GiveDirectly
Image from GiveDirectly.org

GiveDirectly

Our Recommendation

GiveDirectly is highly recommended. For more than fifteen years it has funded independent researchers to follow its money down its own causal chain, and the evidence now reaches survival itself. A randomized study across 653 villages in Kenya, drawing on a census of more than 100,000 births, found that a one-time transfer of about $1,000 led to 48 percent fewer infant deaths and 45 percent fewer deaths before age five. The researchers trace the effect to better maternal health and nutrition, mothers working far less around the time of birth, and a 45 percent increase in hospital deliveries. A researcher quoted in the New York Times called the infant mortality decline a show-stopping result.

The same study reports a limit that a donor should weigh just as carefully. Mortality largely returned to pre-program levels after the transfers ended, and the effect faded when cash arrived more than three months from a birth. The paper is a working paper, revised in January 2026, and has not yet been peer-reviewed.

Around that result sits an unusually complete set of randomized ultimate outcomes: intimate partner violence down 51 percent, life satisfaction up 15 percent, food availability and dietary diversity up, and psychological wellbeing improved across at least six separate trials.

Three things deserve a donor’s attention. GiveDirectly has funded three randomized studies that pit cash against an alternative program at equal cost, and it publishes all of them, including a Rwanda nutrition study in which neither the traditional program nor a cost-equivalent $110 transfer moved the primary outcomes, and only a larger $532 transfer did. It answered the strongest objection to cash at scale with a study designed to find the problem: a village-level trial that measured effects on non-recipients and on local prices. It also publishes an annual report on fraud, harm, and safety incidents.

GiveDirectly's Fierce Certification score is 100/100 points based on our criteria:
✔ Has Ultimate Outcome Goals (20 pts)
✔ Measures Intermediate Outcomes (15 pts)
✔ Measures Ultimate Outcomes (20 pts)
✔ Measures Intermediate and/or Ultimate Counterfactual (20 pts)
✔ Shows Continual Learning & Adaptation (25 pts)

The Social Problem

Extreme poverty is not a single deprivation but a set of them that arrive together, and GiveDirectly frames the problem simply: households do not have enough money, and the aid system’s habitual response is to decide on their behalf what they need.

The consequences run in two chains. The economic chain is familiar: too little income, no savings, no assets, debt on bad terms, no credit, and no capital to start anything. Those conditions keep a household where it is.

The second chain is what the missing money costs a family, and it is where the harm lands. Meals skipped. A pregnant woman who keeps doing heavy field labor into her ninth month and gives birth at home because a hospital delivery is out of reach. A roof of thatch. School fees that cannot be paid. Sickness untreated. The psychological weight of destitution, which shows up in the research as measurable depression and stress. Violence inside the home, which money stress makes worse. And infant deaths that a single transfer turns out to cut nearly in half.

Underneath sits the problem the organization names first: being treated as the object of someone else’s decision instead of the author of your own. In UNHCR’s monitoring of its cash recipients in 44 countries, 90 percent said they preferred cash or a mix of cash and in-kind aid over in-kind assistance alone.

The Solution

GiveDirectly sends money directly to people living in extreme poverty, with no conditions attached.

The mechanism. GiveDirectly identifies households in extreme poverty, often by visible markers such as thatched roofing, enrolls them, and transfers money through mobile money. There are no conditions, no required behaviors, and no in-kind substitution. Transfer sizes in the studied programs range from about $250 to $1,000, as lump sums or monthly installments, with the flagship African poverty relief program at around $1,000.

Delivery economics are part of the model. More than 87 percent of funds reach recipients in the flagship program. The organization co-published research in Nature on using mobile phone data and machine learning to identify the poorest subscribers in Togo during COVID-19, comparing that method with traditional targeting on exclusion errors, welfare impact, and fairness.

Five program lines: poverty relief in Africa, emergency cash for crises and climate disasters, cash in the United States, basic income, and refugee programs. Government partnerships in Malawi and Rwanda integrate cash into national development plans.

Recipient consent is a stated policy. The organization documents recipients only with formal consent, using plain-language forms translated into local languages, and publishes its reasoning about avoiding what it calls poverty pornography.

Mapped against a Theory of Change, the chain is one link long by design: give money, and let recipients decide which consequence of poverty to address first. That design is what makes measurement hard. With an unconditional transfer you cannot predict which outcome will move, so you have to measure all of them, and GiveDirectly has funded trials across an unusually wide range of endpoints for exactly that reason.

Key Outputs

  • More than $1 billion delivered to 2,185,729 people across three continents and 19 countries
  • More than 160,000 donors
  • More than 87 percent of funds delivered to recipients in the flagship program; 85 percent is stated elsewhere on the organization’s site
  • 33 studies of GiveDirectly’s own programs: 24 randomized controlled trials and 9 non-experimental evaluations, of which 13 are still being implemented or analyzed
  • Research partners including Innovations for Poverty Action, IDinsight, the Center for Effective Global Action, the University of Chicago Inclusive Economy Lab, Georgetown University’s Gui2de, the International Food Policy Research Institute, the World Bank, the International Rescue Committee, ideas42, and Save the Children
  • Trials registered on the American Economic Association Randomized Controlled Trial Registry, the Registry for International Development Impact Evaluations, and the ISRCTN Registry
  • The largest randomized controlled trial of unconditional cash transfers, now running in Malawi: $550 to 140,000 adults across two districts, measuring 16 outcome domains including child mortality
  • The largest universal basic income study to date: 12-year, 2-year, and lump-sum arms in Kenya at about $0.75 per day
  • An annual public report on fraud, harm, and safety incidents

Key Intermediate Outcomes

Under our standard, most of GiveDirectly’s best-known figures land here, and nearly all are randomized.

Income, assets, and savings. Annual income up 50 percent and savings up 32 percent. In the original Kenya trial, consumption rose 23 percent and asset holdings 61 percent, mostly livestock, furniture, and metal roofs, while business revenue rose 33 percent without a measurable rise in profits. That trial found no effect on health or education measures. A second endline more than two years after transfers ended found sustained increases in assets, though the authors noted that long-term impacts on other dimensions and potential spillovers remained to be substantiated.

Housing value. Home value up 20 percent. This is an asset value, so we treat it as intermediate outcome. The ultimate outcome would be what that investment does for a family’s safety and health.

Business formation and the local economy. The number of businesses rose 20 percent. The general equilibrium study, published in Econometrica, randomized at the village level and found that each $1,000 transfer generated roughly $2,500 in total economic activity over 27 months, with no significant negative effects on non-recipients and prices rising only 0.1 percent on average, despite transfers exceeding 15 percent of local gross domestic product. Inflation was the strongest theoretical objection to cash at scale, which is why that last finding matters.

Village inequality. Down 29 percent, an indication that gains did not concentrate among the already better off.

Education. Children passing exams up 14 percent.

Care seeking, and the mechanism behind the mortality result. The mortality study found a 45 percent increase in hospital deliveries and a 53 percent reduction in mothers’ labor around the time of birth, with the largest mortality gains among households living close to physician-staffed facilities. Vaccination rates rose 16 percent in Rwanda.

Benchmarking against alternatives. Three randomized studies compared cash with an alternative program at equal cost per beneficiary. Against Rwandan youth job training at $332 per person, cash proved superior across a number of economic outcomes, while training outperformed cash only in producing business knowledge, and combining the two showed little gain in cost-effectiveness. Against the Rwandan Gikuriro nutrition program, neither the program nor a cost-equivalent $110 transfer affected the primary outcomes, and only a larger $532 transfer did. A Congolese youth training comparison is under analysis.

Key Ultimate Outcomes

Infant and child mortality. A randomized study across 653 villages in Kenya, with more than 10,500 households receiving a one-time transfer of about $1,000, found 48 percent fewer infant deaths before age one and 45 percent fewer child deaths before age five. The estimate draws on a new census of more than 100,000 births in the study area and accounts for spillovers onto non-recipient households. It was published as a National Bureau of Economic Research working paper in August 2025 and revised in January 2026, by researchers at the University of California, Berkeley and the University of Oxford.

Three findings in the same paper qualify the headline, and the authors report all of them. Mortality largely returned to pre-program levels after the transfers ended. The effect faded when cash arrived more than three months from a birth. And fertility rose a transient 10 percent. The mortality reductions were largest in households that received transfers after a pregnancy had begun, which the authors use to rule out the possibility that the result reflects which women chose to become pregnant. Survival is an ultimate outcome, and this is a randomized estimate of it at a scale few mortality studies reach. It is also an estimate of what cash does while it is arriving.

Food security. Improved against control in at least six separate trials across Kenya, Uganda, Liberia, Malawi, and Rwanda. Food availability at home rose 13 percent and dietary diversity 12 percent, and childhood growth improved at the larger Rwandan transfer size. Our standard treats food security as an ultimate outcome that higher income is supposed to produce, and here the chain is completed with a comparison group at both ends.

Mental health. Psychological wellbeing improved against randomized controls in the original Kenya trial, the general equilibrium trial, the Rwanda youth benchmarking study, the Kiryandongo refugee trial, the Liberia and Malawi study, and the Nairobi refugee study. Life satisfaction rose 15 percent.

Housing. In the original Kenya trial, recipient households were 24 percentage points more likely to have a non-thatched roof after the program. A metal roof is a known asset that keeps a family dry and safer, which our standard counts as an ultimate outcome.

Continual Learning & Adaptation

It funded studies designed to show cash not being as effective as expected. The three benchmarking trials compare cash head-to-head with an alternative program at equal cost, and the organization publishes every result, including the Rwanda nutrition study in which neither the traditional program nor a cost-equivalent $110 transfer affected the primary outcomes. That is a published null on its own intervention at a realistic transfer size, and it produced an actionable lesson: transfer size matters, and $110 was not enough.

It answered its strongest critique with a trial designed to find the problem. Inflation and spillovers were the main theoretical objections to cash at scale. The general equilibrium study randomized at the village level, measured effects on non-recipients, and reported both the multiplier and the price effect. A 0.1 percent price change counts as evidence because the study could have found the opposite.

It reports what its own results cannot yet support. GiveDirectly’s summary of the second Kenya endline carries the authors’ caveat that long-term impacts on other dimensions and potential spillover effects remain to be substantiated. The mortality paper, likewise, reports that the effect does not persist once transfers end.

Study design evolved from the findings. The original trial tested lump sums against installments and transfers to men against women. Later work tested letting recipients choose timing, compared long-term and short-term universal basic income against a lump sum, and tested whether adding psychological interventions, mentorship, financial planning, or nutrition education improves on cash alone. Several of those cash-plus studies are explicitly designed to find out whether GiveDirectly should be doing more than giving money.

It publishes a fraud and harm report every year. The organization states that the report reflects increased investment in safeguarding, and a public incident register is something few organizations volunteer.

It tests its own explanations to recipients. After moving to a payment system driven by artificial intelligence, it worked with researchers and local teams to test how best to explain that system to recipients, and found that scenario-based examples worked best.

It has moved toward government. Partnerships with Malawi and Rwanda integrate cash into national development plans, which the organization frames as the path to ending extreme poverty faster, not as an expansion of its own footprint.

Its measurement pipeline reaches further. The Malawi trial of 140,000 adults measures 16 outcome domains, including child mortality, intimate partner violence, and civic engagement. The Flint, Michigan study measures child mortality and kindergarten readiness. The Chicago homeless families study measures housing stability and criminal justice contact.

Further Reading

Organization Sources

  • GiveDirectly homepage, source of the 2,185,729 people reached, $1 billion delivered, 87 percent delivery ratio, 160,000 donors, and the 48 percent infant mortality headline. givedirectly.org
  • GiveDirectly, Our Research, the full library of 33 studies with design, registration, transfer size, population, and outcome details for each. Source of the 50 percent income, 61 percent assets, 32 percent savings, 20 percent home value, 16 percent vaccination, 15 percent life satisfaction, 13 percent food availability, 12 percent dietary diversity, 2.5 times local economic activity, 20 percent businesses, 14 percent exam passes, 29 percent inequality, and 51 percent intimate partner violence figures. givedirectly.org/gdresearch
  • GiveDirectly, “Study: Giving cash to mothers cut infant deaths in half.” givedirectly.org/mortality2025
  • GiveDirectly, Research on cash transfers and the Cash Evidence Explorer. Research on cash and Evidence Explorer
  • GiveDirectly, How we prevent, address, and report risks, including the annual fraud and harm report. givedirectly.org/risks
  • GiveDirectly, Financials and the cost breakdown. givedirectly.org/financials and Yes, we have costs
  • GiveDirectly, “Sharing stories of poverty but not poverty porn,” on recipient consent. givedirectly.org/poverty-porn
  • GiveDirectly, “Explain AI/ML systems to people in poverty using real-world scenarios.” givedirectly.org/explain_ai
  • GiveDirectly, government partnership work in Malawi and Rwanda. givedirectly.org/work-with-government-2024

Randomized Evidence

  • Haushofer, J., and Shapiro, J. (2016). “The Short-Term Impact of Unconditional Cash Transfers to the Poor: Experimental Evidence from Kenya.” Quarterly Journal of Economics 131(4): 1973-2042, with a 2018 follow-up endline. Source of the 23 percent consumption, 61 percent asset, 33 percent revenue, and metal roof findings, the null on health and education, and the sustained-assets finding. academic.oup.com
  • Walker, M. W., Shankar, N., Miguel, E., Egger, D., and Killeen, G. (2025, revised 2026). “Can Cash Transfers Save Lives? Evidence from a Large-Scale Experiment in Kenya.” National Bureau of Economic Research Working Paper 34152; also Centre for Economic Policy Research Discussion Paper 20995. Registered as AEARCTR-0000505. Source of the 48 percent infant and 45 percent under-five mortality reductions, the hospital delivery and labor findings, the reversion of mortality after transfers ended, and the transient fertility increase. Not yet peer-reviewed. nber.org
  • Egger, D., Haushofer, J., Miguel, E., Niehaus, P., and Walker, M. (2022). “General Equilibrium Effects of Cash Transfers: Experimental Evidence from Kenya.” Econometrica 90(6): 2603-2643. Source of the $2,500 multiplier, the 0.1 percent price effect, and the finding of no negative spillovers. Wiley
  • Banerjee, A., Faye, M., Krueger, A., Niehaus, P., and Suri, T. (2023). Universal basic income comparison, Kenya. Source of the income, savings, business, exam-pass, and inequality figures. NBER conference paper and the COVID-period paper
  • McIntosh, C., and Zeitlin, A. (2018, 2020). “Benchmarking a Child Nutrition Program Against Cash: Experimental Evidence from Rwanda.” Source of the home value and vaccination figures and the null on cost-equivalent transfers. Manuscript PDF
  • McIntosh, C., and Zeitlin, A. (2020, 2022). “Using Household Grants to Benchmark the Cost Effectiveness of a Youth Employment Program.” Source of the finding that cash proved superior across economic outcomes while training outperformed only on business knowledge. Manuscript PDF
  • Haushofer, J., Ringdal, C., Shapiro, J., and Wang, X. Y. (2019). Intimate partner violence and cash transfers. National Bureau of Economic Research Working Paper 25627. NBER PDF
  • Aiken, E., Bellue, S., Karlan, D., Udry, C., and Blumenstock, J. (2022). “Machine learning and phone data can improve targeting of humanitarian aid.” Nature. nature.com
  • Stein, D., Nshakira Rukundo, E., and Lanthorn, H. (2022). $1,000 to refugees in Kiryandongo, Uganda, with IDinsight. givedirectly.org
  • Jacob, B., Pilkauskas, N., Rhodes, E., Richard, K., and Shaefer, H. L. (2022). Project 100, $1,000 transfers to low-income Americans during COVID-19, more than 25,000 participants. givedirectly.org
  • Kahura et al. (2022). GiveDirectly Uganda endline report, IDinsight. idinsight.org

Outside Coverage and Ratings

  • Abdul Latif Jameel Poverty Action Lab, “Improving Economic and Psychological Well-being through Unconditional Cash Transfer in Kenya.” Summary of the original Kenya trial. povertyactionlab.org
  • Center for Effective Global Action, “Cash Transfers Dramatically Cut Infant Mortality in Kenya, New Research Finds,” 2025. Source of the Siaya County setting and the account of hospital births, rest for pregnant women, and nutrition as likely channels. cega.berkeley.edu
  • New York Times, “Giving $1,000 to poor families lowered infant mortality rates by nearly half,” August 2025. Source of the “show-stopping result” quotation. nytimes.com
  • UNHCR (2022), “Main Outcomes of Cash Assistance in 2021: Findings from Post Distribution Monitoring.” 19,699 household interviews across 44 countries. Source of the 90 percent preference for cash or a cash and in-kind combination. unhcr.org
  • The Economist, “GiveDirectly does what it says on the tin,” January 2024. economist.com
  • The Life You Can Save, GiveDirectly charity assessment. thelifeyoucansave.org

REGULAR POSTS

Be the first to know when a new organization is certified—and why it meets the standard.

By signing up for our mailing list you will receive a quick email when we add a podcast or a new Fierce Certified organization. You'll also be the first to hear about events and workshops we may have from time to time. We won't share your information without your permission.

Great! Please check your inbox and click the confirmation link.
Sorry, something went wrong. Please try again.

Written by

Todd Manwaring