The slogan is older than any of us: tax the rent, not the factory. Tax the land and the location, and stop taxing the work, the wages and the buildings. The argument for it is the strongest in tax economics, and it is also the most oversold.
This piece is not about what land value taxation is, or whether anyone has tried it. That has been done here already: What Is Georgism? sets out Henry George’s single tax, the difference between it and a modern land value tax, and the jurisdiction-by-jurisdiction record, including the finding that no country has ever funded itself from land alone.
What is missing from that record, and from almost every argument made in its favor, is the arithmetic. How large is the cost of taxing work? How large is the benefit of taxing land instead? Who actually bears the new tax, and what happens to them? And how much of a modern state could land rent plausibly carry?
The answers are more interesting than the slogan, and in one respect they cut against it.
Two taxes, and what each one charges you for
Start with what the two taxes are actually doing, because it is not clear from the usual wording.
A tax on production charges you for doing something. Hire somebody and you pay payroll tax. Sell something and you charge VAT, which the buyer pays and the business administers. Build a building or improve a site and the improvement shows up in the assessment. Make a profit and you pay tax on it. In each case the tax is attached to an activity, and activities can be reduced, relocated, postponed or not started.
A tax on land value charges you for holding a location. It does not care what you do with it. Build a factory on it and the tax is the same as if you left it empty, which is the point: the tax is on the site’s value, and that value comes from everything around it - the roads, the schools, the jobs, the transport, the neighbors - not from the owner’s effort.
Now the key asymmetry. The supply of land is fixed. Nobody manufactures more of it in response to a price signal, and nobody takes it out of the country. A tax on something in fixed supply cannot reduce the quantity of it, so the usual argument against taxation - that it distorts behavior and shrinks the thing being taxed - has nothing to operate on.
That is the theory, and it is pretty strong. The question is how much money does it raise.
The cost of taxing work
The number most often attached to this argument comes from Martin Feldstein’s 1999 paper, Tax Avoidance and the Deadweight Loss of the Income Tax, published in the Review of Economics and Statistics.
Its scope matters: the United States federal individual income tax, calibrated to 1994 incomes and rates, single period, with saving excluded. Within that frame, Feldstein’s estimate is that the deadweight loss may be as much as 30 percent of revenue, which he notes is more than ten times Harberger’s classic 1964 estimate. He also finds that the loss from raising existing rates is substantially greater, and may exceed two dollars of lost welfare for every dollar of revenue.
The reason to lead with the range rather than the top figure is that Feldstein’s paper itself records the baseline he is attacking. In Harberger’s own analysis, the deadweight loss of using a tax on labor income instead of a lump-sum tax was about 2.5 percent of the revenue raised.
So the defensible statement is this: estimates of the cost of taxing labor income run from about 2.5 percent of revenue to as much as 30 percent, and the gap is mostly about how much you assume people change their behavior in response. That is a wide range, and it matters. A tax whose cost is 2.5 cents on the dollar is a mild inefficiency. A tax whose cost is 30 cents on the dollar is a machine for destroying value, and the case for replacing it becomes urgent rather than attractive.
Anyone who quotes you the 30 percent without the 2.5 percent is selling you the top of the range. Anyone who quotes the 2.5 percent without the 30 is defending the status quo. Both numbers are in the literature, and the plain description of the literature is a range.
What happens when somewhere actually shifts the tax
Theory will only take you so far here, so let us look at the closest thing to a controlled experiment in the English-speaking world: Pennsylvania.
Since the 1970s, Pennsylvania law has allowed municipalities to tax land at a higher rate than buildings. Pittsburgh went furthest. In 1979 and 1980 it set the rate on land at more than five times the rate on structures.
Wallace Oates and Robert Schwab studied what followed, comparing Pittsburgh’s construction permits with those of 14 other cities between 1960 and 1989. Their finding, published in the National Tax Journal in 1997, was that Pittsburgh experienced in the 1980s a dramatic increase in building activity, far in excess of other cities in the region.
Their own caveat is in the same abstract, and it is the part that gets left out. They attribute the expansion primarily to a shortage of commercial space, with the reliance on increased land taxation playing an important supporting role, by allowing the city to avoid increases in other taxes that could have held development back.
So the Pennsylvania result is real, and it is smaller than the version usually quoted. It is evidence that a split-rate tax can sit alongside a building boom without killing it, and some evidence that it helped. It is not evidence that shifting the tax base causes a boom by itself.
The follow-up work is consistent with that. Florenz Plassmann and T. Nicolaus Tideman examined 15 Pennsylvania municipalities using two-rate property taxes between 1972 and 1994, and found that a higher tax rate on land relative to structures is associated with a statistically significant increase in the number and value of building permits. Note what is not there: no percentage is quoted, because the published abstract does not give one. Figures of 15 percent or 70 percent circulate widely in Georgist literature; neither could be traced to the paper. So we do not use them.
What a working national land tax looks like
Estonia is the most useful live example, because its land tax is national, long-standing and levied on land only.
The base is the cadastral unit, not the building. The rate is set by each municipal council inside statutory bands, is administered by the state tax board, and the revenue is paid in full into local government budgets. In 2025 the bands were 0.1 to 1 percent of land value for residential land and yard land on profit-yielding land, 0.1 to 0.5 percent for profit-yielding land, and 0.1 to 2 percent for other land.
Estonia also does the thing that every serious proposal eventually has to do. It has legislated a protective mechanism capping how far any single cadastral unit’s annual land tax can rise: up to 50 percent compared with the previous year, or 20 euros, in 2025, with local governments setting the cap between 10 and 100 percent for 2026.
That cap is not a detail. It is the legislature admitting, in statute, that the shock its own tax creates is real enough to need a limit.
The part where the evidence splits
Here is the question that decides who wins and loses, and here is where the account has to stop being tidy.
A permanent land tax does not just charge the current owner every year. Because land cannot be moved or made, the whole present value of every future payment gets capitalized into the price of the land at the moment the tax is announced. The person who owns the site on announcement day absorbs the entire future tax in the form of a lower sale price. Everyone who buys afterwards pays a lower price and is unaffected.
For a land tax, that is not an argument against it. It is what makes it a one-off levy on a specific set of people: the owners at the moment of the change.
The best-known test is Danish. Anne Kristine Høj, Mads Rahbek Jørgensen and Poul Schou, in Fiscal Studies in 2018, looked at the land tax changes produced by Denmark’s 2007 municipal reform, which raised rates in some municipalities and cut them in others. Their conclusion: the results imply full capitalization, and the full incidence of a permanent land tax change lies on the owner at the time of the announcement. Future owners, even though they officially pay the tax, are not affected, because they paid a correspondingly lower price.
That is the standard result, and it is the one that makes land taxation attractive to economists: efficient, and the loss falls on a narrowly defined group.
It is also contradicted, recently, by a paper, again using Danish data, and a different tax reform. Ulf Nielsson, Caleb Wroblewski and Anders Yding, in a November 2024 working paper, use bordering municipalities that were forced to reset land tax rates, with an instrumental-variables difference-in-differences design and repeat-sales price indices. Their abstract reports a precise zero effect of land taxes on residential home prices, explicitly rules out full capitalization, puts part of the burden on tenants and future purchasers, and finds null effects on housing development, mobility and homeownership. They read their own results as consistent with limited efficiency costs, but as implying that land taxes are more regressive in their setting than standard models predict.
Both papers are credible. Both are about Denmark. They disagree about the central question of who pays.
Two things follow, and both matter for anyone arguing the case.
The efficiency claim survives both papers. In neither is the land tax shown to distort construction or mobility in a costly way. What does not survive is the confident statement that we know who bears it. If a land tax is fully capitalized, the loss falls on existing owners and nobody else. If it is not capitalized at all, the cost partly lands on tenants, which is the opposite of the usual selling point.
Any argument that requires the first result to be true is an argument resting on a contested finding.
What a British version would actually cost
The most traceable current model for the United Kingdom is Dan Neidle’s, published by Tax Policy Associates on 12 July 2026. Its scope is narrow and should be stated with the figures: England, residential property only, replacing council tax and residential stamp duty land tax, with land assumed to be 55 percent of housing value, a 3 percent discount rate, and capitalization shown separately from the static revenue estimate.
Within that frame:
- A pure land value tax at 1.28 percent of land value replaces both taxes and raises about £56.7 billion a year.
- At least 63 percent of people would be better off in immediate cash terms.
- Administration would cost somewhere between £300 million and £1 billion a year in steady state.
- An average Kensington and Chelsea home would pay about £15,750 a year more, which capitalizes to roughly a 27 percent fall in property prices.
- Blackpool and Hartlepool would see gains in value of 40 to 70 percent.
- An Islington flat bought for £520,000 would move from about £2,000 of council tax to about £5,000 of land value tax, offset by the £16,000 of stamp duty it no longer pays.
Neidle’s own ranking of the difficulties is worth quoting: “valuation is a very serious challenge, but the bigger problems are the very substantial tax increases for some people, particularly in London, and capitalization.”
Those two are the same problem seen from two ends. A revenue-neutral shift is a redistribution between places, and the redistribution is large. That is not an argument against it. It is an argument against pretending it is a technical fix.
How much of the state could land carry
Now the constraint that the slogan ignores, and this is where the piece has to disagree with its own title.
The OECD’s Revenue Statistics 2025, with final data to 2023, records that taxes on property fell from 7.9 percent to 5.1 percent of total tax revenues across the OECD between 1965 and 2023. In Korea, the United States, Israel and the United Kingdom, property tax revenues were more than 10 percent of total tax revenues in 2023.
Two cautions. First, the OECD’s property category is wider than land: it includes recurrent taxes on immovable property, net wealth taxes, inheritance and gift taxes, and taxes on financial and capital transactions. So it overstates a land-only base, which makes it an upper bound. Second, the OECD is an interested party in tax policy, and this site treats its framing with suspicion elsewhere. Here it is being used against our own argument.
The implication is stark. Even the widest property-tax definition, in the OECD country that leans on it hardest, is a little over a tenth of revenue.
So “tax the rent, not the factory” cannot mean replacing the factory taxes. There is not enough land rent in a modern economy to fund a modern state, and any proposal that implies otherwise is not describing a tax shift. It is describing a tax cut with a story attached.
What the arithmetic does support is a partial shift at the margin: use land rent to reduce the taxes with the highest deadweight cost, and accept that most of the state will still be funded by other means.
How to transition is the real argument
If the case is a partial shift, then the interesting questions are all about how you get there, and the real-world examples are instructive precisely because they are cautious.
Estonia caps annual increases. Denmark cut its cap on the land tax rate from 34 to 30 per mille from 2024. And the sympathetic literature has been recommending the same two devices for years: a multi-year phase-in, and a credit to protect owners who are asset-rich and income-poor - the pensioner in a valuable house with no cash to pay a tax based on the value of the land under it.
That last group is the one most arguments in favor of land taxation skip, and it is the one that decides whether the policy survives contact with electorates. A land tax is efficient. Efficiency is not the same as being popular.
What follows, if you take the evidence seriously
Four claims survive, and one does not.
- Survives: the direction is right. Taxes on production carry a real cost, somewhere between 2.5 and 30 percent of revenue depending on assumptions, and land taxes carry a much smaller one. Shifting the margin is defensible.
- Survives: the base is real but small. A national land tax can raise serious money at low rates, because land values are large. It cannot replace the tax system.
- Survives: split-rate taxes have been tried and did not wreck the places that used them. Pennsylvania is the evidence, with the caveat its own authors supplied.
- Survives, but only as one side of a disagreement: full capitalization. Two credible Danish studies, using different reforms, reach opposite conclusions. The claim that existing owners bear the whole cost is the standard result, not a settled one.
- Does not survive: “replace the factory taxes”. The arithmetic does not work.
What would settle it
If you want to know whether a land tax is working, do not ask whether the theory is elegant. Ask four measurable questions.
-
Does construction respond? Pennsylvania says it can, modestly, with a commercial-space shortage doing part of the work. Any serious proposal should be evaluated on permits and completions, not on principle.
-
Does the price fall? If the tax capitalizes into land prices, the fall is the mechanism and the compensation question is immediate. If it does not, the tax is landing somewhere else, and the regressivity finding matters more.
-
Does the base hold? Land cannot leave, but valuations can be contested and reliefs can be expanded. Watch the assessment appeals and the exemptions, because that is how land taxes get quietly hollowed out.
-
Who is paying it in year five? Not in the model, in the receipts. That is the question the Danish literature cannot yet answer and the reason this argument is unfinished.
The slogan is a direction, not a policy. The evidence supports taking a step in that direction, with a phase-in, a credit for the cash-poor, and a candid admission that the beneficiaries and losers can be identified before you start - which is exactly why the politics is hard, and exactly why the argument has to be made with numbers rather than with Henry George’s eloquence.
For the theoretical foundations and the full jurisdictional record, read What Is Georgism? For why incidence is the question that decides every tax argument, Taxes - Who Actually Pays is the primer. And for the opposite approach to the same rent - capping it rather than taxing it - the record is in The Fastest Way to Destroy a City.