Did Sam Altman’s Basic Income Experiment Succeed or Fail?

An article to better inform the discussion of the results of the OpenResearch unconditional basic income pilot experiment

21 min readAug 2, 2024

The results of one of the biggest basic income experiments ever came out in July 2024, and as usual, the nuances of the findings are lost among the voices of those loudly proclaiming basic income doesn’t work. This one is the three-year pilot of Sam Altman’s that provided $1,000 a month to 1,000 people in Texas and Illinois and compared that group to a control group of 2,000 people who got $50 a month. Every participant was between the ages of 21 and 40. In this article, I will explain the nuances and how the results of this pilot provide some new info but mostly replicate the findings of previous experiments going back to the 1970s and only further demonstrate that what’s at stake here is real freedom and the perceived danger it poses to those who benefit from the widespread lack of that freedom.

Before we get into the results, for those who aren’t already familiar with prior findings, here’s the conclusion of a peer-reviewed 2020 study of 38 studies:

“Despite a detailed search, we have not found any evidence of a significant reduction in labor supply. Instead, we found evidence that labor supply increases globally among adults, men and women, young and old, and the existence of some insignificant and functional reductions to the system such as a decrease in workers from the following categories: Children, the elderly, the sick, those with disabilities, women with young children to look after, or young people who continued studying. These reductions do not reduce the overall supply since it is largely offset by increased supply from other members of the community.”

It’s important to understand that the above conclusion includes what are called “saturation” pilots where entire communities receive basic income instead of only individuals spread across a large area. When basic income is provided to people here and there, local economies aren’t stimulated by the spending of the money and new jobs aren’t created by employers needing to hire more employees to meet higher demand. It’s one thing to provide money to an entrepreneur. It’s another to do that and also provide their business lots of customers with money to spend.

Here’s another important thing to consider: imagine a group of one hundred people where half of them slap you twice and the other half don’t slap you at all. The average is one slap per person. Is it fair to treat everyone as if each of them slapped you because the average was one per person? Hold on to that thought.

With all that covered in advance, here are the headline findings from this huge and very impressively ran and researched two-state basic income pilot, and what I believe is the proper context to understand these results as a UBI expert.

Employment can increase or decrease along two measures: the binary state of working a job or not and the number of hours worked. On average, those who got basic income were two percentage points less likely to be employed and worked about 1.3 fewer hours per week. These two numbers are basically where many UBI opponents and skeptics claim with triumph that this pilot was a failure. But let’s go deeper to gain a better understanding of what these numbers tell us.

The Context

First, it’s important we recognize the starting point of this pilot. The payments started getting received in late 2020 when unemployment due to the pandemic was right around its peak. The control group started slightly more employed than the treatment group, and both groups greatly increased their employment. By the end of the three years in late 2023, the control group remained slightly above the treatment group. By any measure, it’s clear just from looking at the chart that basic income did not reduce the employment of people in any way that should be considered a cautionary tale. The employment of both groups greatly increased.

A weekly drop of 1.3 hours works out to about 15 minutes a workday. That’s an extra break or a slightly longer lunch. On an annual basis, it’s equivalent to 8 days a year. That’s a week-long paid vacation. Now consider how the United States compares to the rest of the OECD in regards to paid holidays and paid vacation. An increase of 8 days would still leave the US at the bottom of all OECD nations, still behind Japan by two days. The difference between us and France would be 22 days instead of 30. Does 8 days seem like an economy-destroying bump in leisure time?

With that context in mind, now again remember the results from previous experiments, particularly how employment effects from basic income tend to be concentrated among parents and young adults. Did we see that here too? Indeed we did. It’s also the only places we observed it. The use of an average spread across all participants — just as the earlier thought experiment about one slap per person — hides the reality that there were no significant decreases in employment status and hours worked among childless adults or those over age 30.

From the report:

“Recipients who were single parents at the time of enrollment were about 3.9 percentage points less likely to be employed and worked an average of 2.8 hours less per week than single parent control participants. For recipients who were not single parents at enrollment, we do not find statistically significant effects on employment or hours worked.”

The fact that those with kids responded differently than those without kids mirrors previous findings going all the way back to the 1970s American Income Maintenance Experiments in locations like New Jersey, Indiana, Seattle, and Denver, and also the 5-year MINCOME experiment in Canada. The reason that parents respond differently should be obvious. They aren’t working less. They are switching from paid work to unpaid work. They’re putting their kids first. If this is a response you personally dislike, then you should support affordable child care in addition to basic income to give parents a different choice than to earn less from employment than they pay in the child care that enables employment. If it costs $1,200 a month to earn anything under or barely above $1,200 a month, a job does not make sense.

Next, remember that everyone in this pilot was older than 21 and younger than 40 and that we know from previous basic income experiments that young adults tend to choose more education and less employment.

“There was no statistically significant effect on employment or hours worked for recipients over 30. In contrast, recipients under 30 were roughly 4 percentage points less likely to be employed and worked an average of 1.8 fewer hours per week compared to control participants. We also observe larger effects on formal education among those in this age group, suggesting younger adults may be more likely to use the money to enroll in post-secondary education and work fewer hours while in school, though this alone would not account for the observed differences in employment.”

Those between the ages of 30 and 40 did not see a change in employment status or hours worked. The decrease was only among those under age 30 and a lot of it — not all of it — was due to a switch from paid work to the unpaid work of learning. It reminds me of the Dauphin experiment and how so many kids went back to school to finish that graduation rates exceeded 100% due to dropouts coming back. Is this an effect we don’t want to see? I certainly don’t think so.

So to what degree did this pilot show increases in education and training?

“Recipients were significantly more likely (3.3 percentage points) to have pursued education or job training during the final year of the program-an increase of 14% relative to the average control participant. The effect was greatest for recipients who had the lowest household income at enrollment: these individuals were, on average, 34% more likely to have participated in education or job training during the third year of the program compared to control participants.”

The largest impact on education was seen in those with the lowest incomes. This also wasn’t the only impact to vary by income. By dividing the participants into three categories: (A) those under 100% of the federal poverty line, (B) those between 100% and 200%, and © those between 200% and 300%, it was observed that the bulk of the labor decrease was in the highest income group.

“[Group C] recipients were 4.4 percentage points less likely to be employed and the effect on the hours they and their partners jointly worked was, on average, 3.7 hours less per week. In contrast, the average reduction in joint work hours was -2.1 for the lowest income recipients and -2.8 for the middle income recipients.”

So the higher someone’s income from work, the more likely they were to reduce their time working. Those with the lowest incomes were the least likely to decrease their time working.

The thing about UBI that many people don’t immediately understand is that the net income boost after taxes decreases with income. Yes, everyone gets the same amount of UBI, but everyone pays a different total amount of increased taxes. This pilot didn’t model anything tax related. In reality, those in Group C would have received a smaller income boost after taxes than Groups A and B, and Group B would have received a larger boost than Group C but a smaller boost than Group A.

It all depends on the details of a UBI plan, but to help understand this, just subtract 10% off of earned income. In this case, that would mean Group C getting boosted by $687/mo, not $1,000 a month, and thus the labor impact would have been smaller. Group B would also have been boosted less, netting perhaps about $812/mo, not $1,000 a month.

Someone who knows their economics could then argue that because no tax impacts were modeled, that we should also expect a larger labor impact due to tax rate increases, which is entirely true. However, the higher someone’s income, the less of an impact marginal tax rate increases have on their income. Also, this pilot fully maintained all welfare benefits. In reality, someone would likely get UBI instead of various welfare benefits, and thus face a lower marginal tax rate, not a higher one at the low end of the income spectrum.

The fact that all welfare programs were additional to the UBI is also something to consider as potentially having an impact on both the treatment group and control group. If you are getting $3,000 a month in housing assistance, it does not make sense to increase your income, nor does it make sense to increase your income when receiving a program like SSI that imposes a 50% marginal tax rate and when if lost can mean years of waiting to get back.

Finland’s basic income pilot had this issue too, where people got basic income, but they also got welfare, and even though Finland’s pilot showed a slight increase in days worked per year, it was traditional welfare that kept them from employment despite their basic income, because their welfare was much larger and losable.

Entrepreneurship

Another important impact on employment of basic income is how much it increases entrepreneurship. Based on previous pilots, we know that this impact is kind of huge in developing economies. In Namibia, UBI led to a 301% boost. The India UBI pilot observed three times the entrepreneurship in treatment villages than control villages. The impact on self-employment is smaller in developed economies where there’s greater access to capital, but it varies based on that access. For example, some unpublished polling I was privy to back in 2016 showed that 3% of white respondents said they’d start a new business with UBI compared to 30% of Black respondents who said they would. Clearly there’s less access to capital in the Black community and a lot more business ideas that never see the light of day.

So did entrepreneurship increase in this pilot? Not for everyone on average, but entrepreneurship did increase significantly for Black recipients and women.

“In the third year of the program, Black recipients were 9 percentage points more likely to report ever starting or helping to start a business-a 26% increase from the average for Black control participants.”
“In the third year of the program, female recipients were 5 percentage points more likely to report ever starting or helping to start a business-a 15% increase from the average female control participant.

It’s also interesting to note how the intention to start a business increased year after year for everyone, and by the end of it, recipients were 5 percentage points more likely on average than the control group to want to start a business.

Job Searching

Another fascinating difference between the treatment group and the control group was how each of them went about looking for jobs. Part of the reason for the treatment group being two percentage points more likely to be unemployed on average was because they spent one more month on average looking for a job. Were they being lazy? No. They were being choosy. How do we know that? Because recipients were 5.5 percentage points more likely to indicate that “interesting or meaningful work” was an “essential condition for any job they would accept,” and recipients were also 6 percentage points more likely to be actively searching for a job and 4.5 percentage points more likely to have applied for a job.

These numbers paint a clear picture of basic income recipients highly valuing employment and wanting to find the best job they could. They were the ones doing more job hunting and applying. The control group was just choosing whatever they could find. That’s not best for overall productivity.

Job Creation

Another thing we absolutely have to cover is the difference between a pilot that provides cash to people here and there spread across a large geographic area and one where everyone in an area gets money. The former doesn’t create new jobs unless the recipients create them themselves. The latter does create new jobs and in doing so helps job-seekers find employment.

The best example of this is in Alaska where a small annual UBI has existed since 1982 where for over four decades everyone has received on average $1,500 a year. Researchers curious about the impact on employment concluded that there was a small decrease in full-time employment but that it was counteracted by the spending of the money creating new jobs that resulted in more employment. The net result was full-time employment didn’t go up or down. There was however a 17% increase in part-time employment so that overall employment went up.

Now, imagine that a basic income pilot took place in Alaska and provided $1,500 a year for three years to 1,000 Alaskans spread out all over Alaska. What if it found a small decrease in employment? Would it be accurate to assume that if extended to everyone in Alaska that employment would go down? No, because we already know the opposite is true due to the impacts of the spending of the money on job creation. We only know this because Alaska already has actual UBI.

The qualitative data reflects this too. One of Altman’s pilot participants shared that “between her unreliable transportation, and living in a rural area removed from job opportunities, she struggled to find employment.” Had her entire community been part of the pilot, a new job may have been created for her to find and take.

This is also why there needs to be more saturation site pilots. There’s definitely things we can learn from all the pilots going on, but there’s also a lot we can’t possibly learn about UBI without providing to entire communities. For example, a recent saturation pilot in India where an entire community received UBI resulted in the community forming a labor union. That would not have happened if only one person in that community got basic income.

Job Quality

One of the findings considered most disappointing about this pilot was how researchers expected to see the treatment group find higher quality jobs and just didn’t observe that as measured in a variety of different ways. I admit I expected to see that reflected in the quantitative data too, but it is there in the qualitative data, and it’s pointed to in the geographic mobility results.

Here’s Lisa’s story:

“Lisa is a single mother to three living in Texas. At the start of program, she had no income. Lisa has lupus which kept her out of the workforce at the time, and her short-term disability had been cut off. At the time, her goal was to return to the workforce. Because of the $1,000 per month, Lisa is able to take a job making less money than at her previous job, but with a company that offered more opportunity for growth. Two years later, Lisa is in a salaried position, making over $75k, has gotten two promotions, and thinks she can get another within a year. She loves her job… Lisa was also able to leave her abusive boyfriend and move into her own place, which wouldn’t be possible without the unconditional cash transfers. Her three sons are thriving.”

Lisa used her basic income to get a “worse” job that was actually a better job.

Geographic Mobility

In regards to geographic mobility, another participant named Nikisha reported feeling that “if she had a job her life might be different, but she lives in a rural area and the jobs close by do not pay enough to offset the cost of childcare,” and that “better paying jobs are an hour and a half commute one way.”

Nikisha didn’t move but many UBI recipients did, likely in the hunt for better jobs even if they didn’t end up finding them. Recipients were, on average, 16% more likely to actively search for new housing and 23% more likely to actively search for a new neighborhood than the average control participant. They were also 4.4 percentage points more likely tomove to new neighborhoods.

Again consider actual UBI where Main Street USA sees a lot more consumer spending and more people are able to move to where the jobs they want are. It seems likely that more people would be happier with their jobs than right now.

Drinking and Drug Use

Another key result in Altman’s pilot was usage of drugs and alcohol. It’s very common for people to assume that basic income would lead to more abuse of drugs and alcohol, despite all the existing evidence to the contrary, but this pilot showed very significant positive effects for men especially, who also happen to be the ones most negatively impacted by the status quo.

For male recipients of basic income there was a 41% decrease of being under the influence in dangerous situations, a 45% decrease in drinking interfering with responsibilities, a 35% decrease in drinking causing arguments with others, and a whopping 81% decrease in days using painkillers not prescribed to them.

Considering all the deaths of despair going on as a direct result of alcohol and painkillers, it’s impossible for me to look at numbers like that as anything but a prescription for basic income for better health.

Health Effects

In regards to other health impacts observed in Altman’s pilot, researchers expected more than they found (possibly due to the overwhelming stress of the pandemic followed by significantly higher costs of living) but what they did find was more visits to the dentist and doctors, pointing to more preventative treatment leading to health savings in the future. However, it’s important to note again that the age group was 21 to 40.

Another study of another basic income program was published the same day as the initial studies of Altman’s pilot, and it found hugely positive health impacts. Basic income recipients in Chelsea were a third less likely to need to visit the ER. My hypothesis is that this comes down to age. The average age in Chelsea was 45. It’s likely that just as employment effects vary by age that health effects also vary by age. I should also mention that Dauphin’s huge 5-year saturation pilot found a 9% decrease in hospitalization rates.

Spending Behavior

How people spend their basic income is probably the least interesting thing to me, because we all need to spend money on our basic needs and all the pilots always reflect that, but two things I do want to highlight about what Altman’s pilot found is how the bank savings of basic income recipients grew by 25% and that spending on others increased by 26%. That’s how much more they spent on gifts to friends and family, loans to others, donations to charity, and alimony payments.

On average, recipients spent $310 more per month and most of that was on food, rent, and transportation. They also spent more on child care, which we know is something that better enables employment. Perhaps most importantly though, it was the variety of ways in which recipients spent their money that supports the idea of universal basic income.

“Recipients expressed hundreds of different pressing needs, ranging from a father’s funeral costs, back surgery, diapers for a newborn, car repairs, fines for speeding tickets, unpaid school tuition, and a computer and internet service for distance learning, to impending foreclosures, phone and electricity shutoffs, and money to renew a professional license in order to be able to get a job. In that moment, cash was the only tool that provided the flexibility to address those diverse needs, and cash was responsive when those needs shifted based on recipients’ circumstances and the unique challenges posed by the pandemic and the larger economic context.”

As I frequently stress, cash can be anything. There is no good or service that government can provide that can meet everyone’s needs at all times of need. UBI is a skeleton key to unlock any door, and it’s always there in advance of needing to unlock the door. No applications necessary. No need to prove you really need to open that door, or that the door is important enough to provide help to open.

Basic income is basic freedom.

Real Freedom

I recommend reading all of the qualitative stories to get a better idea of just what basic income really meant to the people in this pilot, but I want to leave you dear reader with this one above all — Lila’s story:

“Four years before the program started, Lila was the victim of a horrific domestic violence attack. ‘I got shot three times in my head, stabbed in the neck. I was completely paralyzed on the right side, totally blind. Um, some of it has come back and I’ve learned to like walk and stuff again, but that has definitely like affected my life so much.’ As a result, she has PTSD, partial blindness and multiple complications, impairing her ability to drive, and making it difficult for her to work or live independently. She applied for disability benefits but was repeatedly denied. When she learns she will be receiving $1,000 a month felt like it was replacing ‘the disability that the government keeps refusing me.’ Unable to work, she feels if it were not for the program, ‘I probably would be homeless without it.’”

Consider Lila’s story when you think of the people in Altman’s pilot who were unemployed and not technically defined by the government as unable to work.

Consider Lila’s story when you think of just how much it can mean to people to escape their abusers, and how without basic income they remain trapped.

If you look at Lila’s employment status data and see an unemployed person using their basic income to be lazy, you aren’t really looking at Lila. You aren’t looking at just how massively our non-universal and conditional safety net is failing people. And you aren’t seeing basic income as the tool of mass emancipation that it is.

Yeah, when provided basic income, people can work less, because they have the freedom to do so. They gain the power to say no to abuse — at home and at work. They gain the freedom to start a business or be a stay-at-home parent. They gain the agency to choose to go back to school or to learn a new skill. They gain the liberty to seek a better job, be it in town or in another. They gain the autonomy to purchase what they themselves decide they need most when they most need it.

Basic income is basic freedom of choice.

Future Basic Income Studies

There will be more reports written about Altman’s study. I’m personally curious to see if engagement in crimes among recipients went down, or if recidivism went down, but I also don’t expect any experiment designed like this to show as big of impacts along these lines than saturation pilots that include entire communities.

What I do expect to continue seeing is study after study showing all kinds of positive results, most especially for those under the poverty line where basic income makes the biggest difference. One of those studies has already been published since the release of Altman’s pilot results.

Another one of the biggest experiments in the US, where over three times as many people received $1,000 a month than in Altman’s pilot, and where the control group was also almost three times larger, found that providing a basic income to impoverished adults with kids in Los Angeles led to across the board improvements. The basic income recipients were more likely to secure full-time employment than to remain unemployed compared to those in the control group.

“They had better employment outcomes, felt safer at home, enrolled their kids in more extracurriculars, and were more likely to leave domestically abusive situations… Within the first 6 months of the program, participants left situations of intimate partner violence more frequently than the control group. In the last 6 months, participants were more likely to establish proximate safety and work toward protecting themselves in the future.”

Clearly, basic income is particularly helpful for those in poverty and helps connect them to jobs they want to take and empowers them to refuse abuse. It is this escape from abuse that is seen over and over again in the qualitative data of Altman’s pilot and it is seen again in the Los Angeles pilot and many other pilots too. In Kenya, physical abuse of domestic partners dropped in half.

Conclusion

Altman’s pilot was an interesting experiment, but as with other pilots, the most interesting stuff to glean requires the nuance of looking closer, looking inside the averages, and considering the total context of all pilots that came before it and are still to come.

It is inside the averages that we see no reductions in work among those who didn’t have kids or who were over the age of 30. Employment reductions were only significant for single parents who cared more for their kids or those between age 21 and 30, many of whom went to college or job training.

The impact in the above groups works out to about 17 days less per year. And in that context the US is the only OECD nation with zero days of paid vacation whereas every other OECD nation has at least 20 days of paid vacation except for Japan and Canada. And don’t only consider this in the context of paid vacation, but also paid parental leave. If a mom of a newborn reduced their employment by a year by using her basic income as paid maternity leave, that would drag up the average unemployment for all parents. The average paid parental leave policy duration in the OECD is 12 months. The US is zero months.

Entrepreneurship also significantly increased for women and Black participants. Drug and alchohol abuse decreased, especially for men, and people moved to places they preferred living. Many women escaped abusive relationships.

The fear is that basic income would mean dramatic decreases in employment, but none of the data supports that fear, not even Altman’s pilot that showed a slight decrease for parents and young adults. Clearly, all of the resources put into making sure that people are working or seeking work is pointless. Clearly, the means-testing that excludes such a huge percentage of people in poverty and causes so much stigma is needlessly harmful. Clearly, all of the administrative overhead going into avoiding just sending people money to spend on anything instead of insisting on anything but cash, is wasteful. So why are we wasting so much money on the paychecks of middlemen bureaucrats instead of just sending that money directly to people to spend in their local economies?

The way we go about the existing safety net is harmful, wasteful, and expensive. It leaves out too many people who fail the tests and it effectively punishes people for increasing their incomes when it’s pulled away as a consequence. And the fact almost everything we do isn’t cash limits the usefulness of it to everyone who simply just needs cash plus the knowledge that cash will always be there next month too. With UBI we can simply prevent any need for more help. A boost of just $1,000 when someone is on the brink of homelessness can reduce it by 88%. We are spending so much more than that treating the results of homelessness.

Here’s one final thought experiment to consider:

If we had done an experiment in a country with UBI, and tested getting rid of it, and the results were that employment didn’t change for those without kids and those over 30, but employment did increase by 17 days a year for those with kids and those under age 30, resulting in parents spending less time with their kids and young adults becoming less educated and skilled, and fewer women and Black individuals becoming entrepreneurs, and men increasing their abuse of women, alcohol, and painkillers, would we consider that a success?

Would we say it demonstrated that keeping UBI was a bad idea?

Want more content like this? Please share it and click the subscribe button

Special thanks to my monthly supporters: Gisele Huff, Haroon Mokhtarzada, Steven Grimm, Judith Bliss, Lowell Aronoff, Katie Moussouris, David Ruark, Tricia Garrett, Zack Sargent, A.W.R., Daryl Smith, Larry Cohen, Keith Smith, Philip Rosedale, Liya Brook, Frederick Weber, John Steinberger, Bridget I Flynn, Laurel gillespie, Dylan Hirsch-Shell, Tom Cooper, Robert Collins, Joanna Zarach, ace bailey, Daragh Ward, Andrew Yang, Peter T Knight, Michael Finney, David Ihnen, Miki Phagan, Albert Daniel Brockman, Natalie Foster, Joe Ballou, Arjun , Christopher Wroth, S, Jocelyn Hockings, Kara Gillies, Faith Stanhope, Mark Donovan, Capitalists for Shared Income, Jason Clark, Chuck Cordes, Thomas Fitzsimmons, Mark Broadgate, Leslie Kausch, Jessica Chew, Braden Ferrin , Juro Antal, Austin Begin, Deanna McHugh, Nikolaus Rath, chris heinz, Zachary Weaver, Jodi Sarda, Rosa Tran, Ryan Ash-Miller, bradzone, John Sullivan, Team TJ, Yang Deng, Yan Xie, Marie janicke, Tim , Warren J Polk, Jeffrey Emmett, Stephen Castro-Starkey, Kev Roberts, Nicolas Pouillard, Walter Schaerer, Eric Skiff, Thomas Welsh, and Laura Ashby. . Also consider making a monthly pledge in support of all my work.