Who Owns the Farmland — Non-Farmers Own 79% of Rented Acres

The central fact about farmland in 2026 is that ownership and operation have come apart. The Tenure, Ownership and Transition of Agricultural Land survey released by USDA in March 2026 found that nearly 348 million acres of American farmland were rented out during 2024, and that 79% of those acres belong to landlords who do not farm [2][47]. Those landlords collected 34.1 billion dollars in rent, and the land and buildings they let out were valued at 1.7 trillion dollars — 47% more than a decade earlier [2]. This is not a marginal feature of American agriculture. It is the dominant tenure arrangement on rented ground, and rented ground is where a large share of the country's grain is grown.

The composition of that landlord class matters more than its size. Individual private owners account for 38% of the land rented out by non-operators, trusts for 27% and family-owned legal entities for 26% [2]. That is a structure built to hold rather than to sell. Trusts and family entities exist precisely to survive the death of the person who set them up, which is why USDA analysts expect very little of this acreage to reach the open market even as its owners age out [47]. The land is not being hoarded by a villain. It is being held by estate-planning instruments that have no mechanism for returning it to the people who work it.

Set that ownership picture against what farming itself now returns. The farm share of each dollar Americans spend on domestically produced food fell to 11.8 cents in 2024, down from 12.1 cents the year before [9]. For food eaten at home the share is 18.5 cents; for food eaten out it is 7.1 cents [9]. The gap between the value of the land and the value of what the land produces is the whole story of this report. An asset generating eleven and a half cents on the consumer dollar was nonetheless worth 1.7 trillion dollars in landlord hands alone, and rising.

Concentration among operators has moved in the same direction for three decades. USDA's own structural analysis found that midsize crop farms of 100 to 999 acres worked 57% of United States cropland in 1987 and only 33% by 2017, while farms of 2,000 acres or more rose from 15% to 41% [37]. The middle of American agriculture has been hollowed out from both ends: the operator side consolidated into fewer, larger businesses, and the ownership side dispersed into trusts and investment vehicles that never set foot on the ground. Those two movements are usually discussed separately. They are the same event seen from opposite sides of a lease.

Europe shows the same structure arriving by a different route. In the 2020 agricultural census, holdings of at least 100 hectares made up 3.6% of all European Union farms but worked 52.5% of its agricultural area [15]. Widen the threshold slightly and the picture sharpens: holdings of 50 hectares or more, 7.5% of the total, managed 68.2% of the area [15]. Roughly three million European farms disappeared in a decade [15]. The European Parliament had already documented the pattern in 2017, when it found that 3.1% of farms controlled 52.2% of the Union's farmland and called for land to be treated as a finite resource rather than an ordinary commodity [17].

The global version of this ledger is thinner but points the same way. The Land Matrix observatory, which tracks large-scale land acquisitions in low- and middle-income countries, has recorded more than 2,200 concluded deals covering over 65 million hectares transferred from local communities to private and public investors since the early 2000s [40]. That is a land area larger than France, moved by contract rather than by conquest, and most of it documented only because an academic consortium chose to count. No government publishes a consolidated register of who owns the world's cropland. The absence of that register is itself a policy choice.

USDA's July 2026 land values report put average United States cropland at 6,020 dollars an acre, up 190 dollars or 3.3% on the year, and the first reading above 6,000 dollars in the history of the series [1]. Farm real estate overall averaged 4,500 dollars, up 3.4%, and pasture 2,000 dollars, up 4.2% [1]. Regionally the Corn Belt averaged 8,590 dollars and the Pacific states 8,440 dollars [1]. Since 2022 the national cropland average has risen 21.8% [1]. These are figures drawn from a survey of roughly 30,000 farm operations, not from listing prices, so they describe transactions that actually cleared.

Cash rents did not follow. The 2026 NASS survey put average cropland cash rent at 160 dollars an acre, one dollar lower than 2025, though still 15% above the 2020 level [39]. Irrigated cropland held at 244 dollars, non-irrigated slipped to 146 dollars and pasture rose to a record 16.50 dollars [39]. A rising asset price against a flat rent is, in any other market, a compressing yield. On the 2026 figures, average cropland rent represents about 2.7% of average cropland value before any allowance for property taxes, drainage or management. That is a capitalisation rate closer to a long-dated government bond than to an operating business.

Globally the appreciation has been sharper still. The Savills Global Farmland Index rose 18% during 2024, its strongest year since 2021, led by a 47% jump in South America and gains of about 10% in Australia and Central Europe [27]. Savills puts the compound annual growth rate of the index at roughly 11% since 2002, with lower volatility than the commodity benchmarks it tracks alongside [27]. Low volatility is the selling point. Farmland is marketed to pension funds and endowments not as a way to produce food but as a real asset with bond-like characteristics and an inflation hedge attached.

The institutional pool has grown accordingly, though it remains small against the whole. Nuveen Natural Capital manages 14 billion dollars of farmland across 2.6 million acres and more than 580 properties in 11 countries, and was ranked the largest farmland manager globally by Pensions and Investments in October 2025 [24]. In the third quarter of that year it launched a private United States farmland real estate investment trust aimed at a multi-billion-dollar portfolio of leased cropland and permanent plantings [25]. PGIM reported 9.3 billion dollars in agriculture debt and 2.2 billion dollars in agriculture equity at the end of 2025 [50]. Individually these are modest numbers. Collectively they establish a bid that does not depend on a harvest.

The individual case that draws most attention is also the least representative. Cascade Investment, the vehicle associated with Bill Gates, holds roughly 275,000 acres, of which about 248,000 are working farmland spread across at least 17 states — the largest private farmland holding in the country and approximately 0.03% of the national total [48]. The disproportion between the coverage that holding receives and the acreage it represents is instructive. A single visible owner of three hundredths of one percent generates more political heat than 348 million rented acres held by trusts nobody has to name [2][48].

What the price level does, regardless of who is buying, is set the terms of entry. At 6,020 dollars an acre, a modest 500-acre Midwestern grain operation represents roughly three million dollars of land before a single machine is bought [1]. Against net farm income forecast at 158.4 billion dollars nationally for 2026, down 2.6% in nominal terms and 5.5% in real terms, and total production expenses of 492.8 billion dollars, the arithmetic of a first purchase does not work on farming cash flow alone [8]. It works on inherited equity, on off-farm wealth, or on institutional capital with a thirty-year horizon.

The 2022 Census of Agriculture put the average age of a United States farm producer at 58.1 years, up 0.6 years from 2017 and continuing a trend that has run unbroken for four decades [4]. The distribution is more alarming than the mean. Producers aged 35 to 64 fell 9% between censuses while those aged 65 and over rose 12%, and the census recorded more producers above 75 than below 35 [4]. An industry with that age profile is not ageing gradually. It is approaching a single-generation handover with no queue of buyers formed behind it.

American Farmland Trust estimates that nearly 300 million acres of cropland, about one-third of all United States farmland, will change hands within twenty years [22]. The National Young Farmers Coalition puts the figure closer to half of all farmland over two decades and identifies secure land access as the single largest barrier facing new entrants, and a disproportionate barrier for Black, Indigenous and other farmers of colour [23]. Separately, between 2001 and 2016 the United States lost or fragmented 11 million acres of agricultural land to development — about 2,000 acres a day — with a further 18.4 million acres projected to convert between 2016 and 2040 [22].

Japan is the same curve twenty years further along, and it is worth studying precisely because it removes the ambiguity. A 2026 survey by the Ministry of Agriculture, Forestry and Fisheries put the country's core agricultural workforce at 986,600 people, below one million for the first time and down 4.8% in a single year [18]. The average age is 67.7 [18]. People aged 70 or over make up 56.5% of that workforce; those in their thirties account for 3.8% and those under 30 for 1.2% [18]. The total has more than halved from 2.05 million in 2010 [18]. Self-employed farmworkers fell 25.1% in five years to 1.02 million [52].

The consequence in Japan is not consolidation but abandonment. Roughly 30% of the country's farmland risks having no farmer at all within a decade, with some regions facing abandonment rates as high as 80%, and 257,000 hectares are already out of production with more than 60% of that judged difficult to restore [19]. Japan's cultivated area has contracted by more than 30% over six decades [21]. Where there is no bidder — institutional, foreign or local — land does not transfer. It reverts. That is the outcome a succession cliff produces when capital declines to show up, and it is not obviously better than the outcome where capital does.

Europe's demography sits between the American and Japanese cases. Farmers under 40 manage just 12% of European Union holdings [16]. Farmers aged 65 or over form an outright majority in Cyprus at 52.3% and Portugal at 51.9%, and account for 45.6% of farmers in Italy, 43.4% in Malta, 41.7% in Romania and 39.7% in Spain [16]. These are not uniformly poor agricultural regions; Italy and Spain are among the Union's largest producers. The pattern tracks the cost of entry rather than the profitability of production, which is why it appears in wealthy member states as reliably as in poorer ones.

Put the three demographic pictures together and the sequencing becomes clear. The United States is at the point where the transfer is about to occur and the buyer is undetermined. Europe is at the point where entry has already become inheritance-dependent, with only 12% of holdings in the hands of anyone under 40 [16]. Japan is past the point where a buyer exists at all, and is spending public money to consolidate plots simply to keep them in production [20]. Each stage follows from the same arithmetic: land priced as an asset cannot be bought out of the income it produces as a farm.

The input side of agriculture is among the most concentrated sectors in the industrial economy, and unusually, it is concentrated on precisely the goods a farmer cannot decline to buy. Bayer, Corteva, Syngenta and BASF together sell 56% of the world's commercial seed, with Bayer alone at 23% and revenues of 11.613 billion dollars [5]. The same four firms sell 61% of global pesticides, led by Syngenta at 25% [5]. Four sectors serving agriculture — seeds, pesticides, machinery and animal pharmaceuticals — now meet the conventional threshold at which four firms holding more than 40% defines an oligopoly [5].

Fertiliser is worse, and the United States nitrogen market is the clearest case. Four-firm concentration in domestic ammonia capacity rose from 50% in 2000 to 70% in 2023 as the number of plants fell from 46 to 33 [6]. The Herfindahl-Hirschman Index for that market rose from 0.09 to 0.201, taking it past the 0.18 threshold at which the 2023 Department of Justice and Federal Trade Commission merger guidelines classify a market as highly concentrated [6]. CF Industries alone holds 39% of capacity, Nutrien 16%, Koch 10% and LSB 5% [6]. Testimony to the Senate Agriculture Committee in May 2026 put nitrogen four-firm concentration at 77% on a 2018 to 2019 basis, and potash and phosphate at 100% [7].

The enforcement record confirms that these markets do not behave competitively. In September 2022 the Federal Trade Commission and a bipartisan group of state attorneys general sued Syngenta and Corteva over loyalty programmes that allegedly paid distributors to limit sales of cheaper generic products, keeping prices elevated after the relevant patents had expired [12]. The products named included rimsulfuron, acetochlor and oxamyl on the Corteva side, and azoxystrobin, mesotrione and metolachlor on the Syngenta side [12]. In June 2026 Corteva settled the parallel farmer class action for 85 million dollars, with a settlement class that could exceed 100,000 growers [13].

Bayer reached a separate accommodation. In late May 2026 the Department of Justice announced that Bayer CropScience had agreed to remove potentially anticompetitive provisions from its seed loyalty programme for a period of years [14]. Machinery followed the same trajectory. The FTC and five states sued Deere and Company in January 2025 over repair restrictions, and in July 2026 obtained a settlement requiring Deere for ten years to give independent repair shops the same diagnostic software and tools it provides to authorised dealers, plus one million dollars towards state costs [10]. Three months earlier Deere had settled a farmer class action on the same subject for 99 million dollars [11].

Meanwhile the farm's own accounts deteriorate. USDA forecasts total production expenses of 492.8 billion dollars for 2026, up 21.2 billion or 4.5% on 2025 and 15.1 billion above its own February projection [8]. Against the February forecast, fuel and oil expenses were revised up 28.8%, fertiliser up 15.3% and livestock purchases up 11.4% [51]. Net farm income is forecast at 158.4 billion dollars, down 2.6% nominally and 5.5% in real terms, and that figure is held up by 47.4 billion dollars of direct government payments, up nearly 70% and potentially the largest nominal transfer ever recorded [8]. Sector debt has passed 600 billion dollars [8].

The distributional question that follows is who ends up holding the support. Direct payments flow to whoever controls the acre, and 79% of rented acres are controlled under a lease from someone who does not farm [2]. Rents on that ground stand 15% above their 2020 level and cropland values 21.8% above their 2022 level, while net farm income has fallen in real terms for the year in which the largest payment package on record was issued [1][8][39]. Whatever the intent of a 47.4 billion dollar transfer, the observable outcome in the same period is an asset that costs more to acquire and an operating margin that does not [8].

The AFIDA report delivered to Congress in January 2026 recorded foreign interests in more than 46 million acres of United States agricultural land as of 31 December 2024, or 3.6% of all privately held agricultural land, with a further 10.6 million acres under long-term lease [3]. The largest single holder is Canada, with about 34% of all reported foreign-held acres, roughly 16.1 million, and much of it forestland rather than cropland [34]. Across the whole foreign-held pool, forestland has historically made up 48% of the acreage, cropland 29% and pasture 18% [34]. The growth is real: foreign holdings rose 214% over four decades [34].

Chinese holdings, the subject of almost all of the political attention, amount to roughly 0.02% of United States farmland on the CSIS assessment [32]. That analysis concludes the credible security concern is narrow and site-specific — parcels adjacent to military installations, where proximity rather than acreage is the variable that matters — and not a function of aggregate ownership [32]. The Union of Concerned Scientists goes further and argues that domestic consolidation of land and inputs represents the larger structural risk to food security, and that legislative attention has been misallocated [33].

The legislative response has nonetheless been enormous. Twenty-eight states now impose some form of restriction on foreign land ownership, most of them enacted since 2023, and proposals were introduced in more than half of all states during 2025 [28]. Of 543 bills introduced by states and Congress since 2021, 340 — 63% — contain provisions restricting Chinese citizens from owning some form of property, and 28 target China explicitly and alone [29]. Twenty-five states considered such legislation during 2026 [29]. Several statutes now reach beyond farmland into forestland, water rights, mining claims and mineral rights [29].

Where a genuine foreign-ownership harm has been documented, it has usually turned out to be a water story wearing a land costume. Arizona terminated groundwater leases held by Fondomonte Arizona, a subsidiary of the Saudi dairy group Almarai, in Butler Valley in 2023 after an investigation found lease violations, and declined to renew three further leases [35]. The company grew alfalfa for export to feed livestock in a kingdom whose own law restricts such water-intensive farming [35]. The instructive part is what happened next: Fondomonte continued pumping on privately held land in La Paz County, where groundwater extraction is unregulated, and the Emirati group Al Dahra grows alfalfa on about 3,000 acres of private ground near Wenden [36].

That sequence contains the general lesson. Cancelling the leases changed the nationality of the pumping, not the volume. The binding constraint in La Paz County is the absence of groundwater regulation, and a domestic buyer with the same crop and the same irrigation licence would draw down the same aquifer at the same rate [36]. Regulating who may own an acre is a proxy instrument. Regulating what may be extracted from it is the direct one, and it is politically harder because it binds constituents as well as foreigners.

The federal posture has hardened regardless. USDA launched a National Farm Security Action Plan in July 2025, strengthened AFIDA enforcement, and opened an online portal for reporting foreign acquisitions on 22 January 2026 [30]. Better reporting is unambiguously good, and the GAO asked for exactly that [46]. But the plan's framing treats foreign acquisition as the security variable, when the measured trend in American land ownership over the same period is a domestic transfer roughly seven times larger — 348 million rented acres held by non-operators against 46 million acres in foreign hands of any kind [2][3].

The policy history matters because almost every idea now being proposed has already been tried somewhere, usually for decades. France created the Sociétés d'aménagement foncier et d'établissement rural in 1960, giving regional bodies a right of pre-emption over agricultural land sales in order to favour working farmers over outside buyers [42]. It is the most interventionist land-market regime in any wealthy democracy, and it has been progressively extended for sixty years because buyers kept finding the way around it [42].

The French record is the most useful evidence available on whether restricting buyers works, because it is long enough to evaluate. In its most recently reviewed year the SAFER network exercised pre-emption on 1,240 sales — 0.4% of the 322,400 transactions notified to it [43]. A right exercised four times in every thousand sales operates mainly as a deterrent and a source of market intelligence, not as an allocation mechanism [43]. The Sempastous law of 2023, which extended prefectural approval to transfers of shares in companies holding agricultural land, exists precisely because the share-transfer route had become the standard way to move land without triggering the older instrument [41][42].

Japan has taken the opposite approach, treating the problem as physical rather than legal. Its farmland banks reassemble scattered plots into workable blocks and lease them to whoever will farm them, and the ministry requested a fourfold budget increase for that function alongside funding for plot enlargement and waterway development, targeting a reduction of more than 60% in rice production costs [20]. This is consolidation as public policy, undertaken deliberately, because the alternative on the ground is abandonment of up to 30% of the national farmland base within a decade [19]. Japan is not choosing between family farms and large operations. It is choosing between large operations and scrub.

Ukraine offers the cleanest natural experiment of the three. A twenty-year sales moratorium ended on 1 July 2021 under an International Monetary Fund loan condition, with individuals capped at 100 hectares and corporate buyers admitted from 2024 at up to 10,000 hectares [49]. The country has 42 million hectares of farmland, of which large agribusinesses already operated about 6 million hectares before the market opened [49]. The staged caps were written specifically to prevent a rapid transfer to agroholdings and foreign capital — an acknowledgement, built into the reform itself, that an open land market and a dispersed ownership structure are not naturally compatible.

The United States has now attempted the input side with more vigour than the land side, and the results are informative. Corteva paid 85 million dollars to more than 100,000 farmers over crop-input loyalty programmes [13]. Bayer agreed with the Department of Justice to strip potentially anticompetitive provisions from its seed loyalty programme [14]. Deere paid 99 million dollars to farmers and then accepted a ten-year obligation to give independent repairers the same diagnostic tools as its dealers [11][10]. Every one of those outcomes is a genuine gain for growers. None of them changed a market share [5].

That is the pattern across all four jurisdictions. Instruments that regulate conduct — loyalty programmes, repair restrictions, disclosure duties — can be enforced and do produce results. Instruments that attempt to regulate who ends up owning an asset run into the same obstacle everywhere: the buyer with the lowest cost of capital wins the auction, and a pre-emption right exercised in 0.4% of sales does not change who that buyer is [43]. Japan's response, subsidising the physical reassembly of land so that somebody will farm it at all, is the only one of the four that addresses the price gap directly, and it does so by accepting consolidation rather than resisting it [20].

The strongest empirical case for consolidation is recent and specific. A 2025 peer-reviewed study of sustainable food systems, designed as a quasi-natural experiment around land-transfer policy, found that large-scale agricultural operations raised grain production resilience through two mechanisms: economies of scale that progressively reduced production costs, and increased professional cooperation among farmers [44]. The same study is candid about its limits, reporting that rural labour outflow and underdeveloped rural financial systems constrain the benefit [44]. It measures grain output stability, which is the variable that determines whether a country eats calories in a bad year.

The case against is built on a different dependent variable. Research on economies of scope argues that diversified smaller operations outperform scale on nutritious food production, climate resilience and the sustainability of agricultural ecosystems, and that consolidation pushes cropping patterns toward monoculture of high-yield commodities [45]. That is a claim about dietary diversity and ecological buffering, not about tonnes of grain. Neither literature has produced a reconciliation on a common metric, which is why the two bodies of evidence are routinely deployed against each other in policy debates where they do not actually collide.

There is a second contested question sitting underneath the first, which is whether institutional capital is actually driving land prices at all. The visible evidence says yes: Nuveen manages 2.6 million acres and launched a private United States farmland REIT in 2025, and the Savills global index rose 18% during 2024 [24][25][27]. The measured evidence is awkward for that reading. The NCREIF index, which tracks the institutionally held pool directly, returned minus 1.03% in 2024 and 0.20% in 2025 [26]. Institutions bidding a market up do not normally post those numbers.

The more parsimonious explanation is that the marginal buyer of American farmland is usually the neighbour. Expansion by existing large operators, financed against appreciated equity in land already owned, accounts for most transactions, with institutions a growing but still secondary bidder — a reading consistent with midsize farms' share of cropland falling from 57% to 33% while large farms rose from 15% to 41% [37]. If that is right, the political focus on outside capital is misdirected for the second time in this report. The consolidating force is internal to farming, and it is funded by the land price itself.

Both readings survive the evidence, and an honest report has to say so. What does not survive is the claim that the question has been settled in either direction. Neither the resilience literature nor the ownership literature has produced the study that would settle it: a common metric applied across farm sizes and ownership structures, tracking output stability, dietary diversity, input cost and ecological condition together over a full commodity cycle. Until that exists, scale advocates and dispersion advocates are each citing real findings about different outcomes.

The policy implication of genuine uncertainty is not paralysis. It is a preference for reversible decisions. Land ownership structures are among the least reversible arrangements a society builds: once 348 million acres sit inside trusts and family entities designed to survive their founders, undoing that takes generations and expropriation powers no democracy wants to use [2][47]. A conduct remedy such as the Deere repair settlement can be revisited in ten years [10]. A transfer of the ownership base cannot. That asymmetry, rather than any finding about yields, is the strongest argument for caution about the current trajectory.

Strip the argument to what can be verified and four numbers carry it. Non-farming landlords own 79% of rented United States cropland, some 348 million acres worth 1.7 trillion dollars [2]. Four firms sell 56% of the world's commercial seed and 61% of its pesticides [5]. The farm share of each consumer food dollar is 11.8 cents [9]. And about 300 million acres, a third of American farmland, will change hands within twenty years [22]. Every one of those figures comes from a government statistical agency or a peer-reviewed count, and none of them is seriously contested.

What the numbers describe is a sector where the returns to ownership and the returns to operation have separated and are now measured on different instruments. In 2025 the institutional farmland index earned 3.05% income and lost 2.80% in capital value, while national cropland values rose to a record [26][1]. Cash rent is 160 dollars against an asset worth 6,020 dollars [39][1]. These are the financial signatures of a store of value, not of a business. Agriculture has been quietly reclassified, and the reclassification happened without a policy decision anywhere.

The demographic timing is what makes this urgent rather than merely interesting. The average American producer is 58.1 and there are more over 75 than under 35 [4]. Japan's core agricultural workforce has fallen below one million at an average age of 67.7, with 56.5% of it aged over 70 [18]. Only 12% of European Union holdings are managed by anyone under 40 [16]. Three of the world's largest agricultural economies will resolve the same succession question within the same twenty-year window, and each will resolve it through whatever bidding structure happens to be in place when the sale occurs.

Against that, the policy response has been aimed almost entirely at nationality. Twenty-eight states restrict foreign ownership [28]; 543 bills have been introduced since 2021, 340 of them naming Chinese citizens [29]; USDA has built a farm-security plan and a reporting portal around foreign acquisition [30]. All of it addresses 3.6% of privately held farmland, and in the Chinese case about 0.02% [3][32]. Meanwhile the GAO rates the data underpinning that entire debate as incomplete and unreliable [46], and no jurisdiction publishes a register of who beneficially owns farmland at all.

Two conclusions follow that the evidence will support. First, the binding constraint on who farms is the gap between what land yields as a farm and what it yields as an asset — a gap visible in the 2.7% rent-to-value ratio and confirmed by every entry barrier the succession literature records [1][39][23]. Second, the instruments currently deployed do not touch that gap. Pre-emption rights reach 0.4% of sales [43], foreign-ownership restrictions reach 3.6% of acreage [3], and 47.4 billion dollars of direct payments flow through to whoever controls the acre [8].

The quiet part of this consolidation is that nobody decided it. There is no policy that says farmland should be owned by trusts, pension funds and the largest neighbouring operation. There is a price level that makes those the only bidders who can clear it, a tax and estate system that rewards holding, a support regime that pays the controller of the acre, and an input structure that takes 73% of the operating cost before the crop is sold [38]. Three hundred million acres will settle that question in the next twenty years [22]. On current instruments, the answer is already determined, and it is not being recorded.