And the cashonistas rejoice

What happens when $1000 of manna falls onto your mobile phone? The GiveDirectly study of unconditional cash to poor farmers in Kenya is out.

The answer? Many good things. Less hunger, more businesses and incomes, long term investments in assets like cattle or roofs. Interestingly, little effect on education. But, then, primary education is mostly free in Kenya so money might not be the big thing holding people back from schooling. The bottom line: people don’t waste it.

The study is extremely well done. There are a lot of different treatments to evaluate for a modest sample–big versus small grants, one-time versus spread out, to men versus women. Fortunately for the authors most of the impacts are big enough they can parse out the impacts, and they are largely promising. It turns out

I think skeptics would be right to make two points.

First, maybe the people who get the cash tell surveyors what they want to hear.This is always a risk. You can get around it by looking at big things you can observe, like new cattle and roofs. Or asking questions indirectly (e.g. measuring expenditures and assets, which respondents might answer less strategically). This satisfies me.

The second concern, from some policymakers–how long does this last? Isn’t far better to invest in children’s health and education? I suppose the answer is that cash is not necessarily a substitute for improving public schools and lowering the cost. Nor is it an easy substitute for establishing clinics or public health campaigns. It’s possible cash would stimulate health and education more, but that’s unproven and (my guess) unlikely.

The kinds of programs we should put squarely in our sights, and consider replacing with cash, are the kinds where we deliver expensive, heavy, cumbersome stuff to people because we think we know what they need–bags of rice, business skills trainings, vocational training, fertilizer, agricultural inputs, and the like.

The policy world needs to run some horse races between cash and these competitors, and also between different kinds of cash (or should I say different strings attached). Cash and strings may or may not win out, but the evidence it might is becoming too overwhelming to ignore.

Yesterday’s Economist article is possibly the single best piece of journalism on cash transfers I’ve seen so far. And by “best” I mean detailed and accurate, representing a large body of academic work well–all traits that, sadly, tend to make the Pulitzer committee fall asleep in bed.

See the cash transfer debate on this blog.

10 Responses

  1. One question on the potential uptick in UCT is whether local systems will start to adapt to it and try to capture it for their own purposes. E.g., it is noted that the selected communities were driven by crunching census data in Kenya; to the extent that political types want to steer resources to their home communities (and claim credit for this), if UCT was expanded in volume to approach anything like aid flows – reaching say 2 million Kenyans households – would we see politicians adjusting the census data so that it directed resources to follow political incentives, or attacking the program for disproportionately helping people of one ethnic group?

  2. @Max Weylandt

    The “ground-up with assistance” approach might be easier, though, if you’ve got a government that’s not really capable of managing a large nation-scale education program without major issues of corruption. At the very least, the people in the local school district are closer to the ground on issues of school quality.

  3. Isn’t time we stopped calling them the patronizing term “cash transfer”, instead let’s call them what the global middle class receives every month if not bi-monthly, Salaries. Let’s not split hairs on it being given for paid work. If you’re shuffling paper/ pie chart or (data) plot then you’re getting an UCT or CCT (have to show up ala Yahoo). So hurray for the salaried poor.

  4. @Naunihal I think bottom-up education can work, in the absence of quality education top-down, but I think it’s an inferior solution. In the US, for example, a great deal of schools’ funding comes from local property taxes, which means that rich areas have well-funded schools and poor areas don’t. This explains (some) of the gross inequalities of the system.

    With a top-down system you can have more consistent quality, which is nice if quality is good. If it’s not, perhaps bottom-up can be a solution at least for a while.

  5. Suvojit: Hypothetically, what’s the difference between people organizing their schooling from the bottom and governments organizing it from the top? In America, actually, the history was of such community schools, and even now schools are controlled at the local level, not the state and definitely not at the national level.

    What I’m saying is that cash in the hands of family + effective local collective action may be a complete substitute for nationally organized schools, we just don’t know yet. Same for clinics, although the argument breaks down for roads.

    I don’t have a dog in this fight at all, I’m just not ready to conclude one way or the other.

  6. Thanks, the Economist article is excellent. One question on the potential uptick in UCT is whether local systems will start to adapt to it and try to capture it for their own purposes. E.g., it is noted that the selected communities were driven by crunching census data in Kenya; to the extent that political types want to steer resources to their home communities (and claim credit for this), if UCT was expanded in volume to approach anything like aid flows – reaching say 2 million Kenyans households – would we see politicians adjusting the census data so that it directed resources to follow political incentives, or attacking the program for disproportionately helping people of one ethnic group? Would mixed communities, say in peri-urban areas, witness increases in conflict over the specific households receiving cash transfers?

    In any case, those are secondary concerns, and certainly your point on which programs UCTs should be seen as competition/replacements for is spot on.

  7. P.S. A great example of how little we know about the lasting effects of even giant expenditures on ‘social’ interventions like schooling and health: Chen, Mu, and Ravallion’s 2009 paper in the Journal of Public Economiics:

    http://www.sciencedirect.com/science/article/pii/S0047272708001710

    Hundreds of thousands of dollars spent on nice-sounding social interventions in villages across poor, remote provinces of China. In the last year of the project, big effects. Five years later, no detectable effects. So “how long does this last” is a question we should be asking equally of GiveDirectly *and* of all other feasible uses of the same money, without preconceptions.

  8. The first concern is spot-on, though I’d go even further: actual behavior, not just reports, might respond to reciprocity norms. (Maybe actual domestic violence declines, not just reports of it, in part because people sense that good behavior increases the probability of future transfers—despite verbal assurances that it’s a one-off.) That would challenge both construct validity (mechanism interpretation) and external validity.

    I’m less sure about the second concern. For many education and health interventions that could be accomplished at the same cost in the same places, we don’t have clear evidence of the long-run effects. We certainly don’t know that the long-run effects of those alternatives are generally far better. (I know you’re not claiming they’re far better—you pose a rhetorical question—but I’d respond more strongly to the rhetorical question. We definitely do not know that the long-run effects of many alternative interventions, just because they feel ‘social’ or ‘collective’, is far better, or any better.) In general we just don’t have long-run impact evidence about much—not GiveDirectly, to be sure, but not about many alternative uses of the same money either.