Note: This page summarizes the rationale behind a GiveWell grant to GiveDirectly. GiveDirectly staff reviewed this page prior to publication.
In a nutshell
In November 2025, GiveWell recommended a grant of approximately $4.6 million to GiveDirectly to pilot three variations of its standard cash transfer program. The goal of this grant is to learn more about variations of GiveDirectly’s program that could meet or exceed our 6x cost-effectiveness bar for livelihoods interventions, with each pilot testing a different lever in our cost-effectiveness model.
The pilots will take place in three different countries, and each will test a different variation of GiveDirectly's program:
- Malawi ($1.9 million): Tests whether pairing business grants with household transfers can increase the economic multiplier.
- Mozambique ($1.5 million): Tests whether targeting the poorest young adults can increase productive investment.
- Uganda ($1.2 million): Tests whether pairing cash with new infrastructure (footbridges) can remove physical constraints to unlock greater impact.
Why we recommended this grant:
- We believe the primary value of this grant is through learning and optionality (more):
- Learning Value: We think the grant will give us information about which models are viable and give us the option to fund the successful models at scale. Our confidence in learning value varies across each pilot.
- Optionality: We estimate the optional value of this grant at 12x our cost-effectiveness benchmark.
- This grant supports our priority of identifying a scalable, cost-effective version of GiveDirectly's program (more).
Our main reservations:
- We may learn less than hoped. The Mozambique and Uganda pilots rely on non-causal research designs that may not give us enough confidence to direct large-scale future funding (more).
- The pilots face operational risks, including dependency on an external partner's construction timeline for Uganda (more).
- Our best guesses for the standalone cost-effectiveness of these pilots (4x–6x) are at or below our 6x bar for livelihoods interventions, meaning the grant's value depends heavily on the "option to scale" succeeding (more).
- We think there is some chance that these pilots would be funded anyway without this grant (more).
- This was a faster-than-usual grant investigation, and it's possible we may have overlooked factors that could affect our decision (more).
Published: May 2026
Background
GiveDirectly is a non-profit focused predominantly on delivering unconditional cash transfers to people living in extreme poverty in low-income countries.1 Its flagship Cash for Poverty Relief program delivers unconditional one-off transfers of ~$1,000 via mobile money to households living in poor regions of low-income countries.2 The program is currently operational in 5 sub-Saharan Africa countries: Kenya, Malawi, Mozambique, Rwanda, and Uganda.3 For our full analysis of GiveDirectly's Cash for Poverty Relief Program, see our intervention report here.
Our best guess is that GiveDirectly's Cash for Poverty Relief Program is approximately 3–4x as cost-effective as our benchmark.4 While this is below our current 6x livelihoods bar, the small gap suggests that program adaptations could plausibly meet our threshold.5 In March 2025, we made a small scoping grant to GiveDirectly to explore variations of its cash transfers program. Based on this scoping, GiveDirectly proposed four pilot concepts. We decided to prioritize three of the four proposed pilots for this grant.6
The grant
Budget for grant activities
This grant provides approximately $4.6 million in total for three pilots in Malawi, Mozambique, and Uganda. Across all three pilots, 70–80% of budgets are allocated to direct recipient transfers.7 The pilot’s budgets and main cost drivers are broken out below:
Malawi:
- Total: $1,898,726
- Recipient transfers: $1,475,000 (78%)
- Indirect costs: $172,611
- Staff: $134,881
Mozambique:
- Total: $1,515,499
- Recipient transfers: $1,095,050 (72%)
- Indirect costs: $137,773
- Staff: $175,543
Uganda:
- Total: $1,197,014
- Recipient transfers: $783,981 (65%)
- Indirect costs: $108,819
- Staff: $242,288
The three pilots
Malawi: Business grants with household transfers
This pilot tests whether combining standard large-scale household cash transfers with targeted grants ($500 or $1,000) to local businesses can amplify local economic growth.8
How it works: The pilot is embedded within an existing randomized control trial in Chiradzulu that GiveWell is co-funding. Approximately 1,850 enterprises will receive the targeted grants, timed to be delivered before household transfers.9 The theory of change is that business grants will help local enterprises expand supply to meet new demand from household transfers, potentially preventing price inflation and increasing the overall economic multiplier.10
Learning approach: We think that this pilot provides the highest learning value because it is embedded in a well-designed RCT. The study will cross-randomize 6,400 enterprises across three treatment arms and use high-frequency panel surveys and bi-monthly censuses of 9,000 firms to measure impacts on profits, firm entry/exit, and multiplier effects.11
Mozambique: Targeting the poorest young adults
This pilot tests a "youth-first" targeting model in Lalaua district, an area with estimated average consumption of approximately $1.04/day.12
How it works: Approximately 2,000 recipients will receive $550 transfers in two tranches (a smaller token payment, followed by a large lump sum). All adults aged 18 to 35 in targeted villages – plus at least one head of household over 35 – will receive a transfer, ensuring every household receives a transfer.13 The theory of change is that since young adults have the steepest barriers to market entry,14 a program targeting young adults will prove more cost-effective than the base lump sum cash program.15
Learning approach: This pilot relies on pre/post surveys at baseline and at 3 months of at least 700 recipients (a 20% sample).16 Surveys will track transfer allocation between productive investments and basic needs, food security, shifts in productivity (e.g. hours worked, productive activities), and asset ownership. There will not be a control group.17
Uganda: Cash with bridges
This pilot tests pairing cash transfers with infrastructure—specifically footbridges built by Fika (formerly Bridges to Prosperity), an organization that focuses on rural transportation infrastructure18
How it works: Approximately 1,216 households in villages near a new bridge site will receive $644 transfers,19 timed to coincide with bridge completion.20 The theory of change is that combining capital (cash) with access (bridge) will provide greater impact than either intervention alone, since new economic opportunities will open at the same time that recipients have the capital to take advantage of them.21
Learning approach: This pilot uses a mixed-methods, descriptive comparison rather than a causal evaluation. A ~30% sample of recipients will be surveyed at baseline, 3 months, and 6 months.22 The 3- and 6-month surveys will measure market participation, use of public services (clinic and school attendance), income, and the baseline survey will measure market access. Qualitative data will include recipient feedback23
The case for the grant
We are recommending this grant for the following reasons:
Learning and optionality
Learning Value
Through this grant, we expect to learn about the operational feasibility, and potential cost-effectiveness, of each pilot. We think this will help us understand whether any of the adaptations tested in these pilots could increase the cost-effectiveness of GiveDirectly’s core program. Our confidence in learning value varies across each pilot:
- Malawi: Because the pilot is embedded in a large-scale RCT, we expect it will provide us with rigorous, causal data on cost-effectiveness and operational feasibility.
- Mozambique and Uganda: These pilots will serve primarily as tests of operational feasibility and social acceptability, with a secondary goal of gathering descriptive data on cost-effectiveness.
Optionality
We estimate the option value of this grant to be 12x our benchmark.24 The pilots could unlock cost-effective funding opportunities with a large amount of room for more funding. If successful, they could also inform future decisions about whether to fund more rigorous research or direct program implementation.
Strategic alignment
This grant supports the strategic goal of our livelihoods portfolio of increasing the cost-effectiveness of cash transfers. We believe GiveDirectly cash transfers are a particularly important intervention to pursue. GiveDirectly has demonstrated scalability with operations across multiple populations and the ability to rapidly reach new populations. This ability to flexibly scale its program up or down makes it a good candidate for receiving large amounts of GiveWell funding in the future, if we are able to identify a version of the program that meets our cost-effectiveness threshold.25
Additionally, our best guess that their standard program is 3–4x our benchmark.26 This means the gap to our 6x bar is relatively small, increasing the likelihood that program adaptations could plausibly meet our threshold.
Risks and reservations
Our main reservations about this grant are as follows:
Limitations on learning
The Mozambique and Uganda pilots both use non-randomized study designs to evaluate the impact of their programs. These results may not give us enough confidence to direct large-scale funding in the future, even if descriptive signals are positive.
- The targeting pilot in Mozambique relies on pre/post surveys of a recipient sample and explicitly lacks a control group.
- The bridges pilot in Uganda relies on pre/post surveys of a recipient sample and uses a "descriptive comparison" against historical data from Fika on the impact of bridges without cash.
We have accounted for this skepticism in our optionality model with a "probability we update post-pilot" parameter. We estimate 60% for Malawi (given RCT evidence), 40% for Mozambique, and 25% for Uganda.27
Operational risks
The pilots are more complex than standard cash transfers and depend on factors outside GiveDirectly's full control:
- Malawi: The pilot requires enterprise grants to be delivered before household grants; eligibility verification delays could cause both types of grants to be delivered simultaneously, which would make it more difficult to test the pilot's core premise.28
- Uganda: The pilot depends on Fika (formerly Bridgers to Prosperity) completing bridge construction on time. Payments must be "timed precisely", and any slippage from B2P could compromise the pilot's design.29
We believe these risks are manageable and typical for early-stage pilots.
Cost-effectiveness uncertainty
Our best guesses for the standalone cost-effectiveness of each pilot are:30
| Pilot | GiveWell best guess | GiveDirectly best guess |
|---|---|---|
| Malawi (business grants) | 6x | 8x |
| Mozambique (targeting poorest) | 5x | 8x |
| Uganda (cash + bridges) | 4x | 6x |
These estimates are at or below our 6x bar. This means that most of the grant's value is derived from the “option to scale.”
Funging risk
Our cost-effectiveness estimates often include an adjustment for “funging,” to account for how our grants might impact other funders.31 In this case, we think that GiveDirectly is motivated to experiment and might find funding for these pilots even if we don’t fund them. We account for this with a -33% funging adjustment in our optionality model.32
Quick investigation
This was a faster-than-usual GiveWell grant investigation. Because of this, it's possible we may have overlooked details that could have led us to make a different funding decision, or our learning agendas may not be as well developed as they could be.
Plans for follow up
Our follow-up will be tied to key operational milestones, and will consist of grantee check-in calls and confirmation of operational progress to allow for future funding or research decisions as pilot data becomes available.
Internal forecasts
For this grant, we are recording the following forecasts:
| Confidence | Prediction | By time |
|---|---|---|
| 57% | We direct additional funding toward the supply-side (Malawi) cash program within 6 months of the pilot ending. | June 2029 |
| 36% | We direct additional funding toward the targeting (Mozambique) program within 6 months of the pilot ending. | July 2027 |
| 20% | We direct additional funding toward the bridges (Uganda) program within 6 months of the pilot ending. | July 2027 |
| 75% | The Malawi pilot successfully delivers >90% of enterprise grants at least 2 weeks before the corresponding household transfers, by June 30, 2026 | June 2026 |
Our process
After our GiveDirectly update,33 we gave GiveDirectly a small scoping grant to figure out if there were variations of their “base” unconditional cash transfer program that could increase cost-effectiveness. The final deliverable of that scoping grant was a proposal for four pilots. We decided to prioritize three of the four proposed pilots.
- This was a very quick investigation which involved review of GiveDirectly materials, asking follow up questions, and developing a cost-effectiveness, optionality, and value of information model which received internal review.
Sources
- 1“GiveDirectly is a nonprofit that lets donors like you send money directly to the world's poorest households.” GiveDirectly, About GiveDirectly.
- 2“Send large, one-time cash transfers directly to families in extreme poverty… Every household in a village receives ~$1,000 with no strings attached… We primarily use government data to target villages where most or all residents are living in extreme poverty, defined by the World Bank as earning under $3/day… [T]he funds are sent in two parts through a SMS-enabled banking technology called mobile money.." GiveDirectly, Cash for poverty relief.
- 3"As of February 2024, our poverty relief program is currently running in Kenya, Malawi, Mozambique, Rwanda, and Uganda. Since 2009, we’ve reached over 440,000 households in rural villages across Africa through this program." GiveDirectly, Cash for poverty relief.
- 4“As of October 2024, our best guess is that donations to GiveDirectly’s Cash for Poverty Relief program are ~3-4x more cost-effective than we had previously estimated, and around ~30-40% as cost-effective as our marginal funding opportunity.” GiveWell, GiveDirectly’s cash for poverty relief program.
- 5Our current bar for livelihoods programs is 6x our benchmark. Additionally, we use slightly different moral weights; for these analyses we assume that saving a life is about 50 times more valuable than doubling a person’s income for a year, compared to our standard 100x. Read more about our livelihoods work in our blog post: GiveWell, GiveWell expands work on livelihoods program, 2025.
- 6GiveDirectly proposal, 2025 (unpublished)
- 7GiveDirectly budget, 2025 (unpublished)
- 8GiveDirectly proposal, 2025 (unpublished)
- 9GiveDirectly and GiveWell internal discussions, 2025 (unpublished)
- 10GiveDirectly proposal, 2025 (unpublished)
- 11GiveDirectly proposal, 2025 (unpublished)
- 12 GiveDirectly and GiveWell internal discussions, 2025 (unpublished)
- 13 GiveDirectly proposal, 2025 (unpublished)
- 14 GiveDirectly proposal, 2025 (unpublished)
- 15 GiveDirectly proposal, 2025 (unpublished)
- 16 GiveDirectly and GiveWell internal discussions, 2025 (unpublished)
- 17 GiveDirectly and GiveWell internal discussions, 2025 (unpublished)
- 18“We enable the construction, design, and deployment of Rural Networked Transport Infrastructure (RNTI) at scale. RNTI encompasses rural roads, bridges, footbridges, and associated drainage, trails, paths, and waterways that link rural communities to one another and to higher-order roads and service centers.” Fika, "Who We Are"
- 19 GiveDirectly proposal, 2025 (unpublished)
- 20 GiveDirectly proposal, 2025 (unpublished)
- 21 GiveDirectly proposal, 2025 (unpublished)
- 22 GiveDirectly proposal, 2025 (unpublished)
- 23 GiveDirectly proposal, 2025 (unpublished)
- 24See our optionality cost-effectiveness analysis of GiveDirectly’s pilots, “Summary” tab.
- 25 See “The Case for the grant”, GiveWell, “GiveDirectly — Scoping Grant for New Program Variations (March 2025)”.
- 26“As of October 2024, our best guess is that donations to GiveDirectly’s Cash for Poverty Relief program are ~3-4x more cost-effective than we had previously estimated, and around ~30-40% as cost-effective as our marginal funding opportunity.” GiveDirectly’s Cash for Poverty Relief Program.
- 27See our optionality cost-effectiveness analysis of GiveDirectly’s pilots, row “Probability that we direct funds toward program post-pilot (assuming no failure)”.
- 28 GiveDirectly proposal, 2025 (unpublished)
- 29 GiveDirectly proposal, 2025 (unpublished)
- 30See the GiveDirectly >6x BOTEC, “Summary” tab.
- 31Read more about funging here: GiveWell, How we produce impact estimates - What is funging, and how does it affect impact?
- 32See our optionality cost-effectiveness analysis of GiveDirectly’s pilots, row “Funging adjustment: risk that pilot/study is funded without us.”
- 33“As of October 2024, our best guess is that donations to GiveDirectly’s Cash for Poverty Relief program are ~3-4x more cost-effective than we had previously estimated, and around ~30-40% as cost-effective as our marginal funding opportunity.” GiveWell, GiveDirectly’s Cash for Poverty Relief Program