UBI vs. negative income tax vs. EITC: three ways to replace the minimum wage

Our argument against the minimum wage has a second half that critics are right to demand. If you take away the wage floor, what keeps a low-wage household's income up? Milton Friedman's answer in Capitalism and Freedom (1962) was a negative income tax. Today the debate has three main contenders: a universal basic income (UBI), a negative income tax (NIT), and an expanded Earned Income Tax Credit (EITC). All three put cash in people's hands through the public budget instead of through the price of an hour of labor. Where they differ is in who gets paid, how fast the help fades as earnings rise, and what that does to the incentive to work.

How each one works

Universal basic income. Everyone gets the same payment, say $1,000 a month, whatever they earn. Nothing is withdrawn as earnings rise, so the UBI itself has no phase-out. The money has to be raised somewhere, though, and the taxes that pay for it are where the marginal rates show up.

Negative income tax. A household with no income gets a guarantee, G. Each dollar earned reduces the payment by a fraction t, the phase-out rate. The payment reaches zero at the break-even income, G ÷ t. Friedman's own illustration used a 50% rate, so a worker always keeps at least half of each extra dollar. Our negative income tax calculator lets you set G and t yourself.

Earned Income Tax Credit. The EITC turns the NIT upside down at the bottom. It pays nothing at zero earnings, then adds to each dollar earned up to a plateau, then phases out. Under the IRS's 2026 parameters (Rev. Proc. 2025-32), a single parent with two children gets 40 cents per dollar earned up to $18,290 of earnings, which is the $7,316 maximum. The credit holds at that level until income reaches $23,890 and then falls by about 21 cents per dollar until it runs out at $58,629. For a worker with no children, the maximum is just $664.

The surprising part: a UBI with a flat tax is an NIT

UBI and NIT sound like rival philosophies: one universal, one means-tested. On paper, though, a UBI paid for with a flat income tax gives every household exactly the same budget as an NIT with the same guarantee and a phase-out rate equal to that tax rate. Take a $12,000-a-year UBI funded by a 50% flat tax on earnings. Compare it with an NIT that has a $12,000 guarantee and a 50% rate, in which households above the $24,000 break-even pay 50% on each dollar above it.

The formula is the same in both columns: net income = earnings + $12,000 − 0.5 × earnings. Every household ends up in the same place, and every household faces the same 50% marginal rate. The only difference is the gross flows. The UBI sends a $12,000 check to the $40,000 earner and takes back $20,000. The NIT nets the two and bills her $8,000. That makes the UBI look far more expensive on a budget table, even though the net transfers are identical.

Three caveats keep this from being a mere accounting trick. First, real UBI proposals are rarely financed this way. Most layer the payment on top of the existing tax schedule, so they are not equivalent to any particular NIT. Second, timing differs. A UBI arrives every month by default, while an NIT run through the annual tax return pays after the fact unless it makes advance payments, and advance payments bring back the reconciliation problems a UBI avoids. Third, the label shapes the politics. A universal check may be harder to cut and easier to defend than a means-tested one, or it may be easier to attack as paying the comfortable. Any honest comparison has to separate the budget arithmetic, which can be identical, from the delivery and politics, which aren't.

Side by side

The trade-off at the heart of this table is sometimes called the iron triangle: a generous guarantee, a low phase-out rate and a contained cost. You can have any two. A UBI gives up containment, at least in gross terms. An NIT has to choose a point on the triangle explicitly. The EITC dodges the trade-off at the bottom by offering no guarantee at all. That makes it cheap and pro-work, and it leaves out people with no earnings.

What the evidence says

The 1970s negative income tax experiments

Between 1968 and 1982, the federal government ran four randomized income-maintenance experiments, in New Jersey and Pennsylvania, in rural Iowa and North Carolina, in Gary, Indiana, and in Seattle and Denver. As we summarize in the calculator's explainer, work effort fell but did not collapse. Reductions were modest for primary earners and larger for second earners and young people. More generous designs were generally associated with larger responses. Robert Moffitt's 2003 review in the Journal of Economic Perspectives adds a subtler point. Lowering the phase-out rate does not automatically raise work. It extends eligibility to families further up the income scale, and some of them cut their hours. The overall effect on labor supply can go either way.

OpenResearch: $1,000 a month for three years

The largest modern test is the OpenResearch Unconditional Cash Study, analyzed by Eva Vivalt, Elizabeth Rhodes, Alexander Bartik, David Broockman, Patrick Krause and Sarah Miller in NBER Working Paper 32719. Starting in November 2020, 1,000 people aged 21 to 40 with household incomes up to 300% of the federal poverty level, in northern Illinois and north-central Texas, received $1,000 a month for three years. A control group of 2,000 received $50 a month. The transfer was about a 40% boost to the average participant's household income.

The July 2024 version of the paper found that income excluding the transfers fell by about $1,500 a year relative to the control group. Labor force participation fell by 2.0 percentage points, and participants worked about 1.3 to 1.4 fewer hours a week. Their partners cut hours by a similar amount. Leisure was the largest gain in time use. The authors found no improvement in job quality, and their confidence intervals rule out even small improvements. They conclude that the labor supply reduction is moderate and "does not appear offset by other productive activities." Later revisions of the paper, the latest dated August 2026, report somewhat larger effects: about $1,900 a year in lost non-transfer income and a 4.2-point drop in participation.

This is the cleanest evidence we have on a UBI-style payment in the United States. It does not show people quitting en masse. It does show that unconditional cash buys some reduction in work, mostly taken as leisure rather than training or a better job search.

Stockton's SEED

The Stockton Economic Empowerment Demonstration paid 125 residents $500 a month, unconditionally, starting in February 2019. Its first-year analysis covered the pre-COVID period through February 2020. Full-time employment among recipients rose from 28% to 40% that year, compared with 32% to 37% for a 200-person control group. The recipients also reported less depression and anxiety. The evaluators caution that the findings "cannot point to larger labor market trends." The core research sample was 100 people and the employment measures were self-reported. SEED is encouraging, but it is too small to set against the OpenResearch results.

Alaska's Permanent Fund Dividend

Alaska has paid every resident an annual dividend since 1982, and the amount varies from year to year. It is the closest thing in the United States to a permanent, universal cash payment. Damon Jones and Ioana Marinescu (American Economic Journal: Economic Policy, 2022) compare Alaska with a synthetic control built from other states. They find no effect on employment, and a 1.8-percentage-point (17%) increase in part-time work. Their explanation is a general-equilibrium one: the dividend boosts local spending, which supports demand for workers in sectors that sell locally. That is a genuine point for UBI advocates. It also comes with limits. The dividend is far smaller than a living-income UBI would be, and an aggregate state-level comparison cannot tell us how a much larger, tax-financed national payment would behave.

The EITC

The EITC evidence points the other way, toward more work. Nada Eissa and Jeffrey Liebman (Quarterly Journal of Economics, 1996) studied the Tax Reform Act of 1986, which expanded the credit. Comparing 1984–86 with 1988–89, single women with children increased their labor force participation by 1.4 percentage points relative to single women without children, from a base of 73.1%. They found no effect on the hours of mothers who were already working.

V. Joseph Hotz and John Karl Scholz's survey of the EITC concludes that the credit consistently raises labor force participation, with smaller negative effects on hours for some people already working, particularly secondary earners. Their later work with Charles Mullin (NBER Working Paper 11968) matched California welfare, unemployment-insurance and tax records. It found that the 1990s expansions, which favored families with two or more children, raised their employment by as much as 3.4 percentage points relative to one-child families.

The EITC has known weaknesses. About one in five eligible households doesn't claim it. Jesse Rothstein's incidence analysis finds that part of its value passes to employers through lower wages, and that an NIT does better on this score. We cover both in Minimum wage vs. EITC and Living wage vs. minimum wage.

Where we come down

We prefer an EITC-style or NIT-style earnings supplement over a UBI as the replacement for the minimum wage. Here is why.

• It does the minimum wage's job without the minimum wage's damage. A wage floor tries to raise the income of low-wage workers by raising what an employer must pay for an hour of work. An earnings supplement raises what the worker takes home for that hour, so the cost of hiring doesn't rise. The entry-level jobs that Friedman worried a wage floor would destroy stay open. That is the core of Friedman's argument.
• It puts the money where the need is. A UBI spends most of its gross budget on households that don't need it, then claws it back through taxes. A supplement aimed at low earners delivers the same help to the bottom of the distribution at a fraction of the gross cost, which leaves room to make it generous.
• The work evidence favors it. The EITC literature consistently finds higher participation. The best modern UBI-style trial finds moderate reductions in work that aren't offset by training or better jobs. If the policy goal is to replace a wage floor, which exists to support working people, a design that pulls people into work fits better than one that pays them to step back.

The UBI case has real force on two points, and they shape what we would build. First, the EITC pays nothing to someone with no earnings, so a pure earnings supplement leaves people who can't work, or can't find work, with no help. Second, the EITC's take-up gap and once-a-year delivery are real failings. Our answer is a hybrid. Use an EITC-style phase-in and a much stronger credit for workers without children, who now get almost nothing. Pair it with a modest NIT-style floor for those with no earnings, with a phase-out rate kept well below 50% so that work always pays. And deliver it through the year, not just at tax time.

Every one of these designs runs into the iron triangle, and none of them is free. But all three are honest about what they cost, because the cost shows up in the budget instead of being hidden in lost jobs. That alone makes any of them a better way to support low earners than a minimum wage. Try your own numbers in the negative income tax calculator.