[This article is a draft chapter of my book, A Critical Discussion of UBI Experiments, adapted as a blog post]

Like the experiments the 1960s and ’70s, the current round of experiments appears at a time when concern about poverty and inequality is rising and people are rethinking the existing redistributive strategy. The context is otherwise very different. The welfare state has been under attack and greatly pared back in many countries since the 1970s while it has been gradually expanding in many countries from the 1930s to the 1970s. The concern that automation disrupts the labor force that played a small but significant part the 1960s BIG movement, now plays a far larger role in the debate today. The two U.S. experiments are both largely funded by tech entrepreneurs who are particularly concerned this issue. One might think that the increased concern with automation would decrease the concern that UBI might decease work effort, but this does not seem to be the case for all of the experiment. Many still seem tacitly to assume that decreased work effort is necessarily a bad thing.

The current round of experiments is taking place in a much wider context. Including the Namibian and Indian projects that were completed several years ago, the current round involves experiments in four different continents, in very wealthy and much less wealthy countries, and in countries with very strong or with rather weak welfare systems. The different contexts make different testing opportunities possible, but they also bring in new constraints, because researchers have to comply with local laws which can significantly constrain the project. This is particularly important in Europe where experiments have to comply with national and European Union law.

Researchers in different political contexts are understandably interested in very different questions, but they should be aware of the experience in other countries for at least three reasons. First, they might learn how to defend their experiments from criticism that they had not expected in their political context. Second, researchers might consider attempting to replicate each other’s findings with different methods and/or in different circumstances. Third, researchers might try to look for things that other experiments have neglected to examine.

Researchers today obviously have access to much more sophisticated computer statistics programs, but the logistical and financial difficulties of distributing cash to hundreds or thousands of people remain. Therefore, the experiments today are, for the most part, comparable in size and scope to the 1970s experiments. Only in less wealthy countries have significantly larger experiments become feasible.

The next several sections give a brief overview of seven current or proposed experiments on or closely relating to UBI. Other possible experiments of varying connectedness to UBI are being discussed and might be implemented, but I have no accurate information about them at this time. And so, I’ll limit my overview to these seven with a brief discussion of how their findings might be viewed.

GiveDirectly in Kenya

GiveDirectly is a U.S. non-profit organization that has recently established the world’s largest UBI experiment in Kenya. The project is motivated largely be the desire for an evidence-based approached to international charity and development aid, and the belief that evidence so far indicates that the poorest people in the world find cash is extremely helpful. The experiment will involve tens of thousands of people across dozens of villages for several years. It will combine the techniques of RCTs and saturation studies with a significant number of control and experimental villages. The project is able to be so large both because GiveDirectly has raised a lot of money and because Kenya has such deep poverty. Some villages will receive a UBI of as little as US$0.50 per day. Others will receive $1 or perhaps more.

The low level of the UBI in the GiveDirectly project is necessary because of the great poverty and inequality in Kenya. Many of the villages where GiveDirectly operates have average incomes less than $1 per day. If GiveDirectly were to give everyone in one village $2 per day, they could easily make that village four-times-richer than the control or non-participating village down the road. This could create animosity and resistance to the program. Until they can afford the give the grant to everyone in Kenya, it has to be small.

But the small size of the grant makes a very large study possible. Researchers for GiveDirectly are able to combine RCT and saturation techniques and to run a fairly long-term study that is like to produce a great deal of valuable data about how UBI affects various quality-of-life indicators. Although the effects of a very small UBI on severely impoverished villages in Kenya might not tell us a lot about low a large UBI will work in wealthier nations, this study promises to provide a great deal of useful information about how UBI will work in lesser developed countries.

Finland

Olli Kangas of Kela

As I write, Finland is in the middle of a small-scale, two-year UBI experiment, which is being conducted by Kela, the Finnish Social Insurance Institution. It involves about 2,000 participants between ages 25 and 58, selected by a nationwide random sample of people receiving unemployment benefits. The experiment replaces unemployment insurance benefits of €560 per month with a UBI of the same size. The Finnish parliament rewrote the law to make participation in the experiment mandatory for unemployment benefit recipients who were selected.

The Finnish effort has been criticized because the UBI is so low and because, being drawn from people receiving unemployment benefits, it incorporates the conditions of eligibility attached to those unemployment benefits. Kela responded that it simply does not have the budget to conduct an experiment across a large selection of low-income individuals.[i]

The make-up of the Finish experiment has at least two advantages as a UBI test. First, the low-level of the grant makes it comparable to the existing program, eliminating problems of distinguishing the effects of the size and type of program under investigation (as discussed in Chapter 4 of my book). Second, even though people had to be eligible for unemployment benefits to be selected for the study, once they were assigned to the experimental group, all or most conditions were eliminated. Therefore, although the study is not designed to examine how a large UBI would affect a large cross-section of the public, it is well designed to examine how a small UBI would affect people currently on unemployment benefits. And that kind of study reveal a great deal of useful information about UBI.

The stated goal of the Finnish experiment is, “To obtain information on the effects of a basic income on employment.”[ii] This concern is very similar to what became the focus of the four U.S. experiments in the 1970s, but the design and focus of the study makes it very different. One of the motivations of the experiment is the fear that Finland’s long-term unemployment insurance eligibility criteria created significant disincentives to work.

Because the Finnish project tests UBI only on people currently receiving unemployment benefits (that is, people currently not working), and because UBI eliminates eligibility criteria that might inhibit unemployed people from taking jobs, the study might find that UBI increases employment among study participants. The study does not increase marginal tax rates for participants and so it will provide a much higher overall income for low-income workers in the study,[iii] but it will be expensive to replicate that program design on a national scale.

Canada

Issues such as poverty, inequality, and the complexity of the social insurance system have inspired the Canadian experiment. The Ontario government is conducting an experiment at three sites in Ontario: Hamilton, Thunder Bay, and Lindsay, and might later include an additional study at a First Nations community. The study so far involves an experimental group of up to 4,000 low-income people aged 18 to 64. One of sites has been described as a “quasi-saturation site,” but I have been unable to clarify that that means. Researchers hope to examine the NIT’s effects on quality-of-life indicators as well as work behavior, education, and entrepreneurship.[iv]

Although the people conducting the study call it a “basic income,” it is a negative income tax that is conditional not only on household income, but also on household size. Single people receive a maximum of C$16,989 per year while couples receive a maximum of C$24,027 and both face a take-back rate of 50% of earned income.[v]

The 6th Chapter of my book explained that the inclusion of a marginal tax rate is an element of the NIT model, but it is needed to approximate the impact of marginal tax rates on recipients. The fact that the maximum benefit for a couple is not simply double the maximum benefit for an individual is a form of conditionality that departs from the UBI model in a way that is not strictly necessary for the purpose of conducting experiments. That is, unlike the UBI model in which individuals receive the same amount regardless of whether they live in small or large households, in the Ontario study two people living together receive considerably less than two people living separately. The motivation for this conditionality is probably to save money. Two people living together can live more cheaply than two people living apart. By including this condition the program can provide a poverty-level BIG at a lower cost, but they create an incentive for people to live apart, and might create a situation in which recipients pretend to live apart.

Y Combinator in the United States

Y Combinator Research (YCR) the nonprofit arm of Y Combinator—a private venture capital firm in the United States. It is run by tech entrepreneurs who are very motivated by the automation issue. Basic Income has become a major focus of YCR’s research, and it has taken on the effort to fund a large-scale UBI project with purely private funds.

Originally planned for Oakland, California, the organizers decided to move the experiment to two other states not yet announced. The experimental group will involve at least 1,000 people who will receive $1,000 per month for 3-to-5 years. More subjects will be included if funding allows. The experimental group will involve people aged 21 and 40 with total household incomes (in the year before enrollment) below the median income in their local community. Although researchers will gather data on how participants use their time and money, they will focus on the impact of UBI on social and physiological well-being—using both subjective and objective measures. The initial project proposal makes no mention of phasing out the grant as income rises.[vi] Therefore, YCR is testing a true UBI, but like the Finnish study, the YCR study implicitly assumes that recipients will face no higher marginal tax rates under a UBI system than they do now.

The Netherlands

The Netherlands experiment is a bit unusual for the times. While politicians in Greece, Italy, Spain, and several other places today are promoting proposals that are called “basic income” even though they share little with the basic income model, the Netherlands is experimenting with something that they do not call “basic income” even though it takes a significant step in the direction of basic income. The experiment seems to be motived in part by dissatisfaction with so-called “active labor-market policies” that are in place in the Netherlands and several other countries. These policies allow people to keep some benefits while in work, but subject them to harsh sanctions if they fail to search for work or to remain in work if they find it.[vii] These policies have proven to be cost-ineffective and often allow employers to capture some of the benefit intended for low wage workers.[viii]

Although the Dutch experiment is limited to welfare recipients under the current system, it frees people from job requirements of the current system and allows them to keep some of their benefits as they earn. These are two important features of a UBI. Because the cost-effectiveness record of active labor-market policies is so poor, this experiment could show that these steps in the direction of UBI will prove to be a more cost-effective means of achieving some of the ends of active labor-market policies.[ix]

The Dutch experiment is sometimes conceived of as a “trust” experiment because the existing system makes caseworkers responsible for enforcing rather draconian sanctions on recipients fostering distrust on both sides. Yet, this experiment conceptualizes “trust” in terms of fulfilling the obligations of a recipient of conventional social assistance—primarily to take work if they find it. In that sense they are not directly related to UBI, which is often conceived as a rejection of such obligations.

The Dutch experiment is actually several experiments that will take place in several different municipalities across the country—made possible by a 2015 law allowing experimentation at the municipal level. The experiments, launched in late 2017 and expected to last for two years, will study the effects on labor market and social participation, health and well-being of allowing social assistance claimants to maintain at least some of their benefits as their income rises while exempting them from the legal duties of seeking work and/or participating in training activities. The experiments involve several different experimental groups eligible for slightly different policies. Recipients are randomly assigned to the control group or one of the experimental groups in their municipality.[x]

Stockton, California

The city of Stockton, California has secured funding from private non-profits to launch a small-scale UBI project with about 100 participants receiving $500 a month for approximately 18 months. Like Y Combinator, major funders of the Stockton project are also largely involved in the tech industry and motivated by the automation issue.

Although the project has received a great deal of media attention, it is in the early planning stages and few details have been announced. The project is not called “the Stockton experiment” but “the Stockton Economic Empowerment Demonstration” (SEED). The organizers do not claim to be planning a “scientific experiment,” but a “a guaranteed income demonstration,” which could be taken as indication that it is aimed not to gather rigorous data but to present useful but possibly anecdotal evidence to further UBI politically.[xi] There is nothing wrong with conducting a smaller-scale and/or a less-rigorous study, and all the difficulties of clearly communicating what it does and does not say about the implementation of a full, nationwide UBI still apply.

Scotland

The Scottish government has committed funds to conduct a full-scale UBI experiment, but it is currently in the planning stages and few if any details about the experiment have been announced yet.[xii]

Will we refight the last war?

Earlier chapters of my book showed, in the 1970s, BIG opponents focused on two findings of the UBI experiments: the relative decline in hours worked and possible but controversial finding of a correlation with increased divorce rate. Opponents framed those issues in very extreme ways to make the findings appear definitive against BIG: any decline in work effort, no matter how small and no matter that it might be counteracted by other policies was taken not only as a “bad” thing, but bad enough to be a definitive reason to consider the policy a failure. Any decrease in the divorce rate was considered “good,” even if divorce was inhibited by keeping unhappy women financially dependent on men.

Will something like this happen again when these seven experiments start releasing their findings? It will probably not happen in the exact same way. Much of the discussion of the 1970s experiments was particular to the time and place: supply-side economics was on the rise within academia; the War on Poverty had decline in popularity politically; and politicians who vilified the poor were on the rise. But it is almost certain that less conscientious supporters and opponents will attempt to seize on whatever findings they can, framing them in whatever way necessary to spin the discussion in their favor. More conscientious participants of the discussion—whether directly involved in the experiments or not—with the benefit of past experience need to be ready this time.

I doubt the divorce issue will come back, but because the vilification of any non-wealthy person who balks and long hours for low pay is such a perennial favorite of the opponents of virtually any redistributive measure, people need to be ready for this sort of framing of the work-effort issue even if they do not expect it in their political context. It was not a major issue in India or Namibia because in those areas UBI was associated with increase work time. Similar results are expected in Kenya. The Finnish and Dutch experiments draw their samples in a way that is less likely to show a negative correlation between UBI and labor effort and may even show a positive correlation. This is so because conditional programs have a poverty trap that discourages people who don’t meet the conditions from leaving the labor force but encourages those who do meet the conditions to remain out of the labor force. By relieving the conditions, UBI is likely to be correlated with less work for those who had not been eligible and more work for those who had been eligible for redistribution under the conditional system. Most U.S. NIT experiments of the 1970s focused on people who had not been eligible for the largest redistributive programs, and so they were correlated with decreases in labor effort. The Finnish and Dutch experiments focus on people who are eligible for redistributive programs and so they might be correlated with increased work effort.

The other four experiments might now negative correlations and people involved should be consider ways to preempt or counteract any spin based on that correlation. Later chapters of my book consider how.

Of course, there are many other issues that people might use to spin the results of new UBI experiments. The issues will vary significantly by time and place. Knowing the specific political context and the international experience will help people preempt and/or counteract spin.

Dauphin, Manitoba: “the town without poverty”

Notes: contact me for full references:

 

[i] {Kangas, 2017 #1424}; {Kangas, 2016 #1425}

[ii] {Kangas, 2017 #1426}

[iii] {Kangas, 2017 #1426}

[iv] {Ministry-of-Community-and-Social-Services, 2018 #1433}; {Forget, 2016 #1427}

[v] {Ministry-of-Community-and-Social-Services, 2018 #1433}; {Forget, 2016 #1427}

[vi] {Y-Combinator-Research, 2017 #1428}

[vii] Loek Groot and Robert van der Veen, remarks made and the workshop on Basic Income experiments held at the Center for International and Regional Studies, Georgetown-University Qatar, March 26, 2018

[viii] {Bouquin, 2005 #303}

[ix] Loek Groot and Robert van der Veen, remarks made and the workshop on Basic Income experiments held at the Center for International and Regional Studies, Georgetown-University Qatar, March 26, 2018

[x]{McFarland, 2017 #1431}; {Groot, 2016 #1429};

[xi] {SEED, 2018 #1432}

[xii] {McFarland, 2017 #1431}