Giving out cash is a great way to fight poverty. This approach might be even better.

Just giving out cash to poor people is a pretty good way to make them less poor.

That might seem obvious, but it wasn’t a commonly held viewpoint in development charities until relatively recently. Jacquelline Fuller, who runs Google’s philanthropic arm, has said that when she first pitched one of her bosses on supporting GiveDirectly (a charity doing unrestricted cash transfers), he replied, “You must be smoking crack.”

But in part due to groups like GiveDirectly, and in even larger part due to the success of government programs like Brazil’s Bolsa Familia and Kenya’s cash program for orphans and vulnerable children, that stigma has dissipated. Cash is cool now, at least in some corners.

And for good reason. The most common arguments against giving out cash — that it’s wasted on drugs and alcohol, or makes recipients stop working — have been debunked in repeated studies, and a review of hundreds of studies measuring dozens of different outcomes suggests that cash programs can increase food consumption, boost school attendance, and improve nutrition. If nothing else, cash just mechanically makes people less poor. It’s not a cure-all and has real limitations, but it’s pretty good, and “pretty good” can be hard to find in international development.

One advantage of having a pretty good rough-and-ready way to help poor people abroad is that it gives you something to test against. This is called “cash benchmarking,” and it’s something that cash fans, like GiveDirectly’s co-founder Paul Niehaus, have promoted for years. The idea is that because cash works reasonably well, respects the independence of recipients, and is relatively easy to hand out at minimal administrative expense, aid agencies should test programs to see if they meet their objectives better than cash would. If they don’t, that’s a pretty good argument to either improve the program or switch to cash.

USAID, the American foreign aid agency, made news in October by testing a nutrition program against cash. The two performed about equally well, with maybe a slight advantage to the cost-equivalent cash program; a much bigger cash program had really outstanding impacts.

But as a number of development professionals pointed out after I profiled the USAID program, that’s not the full story. At least two other studies have compared complex non-cash aid programs to cash — and found that certain non-cash alternatives actually beat cash.

Both studies involve programs commonly known in the development world as “ultra-poor graduation” programs, as they’re meant to “graduate” beneficiaries out of extreme poverty.

Graduation programs try to target the very poorest people in already very poor countries. Instead of only giving cash, they give valuable assets (which could be money but could also be an animal like a goat or cow, or equipment like a bicycle or sewing machine) as well as training, mentoring, and ongoing support (and sometimes some cash too, to buy food and keep people going). The hope is that giving some startup capital and some business skills helps recipients build a small ongoing enterprise — a small vegetable or dairy farming operation, say, or a bicycle messenger service, or a seamstress shop. That, in turn, is meant to enable a durable escape from poverty.

Think of it as the “teach someone to fish” approach, to cash’s “just give them the fish already” approach — though it’s still more “give them a fish” than microfinance, which took a similar approach but forced beneficiaries to pay back the asset grant with interest.

I’ve generally been skeptical of “teach someone to fish” approaches to development, not because they’re a bad idea conceptually but because we’ve often struggled to figure out how to teach people to fish effectively.

But recent research has suggested the graduation approach is promising. A massive randomized study published in 2015 by a murderer’s row of prominent development economists — including Northwestern’s Dean Karlan and MIT’s Abhijit Banerjee and Esther Duflo, among others — found that a graduation program tested in Ethiopia, Ghana, Honduras, India, Pakistan, and Peru significantly increased income and savings, reduced hunger and missed meals, and improved mental health, on average. It worked in every country but Honduras, where people fell behind when the chickens they were given died of disease.

Longer-term studies have also found promising results. A large-scale trial in Bangladesh covering more than 21,000 households found that even at a seven-year follow-up, graduation programs reduce poverty and increase assets among recipients. A non-randomized paper, also in Bangladesh, found that a graduation program still had large effects after six years, and while they dissipated a bit by nine years out, they didn’t evaporate — income was still significantly higher.

The two newer studies also reach promising conclusions — and in particular find that the graduation model outperforms a simple one-off cash transfer. The first study, conducted in Uganda by Oxford’s Richard Sedlmayr, UChicago’s Anuj Shah, and Save the Children’s Munshi Sulaiman, evaluates a program run by a long-running group called Village Enterprise in Uganda, which has run more than 100,000 trainings over the past three decades.

“It looks a lot like ultra-poor graduation, but is less intensive, and therefore, less expensive,” Innovations for Poverty Action, the development economics research group that conducted the evaluation, explains. It costs about a third as much and runs for only one year, whereas graduation approaches generally provide training and mentoring over two years.

The people targeted were extremely poor. More than 95 percent engaged in subsistence farming to get by, and 87 percent lived on under $2 a day. They were randomly given either nothing; a cash transfer of $119.40, paid once; or the Village Enterprise program, which included business training, a capital grant in cash to use to start a business, mentoring from other businesspeople, and a “business savings group” to help entrepreneurs in the community borrow money to invest and save profits. (Some groups also got variations on the Village Enterprise program, or a cash transfer plus some therapy-like sessions.)

So what did the researchers find? The short answer is that the full Village Enterprise program appeared to work: Recipients consumed $7.33 more per year per person (which may not sound like a lot but is for ultra-poor families), and in particular were able to buy more food. The researchers found that the result was driven by recipients earning more self-employment income. That’s exactly what the program is meant to do: give people training and assets so they can start small enterprises and support themselves.

Cash was another story. “Contrary to expectations,” the authors write, “consumption estimates are markedly negative among cash transfer program beneficiaries.”

Let that sink in for a second. Giving people cash led them to spend less, not more, money.

The money, instead, appeared to be used for loan repayment. That can be really good. A lot of very poor people in Africa have very high interest rates on debt, and it might make more sense for them to pay down debts first before starting businesses if those businesses don’t have high returns. But it meant that their day-to-day standard of living didn’t improve as much as it did among the Village Enterprise recipients, whose consumption went up.

“For reasons we cannot fully explain, transfer recipients appeared to derive less economic value from their assets than microenterprise beneficiaries did,” the study authors write. They speculate that, “left to themselves — without training and mentorship — beneficiaries struggled to make productive investments, maintain them, and derive sustained value from them.”

This is just one study, and even the authors caution skepticism when interpreting the results: “The point estimates of the cash arm are puzzling and could warrant some suspicion.” Cash isn’t supposed to make you spend less money, after all. It’s one data point suggesting that a more comprehensive anti-poverty intervention could outperform cash.

The second study, authored by Munshi Sulaiman from the first study, along with University of Illinois’s Reajul Chowdhury and UC Berkeley’s Elliott Collins and Ethan Ligon, looks at a similar “graduation” program pilot in South Sudan, implemented by BRAC, the Bangladesh-based development group that pioneered this whole approach.

The study assigned 250 households near the BRAC office in the city of Yei to a full graduation program, with training in business skills and specific trades (like gardening or rearing livestock), transfers of assets like goats or gardening equipment, support meetings with other participants, and transfers of food to help support participants as they got their new businesses up and running. Another 125 households were assigned to receive a cash transfer equal to the amount spent on business assets and food for people in the graduation program. Another 274 were a pure control group.

Unlike the Uganda study, the South Sudan trial found that both the graduation approach and the cash transfers increased consumption in the near term, particularly spending on food, with the effect about the same for each. But only recipients of the graduation program had significantly higher assets. Cash recipients spent a little, saved a little, and shifted a bit away from growing crops toward other professions. But they didn’t set up durable enterprises.

“Cash transfers appear to increase consumption and possibly shift investment from agriculture to non-farm activities, without a related increase in wealth or income,” the authors summarize. “Conversely, the [graduation] program increased wealth and directly shifted work from agriculture to livestock, with increased consumption in the short run.”

The South Sudan results are a little easier to make sense of than the Uganda results. It stands to reason that cash should increase spending on basic necessities like food in the short run, as the South Sudan study found. But it’s also totally believable that the results of both would be short-lived.

Think about what you would do with a sudden cash infusion, perhaps one as big as your normal annual income. Would you start a small business — or find ways to spend it or save it without fundamentally transforming your life and job? Indeed, a study released this year comparing graduation to a strict asset grant without training found that the training and mentoring is necessary to get gains in consumption.

Again, these are only two studies, yet to go through peer review. And like any approach, ultra-poor graduation has its limits. A randomized trial in India, published in 2012, found no benefits from a graduation program because it mostly moved people out of regular agricultural jobs where they could’ve earned just as much. In other words, when the economy is functional enough that other paths to escape poverty exists, adding a graduation option might not help much.

Skeptics of the graduation approach also note that describing the results as “escaping poverty” could be a stretch. As Stephen Kidd and Diloá Bailey-Athias note in one skeptical piece, the large-scale six-country graduation study from Dean Karlan, Abhijit Banerjee, Esther Duflo, and their collaborators found that consumption only grew by between 4 and 12 cents per person per day. That’s meaningful for the people affected, many of whom were living on under $1 a day, but even the study authors conceded that “the average effects are not very large and do not correspond to our intuitive sense of what it would mean to be liberated from the trap of poverty.”

We need more comparative work, over a longer time horizon, to get a fuller picture of how cash stacks up versus a TUP approach.

But if the finding that transfers to the ultra-poor outperforms cash holds up, that’s not necessarily a bad result for cash as a movement. Indeed, it’s an illustration of how having cash as a benchmark can make it easier to identify really good interventions that go beyond just handing out money.

Without a direct comparison to cash, it was easy to look at studies of TUP programs, even massive ones like the six-country study, and wonder if it could all be so much simpler. TUP is an approach that pairs a raw transfer of assets, in the form of either money or specific goods, with training, mentoring, and sometimes savings incentives and food support. It makes sense that that would help, but it’s also complicated to implement, and if transferring cash assets alone does as much good, then scaling up cash programs instead sounds like a better option. It’s just less administratively difficult.

But that isn’t, so far, the finding. And if TUP’s advantage holds up, that means we’ll have found something really special: a targeted intervention, tested in a number of very different countries, that helps more people out of poverty at a lower cost than cash. That’s a high bar; cash is already pretty good. And finding stuff that clears such a high bar is a big deal.

Sign up for the Future Perfect newsletter. Twice a week, you’ll get a roundup of ideas and solutions for tackling our biggest challenges: improving public health, decreasing human and animal suffering, easing catastrophic risks, and — to put it simply — getting better at doing good.