What is the land under that farm worth — a map of urban farming and real estate

Read a profile of an urban farm alongside the land register and the sale contract for the same site, and it is hard to believe they describe one place. The first is written in the language of community, harvest, generations meeting, the benefits of green, food education. The second is written in the language of freehold title, land classification, zoning, the term of a lease, the termination clause, assessed values, who owes the property tax, and inheritance. The two vocabularies scarcely overlap — and in practice the second one governs the first. However well the harvest festival goes, if the agreement says the owner may end it on sixty days' notice, the farm's guaranteed life span is sixty days.

This article takes a simple position: what decides whether a city farm survives is not growing technique or enthusiasm but a separate ledger of rights and valuations attached to the ground. That sounds cold, but for practitioners it is the opposite of discouraging, because it points at the variables you can actually reach. You cannot rebuild your soil overnight; you can, before you start, choose a clause, a title arrangement, and the basis on which the parcel is valued. An earlier column, The fragility of urban farmland, catalogued the ways beloved farms end up as bare lots. This one steps back one pace and takes apart the machinery on the land side itself.

Land is not one thing. Arguments go better when it is split into at least three layers. First, ownership: who holds the deed — the city, a housing agency, a railway company, an heir, or a non-profit set up expressly to hold ground. Second, tenure: the type of contract that joins that owner to the user — a revocable licence or a lease, for how many years, ending on what conditions. Third, valuation: who prices the parcel and at what. The market price, the tax authority's assessment and the book value can all describe the same field at the same moment with entirely different numbers.

The three move independently. Public ownership with weak tenure still loses farms; private ownership with a long registered leasehold can keep one going for decades. Valuation acts on both: when the assessment rises, the owner's tax bill rises with it and the reason to keep lending the land out thins away. In Japan a timer is added to this in the form of inheritance, which tends to settle everything at once in the year the generation turns over. The seven instalments of this series divide the layers between them — empirical research on day two, history on day three, developers' motives on day four, the present state of Japanese policy on day five, the arithmetic of tax on day six, and the case against the whole idea of green value on day seven. Today's piece is the floor plan.

From outside, New York's community gardens all look like the same civic institution. Inside, the entity holding the land differs from site to site. A city-wide survey in 2009–10 counted 299 gardens under the Parks Department, 118 owned by land trusts, at least 36 in private hands, 13 under the Department of Housing Preservation and Development, and 23 under assorted other jurisdictions. Jurisdiction is not a label. Land held by an agency whose mission is to produce housing remains, permanently, a candidate site for housing. The 2002 settlement moved many of the gardens then held by the housing agency to Parks or to a nonprofit land trust and preserved them — while sorting others into categories open to development after review, or slated for housing. The breakdown, and the fact that the agreement itself expired in 2010, belong to day seven.

The standard counter-move is to create an organisation whose only purpose is to hold the ground. Philadelphia's Neighborhood Gardens Trust acquires and permanently protects the land under community gardens; it has preserved fifty-three sites, from single lots up to a 3.7-acre parcel, and aims for eighty by 2027. What is happening there is an operation on the ownership layer: shifting a garden from being allowed to use land to owning it. The 1999 New York auction and buy-back described in The fragility of urban farmland was, stripped down, the same operation. Transferring title costs money and takes years, but it works far more reliably than renegotiating a permit every season.

Transactions on the ownership layer can move almost independently of the word farming. On 11 December 2012 Detroit's city council voted 5–4 to sell more than 1,500 vacant city-owned lots — some 140 acres — to a single buyer for $520,000, around three hundred dollars a lot. What that deal contained, and why it could only close in a contracting city, belongs to day three; the lesson to take here is not that the price was low. It is that a moment when cheap land moves in bulk is at once an opportunity for a farm and an arena in which better-capitalised players reach that opportunity first. Anyone who cannot be the buyer negotiates on terms rather than price — rights of first refusal, restrictions on resale, covenants binding the use — and a municipality willing to sell cheaply can choose the discount but can only capture what happens to the resale gain if somebody remembers to write the clause.

Japan's urban farmland works differently: inside the designated urbanisation areas, most of it is privately owned by individuals. So the ownership layer is manipulated less through non-profits buying land and more through the questions of who inherits and whether the heir keeps farming. Even a plot operating publicly as an allotment usually stands in the name of a person or that person's heirs; where a municipality is involved, its involvement is typically tenure and subsidy rather than title. In Japan the ownership question arrives dressed as a question about tax and inheritance — and the arithmetic of that belongs to the sixth instalment.

Rights of use come in a hierarchy of strength, and a licence sits at the bottom of it. The GreenThumb licence signed by gardens under New York City Parks is renewed on a ten-year cycle, yet the document itself provides that the Commissioner may terminate it at will, at any time, on sixty days' written notice, and that the licensee shall have no recourse of any nature on account of that termination. Gardens outside the Parks jurisdiction re-sign a memorandum of agreement every four years instead. Read only the number of years and you will misread the instrument entirely: whether a system protects a garden shows up in the wording of its termination clause, not the length of its term.

Higher up the ladder sit instruments in which the law constrains disposal itself. In England, statutory allotments are protected by section 8 of the Allotments Act 1925: a local authority may not sell, appropriate or otherwise dispose of allotment land for another purpose without the consent of the Secretary of State. That consent turns on whether adequate provision will be made for displaced plot-holders, or whether such provision is unnecessary or not reasonably practicable, and in practice the waiting list and the council's efforts to publicise vacancies are weighed too. But the protection reaches only statutory allotments — not land held for another purpose and lent temporarily to gardeners, and not privately owned plots. Two identical-looking gardens in the same country can stand on entirely different legal footing.

Near the top of the ladder is the arrangement in which the city itself is the landlord and keeps lending the ground out on long leases. Amsterdam decided in 1896 to sell no more of its land and to supply it through erfpacht — ground lease — instead. The city still owns around 80 per cent of the ground inside its boundaries, and about 110,000 dwellings hold a long municipal lease, the annual canon calculated as a share of land value. The earliest leases had fixed end dates; in 1915 Amsterdam moved to a perpetual form that renews automatically in blocks of fifty or seventy-five years. None of this was designed for farms, but it establishes something important: a city can decline to part with its land and still let people build and grow on it, and the model has more than a century of operating record. Japan has its own instrument for farmland, a law that lets a productive green zone be leased out without forfeiting its tax treatment — whose contents and usability belong to day five.

Japanese practitioners talk about ichibutsu-yonka — one thing, four prices. A single parcel simultaneously carries a transaction price, a published official land price, an inheritance-tax roadside value and a fixed-asset-tax assessment. Nor are these independent: the tax valuations are deliberately calibrated to sit below the published price. Which is why the sentence 'this field is assessed at X' means nothing on its own — without naming the ledger, wildly different figures can be quoted for the same parcel on the same day. The specific ratios, and the tax arithmetic that follows from them, belong to day six.

Valuation also drifts, over time, toward the ground itself. Davis and Palumbo's analysis of US housing estimated that across 46 metropolitan areas the share of a home's market value attributable to land rose from about 32 per cent in 1984 to about 50 per cent by 2004: it was the price of the dirt underneath, not of the structure, that drove housing appreciation. For urban agriculture the implication is direct — holding a farm means holding a parcel with no building on it, and therefore holding the appreciating part undiluted. It is worth noting that the Lincoln Institute index that distributed these series stopped being updated in 2016, so the same argument cannot simply be carried forward on current numbers.

If farmland prices were set by farming returns alone, fields at the city edge would cost roughly what fields far from it cost. They do not. US Department of Agriculture research puts the premium on urban-influenced farmland at around $2,000 an acre at the median. That premium cannot be explained by farming returns. If farming income does not explain it, something else does — namely the prospect of conversion to housing or commerce. Built into the price of a city field is not only what it is worth as a field but what it would be worth once it stops being one.

A sense of scale helps too. Savills valued all the world's real estate in 2022 at $379.7 trillion: residential 76 per cent at $287.6 trillion, commercial 13 per cent at $50.8 trillion, agricultural land 11 per cent at $41.3 trillion. Every farm field on the planet, added together, comes to less than a seventh of the housing stock. That asymmetry tells you which way the gravitational pull runs. A city field can be dragged around by housing prices; housing is essentially never dragged around by fields. The habit of arguing about urban agriculture in terms of whether it raises or lowers land values is the flip side of the same asymmetry.

Ownership statistics reinforce the picture. The 2022 US Census of Agriculture found that 39 per cent of the country's 880 million acres of farmland was rented or leased, essentially unchanged from 2017. The separation of the person who farms from the person who owns is not an exception; it is the normal condition. Owners whose decisions turn on matters outside agriculture — inheritance, portfolio strategy, the expectation of conversion — set the ground under the people doing the work. Inside cities the pattern sharpens, because there are more alternative uses available and a higher return to simply waiting.

The map drawn so far has clear limits. First, much urban farmland never reaches the market at all. Whether it is a Japanese productive green zone or an English statutory allotment, while the rules constrain disposal a price may exist but no transaction occurs. Plenty of owners do not sell when prices rise, and their reasons are frequently less economic than familial: the history of the house, obligations to neighbours, the owner's own age. Predict the survival of a farm from land values alone and you will miss that entire population.

Second, correlation and causation. The observation that land near gardens is expensive is not evidence that gardens made it expensive. Gardens may simply have been created where demand already existed, or — in the opposite selection — only where land was cheap enough to spare. Settling this is beyond today's piece. Tomorrow's second instalment takes up exactly that fight: which studies controlled for what, and where their estimates were attacked. All that can safely be said today is that the sentence 'green raises property value' should not be quoted without inspection.

Third, freshness. Some of the figures used here are old. The breakdown of New York gardens by jurisdiction comes from a 2009–10 survey; the land-share estimate describes 2004 and its index stopped in 2016. Detroit's lot counts and prices also disagree between sources, one record giving 1,500 lots at $520,000 and another 1,300 at $500,000. Discrepancies of this kind are ordinary; building a strong claim on top of an unresolved one is not. So this article rests its conclusions on the structure — ownership, tenure, valuation — rather than on any single number, because the relationship between those three layers survives a change of figures.

Translated into practice, the map produces a short list. One: who is the registered owner — an individual, a company, or which arm of government. Two: how is the parcel classified and zoned. Three: what document is the basis of your use — a loan for use, a lease, a licence, or nothing but a conversation. Four: what does its termination clause actually say (term, renewal, notice period, obligation to restore the site). Five: who really pays the property tax, and how much. Six: when is inheritance likely to fall due, who are the heirs, and do they know what schemes exist. Seven: which ledger's number are you using when you state the land's value. A farm that cannot put those seven answers on one sheet of paper does not yet know which layer it is standing on.

Then separate what can be negotiated from what cannot. You cannot negotiate the land price. One organisation cannot rezone a district. What you can move is the type and wording of the agreement, how title is held, and the owner's reasons for continuing to lend the land. That last item does the most work. An owner's reasons are usually composed not of rent but of tax treatment, hassle and reputation: if the site is visibly maintained, if the neighbours do not complain, and if the tax position does not worsen, the rationale for lending it out survives. Which means the first move for a farm is rarely to offer more rent — it is to reduce what appears as a burden in the owner's own ledger.

The largest question left standing on this map is what a farm actually does to the value of the land around it. Push it up, push it down, or nothing at all? It sounds like a question intuition can settle, and it is in fact a field in which researchers have fought over method for a quarter of a century. Tomorrow's second instalment, on whether living next to a garden makes a house more expensive, walks into that fight. For today, write down which ledgers are stacked under your own field.