Poor Families Given $709 With No Strings Attached Spent Less on Alcohol, Not More

A Fear That Shapes Every Dollar of Aid

When policymakers discuss giving money directly to poor people, a single objection dominates every other concern. They'll drink it. They'll smoke it. They'll blow it on things that make their lives worse, not better. This assumption runs so deep it rarely gets stated as a hypothesis. It operates as a premise, embedded in the architecture of virtually every aid program on Earth. Conditional cash transfers require school attendance. In-kind programs deliver livestock instead of currency. Food stamps restrict purchasing categories. Aid program design assumes the poor cannot be trusted with choices.

A randomized controlled trial in rural Kenya tested that assumption head-on. What researchers found embarrassed it.

$709 to a Family Under a Thatched Roof

Between 2011 and 2013, GiveDirectly sent unconditional cash transfers averaging $709 PPP to 503 randomly selected households in western Kenya's Rarieda district. Recipients were among the region's poorest, identified by a single criterion: they lived under thatched roofs. Cash arrived via M-Pesa, Kenya's mobile money platform. No conditions. No monitoring. No restrictions on spending. Recipients could buy a metal roof, feed their children, start a business, or spend every shilling at a bar. How they used the money was entirely up to them.

Johannes Haushofer of Princeton University and Jeremy Shapiro of Nairobi's Busara Center for Behavioral Economics designed the trial with unusual rigor. Randomization operated at two levels: first selecting 60 of 120 villages for treatment, then choosing which households within those villages would receive transfers. Within that group, they further randomized whether the wife or husband received the payment, whether it arrived as a lump sum or nine monthly installments, and whether the amount totaled $404 or $1,525 PPP. Across treatment, spillover, and pure control groups, 1,440 households were tracked through baseline and endline surveys that measured everything from roof material to cortisol levels.

Where Did It Go?

Nine months after transfers began, treated households had increased monthly consumption from $158 to $193 PPP. Food security improved substantially. Asset holdings jumped by $302 PPP against a control mean of $495 PPP, representing a 61% increase. Many families replaced their thatched roofs with metal ones, a durable investment that protects against rain damage and signals economic stability in rural Kenya.

Monthly revenue from agriculture, livestock, and small enterprises climbed by $16 PPP. Households were not simply spending the transfer down to zero. They were channeling portions of it into productive activities generating ongoing returns.

And alcohol? Tobacco? Treatment effects on temptation goods were negative. Not zero. Negative. Recipients spent slightly less on alcohol and tobacco than the control group, though the researchers note this difference was not statistically significant. What matters: across a trial specifically designed with power to detect it, no evidence appeared that unconditional cash increased spending on the items paternalistic program design most fears.

Inside Their Heads, Not Just Their Wallets

Researchers measured more than purchasing behavior. Psychological well-being improved by 0.16 standard deviations for happiness, 0.17 for life satisfaction, and 0.26 for stress reduction. Depression scores fell. In a population where baseline well-being is compressed by chronic poverty, these are not small numbers.

Cortisol data told a more nuanced story. Average cortisol levels showed no overall treatment effect, suggesting the biochemical stress response may be less sensitive to income shocks than self-reported well-being. But subgroup analysis revealed significantly lower cortisol when transfers went to wives rather than husbands, when they arrived as lump sums rather than installments, and when they were large rather than small. Poverty relief can change brain chemistry. How you structure the relief determines the pathway.

Nineteen Studies Say It Wasn't a Fluke

Haushofer and Shapiro's result was not isolated. David Evans and Anna Popova of the World Bank reviewed 19 studies spanning Latin America, Africa, and Asia, examining cash transfers and temptation goods. Published in Economic Development and Cultural Change, their meta-analysis found a pooled effect of −0.18 standard deviations on temptation spending. Across nearly two decades of evidence and multiple continents, cash transfers did not increase spending on alcohol or tobacco. If anything, the trend pointed slightly downward.

A 2019 replication study by Wang and colleagues at 3ie re-analyzed the original data using alternative statistical models and confirmed findings across all eight outcome domains. In 2022, Egger and colleagues published a general equilibrium study in Econometrica covering 10,500 households. They found a local transfer multiplier of 2.5, meaning each dollar transferred generated $2.50 in local economic activity, with positive spillovers even reaching households that received nothing.

Strongest Counterargument

A credible objection concerns duration rather than behavior. Haushofer and Shapiro measured outcomes nine months after transfers began. Poverty is chronic. A one-time transfer may produce a consumption spike that fades. Nobody can say from this study whether families who built metal roofs saw sustained income gains five years later, or whether the psychological benefits persisted once the money ran out. Short-term RCTs are powerful tools for causal identification but poor tools for modeling lifetime trajectories. If effects dissipate within two years, the cost-effectiveness calculus shifts substantially.

What We Didn't Prove

This trial ran in one district of rural Kenya, among families living under thatched roofs. Extrapolating to urban populations, different cultures, or middle-income countries requires caution. Null results on temptation goods are real but imprecise: the study lacked power to detect small increases in alcohol spending, and the researchers were transparent about this limitation. No evidence of increased spending is not proof of zero effect. Health and educational outcomes showed no improvement, raising questions about whether cash alone can address structural barriers in those domains. Psychological benefits were self-reported; cortisol provided only partial biochemical corroboration. A minor erratum published in 2017 corrected computational errors that did not materially alter results, but it serves as a reminder that even well-designed trials require ongoing scrutiny.

What You Can Do

If you donate to poverty-reduction organizations, consider GiveDirectly or similar direct-transfer programs, which consistently rank among the most cost-effective interventions evaluated by GiveWell. When you encounter arguments against direct cash transfers, ask for the evidence behind them. Comprehensive reviews spanning two decades show the "they'll waste it" concern lacks empirical support across multiple countries. For policymakers, conditions attached to cash transfers may add administrative cost without adding value. At minimum, the burden of proof has shifted: if you want to restrict how the poor spend money, you need evidence that restrictions improve outcomes. So far, that evidence remains thin.