Impactism
日本語 Join Impactism
Manifesto The Impact Standard Foundations Objections Read For organizations 日本語
Join Impactism
A corner where two white walls meet a bare concrete floor
Photo by Julia Taubitz on Unsplash

Essay

Universal Basic Income: What the Results Show

September 24, 20269 min read

A universal basic income is money paid regularly to individuals with no work test and no conditions attached. Several serious experiments have now reported, from Finland, California, Kenya, Manitoba and two American states. The results are more consistent than the argument about them suggests. Unconditional cash improves health and mental health. It does not stop people working, though in the largest and longest study it did reduce work by a little over an hour a week.

Here is each one, with what it found and what it cost.

Finland, 2017 and 2018

Finland ran the cleanest design. Kela, the national social insurance institution, drew a random sample from everyone receiving an unemployment benefit in November 2016 and paid 2,000 of them a basic income of 560 euros a month, tax-free, for two years from 1 January 2017. The money kept coming whether or not they found work (Ministry of Social Affairs and Health 2020).

Over the twelve months to October 2018, recipients worked six days more than the control group, averaging 78 days of employment. The government called the employment effects small, and they were.

The other half of the result is the half that gets quoted less. Survey respondents on basic income were more satisfied with their lives, reported less mental strain, less depression, less sadness and less loneliness, and trusted other people and their institutions more than the control group did.

A randomised trial in a rich welfare state found that paying people unconditionally made them slightly more employed and considerably less miserable.

Stockton, California

Stockton’s mayor Michael Tubbs launched SEED in February 2019. It gave 125 residents 500 dollars a month for 24 months, unconditional, with no work requirement, evaluated as a randomised controlled trial by Stacia West of the University of Tennessee and Amy Castro Baker of the University of Pennsylvania (West & Castro Baker 2021).

In February 2019, 28% of recipients held full-time jobs. A year later, 40% did. Over the same year the control group went from 32% to 37%. The researchers were careful not to read a labour market trend into 125 people, and their qualitative work explains the mechanism: the cash let people drop a shift to sit an interview, finish a course, or apply for something better than the job they were surviving in.

On the Kessler 10, a standard screen for psychological distress, the treatment group improved and the control group did not.

Kenya

The largest trial is GiveDirectly’s in rural Kenya, running since 2018 across roughly 200 villages, studied by Tavneet Suri of MIT with Abhijit Banerjee, Alan Krueger, Paul Niehaus and Michael Faye. Villages were assigned to a long-term basic income of 22.50 dollars a month for twelve years, a short-term version of the same payment for two years, a single lump sum of 500 dollars, or nothing (GiveDirectly 2023).

Treated households did not work less. What changed was the composition. Hours of wage labour fell and hours of non-agricultural self-employment rose by slightly more. People stopped selling their labour to someone else and started running something of their own.

The lump sum beat the drip for starting a business, which is a design finding rather than a philosophical one, and it matters for anyone writing a real policy.

Two American states

The study that complicates the picture is also the best resourced. OpenResearch randomised 1,000 low-income people into receiving 1,000 dollars a month, unconditionally, for three years, against a control group of 2,000 receiving 50 dollars a month, in Texas and Illinois. The evaluation used surveys, administrative records and a phone app (Vivalt, Rhodes, Bartik, Broockman, Krause & Miller 2024).

Labour market participation fell 4.2 percentage points. Individual income excluding the transfer fell by about 1,900 dollars a year. Recipients worked around 1.3 fewer hours a week than the control group across the three years, and their partners cut hours by a comparable amount (OpenResearch).

That is a real cost and we are not going to soften it. Anyone arguing that a basic income is free of labour supply effects has to answer this study, and the answer is that the effects are modest, not absent.

The same data also carries the finding we find hardest to stop thinking about. Recipients were 6 percentage points more likely to be actively looking for work and 4.5 points more likely to have applied for something, while spending longer between jobs and sending fewer applications. They were choosier. And they were 5.5 percentage points more likely to say they were prioritising interesting or meaningful work.

Some quit to homeschool a child. Some went back to study. Some took time to deal with a health problem they had been carrying. Some rested.

Dauphin, Manitoba

The oldest evidence is Canadian. Mincome ran from 1974 to 1979, and in the town of Dauphin every household was eligible, which makes it the only saturation experiment anyone has run. The records sat unexamined for thirty years until Evelyn Forget went back to the provincial health administration data (Forget 2011).

Hospitalisation among participants fell 8.5% relative to controls, concentrated in accidents, injuries and mental health. Contacts with physicians for mental health complaints fell. More adolescents finished grade 12. Forget found no change in fertility, family break-up or birth outcomes.

This is quasi-experimental work on administrative data rather than a trial, and it is thirty-five years after the fact. It is also the only study in this list that looked at what happens to a whole community, and what it found was a town that went to hospital less.

What the five add up to

Cash does not make people stop working. It makes some of them work a bit less, mostly at the edges, and it makes many of them choose differently. Health improves, in self-report and in hospital admissions. Financial volatility falls. Trust rises. None of that is in serious dispute any more.

What remains unsolved is the money. None of these tested a full livable floor in a rich country, paid for out of general taxation, at national scale, for good. Finland paid 560 euros to unemployed people already inside a generous welfare state. Kenya paid 22.50 dollars a month in a place where that goes a long way. The American studies were philanthropically funded and time-limited, and a three-year payment is a different thing from a permanent right, because you cannot rebuild a life around something you know ends.

So we hold the floor as a possibility rather than a promise. We have written about why we think a society should build one before it needs one, and about how little the evidence lets anyone claim, in Start Before We Need It.

The idea is old and it is not radical in the way people assume. Japan wrote a version of it into its constitution in 1947. Article 25 gives everyone the right to maintain “the minimum standards of wholesome and cultured living” (Constitution of Japan). Cultured, not merely fed. The same pairing sits in international law, where an adequate standard of living and the right to take part in cultural life are neighbouring articles of the ICESCR (Arts. 11 and 15).

Nobody is proposing to abolish money here. A floor sits underneath the market and leaves everything stacked above it alone. Expensive things still cost money, and some people still end up much richer than others.

What none of them measured

Run down the outcome variables in all five studies. Employment. Hours. Earnings. Consumption. Food security. Hospital admissions. Depression scores. Trust.

Not one of them asked whether the money helped anybody live a life that mattered to other people.

That is not a criticism. Those are the outcomes a finance ministry will accept, and getting them measured was hard, careful work. But it means the entire evidence base for a floor is an argument about whether the floor is affordable and whether it breaks the labour market. It is silent on what the floor is for.

The single exception is that survey item from Texas and Illinois. People with a guaranteed 1,000 dollars a month became more likely to say they were prioritising meaningful work. It is self-reported, it is one question, and we would not build a policy on it. It is still the closest anyone has come to measuring the thing we actually care about.

A floor is not a yardstick

Money past a certain point buys much less than people expect, and we went through that research carefully in The Log Curve. A floor deals with the part of the curve where money does enormous work, which is exactly why it matters, and why the argument for it is strong even before anyone gets to meaning.

But security is not purpose. Keynes saw this in 1930 and it is the argument of What Keynes Promised Us: solve the economic problem and you do not hand people peace, you hand them a vacancy where the shape of their days used to be. A basic income solves the economic problem for one household. It answers nothing about what that household should do on Tuesday.

What a floor buys is 余裕, the room to breathe. A person with no savings and no idea how next month works cannot afford the question of what their life is for, because they are busy keeping it afloat. Give them the floor and the question becomes available to them. It does not become answered.

The answering is culture, and culture is not a transfer payment. It is what a society notices and repeats and admires, and at the moment it mostly admires money and lately attention. That can change, and changing it does not require a single line of fiscal legislation.

Build the floor, on the evidence, honestly stated. Then there is a second question, and the money never touches it.

Make impact, not money, the measure of a life.

Sources

← All essays

Impactism is free — no paywall, no membership fee, and it will stay that way. If this argument was worth your time, donations are what pay for it. One-time or monthly, any amount.

Support the argument