A Way Forward, Together

Paying for It

By AFK
First edition, October 26, 2026

The short version

One of our principles is that every idea has to say what it costs, who pays and what we give up. This is where we do that for the whole plan, in one budget. It covers what each commitment costs, where every dollar comes from, the rule that keeps it honest, and what it means for families and for the national debt. Nothing is counted twice.

How it connects. The goals behind every line are in A Way Forward, Together, and each commitment is laid out in depth in the commitments in full. Healthcare has its own trust fund and its own plan, Healthcare for Everyone. How people and markets are likely to react to each tax is worked through in Following the chain.

All figures are per year and in 2026 dollars after inflation. Most are our own estimates, and they need an independent model before anyone should lean on them.

Where we're starting

So we start out $2 trillion a year in the hole. That's why every new program has to be paid for with real revenue. It's why savings only count once they actually show up. And it's why the debt gets its own section below.

What it costs

These are the new commitments that need money, and what each one costs once it's fully running. The numbers in parentheses are the commitments they belong to.

In all: the services (everything except the basic income, healthcare and leave) come to about $0.5 to $1.0 trillion, or about $0.6 trillion if you take the middle of each range. Leaving out healthcare, new yearly spending is about $2.1 trillion at launch, while the services are just getting started. It rises to about $4 trillion once everything is at full size. Healthcare's $2.5 trillion runs through its own trust fund and mostly replaces what people pay now.

One-time costs of the switch

These come once, mostly between 2027 and 2032. They're easy to forget because they don't repeat.

Healthcare: its own trust fund

The full plan is in Healthcare for Everyone. Everyone is covered in two waves, aimed at January 2029 and January 2030, and each wave starts only when it passes its readiness tests. Everything medically needed is paid in full. Here's the short of it, using our central estimate:

How we pay for it

The money comes in four layers. Each one has a job, and each is matched to what it pays for.

Layer 1: fairer taxes at the top

These ask more of the people and companies that have gained the most. None of them tax paper gains.

Layer 2: a broad but modest tax on spending

Layer 3: a public share of AI and automation

Most federal money comes from taxing work. If AI does more of the work, that base shrinks just when the need for a basic income grows. This layer lets the public share in the gains.

Layer 4: sources that do one job each

Savings: upside we don't count

We've said we'd show what happens if savings come in lower than hoped. These are real targets with real plans behind them, but none of them counts as money until it shows up in the books.

Where each dollar goes

The rule that keeps it honest

The plan puts programs first. We'd rather borrow in the open in the early years than make families wait. But that borrowing has a limit, and the rules are written into law. An independent office (Treasury's actuaries, checked by the Congressional Budget Office) applies them every year.

What happens, depending on AI

Nobody knows how much AI will add to the economy, so we tested the plan against four cases. These count everything except healthcare, which runs through its own trust fund. That means today's budget, the basic income, the services, the ordinary taxes from AI-driven growth, the one-time costs of the switch and the allowance for stacked taxes at the top. On today's path, with no plan at all, the national debt goes from about 106% of GDP in 2029 to about 144% by 2046 (based on CBO's baseline).

If you leave out the ordinary taxes from AI growth, as a cautious reader might, every date moves several years later.

Year by year. Here's where each case stands every few years: the adult amount, the VAT, how much of the services are in, the plan's own added debt and the national debt, both as a share of the economy.

The honest gap.

What it means for households

Here we add up every layer for four example households, in the first year with both the basic income and full healthcare. That's the basic income and the taxes that pay for it, the health contribution against today's premiums and out-of-pocket costs, the carbon fee and its checks, the worker's half of the leave premium and Social Security above the cap. "Before raises" leaves out the money employers save on premiums. "After raises" assumes employers pass those savings on as pay, which is what most economists expect within a few years, and counts the taxes on that pay.

In the slow-ramp case (adults at $500 and a 22% VAT in the mid-2030s), the $80,000 family is still about $20,100 ahead after raises, the $60,000 single worker is about even and the $150,000 couple is about $2,700 behind.

What it tells us.

Prices: a 10% VAT raises prices once, by roughly 5%. If the backstop raises the VAT, prices go up again. Social Security and other benefits tied to prices are protected, and the plan pays for that.

All the taxes, added up

A combined marginal rate is how much of one extra dollar of pay goes to taxes of every kind. That's federal and state income tax, Social Security and Medicare, the health contribution, the employer's payroll taxes (which come out of what a job can pay) and the VAT on whatever the dollar buys. It's what people feel when they decide whether to pick up an extra shift or take a raise. These are for a single worker in Texas (no state income tax) and in California. Each line shows today, then the plan with a 10% VAT, then with a 22% VAT. They leave out state and local sales taxes, which apply either way.

What drives it in the middle. For a $60,000 worker, the health contribution (8% from the worker and 6.5% from the employer) is the biggest piece, and the VAT adds about 6 points. Keeping the full standard deduction below $100,000 keeps them in the 12% bracket even with the basic income counted as income. The steepest spot for middle earners is between $125,000 and $150,000 single, where the health rate goes to 16% and the deduction phases down.

What pulls the other way.

What it means at the top. Above about $250,000 the plan takes about 65 to 77 cents of the next dollar, and close to 87 cents in California above $1 million. That's right at the point where higher rates stop raising more money, because high earners shift and put off income instead. So we keep the top rates where they are, ask more of the top by widening what gets taxed, and count about $190 billion a year less from the top than the separate estimates add up to.

What we'll watch. Hours worked and people taking jobs, especially second earners and people close to retirement. Reported income at the top compared with the forecast. And how much of employers' premium savings actually shows up in pay.

What people and employers stop paying

Looking only at the federal budget is misleading, because families and employers already pay large amounts that this plan replaces.

Most of the new federal health cost replaces what people and employers pay now. The real question isn't whether the country can afford it. It's who pays.

Employers. The average employer pays about $20,000 a year for a worker with family coverage and about $7,000 for single coverage, plus workers' comp (KFF 2025). Under full single payer that's replaced by a 6.5% payroll contribution, which leaves business as a whole about even. Today that's about $750 billion in premiums and $40 billion in workers' comp medical costs, against about $12 trillion of wages. Employers that offer good coverage now will save, and a tax credit rewards passing those savings on as higher pay. Employers that offer no coverage now, and those with very highly paid staff, will pay more. The plan also raises the corporate rate to 28% and adds the VAT, which businesses collect but customers mostly pay. And employers pay half of the leave premium, about 0.5% of wages.

Total taxes. The basic income package takes total U.S. taxes from about 25.6% to roughly 29% of the economy at a 10% VAT. That's still below the 34% average for wealthy countries. At a 22% VAT it's near that average (OECD Revenue Statistics). Full single payer, which swaps premiums for taxes, pushes the total higher, toward the levels of France or the Nordic countries. But families stop paying premiums and most out-of-pocket costs.

The national debt

The deficit is about 6.2% of GDP today. On today's path, with no plan at all, the debt grows from about 106% of GDP in 2029 to about 144% by 2046.

The order we'd go in: pay for the agreed benefits and services first. Keep reserves for the payments, invest in things that build capacity and always pay the interest on the debt. Then use lasting gains to bring the debt down as a share of the economy, with an automatic pause during recessions. The debt doesn't have to be paid off to be sustainable. What matters is that it shrinks compared with the economy over time.

What the whole plan does to the debt by 2046, by our estimate:

The honest part: the plan spends most of AI's dividend on people and services. With the same growth and no plan, the debt would fall further, to about 96% of GDP by 2046 in the substantial case. That's the trade we're making: programs first. Holding the debt steady as a share of the economy takes lasting fixes to today's budget worth about 1.5% of GDP if AI's gains are substantial and about 2% if they're modest. We haven't settled on those yet. These are the candidates:

What it won't do: count borrowed money as revenue, or count the America Fund's shares as cash.

Each source in detail

The VAT and the smaller deduction

Capital gains: reward owners, not flippers

A small toll on speed trading and big short bets

This taxes trading instead of banning it. Ordinary investors who buy and hold would barely notice.

The automation tax

The tax follows the companies that gain from replacing workers, not just the companies that sell AI. A bank or carmaker that cuts thousands of jobs with AI tools gains as much as the company that sold them the tools.

The America Fund

How it works. Each year, every public company worth over $1 billion issues new shares equal to 1.5% of its stock and puts them in the America Fund. It phases in between $1 billion and $2 billion so there's no cliff. That covers about 1,920 companies holding almost all the value of U.S. public stock. Private companies are left out, but large ones still pay the automation tax.

What it pays. The fund keeps its shares and pays out 80% of its cash income (dividends after costs), keeping 20% as a reserve. That cash is folded into the monthly basic income for every adult and child, not paid separately. Here's how much of each person's monthly payment it covers:

The fund pays part of the basic income people already get. For example, in 2037 about $38 of an adult's $1,000 would come from the fund, and about $38 of a child's $250, with the rest coming from taxes. None of it goes to Trump Accounts or any other separate account. Only cash actually paid out counts in the budget, never what the fund is worth on paper.

Why hold instead of sell. Selling a 4% slice each year would pay about $85 a month by year 10. But it would turn the fund's growth into cash instead of building a permanent public endowment worth roughly $28 trillion. Borrowing against the shares would pay more early on, but it would leave about $5.4 trillion of fund debt, and a market crash late in the game could cut payments to almost nothing.

Who pays. The company pays no cash, but existing owners own about 1.5% less of each company every year, which is about 1.5 points of return. The top 10% of households own about 87% of U.S. stock, and foreign investors own about 18%. A worker with $100,000 in a 401(k) loses about $1,500 a year in value. That's why we have to show that whole households come out ahead, not just assume it.

Who controls the shares.

If the courts block it. A required share transfer could be challenged as a taking of property. The law includes a fallback: large companies instead pay a cash tax on profits that raises about the same amount. Until the fund survives its first court test, we treat its payout as at risk, and the rule would catch any shortfall.

What we'd watch. Listings and IPOs, companies moving their legal home overseas, new stock issued by young, fast-growing companies, and how the fund votes. If IPOs fall sharply compared with similar countries, lower the rate or raise the size line first.

The precedents. Alaska's dividend, created by a Republican governor, Norway's fund and the 2025 federal stakes in Intel, MP Materials, Lithium Americas and Trilogy Metals show that public ownership can run by rules and be popular in both parties. They don't prove the returns we're projecting here.

A public development bank

Separate from the basic income, a public development bank makes loans for housing, power, water and transportation projects that can pay them back. Because the loans get repaid, it builds capacity without adding to the yearly budget the way a grant would. How much money it starts with is still being worked out.

Sources we looked at but don't count

Keeping every program

The basic income goes on top of everything, automatically. Nobody has to choose between the payment and the benefits they already get. The law has to say the basic income doesn't count as income or assets for:

Alaska's dividend, the federal stimulus checks and the 2021 child tax credit payments were all left out this way. If overlapping programs are ever merged, every household has to end up at least as well off.

How to read the numbers

Four numbers, kept separate. It's easy to describe a basic income in a way that makes it sound impossible, or free. Neither is true.

One set of assumptions. Every companion document uses these. If a number somewhere else disagrees, this list wins until it's changed here.

What we learned from earlier plans

Andrew Yang made basic income famous with his 2020 Freedom Dividend, $1,000 a month for every adult citizen. Scott Santens, a writer and advocate since 2013, has spent more than a decade working out how to pay for one. Their work, and what independent analysts found when they checked it, shaped this plan in five ways.

Where we part ways. Santens leans partly on modern monetary theory, the idea that a government that prints its own currency is limited by inflation, not by money. Most economists dispute it. Even its backers agree that when everyone who wants a job has one, new spending has to be matched by taxes. So this plan pays its own way, and any borrowing is capped by a rule. Yang paid children nothing and Santens pays them the same as adults. We pay children $250 from the first payment and start adults at $500, rising as the money comes in.

Who else has made this case. The idea has backers from every direction, which helps show it isn't a partisan idea.

What we can conclude, and what it depends on

  1. Start at $500 a month for adults. The basic income can start in 2029 at $500 per adult and 250perchild(125 from a family's fifth child on), with a 10% VAT and limited borrowing that's out in the open. In the slower cases the VAT reaches 22% around 2034.
  2. When adults reach $1,000 depends on AI. About 2042 with substantial gains and about 2037 with transformative ones. With modest gains, not before the 2050s.
  3. Healthcare for everyone is mostly a swap, not a new cost. Taking insurers out saves about $0.62 trillion a year, and covering everyone in full costs a little more than that. So national health spending goes up about 4%, all through its own trust fund.
  4. Services can be fully in place by 2033 to 2036 if AI's gains are substantial or better. With modest gains, the brake holds them at 60%.
  5. The taxes add up to a lot on each extra dollar. The combined rate roughly doubles for middle earners and gets close to the limit at the top. We keep the middle lighter, with an 8% health rate up to $125,000 single and the full standard deduction below $100,000, so most working households come out ahead, including middle earners without children. The steepest rates fall between $125,000 and $150,000 single.
  6. The national debt only shrinks compared with the economy if AI's gains are big. Otherwise it needs about 1.5% to 2% of GDP of lasting fixes to today's budget. The full path goes to independent scoring before anyone should rely on it.

Still open

Stress tests still to run.

Sources

Budget, costs and the AI cases

The AI share and markets

Earlier basic income plans

Something to add or correct on this page? Send it in.