What the Cash Transfer Evidence Actually Says

Few propositions in development economics have been tested as often, or as adversarially, as the idea that giving poor people money makes them better off. The hypothesis attracted scepticism precisely because it sounded naive, and that scepticism funded an enormous body of randomised evidence across Kenya, Malawi, Mexico, Brazil, Indonesia and beyond. The evidence has now largely arrived, and it is less ambiguous than the debate around it suggests.

The headline findings are consistent. Recipients do not, on average, spend transfers on alcohol or tobacco; several meta-analyses find consumption of both falls. Investment in productive assets rises. School attendance rises, and so does the use of health services. Effects on nutrition are real but smaller than advocates hoped, and effects on income persist for years after the transfer ends in some settings while fading in others.

Where the picture genuinely complicates is in general equilibrium. A transfer to one household is income; transfers to an entire district are a demand shock, and prices respond. The best studies find modest local inflation that erodes some of the gain for non-recipients, alongside spillover benefits as recipients buy from neighbours. Both effects are real, and neither is large enough to reverse the direct result — but a programme designed as though the second-round effects do not exist will overpromise.

The more interesting disagreement is now about design rather than principle. Lump sums outperform equivalent streams for asset purchases; streams outperform lump sums for smoothing consumption. Conditions attached to schooling produce attendance gains, but a substantial share of those gains appear in unconditional programmes too, which raises the question of whether the monitoring apparatus is worth its cost. Targeting eats a meaningful fraction of budgets and misses a meaningful fraction of the poor.

None of this makes cash a development strategy. It is a floor, not a ladder: it does not build a clinic, train a nurse, or fix the road that gets a crop to market, and the countries that have grown out of poverty did so with structural transformation, not stipends. The mistake in both directions is category error — treating a proven instrument for reducing immediate deprivation as either a panacea or a distraction.

The evidence deserves better than the argument it keeps getting. It says, with unusual clarity, that poor people are not poor because they are bad at managing money. That finding was always the real subject of the controversy.