What is a negative income tax?

A negative income tax (NIT) is a system where people who earn below a set threshold receive payments from the government instead of paying taxes into it. The word "negative" simply describes the direction the money flows: rather than income flowing up to the government, it flows down to the individual. Above the threshold, normal income taxes apply. Below it, the government makes up a portion of the gap. The result is a guaranteed income floor for everyone in the economy, one that phases out automatically as your earnings rise.

The idea is more than six decades old, but it has acquired new urgency. As AI systems automate an expanding range of cognitive and physical work, economists and policymakers are revisiting the question of how to maintain income floors in an economy where the relationship between labor and earnings may look fundamentally different within a generation. The NIT is one of the most seriously considered answers, and understanding how it works is a prerequisite for thinking clearly about that future.

Where the Idea Came From

The negative income tax is most closely associated with Milton Friedman, who proposed it in his 1962 book Capitalism and Freedom. Friedman was a libertarian economist, and his motivation was partly to replace the sprawling tangle of means-tested welfare programs (food stamps, housing subsidies, medical assistance) with a single, clean cash transfer. In his view, cash was more respectful of individual autonomy than in-kind benefits, more efficient to administer, and less prone to bureaucratic distortion.

What's striking about the idea's history is how broadly it was endorsed across ideological lines. Friedrich Hayek, Friedman's fellow free-market economist, supported a guaranteed minimum income. So did Martin Luther King Jr., who called for a version of it in his 1967 book Where Do We Go from Here: Chaos or Community?, writing that "the solution to poverty is to abolish it directly by a now widely discussed measure: the guaranteed income." President Nixon proposed a partial version called the Family Assistance Plan in 1969, which passed the House before dying in the Senate. Few policy ideas have attracted support simultaneously from Milton Friedman and Martin Luther King.

How It Actually Works

The mechanics are straightforward. You set two numbers: a threshold income and a phase-out rate.

Say the threshold is $30,000 and the phase-out rate is 50%. Here's what that looks like for different earners:

• You earn $0: you receive 50% × $30,000 = $15,000 from the government
• You earn $10,000: your gap is $20,000, so you receive 50% × $20,000 = $10,000
• You earn $20,000: gap is $10,000, you receive $5,000
• You earn $30,000: you hit the threshold, no payment either direction
• You earn $40,000: you pay taxes normally

The critical design feature is that you always benefit from earning more. If you go from earning nothing to earning $10,000, your total income goes from $15,000 to $20,000, you gained $5,000. Under some traditional welfare programs, earning more can actually leave you worse off because you lose benefits faster than you gain income. The NIT eliminates that trap by construction.

The exact numbers (threshold, phase-out rate, who qualifies) are policy choices that determine both the generosity and the cost of the program. There's no single "correct" NIT; different versions produce different tradeoffs between adequacy, work incentives, and fiscal cost.

How It Differs from Universal Basic Income

The NIT is often confused with Universal Basic Income (UBI), and the two ideas share similar goals, but they work differently.

UBI pays a flat amount to everyone, regardless of income. A billionaire receives the same payment as someone who earns nothing. The appeal is universality and simplicity, no means testing, no bureaucracy, no stigma. The drawback is cost: if you pay $1,000/month to every American adult, that's roughly $2.5 trillion per year before any other spending.

The NIT concentrates the same money on lower earners. Because payments phase out as income rises, a dollar of program spending does more targeted work. In mathematical terms, a UBI and an NIT with equivalent parameters are actually equivalent in their net effect on people's take-home income, the difference is mainly in how you account for the flow of money and who the program is politically visible to. In practice, though, an NIT is typically designed to be less expensive by targeting a narrower population.

Has It Been Tried?

The most direct tests came in a series of US government experiments in the 1970s. Four randomized controlled trials (in New Jersey, rural Iowa, Gary (Indiana), and Seattle and Denver) enrolled roughly 8,500 families and tracked their behavior over several years. The Seattle-Denver Income Maintenance Experiment (SIME/DIME) was the largest and most rigorous.

The headline finding was that recipients reduced their hours worked by a modest amount, estimates ranged from roughly 5–7% for primary earners. This was seized on by critics as evidence the NIT would reduce work effort. But the effects were concentrated among secondary earners (mainly wives) and teenagers, two groups whose reduced work hours weren't necessarily bad outcomes. Wives spending fewer hours at paid jobs to care for children, or teenagers staying in school longer, weren't obvious economic disasters.

Canada's Mincome experiment in Manitoba ran from 1974 to 1979. Its records were eventually archived and largely forgotten until economist Evelyn Forget reanalyzed them in 2011. Her findings: hospitalization rates in the test town dropped 8.5% during the experiment, high school completion rates rose, and mental health indicators improved. The experiment hadn't primarily been designed to measure these outcomes, which is why they were missed at the time.

No country has implemented a full national NIT, though the US Earned Income Tax Credit (EITC) functions as a partial version, it pays money to low-income workers and phases out as income rises. The EITC currently reaches about 23 million American families and is one of the largest anti-poverty programs in the country. It is also one of the least controversial, with support from both parties, partly because it is tied to work.

The Core Arguments For It

It eliminates the welfare trap. Traditional means-tested programs often create a cliff: earn too much and you lose your benefits, sometimes faster than you gain income. The NIT's gradual phase-out avoids this by design.

It replaces bureaucracy with cash. Friedman's original appeal was administrative efficiency. Rather than funding agencies to determine eligibility for dozens of separate programs, you run one calculation through the tax system. The government already has income data; the NIT uses it.

It preserves dignity. Cash transfers don't prescribe how recipients spend their money. Food stamps can only buy food. Housing vouchers can only pay for housing. Cash lets people allocate according to their actual priorities, which treats them as capable adults.

It's immediate. The day an NIT becomes law, there's a floor. Unlike investment-based approaches such as universal basic capital, which take time to generate returns, an NIT delivers income the moment someone's earnings fall below the threshold.

The Core Arguments Against It

Cost. A generous NIT (one that actually keeps people out of poverty) is expensive. Setting the threshold at the US federal poverty line for a family of four (~$32,000) with a 50% phase-out rate would cost hundreds of billions annually, depending on design. Finding that money requires either cutting other programs or raising taxes.

Political fragility. A government payment is only as durable as the political will to maintain it. Right now, people derive income from their own labor, a form of security that no administration can revoke. If labor income declines and most people's floor comes from a government program, they become vulnerable to whoever controls that program. This is a genuine concern that Friedman himself acknowledged, and one that has grown more pointed as income inequality has increased.

Work disincentives. The 1970s experiments showed modest reductions in work effort. The economic theory is clear: any guaranteed income reduces the urgency of working. Whether this is a bug or a feature depends on your values, but it's a real effect that policy designers have to account for.

It doesn't address wealth. The NIT addresses income, not assets. If the concern is that AI-generated wealth will increasingly accrue to capital owners rather than workers, an income-based solution may miss the fundamental shift. You can give someone a guaranteed income and still have a society where productive assets are concentrated in a very small number of hands.

Why This Is Getting Renewed Attention

The NIT sits in an unusual political position: it has theoretical support from economists across the spectrum, a reasonable empirical track record from the experiments of the 1970s, and essentially no active political momentum toward full implementation in any major economy. The closest living version (the EITC) is deliberately limited to workers, which reflects the political reality that unconditional cash transfers remain controversial even when the economic case for them is strong.

What has changed is the urgency of the underlying question. Economists studying the labor market effects of AI, including Alex Imas at Google DeepMind and Phil Trammell at Epoch AI, have noted that if AI automates substantial portions of knowledge work, the NIT's core feature becomes far more important than it has historically been: its immediacy. Unlike investment-based redistribution models, which require time to generate returns, an NIT delivers income the day it becomes law. In a scenario where displacement happens faster than new job categories emerge, that speed matters enormously.

The question of how to maintain income floors in an economy that may look very different in ten years is not primarily a technical one. The mechanics of an NIT are well understood. What's less understood (and what makes this a live question rather than a settled one) is whether democratic societies can build the political consensus to implement and sustain something like it before the need becomes acute, rather than after. That challenge is as much about values and governance as it is about economics. On what an NIT might look like in an AI economy, there is a great deal more to say.

Frequently Asked Questions