it did not receive this grant? Would another funder step in to support it, or would it go unfunded? Would it reach fewer people? Through this line of questioning, we develop a best guess of what would happen in a scenario without GiveWell-directed funding, which we refer to as "the counterfactual."

Understanding the counterfactual to a grant is necessary for estimating the expected impact of that grant. If, for example, we expect a program to reach 1 million children total in the counterfactual, and we expect the same program to reach 1.5 million children total if it receives a GiveWell-directed grant, then the expected impact of that grant is an additional 500,000 children reached. We build our best guesses of counterfactual program outcomes into our cost-effectiveness models.

However, using the counterfactual to evaluate a grant is rarely as straightforward as the simplified example described above. Our best guesses about what would happen without GiveWell-directed funding are necessarily uncertain, but they can substantially affect our estimate of a grant's cost-effectiveness.

Two counterfactuals to AMF's support

Our understanding is that most countries aim to conduct mass distributions of LLINs every three years. This is based on evidence that LLIN efficacy declines three years after distribution, suggesting that they should be replaced after that interval. Our review of the evidence has similarly led us to conclude that LLINs provide strong protection from malaria for 2-3 years on average.7

To support these periodic distributions, our understanding is that countries primarily receive funding from the Global Fund to Fight AIDS, Tuberculosis and Malaria ("the Global Fund") and from the President's Malaria Initiative (PMI). The Global Fund provides funding to countries on a three-year cycle; the current cycle is running from 2021-23.

AMF has historically supported countries that have scheduled distributions every three years but do not have enough funding from other funders to reach everyone in the country with an LLIN. By purchasing additional LLINs to be allocated during these distributions, we believe AMF causes more people to receive an LLIN than otherwise would have. To evaluate past grants to AMF, we have modeled the cost-effectiveness of increasing the number of LLIN recipients compared to the counterfactual.

In DRC and Nigeria, however, the situations are different:

  • In DRC, distributions occur at the province level. The Global Fund and PMI do not have sufficient funding to support an LLIN distribution in all of the country's provinces during the current 2021-23 cycle. In previous funding cycles, LLIN distributions have been similarly underfunded and have been substantially delayed. Based on this track record, we believe that without additional funding, distributions are likely to be delayed to an average of 39 months after the most recent distribution in each province. In contrast, AMF believes that distributions in DRC should occur every 30 months—even more frequently than every three years—based on evidence that LLIN efficacy declines more rapidly in DRC. This approach is also supported by the Global Fund and by DRC's Programme National de Lutte Contre le Paludisme (PNLP), or national malaria control program.8 By advocating for this scheduling shift and by contributing LLIN funding, we believe AMF will enable a shift to this policy during the 2021-23 cycle. At present, 17 of DRC's provinces are scheduled to conduct their next distributions 30 months after their most recent distributions.9
  • In Nigeria, distributions occur at the state level. The Global Fund and PMI each support a different set of states. For two of the eleven states supported by PMI — Bauchi and Akwa Ibom — PMI will not have funding available to support LLIN distributions until 2022, around four years after the most recent distributions in those states. By contributing LLIN funding to these two states, we believe AMF will enable distributions to occur in 2021, three years after their most recent distributions.

In both cases, we believe AMF's funding will cause people to receive an LLIN earlier than they otherwise would have. This should reduce the number of people using nets that are too worn out to adequately protect them from malaria.

Modelling the counterfactual in DRC and Nigeria

To evaluate these grants to AMF, we updated our existing cost-effectiveness model. Now, instead of modeling the cost-effectiveness of increasing LLIN recipients, we model the cost-effectiveness of reducing the interval between LLIN distributions compared to the counterfactual.10

To do this, we developed a new parameter that adjusts for the change in counterfactual. This adjustment is based on our best guesses of what would happen with or without AMF's support, about which we are uncertain. Some of our open questions include:

  • How would the government or other funders allocate LLINs in the counterfactual? Would they cap the number of LLINs each household receives, in order to preserve scarce LLINs, or would they delay distributions?
  • How much efficacy would LLINs have lost between the time when LLINs will be distributed with AMF's support and the time when they would have been distributed in the counterfactual?
  • What will households do with LLINs that they already own when they receive new ones?

Compared to a counterfactual of increasing LLIN recipients, this new adjustment reduces our estimate of the cost-effectiveness of AMF's support to countries where this is relevant. This is because for each person reached, instead of replacing three years' worth of no LLIN usage with three years' worth of LLIN usage, AMF is replacing roughly one year's worth of usage of an LLIN that provides partial protection with one year's worth of usage of a new LLIN. Even after this reduction, however, we believe that support to DRC and Nigeria, which have high malaria prevalence but are significantly underfunded, is highly cost-effective. We plan to do more work to refine this adjustment, including getting feedback from AMF, in order to evaluate future grants to AMF that will reduce the interval between LLIN distributions.

Case for these grants

  • The opportunities in Nigeria and DRC appear to meet our bar for cost-effectiveness. We estimate that AMF's work in Nigeria in 2021 is 15x cash.11 In DRC, we estimate that the next $20 million AMF spends will be 13x cash12 and that the last $20 million of the funding gap in DRC in the 2021-23 funding cycle will be 10x cash.13 We modeled these gaps separately because we believe that as AMF directs more funding to DRC, the likelihood that it is displacing funding that the Global Fund would have otherwise allocated there increases. More details on our estimates of cost-effectiveness in DRC are in this footnote.14
  • As noted above, the opportunity in Nigeria is time-sensitive: nets are scheduled to be delivered in 2021 (though delays in LLIN campaigns are common, so delivery in early 2022 may be more likely.)
  • The opportunity in DRC may be time-sensitive. The Global Fund, which provides funding on 3-year cycles, supports a large proportion of the nets distributed in DRC. As part of the 2021-23 funding cycle, DRC received some funding for nets, but not enough to cover all areas of need over the given time period. AMF has told us that if it delays committing funding for DRC in 2022 and 2023, the government may decide to triage the planned distributions in 2021 and 2022 in order to save some funding for 2023. This would remove some highly cost-effective giving opportunities by, for example, capping the number of nets each household can receive.15
  • In November, we wrote that we wanted to review recent monitoring results from DRC before directing more funding there. We have since reviewed the first round of monitoring for an AMF-funded LLIN campaign in DRC,16 and spoke with AMF and the partner organization that implemented the survey to understand details about how it was conducted.17 We concluded that the survey was of similar quality to AMF's surveys from other countries and showed evidence that a high proportion of LLINs reached intended recipients.18

Risks and reservations

  • As mentioned above, we are modeling AMF's counterfactual impact in DRC and Nigeria differently from how we've modeled its impact in other countries. While we feel confident enough in our model to believe that AMF's work in DRC and Nigeria is above our bar for cost-effectiveness, our work on modeling the impact of reducing the interval between net distributions is in a preliminary stage, and our estimate of the cost-effectiveness of work in these countries may increase or decrease as we learn more.
  • We project that AMF will raise about $25 million over the next year.19 We are not counting that funding as available to fill these funding gaps because we believe, due to the time-sensitivities described above, that it is better for AMF to commit the funding for these gaps now.20 This means that one effect of the funding we have allocated to AMF in this funding round is to free up AMF's future revenue for support to other distributions. While we don't know exactly how AMF will prioritize using the funds it raises over the next year, the opportunities we expect it to consider seem generally promising to us. These include:
    • Filling the remaining gap in DRC 2023
    • Funding distributions in Nigeria in 2023 to reduce the interval between distributions
    • Renewing funding in countries where AMF has worked in the past:
      • Togo (9x cash21) and Uganda (17x cash22): AMF last funded distributions in Togo and Uganda in 2020. Our understanding is that distributions in these countries occur every three years, so there may be a need for AMF to support these countries again in 2023.
      • DRC in 2024: We guess that AMF will be looking to continue funding DRC past 2023.
    • Potentially funding work in a country AMF hasn't worked in before. AMF requested that we keep the country confidential at this stage in its discussions with the government's malaria program. This country has malaria rates that differ significantly across the country, and AMF is investigating whether there are any gaps in areas that have malaria rates similar to those in other countries where it works.23

Malaria Consortium's seasonal malaria chemoprevention (SMC) program

Malaria Consortium's SMC program is receiving $23.4 million to support its programs in 2023. Malaria Consortium supports SMC in Nigeria, Burkina Faso, Togo, and Chad. This will cover part of the funding gap for its 2023 work in these countries.24

The following table assumes that Malaria Consortium will allocate funding proportionally according to the budget size of each program, excluding Chad (as we wrote in November, we are discussing with Malaria Consortium the possibility of scaling down its work in Chad).

Country Cost-effectiveness25 2023 funding gaps (millions) Amount allocated January 2021 (millions) Remaining funding gaps (millions)
Burkina Faso 15x cash $8.8 $5.7 $3.1
Chad 6x cash $7.8 $0.0 $7.8
Nigeria 13x cash $25.4 $16.4 $9.0
Togo 9x cash $2.1 $1.3 $0.7
Total $44.1 $23.4 $20.7

Case for the grant

This funding gap is cost-effective, and we have less uncertainty about our cost-effectiveness estimates than we do for AMF's funding gaps discussed above. We wrote in more detail about our reasons for recommending SMC in November.

Risks and reservations

This opportunity is not yet time-sensitive, though it will become time-sensitive by the end of 2021. We could keep our options open by waiting to decide how to allocate this funding, but we strongly expect to direct at least this much funding to Malaria Consortium in 2021, so we see limited benefit to waiting. We also think it's possible that having this funding sooner could have subtle positive effects on Malaria Consortium's planning and interest in expanding to new locations.


Document Source
GiveWell, 2020 cost-effectiveness analysis — version 2 Source
GiveWell, 2020 cost-effectiveness analysis — version 2 [change 18 archive] Source
GiveWell, Counterfactual net spacing adjustment calculations Source
GiveWell, DRC nets durability Source
GiveWell, Room for more funding analysis for AMF, January 2021 Source
GiveWell, Room for more funding analysis for Malaria Consortium's SMC program, 2020 Source
GiveWell, Summary of AMF PDM results and methods, 2021 Source
GiveWell's non-verbatim summary of a conversation with Deogratias Cibinda, December 3, 2020 Source