Proposals for universal basic income (UBI) programs—rightly understood as unconditional equal transfer payments to all—have been gaining political traction and capturing the popular imagination recently. New York entrepreneur and dark horse Democratic primary candidate Andrew Yang has made a $1,000 per month UBI proposal the central plank of his campaign. Silicon Valley entrepreneurs like UBI plans because they see them as facilitating disruptive innovations involving automation that may put many people, at least temporarily, out of work. McKinsey forecasts that between 75 million and 375 million workers globally and between 16 and 54 million in the United States will have to try to find new jobs over the next decade, because 60 percent of existing jobs will have on average 30 percent of their current activities automated. Most threatened are office and manufacturing workers and drivers. The wide range of possible forecasts underlines the degree of uncertainty and consequently the anxiety that many workers feel. Nobody really knows how many jobs will be lost and, more importantly, how many of those laid off will find it impossible to find new jobs or update their skills.

Given the anger, fear, and backlash that followed the last recession, the stakes are high for finding the right policy mix for responding to disruptive innovation. From a political and social perspective, anxieties over jobs and status are even more important than actual economic reality. Recent research has shown that, though the economy and employment prospects have been improving, millennials have grown more pessimistic about their job prospects. If workers who have safe jobs feel anxiety, they will behave as if they are facing economic oblivion and social ostracism. Welfare cannot alleviate such anxiety. But perhaps a well-designed UBI program can.

UBI programs—meaning unconditional, direct, and universal payments—offer greater personal security and dignity than means-tested welfare programs because they are unconditional and have no time limits. Furthermore, by eliminating the bureaucratic layers that test the means of welfare beneficiaries and control and monitor the distribution of welfare, UBI programs would shift funds from administrative intermediaries and put them directly in the hands of beneficiaries. They may even make parts of the welfare system redundant—for example, the enactment and monitoring of minimum wage laws. Universal basic income may encourage economic risk taking, entrepreneurship, mobility, prosperity, and professional self-fulfillment, while reducing envy, fear, and resentment, since everybody would receive exactly the same transfer payment.

In an era when old political ideologies seem hard pressed to offer viable new policy ideas, UBI is an idea that is cutting through the traditional Left-Right lines of polarization in interesting ways. Although it is being explored today mostly on the center-Left, it has a chance at winning bipartisan support as a better alternative to the welfare state. In fact, a generation ago, the push for UBI came from the libertarian-conservative Right, counting among its proponents the likes of Milton Friedman, Donald Rumsfeld, Dick Cheney, and—within the context of supporting a carbon tax to pay for UBI—James Baker, Henry Paulson, and George Shultz.

Contemporary critics of UBI, such as the Democrats’ 2016 nominee Hillary Clinton and current 2020 Democratic frontrunner Joe Biden, retort that such programs would be prohibitively expensive and would require a combination of growth-killing high taxation, government debt, and inflation. Others argue UBI may create incentives for able workers to go idle rather than take low paying or otherwise unfulfilling jobs. Consequently, the economy may become less competitive, with more poets and potters and fewer sanitary and social workers. Employers would have to either raise wages beyond those offered in countries without UBI or, when possible, further outsource those jobs. Undoubtedly, in a UBI universe, there would be new risk-taking entrepreneurs who would generate technological progress, values, and jobs absent from our universe. There would also be able workers who work in our universe but would prefer to live on the public purse in a UBI universe. The question is whether entrepreneurship and self-fulfillment would compensate for the costs of idleness.

Advocates and critics of UBI have theorized and speculated about these questions, and about the virtues and vices of UBI. There is also a growing literature on the topic, including, of course, an academic journal: Basic Income Studies.¹ Several pilot programs have been tried recently around the world and in the United States, and more are planned by public and private institutions. On limited scales, pilots cannot do much harm, even if worst-case scenarios are realized. But this raises the question: What kind of pilot programs can give us good information about the prospects of UBI programs on a larger, or even national, scale? This is an important question to get right, if we are to go beyond theory and speculation and have a serious public policy debate about UBI. What follows is thus a brief sketch of the kinds of pilot proposals that can generate useful knowledge about the costs and benefits of UBI.

Well-designed and executed pilot programs should generate hard evidence for estimating the costs and benefits of UBI in comparison with those of welfare as we know it. Over time, pilots may measure the extent to which UBI encourages the assumption of economic risks: changing jobs, starting businesses, acquiring education and professional training, moving to a new house or city, and so on. In the even longer term, it may be possible to measure whether this risk-taking pays off and provides individuals with greater incomes and the state with higher tax revenues to pay for UBI.

Less tangibly, pilots may measure how much safer and self-fulfilled—and how much less envious—people feel when they benefit from UBI. It may be possible to discover correlations between increased personal security and tolerance, reductions in political extremism, and less drug use. We could find that families are strengthened by removing financial stressors, or weakened when partners are released from economic dependency, and so on.

However, many of the pilot programs going under the name UBI these days cannot generate useful evidence. For example, a 2017-18 pilot in Finland that received heavy press coverage conscripted 2,000 randomly selected unemployed Finns to receive €560 per month for two years—an amount comparable to standard Finnish unemployment benefits. If unemployed beneficiaries found employment while enrolled in the program, they continued to receive the €560. Program participants were thus released of the onerous and for some humiliating bureaucratic procedures necessary for receiving unemployment benefits. After two years, the researchers concluded that the experiment had had no effect on rates of employment, though they did find that participants enjoyed improved psychological well-being on various scales by about 10 percent. Now, however one chooses to judge this program—successful or not—this was not in fact a UBI pilot but rather an experiment with a new system for distributing unemployment benefits designed to eliminate disincentives for working low-wage jobs. Perhaps it didn’t work because the previous system of unemployment benefits did not disincentivize recipients from finding work to begin with, or perhaps it was because the misalignment between the Finnish labor market and the unemployed was so extreme that the added incentives and efforts to find work didn’t yield results. But however one chooses to read the outcomes, the program was not true UBI. By limiting the pilot to a random sample of unemployed over a limited time horizon, rather than extending it to the whole of society, indefinitely, and without considering the system of taxation, they were testing something other than a genuine UBI program.

The Finnish pilot, at least, worked with a random, compulsory, and sufficiently large (and therefore representative) sample. Last year’s pilot program in Stockton, California, had no such representative sample. The Stockton experiment, which was managed by local government and paid for by private foundations like the Economic Security Project, chose 130 adults who live in the city’s lower-income neighborhoods (Stockton is one of the poorest cities in California) to receive a monthly stipend of $500 for a period of 18 months in the form of debit cards, so researchers could monitor their expenses. One could perhaps make a case that such a program could measure demand patterns among the very poor, but its results can tell us nothing about Universal Basic Income. Similarly, in Jackson, Mississippi, another “UBI” program is set to give 15 African American mothers $1,000 per month, in addition to leadership training and counseling with a social worker. A noble philanthropic program, perhaps—but it’s not UBI.

The legendary Silicon Valley high-tech incubator Y Combinator ran a pilot in Oakland, California, that gave 100 families $1,500 per month each, while a control group received $50 per month for its cooperation in the research. This program ran into difficulty in recruitment and engagement, especially for the control group. Recently, Y Combinator hired the University of Michigan’s Survey Research Center to set up a different study of a random sample of 1,000 residences in neighborhoods where the median household is $46,000 or less in two states. Each household will receive $1,000 per month for three to five years. A control group of 2,000 households will receive $50 per month for their troubles. While this program is better than the previous pilot, it is still far from universal and so is merely yet another experiment with means-tested welfare.

The University of Michigan Survey Research Center also collects data for another project entitled “Baby’s First Years,” led from the University of California Irvine. The project intends to recruit 1,000 low-income mothers from maternity hospitals in four cities, half of whom will receive an unconditional $333 per month, while the control group will receive $20 per month. This is again an experiment with a welfare program, not UBI. Such research may even be somewhat redundant as many countries around the world have enacted child allowances for parents, either because they faced demographic declines due to sub-replacement fertility levels, or because governments wished to encourage parents to increase the population.

Finally, Massachusetts’s State Senate is considering two draft proposal for UBI pilots. One proposal, on which I have had some unofficial input, is still in the works. The other proposes three studies of 100 individuals each in three different kinds of communities in the state of Massachusetts that should include minorities (though that goal may be somewhat challenging in rural western Massachusetts). Each program participant is to receive a flat $1,000 per month.

Some of these pilots are more statistically valid than others, but because they are all directed at various disadvantaged sectors of society they are, at best, pilots for new welfare programs. At worst, they may even seem like little more than populist publicity stunts marketed as “money for nothing” or “free checks in the mail.” Though not intended as such by their organizers, they are reminiscent of the kinds of lotteries that populist regimes like Peronist Argentina organized, in which a few well-publicized individuals were chosen at random to receive rewards and thus spread hope for similar luck among the rest of the struggling poor.

Obviously, useful UBI pilots must be based on representative samples, divided randomly between experimental and control groups. The “treatment” of the experimental group should be the closest possible simulation of universal basic income. The pilot should measure the effects of UBI on the experimental group by comparing it with the control group and with its own baseline. The groups must be composed of households rather than individuals, because an individual with UBI whose family is not covered will make decisions based on nuisance variables, the total non-UBI condition of the household.

The basic income should be enough to live on. It may be based on an index of cost of living, the minimum wage, Social Security, or a combination of them. UBI should be guaranteed for a lifetime. Cutting corners, offering a few hundred dollars per month for a couple of years to means-tested poor or to unemployed volunteers cannot stimulate the kind of life-changing decisions and emotions that UBI promises. Most significantly, none of the above pilots attempted to simulate the financing of UBI which would obviously require rewriting the tax system. The misleading intuition was that UBI should benefit the poor, so the middle and upper classes can be safely excluded from the pilots. This misses what distinguished UBI from the welfare state.

A simple formula may determine how much each household should receive from, or contribute to, a pilot scheme (please excuse the accounting, but I am writing this after submitting my tax return so I got used to the genre). That formula should start by adding the household’s universal basic income to its actual gross income. Let’s call it Gross UBI Income. Then we should compute how much in taxes the household would be required to pay in a UBI universe (presupposing that we’ve modeled a taxation system that could pay for UBI). Let’s call that UBI Taxation. The balance between gross UBI Income and UBI Taxation is UBI Net Income:

Since pilots take place in the real world, pilot participants will still pay taxes and may receive welfare benefits where they live. If they pay more in taxes than they receive in welfare, the balance should be deducted from the UBI taxes and their UBI net income will increase. If they receive more in welfare than they pay in taxes, this balance should be added to the taxes they pay in the pilot, and their UBI net income should decrease.

This formula is flexible enough to accommodate different pilot models that experiment with different personal taxation systems such as flat or progressive taxation. They may also model welfare systems co-existing with UBI. For example they may assume privatized services such as health or education that will necessitate high UBI net income and higher taxes to pay for it; or they could assume socialized services, paid from taxation and managed by the state, so UBI net income would be lower, but taxes would be needed to pay for social services.

In this way researchers would have to consider from the start how they plan to pay for UBI. What sort of tax system could sustain such a wide safety net? Practically, it would allow pilots to include the middle and upper classes, because, although the UBI guarantee is universal and no household’s UBI net can fall below the minimum level, different households would receive different sums of money from the pilot or be required to pay different sums (in other words, they would have negative UBI net in the above formula), depending on their gross actual income and actual tax payments. In this way, the pilots would not be likely to see payments being transferred to the rich unless they were to stop receiving any income from other sources.

Ideally, UBI pilots could operate with minimal external funding to cover the research and not the transfer payments and in this way prove or disprove their viability. In practice, external funding will likely be required for two reasons: Though a successful UBI scheme would replace the welfare state in whole or in part, the subjects of a pilot will continue to pay into existing welfare systems. The taxes paid into the existing welfare system would have to be deducted from the taxes collected by the UBI scheme, and the value of the services provided by the existing welfare system would have to be deducted from UBI net income. But the overhead gap between bureaucratically managed welfare programs and a relatively low-overhead UBI program would still have to be covered. Secondly, randomly chosen participants who would have to pay into the scheme more than the net income they would receive might understandably be unenthusiastic about participating in a program that, in effect, will offer them little more than a guaranteed minimum income and the moral satisfaction of participation. Of course, it is also possible that the whole thing is unsustainable because it is too expensive, but that is what we need pilot programs to find out.

Pilot programs cannot answer all the interesting questions about UBI, especially on the macro-economic level. For example, how would they affect labor costs, or the global distribution of labor and production in a world where some labor markets would have UBI and others would not? Other interesting questions could only be answered over the very long term. Still, there is plenty to be learned from a well-designed UBI pilot that does not misinterpret the idea as yet another welfare entitlement program directed at means-tested volunteers; that is sufficiently funded; and that has statistically meaningful samplings. Without these things UBI pilots are little more than populist gimmicks.