GiveDirectly

Organisation

GiveDirectly is a non-profit founded in 2009 by development economists from Harvard and MIT. It delivers unconditional cash transfers to people living in extreme poverty: recipients get money with no conditions attached and decide for themselves how to use it. GiveDirectly reports having delivered more than $1 billion to over 2 million people since it was founded, and currently works in Kenya, Malawi, Mozambique, Rwanda, Uganda and the Democratic Republic of the Congo.¹

GiveDirectly runs several programmes, including Cash for Poverty Relief. It absorbs the majority of the organisation's funding (see GiveWell's assessment.²). GiveWell names five countries for this programme: Kenya, Malawi, Mozambique, Rwanda and Uganda.² Alongside it, the organisation runs emergency cash after disasters, basic income programmes, transfers to refugees, and its own programmes in the United States.¹ Effektiv Spenden, which processes tax-deductible donations from Germany to GiveDirectly, has agreed with GiveDirectly that these funds go exclusively to programmes in developing countries; US programmes are excluded as a result.³

The problem: extreme poverty

On the World Bank's most recently updated estimate, about 847 million people were living below the international poverty line in 2024, 10.4% of the world's population.⁴ The line itself was raised in June 2025 from $2.15 to $3.00 per day (2021 prices). For 2022 that moved the headcount from 713 million to 838 million. The driver was new purchasing-power data and, above all, substantially higher national poverty lines in several low-income countries, not a sudden rise in poverty.⁵

Extreme poverty is often bound up with poor health, malnutrition, no access to education and few economic opportunities. For sub-Saharan Africa the World Bank puts the 2022 poverty rate at 45.5%, the highest of any world region.⁵

Since 1990, the number of people living in extreme poverty has fallen from about 2.3 billion to roughly 831 million today, even as its geography has shifted. Countries such as China and Indonesia have pushed their poverty rates below 10%, while most people in extreme poverty now live in Sub-Saharan Africa, in some countries, including the Democratic Republic of the Congo, Mozambique and Malawi, more than half the population. Current projections see the number fall to about 793 million by 2030 before rising again.⁶

Approach

GiveDirectly picks regions with high absolute poverty rates, and within them villages where it expects more than 80% of residents to live below the poverty line. Inside those villages it does not means-test further: paying the whole village has been the default design since 2017, where previously roof material served as a rough proxy for poverty.²

GiveWell's October 2024 report put the transfer at about $1,000 per household, and $550 per adult in Malawi.² GiveDirectly's current programme page now states $650–$950 per adult, roughly what a person lives on in a year, paid in two tranches of about $100 and then about $650 two months later.⁷ Payment runs over mobile money; recipients without a phone can get one, with the cost deducted from the transfer.²

Afterwards every recipient is called to confirm receipt. An organisationally separate audit team follows up on suspected cases, and complaints run through toll-free hotlines.²

Impact

In October 2024 GiveWell revised its cost-effectiveness estimate for the programme upwards by a factor of three to four, mainly on new evidence about effects on neighbouring households and on child mortality. That puts the programme at about 30 to 40% of the impact GiveWell expects from its best alternative funding opportunity.² Its bar for livelihoods programmes is six times its own benchmark, which is why GiveDirectly is not one of its top charities.⁸ Unconditional cash transfers are, however, the benchmark GiveWell measures other programmes against.

Per $1 million of programme spending, GiveWell estimates reach at 740 to 970 households, about 3,300 to 4,100 people.² The evidence in detail:

• Recipient household consumption rises sharply just after the transfer and falls away over the following years. That pattern holds across seven studies of five randomised evaluations.²
• Neighbouring households that receive nothing also consume more, because the money is spent locally. A randomised trial in Kenya measures 13% higher consumption among non-recipients 18 months after payment, with price inflation of only 0.1%.²
• A randomised trial in Kenya finds a 45% reduction in mortality among children under five. GiveWell halves that effect in its own model and does not consider cash transfers a cost-effective way to save lives.²
• No measurable increase in spending on alcohol or tobacco was found in any of the three studies that looked for one.²

How much of the money arrives depends on what it is measured against. GiveDirectly reports that in 2025 more than 87% of donations to this programme reached recipients as cash.⁷ Measured against the total cost of a country programme, GiveWell puts the transfer share at 71 to 85%.²

Transparency

GiveDirectly publishes its finances, its research, and since 2024 an annual report on fraud, abuse and safety incidents. The 2025 edition gives figures and names what went wrong.⁹

• In 2025 the organisation reports delivering $146 million to 439,000 people across 10 countries. Known fraud losses came to 0.27%, against 0.19% the year before and its own target of under 0.5%.⁹
• Two programme designs with less direct recipient contact did considerably worse. In the Democratic Republic of the Congo, remote targeting combined with remote payment produced a leakage rate of 2.98%. In Malawi a shortened enrolment process saw fraudulent enrolments spike to 9.8% before additional verification brought it down to 0.8%.⁹
• About 985 cases were reported of third parties impersonating GiveDirectly, mostly over WhatsApp in Kenya and Uganda, up from more than 400 in 2024.⁹
• 1,772 reports concerned recipient safeguarding, four per 1,000 people paid against six the year before. 74% of them were conflicts within households.⁹
• At a local partner organisation, an investigation established that five staff had sexually harassed, financially exploited or intimidated nine recipients. The funds were repaid, those involved were dismissed and the partnership was ended.⁹
• The organisation misses its own deadlines: 77% of safeguarding cases and 87% of fraud cases closed within 30 days, against a 90% target.⁹

The worst case was also self-disclosed. In 2023 GiveDirectly reported that staff in the Democratic Republic of the Congo had embezzled around $900,000 over six months, depriving more than 1,700 families of their transfers. It halted the programme, published a root-cause analysis, and put the share of that year's delivered money lost to fraud at about 1.1%, the highest in its history.¹⁰

Current projects

In November 2025 GiveWell granted about $4.6 million for three pilots testing whether variations on the core programme can reach its funding bar for livelihoods programmes. Taken individually, GiveWell estimates the three pilots at four, five and six times its benchmark; GiveDirectly estimates them at six to eight times.⁸

The core programme continues in the countries named above; GiveDirectly gives a delivery target of $250 million for 2026.⁷ Emergency cash went out in 2025 after the earthquake in the Philippines and the hurricane in Jamaica, among others.⁹ In the United States it runs its own programmes, including Rx Kids in Michigan and Ohio and In Her Hands in Georgia.¹¹

Why the Wealth for the World Fund funds GiveDirectly

Few anti-poverty interventions are as well studied as the unconditional cash transfer, and no organisation has contributed more to that record: five randomised evaluations of its own core programme, with follow-ups running years out. That is why this programme is the benchmark GiveWell measures other programmes against, and why GiveDirectly is on this list even though it does not clear the livelihoods bar. If you cannot find something better evidenced than cash, give cash.

Then there is what the organisation publishes about itself. Its annual fraud and abuse report gives loss rates by programme, names two of its own design failures, and concedes that it missed its own case-closure targets. An organisation that publishes its worst numbers can be checked on its best ones. The Wealth for the World Fund pays out to GiveDirectly because both things hold at once: the strongest evidence base in the field, and a willingness to document its own failures.

Footnotes

1. GiveDirectly, "About GiveDirectly". givedirectly.org ↩ ↩²

2. GiveWell, "GiveDirectly's Cash for Poverty Relief Program". givewell.org ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³

3. Effektiv Spenden, "GiveDirectly". effektiv-spenden.org ↩

4. World Bank, "March 2026 global poverty update from the World Bank". blogs.worldbank.org ↩

5. World Bank, "The World Bank's new global poverty lines in 2021 prices"; World Bank, "June 2025 global poverty update from the World Bank". blogs.worldbank.org; blogs.worldbank.org ↩ ↩²

6. Max Roser, "The end of progress against extreme poverty?", Our World in Data. ourworldindata.org ↩

7. GiveDirectly, "Cash for Poverty Relief". givedirectly.org ↩ ↩² ↩³

8. GiveWell, "GiveDirectly, Pilots of Cash Transfer Program Variations, November 2025". givewell.org ↩ ↩² ↩³ ↩⁴ ↩⁵

9. GiveDirectly, "Report: risks we faced delivering cash in 2025". givedirectly.org ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸

10. GiveDirectly, "Fraud in D.R.C. – our apology and response". givedirectly.org ↩

11. GiveDirectly, "GiveDirectly in the U.S.". givedirectly.org ↩