A charity's radical experiment: giving 6,000 Kenyans enough money to escape poverty for a decade

GiveDirectly, a charity that gives money directly to poor people in Kenya and Uganda, is launching a big new project: a basic income.

A basic income — also called a universal basic income (UBI), guaranteed minimum income, citizens' dividend, demogrant, etc. — is a regular payment to a group of people just for being alive. Normally, basic income proposals call for the payments to be administered by the government, but there's nothing in principle stopping a nonprofit like GiveDirectly from doing it.

So it's giving the policy a shot, and will give about 6,000 people in Kenya a guaranteed flow of cash for the next 10 years or more. In doing so, GiveDirectly is testing out an idea that's rapidly gaining interest in Finland, Silicon Valley, and Ontario, Canada, and could radically transform welfare policy in both rich and poor countries in the future.

More than that, it's creating what is perhaps the first true universal basic income in recorded history. There have been previous policies that are at least somewhat like this. But GiveDirectly's introduction of a universal payment for whole villages over a long, long period, set at a level of basic subsistence, is truly historic.

How GiveDirectly plans to launch its basic income

GiveDirectly currently uses M-PESA and MTN, cellphone-based payment systems popular in East Africa, to distribute big lump-sum payments to desperately poor people in Kenya and Uganda. So far, it's gotten results: A randomized evaluation of the charity found that recipients ate more and experienced less hunger, invested in expensive but worthwhile assets like iron roofs and farm animals, and reported higher psychological well-being. They were less hungry, richer, and all-around happier.

For its basic income project, the group will randomly select dozens of villages in Kenya (it already has a specific region selected) with about 6,000 people in them total and, starting at the end of this year, provide every current resident with a basic income for the next 10 years, potentially continuing even after that. The group is still finalizing details, but the payment is expected to be about $0.70 to $1.10 per person per day. It will likely vary from village to village to allow for more testing. More than 15,000 people total will get some form of cash transfer from the project, including the 6,000 getting a full UBI.

While $250 to $400 a year is a pittance by US standards, it's as much as or more than the recipients ordinarily make annually. "These recipients are some of the most vulnerable people in the world, living on the U.S. equivalent of less than a dollar a day," GiveDirectly co-founders Michael Faye and Paul Niehaus write.

The group is planning to randomize which villages get the grant and which will serve as "controls." That lets the project double as an experiment generating scientifically valid results about what basic income policies do when implemented over the long run for entire villages. GiveDirectly is bringing in MIT economist Abhijit Banerjee, a co-founder of the Abdul Latif Jameel Poverty Action Lab (J-PAL) and one of the most-respected development economists in the world, to run the evaluation. Also involved is Albert Wenger, a partner at the venture capital firm Union Square Ventures and an angel investor in companies like Etsy and Tumblr, who's donating to the effort.

GiveDirectly expects results well before the 10 years are up. Even within a year or two, consumption patterns and other behavior will likely change in the villages getting the UBI, in large part because participants know the payments will continue for the foreseeable future. That will likely lead to planning and work shifts that wouldn't take place in a temporary trial.

What makes the GiveDirectly experiment different

GiveDirectly's is one of at least five basic income studies happening at the moment. The Dutch city of Utrecht is currently giving some existing welfare recipients a basic income instead of their usual benefit package, Finland is planning a two-year trial testing of a variety of different basic income approaches, the Silicon Valley investment firm Y Combinator is funding an experiment, and the government of Ontario is doing a pilot too.

But the GiveDirectly study could prove to be the most important and informative of the bunch. Unlike every study to date, which lasted a few years at most, GiveDirectly's basic income will be paid out for a decade or more. The plan is to raise $30 million in initial capital for the experiment. As this chart from GiveDirectly points out, the only long-term basic income–like payout to be studied to date is a casino dividend paid by the Eastern Band of Cherokees, which is obviously a rather different situation; it also wasn't a randomized experiment, limiting its power:

(GiveDirectly)

The trial is also unusual in being truly universal. The American basic income trials targeted existing welfare recipients; they weren't available to everyone in the community. Same for the current trial in Utrecht. The Y Combinator trial sounds like it won't target a certain geographic area but rather a set number of people chosen at random, which would similarly limit its ability to test the policy's effects on an entire community. The GiveDirectly basic income would be truly universal, providing an opportunity to see village-wide effects and macroeconomic changes.

The GiveDirectly study also stands out for targeting a developing country. There's been some studies on basic incomes in poor nations, notably a non-randomized effort in Namibia and a trial by veteran basic income advocate Guy Standing in India, but most past experiments were conducted in rich countries. That's somewhat perverse, as the combination of a dramatically lower poverty line and the availability of NGO/foreign aid funds makes basic income more sustainable and affordable in developing countries than in rich ones.

What GiveDirectly is hoping to learn

While basic income itself has not been studied much in poor countries, cash transfers in general definitely have. There's a large empirical literature on cash transfer programs now, and the results are very positive, with various studies finding that just handing out money increases consumption, encourages investments in important assets like metal roofs, encourages more people to start working, boosts earnings, and doesn't lead to more spending on things like alcohol or tobacco.

This had led to a quiet cash revolution in development circles, as aid agencies, nonprofits, and the like have become considerably more sympathetic to cash as an intervention and new charities devoted to cash grants, like GiveDirectly, have gained ground. And it's led to a proliferation of "conditional cash transfer" programs in poor and middle-income countries, which like basic incomes provide a steady stream of cash to poor households but unlike them require recipients to do certain beneficial activities (get vaccinated, send their kids to school, etc.).

So what's the additional benefit of studying basic income, if we know so much about cash in general? One reason, Faye says, is the "income" piece of things. So far, GiveDirectly has focused on doing lump-sum payments rather than recurring ones. This has some obvious disadvantages relative to the regular payments of a basic income, which allow recipients to spread out their spending, ease the mental strain of saving and budgeting, provide a safety net in case things go wrong, etc. There's a reason people tend to prefer regular paychecks to one big check every year.

But lump-sum payments also have a lot going for them. For one thing, they're less paternalistic. Regular basic income payments assume that recipients need the government to ration out access to money, that they can't be trusted with a big, massive grant all at once, even if that grant is the same size as the total basic income checks they'd get.

More importantly from GiveDirectly's perspective, lump-sum transfers could be more effective at spurring actual development. The cash transfer research has tended to find that cash grants have high rates of return. That is, people getting the money also increase their income from business and agriculture, multiplying the money's impact. GiveDirectly's lump-sum grants, averaging $707, produced about $200 more in business and agriculture income, for a rate of return of 28 percent. A recent randomized study in Uganda got rates of return to cash of 30 to 49 percent.

That raises the stakes on the income-versus-lump-sum debate. If the rate of return on basic income is equal to, or lower than, the return for lump-sum payments, then that would seem to suggest you could generate more economic growth, faster, by just giving poor people all the money at once. If you give someone $5,000 all at once rather than $350 a year, then you'll get more returns earlier. But you have to weigh that against the other benefits of paying as income: insurance against future poverty, lower mental stress, etc.

The GiveDirectly study will enable Banerjee and the rest of the evaluation team to weigh those factors. Faye also says the experiment plans on giving at least some participants a choice between a lump-sum and regular payment, in part to gauge which they'd prefer. The results should provide a great lens into this particular question, and help clarify whether cash programs should aim to get as much money out as possible, as quickly as possible, or if regular payments are preferable.

Here's a simple solution to poverty