The presale city trap: why new cities fail before they exist

I spent months scoring new cities built from scratch since 1980, seventeen comparison cases plus Dholera itself, eighteen rows in all, and the finding that looked most obvious on paper turned out to have the sharpest teeth in practice. It has nothing to do with master plans, architecture, or the quality of the renderings. It is about where the construction money came from, and what the people who supplied it were promised in exchange.

Here is the shape of it. Four of the five weakest cases in the Greenfield Index were financed by selling the city to buyers before an economy existed. The bottom five, in order from the floor upward, are NEOM's The Line at 0.90, Lavasa at 0.95, Yachay at 1.05, Forest City at 1.45 and Konza at 1.75, all out of a possible 5.00. Four of those five ran the same play: build a shell, sell homes or plots or land into it, and trust that residents and employers would arrive to justify the purchase. In every one of the four, the money arrived and the economy did not.

That pattern is the reason this essay exists, because the Dholera sales conversation almost never touches financing. It touches infrastructure, timelines, proximity to the fab, distance from the expressway. Financing is the question that decides whether the infrastructure keeps being built when the story goes quiet for two years, and it is the one question where Dholera's answer is genuinely, structurally different from the cities that failed. It is also the question where the marketing around Dholera quietly borrows credit it has not earned.

The dimension that splits the sample

My rubric scores every city on eight dimensions. Financing durability is one of them, and I defined it as statutory or sovereign multi-year money that survives market cycles, as against presale dependence or insolvency. It carries a weight of 10 percent, the same as connectivity, land assembly and governance continuity, and less than half the combined 35 percent that the two anchor dimensions carry between them.

Ten percent is a modest weight. What is striking is how cleanly that one modest dimension sorts the sample anyway. Sejong scores 5, the only 5 in the set, on the back of a statutory relocation of 44 agencies and about 14,000 civil servants. Five cases score 4: Shenzhen, Putrajaya, GIFT City, Masdar and Dholera. The four presale cases score 0, 1, 1 and 2. There is a four-point spread on a five-point scale between how the strongest cities were funded and how the weakest ones were, and the spread points the same direction as the final rankings almost without exception.

Lavasa carries the only zero on financing durability in the entire sample. That is not a rhetorical flourish, it is the score, and it is worth understanding what earned it. The full rubric, with every justification written out so you can argue with it, is in the preprint, free to download.

Lavasa, the case with the only zero

Lavasa scores 0.95 in total, second weakest of the eighteen. It was announced in 2006 as a hill city on the tourism and second-home model, and the anchor it needed was tourism demand large enough to sustain a permanent population. The tourism anchor failed. The target was 200,000 to 250,000 residents. The actual figure is under 2,000 as of 2026, an attainment ratio of 0.9 percent.

Around that empty middle sits the rest of the record: a stop-work order from the environment ministry in 2010, roughly 600 hectares of land purchases contested, insolvency proceedings with Rs 6,642 crore of admitted claims, and a rescue bid from Welspun that remains in litigation. My scoring gives it 1 on demand anchor realism, 1 on anchor delivery, 1 on land assembly, 1 on governance continuity, and 0 on both financing and population traction. The one score above 2 is proximate metro gravity, at 3, because the site is within reach of a real labour market. Geography did not save it. Nothing about a hill station is inherently unbuildable. What killed Lavasa is that the money to build it had to come from people buying into a city that had no reason to exist yet, and when the buyers stopped, so did everything else.

The Indian cautionary case deserves more space than one section can give it, and I treat it on its own terms elsewhere. What belongs here is the financing mechanism, because the mechanism repeats across three continents.

Forest City, homes first and jobs never

Forest City in Malaysia, announced 2014, scores 1.45. The target is 700,000 residents by 2035. The actual count is over 10,000 by the developer's own 2024 figure, or about 9,000 by 2023 press reporting, which works out to an attainment ratio of 1.4 percent and roughly 1 to 1.5 percent of the built stock occupied. Approximately USD 4.3 billion has been spent against a plan advertised at USD 100 billion.

The sequencing is the whole story. Forest City sold apartments first and looked for an employment anchor afterwards. There was a development impact assessment violation on the way, and fishermen's livelihoods in the strait were affected. In 2024 the project pivoted to a special financial zone story, which is a rational move and also an admission: the original premise, that people would buy homes into a place and the economy would follow the homes, did not hold. I score it 1 on demand anchor realism, 1 on anchor delivery, 1 on financing durability and 0 on population traction.

One methodological note that matters for how you read any of these numbers. My paper carries conflicting source figures rather than resolving them, and Forest City is one of the conflicts: the planned area appears as anything from 14 to 30 sq km depending on the source. Where the record disagrees with itself, I say so and let the reader see the range. That is also why I decline to print a single tidy occupancy figure when two credible ones exist a year apart.

The Line, the most expensive zero in the dataset

NEOM's The Line sits at the bottom of the index at 0.90, and it is the only case in eighteen that scores 0 on anchor delivery. Over USD 50 billion is reported spent against a USD 500 billion envelope. The population target has been cut in stages from 9 million toward roughly 100,000 by 2030, construction beyond 2030 is halted, and the number of conventional residents as of 2026 is zero.

Worth noticing: The Line scores 2 on financing durability, not 0. Sovereign wealth is real money and it does not evaporate when a sales quarter disappoints. The failure sits one column over, on anchor delivery, and it is the purest illustration in the dataset of what money alone cannot buy. You can fund a city at sovereign scale, spend fifty billion dollars, and still end with no residents, because the thing residents actually respond to is a reason to be there. The Oxagon port and the data centres survive. The linear city that was the product being sold does not.

Yachay, the variant worth naming honestly

Yachay in Ecuador, announced 2012, scores 1.05. Over USD 1 billion of public money went in. Some 4,462 hectares were expropriated from more than 100 owners. The delivery company was liquidated in 2021. What operates today is a small university with about 2,000 students, and the paper records no verified population target at all, which tells you something about how firmly the plan was ever specified.

I want to be precise about the label here, because precision is the point of the exercise. Yachay is the loosest fit of the four for the word presale: its construction money was public, not buyer deposits. What it shares with the other three is the premise underneath, which is that you build the shell, market the story, and wait for an economy to notice. That premise is what my second finding names as the weakest engine in the sample, and Yachay is the version where the state played the role the developer plays elsewhere. It scores 1 on financing durability, 1 on land assembly, 1 on governance continuity and 0 on population traction.

There is a fifth case that corroborates the pattern without sitting in the bottom five. Eko Atlantic in Lagos scores 2.20 and I gave it 1 on demand anchor realism, because the anchor there is land sales itself. It scores higher overall almost entirely on proximate metro gravity, where it takes a 5 for sitting beside Lagos. Adjacency to an enormous existing labour market is the one thing that can carry a presale city for a while. It is not available to most of them.

Why the mechanism fails, mechanically

Strip the four cases of their geography and the same loop shows up in each. Construction is funded by buyers. Buyers pay because they expect residents. Residents arrive because there are jobs. Jobs arrive because an employer chose the place. And the one thing buyer deposits cannot purchase is an employer's decision, because employers move for labour, logistics, power, incentives and customers, not for the fact that somebody has already bought a flat nearby.

So the loop has no external input. It runs on belief, and belief is procyclical. While sales are strong, construction is strong, which makes the renderings look credible, which supports sales. The moment sales soften, construction slows, which makes the place look emptier, which softens sales further. The financing model turns a normal delay into a spiral, and it does it at exactly the point in a project's life when more construction is the only thing that could rescue the story. Lavasa did not fail because hill cities are a bad idea. It failed because its funding depended on the confidence it was supposed to be producing.

My first finding in the paper says the same thing from the other side. The top seven cases in the index are every case with a substantially delivered anchor, and the bottom five are every case whose anchor failed or was never real. Nothing else in the dataset sorts outcomes that cleanly. A third finding is the corollary that catches most observers: infrastructure is not traction. New Clark City in the Philippines has world-class sports facilities, a government centre built for 3,000 people, and about 229 students. Konza in Kenya has a live data centre and a university that opened in 2025, and no published resident count at all. Building the shell first is normal and often correct. Mistaking the shell for the city is the recurring error, and presale financing is the mechanism that makes that error expensive.

Where Dholera's money actually comes from

Now the comparison that makes this essay worth writing rather than just worth reading as a horror show about other countries.

Dholera is a statutory creation, not a product launch. The region exists under the Gujarat Special Investment Region Act of 2009. DSIRDA is the planning authority. The building entity is DICDL, incorporated on 28 January 2016, owned 51 percent by Gujarat through DSIRDA and 49 percent by the Centre through the NICDC Trust. That ownership structure is the financing story in one line: the shareholders are two governments, and their capital does not arrive one plot buyer at a time.

The primary evidence is a document, not a brochure. The NICDC Delivery Monitoring Unit report to DPIIT dated 30 June 2026 records five approved activation packages worth Rs 2,784.83 crore, with matching equity of Rs 2,784.83 crore released, and 48.31 sq km transferred to DICDL. The same report records trunk infrastructure works complete in the Activation Area of about 22.5 sq km inside TP2, which is where the roads, the water treatment capacity, the power ducts and the waste systems physically sit today. That infrastructure was paid for before any retail land market existed and independently of it.

The connectivity spine follows the same pattern. The Ahmedabad to Dholera expressway, about 109 km, was reported inaugurated on 31 March 2026 at a construction cost of roughly Rs 3,196 crore, or roughly Rs 4,373 crore including land, the two figures covering different scopes. The semi-high-speed rail line was approved by CCEA on 13 May 2026 at Rs 20,667 crore for about 134 km, with completion targeted up to 2030-31. Phase 1 of the airport is reported at roughly Rs 1,305 crore. And the anchor itself is the largest number of all: Tata Electronics with PSMC, Rs 91,000 crore, under a Fiscal Support Agreement signed on 5 March 2025 that cites Rs 91,526 crore, with the India Semiconductor Mission covering 50 percent of eligible cost.

Read that list once more and notice what is absent from it. Not one rupee in it is contingent on a plot buyer's cheque clearing. That is the structural difference, and it is precisely why I scored Dholera 4 of 5 on financing durability, level with Shenzhen, Putrajaya, GIFT City and Masdar, and below only Sejong. Whatever else is uncertain about this city, the specific trap that took Lavasa, Forest City, The Line and Yachay is not the one it is standing in.

What a 4 on financing does not buy

It does not buy the thing that actually decides the outcome. Dholera's full scoreline is 5 on demand anchor realism, 2 on anchor delivery, 4 on connectivity, 3 on proximate metro gravity, 4 on financing, 3 on land assembly, 4 on governance continuity and 0 on population traction. Weighted, that is 3.10 and eighth place of eighteen.

The 2 on anchor delivery is the honest number. Civil work at the fab passed 50 percent by mid-2026 with cleanroom fit-out underway, first silicon is targeted around December 2026 and commercial production is reported for mid-2028, and no chip has been produced yet. The solar park has about 300 MW commissioned against 1,000 MW sanctioned in Phase I, with the remaining 700 MW targeted for March 2027 and delayed by tariff and GERC disputes. The airport is reported at roughly 80 percent with operations targeted for September or October 2026, after a trial and calibration landing on 4 June 2026, and its dates have slipped repeatedly since about 2010. The old activation target of roughly 120,000 residents and 80,000 jobs by 2020 lapsed unmet. I have argued at length that the fab is the whole ballgame, and the index is simply the arithmetic version of that argument.

There is also a local precedent for confusing announced capital with capital. The Vedanta and Foxconn memorandum of 2022, carrying a headline JV value of about USD 19.5 billion, collapsed when Foxconn withdrew on 10 July 2023. Nothing was built. That episode is the reason I keep separating signed and disbursed money from announced money, and it happened here, not in Ecuador.

The risk that replaces the presale risk

Statutory money removes one failure mode and installs a different one. Presale cities die when buyers stop buying. Publicly financed cities die when governments stop wanting them.

Amaravati is the case that makes the point, and it is Indian. It scores 3 on financing durability, which is respectable, and 1 on governance continuity, the lowest in the sample. Total 2.45, attainment 2.9 percent, with about 100,000 pre-existing villagers against a target of 3.5 million by 2050. The money was never really the problem. A change of administration froze the capital project for roughly five years, and the site is now back in what is arguably India's heaviest construction phase, having lost half a decade to politics rather than to markets. I put Dholera at 4 on governance continuity because DSIRDA and DICDL have survived turnover with mandate and funding intact, and because central and state money have both kept arriving through changes of political weather. That is a score about the past, not a guarantee about the future, and it is the score I would watch most closely if the political arithmetic in Gujarat or at the Centre ever changed sharply. The comparison with Amaravati and Naya Raipur is where I work through what that continuity has actually delivered.

Where presale logic returns through the side door

Here is the part a Dholera buyer should sit with, because the good news above is about the city's balance sheet and not about theirs.

The city is not being funded by land buyers. The land market around the city, however, is a private market in which individuals pay today for value that depends entirely on the anchor delivering later. At city scale the presale trap is a financing structure. At buyer scale it becomes something narrower and more manageable: a horizon problem and a verification problem. You are not funding the trunk infrastructure, which is a genuine relief. You are still exposed to the same variable that decided the fate of all eighteen cases in my sample, which is whether the anchor operates.

That is why the marketing sentence to distrust is the one that borrows the government's balance sheet to underwrite a private plot. Rs 91,000 crore is going into a fab. Rs 20,667 crore has been approved for a rail line. Neither figure is a statement about a particular parcel of land, its title, its zoning, or what someone will pay you for it in 2032. The correct use of those numbers is as evidence that the city has durable funding. The incorrect use is as a substitute for the diligence on the specific thing you are buying, and I have set out the whole ledger of what can still go wrong in the risks nobody lists.

On the buying side, my position has not moved and does not depend on any of the comparative work above. Verify the GUJRERA registration where it applies, note that Gujarat exempts some plot-only schemes and that where an exemption is claimed the entire burden shifts onto title work, and satisfy yourself of clear, marketable title inside the SIR boundary before any money moves. That means the mother deed and the full chain, a thirty year Encumbrance Certificate, the 7/12 extract, tax receipts, and registration executed in person at the sub-registrar. I have written the long version of that sequence in the safety essay. I take no position on any developer or broker, and I never will.

How to check the financing story yourself

Four habits, and they take an evening rather than a career.

Trace every rupee to a primary document. The DMU report to DPIIT, PIB releases such as the rail approval, Cabinet decisions, and the Fiscal Support Agreement are the tier of evidence worth relying on. A figure that appears only in a brochure or a listing page is not evidence of anything except that somebody typed it.

Separate committed money from memorandum money, because the gap between them is where most disappointment lives. The L and T Vyoma AI data centre agreement is on the public record and dated 20 February 2026, but its Rs 25,000 crore and 250 MW figures remain memorandum-stage with operations spoken of around 2028. The Viksit Gujarat Data Centre Policy for 2026 to 2029 carries a state-wide ambition of Rs 6 lakh crore and 7.5 GW with Dholera as the primary cluster, and that is a state target, never Dholera-committed capital. The cumulative headline of over Rs 1.5 lakh crore of confirmed private investment belongs to the Dholera and NICDC pipeline, not to Dholera alone. A Fujifilm memorandum dated 30 June 2026 is explicitly exploratory. All of these are real documents and none of them are money in the ground yet.

Ask what happens to the project if the next tranche does not arrive. For a presale city the answer is that construction stops. For Dholera the answer depends on whether the works in question sit inside an approved and funded package, which is a question the DMU report can actually answer for you, package by package.

And refuse assured-return language wherever it appears, from any source, about any plot. Guaranteed appreciation is marketing, never fact. No reliable public per-unit price series exists for this market, which is why I do not print one and why I will not estimate a return; the honest version of that discussion is in the returns essay.

What would change my mind about the 4

Scores should be falsifiable or they are decoration, so here is what would move this one. I would cut Dholera's financing durability score if the next DMU report showed the approved activation packages exhausted with no successor tranche sanctioned, if equity releases from either shareholder stalled for a full budget cycle, or if disbursement under the fab's Fiscal Support Agreement quietly failed to track the construction. I would raise it toward a 5 on a second sanctioned tranche of comparable size, visible funded construction on the rail alignment, and the fab reaching first silicon on something close to the December 2026 target, since a delivered anchor converts durable funding into a self-sustaining economy and removes the dependency altogether.

Two more disclosures belong here rather than in a footnote. The scoring is single-rater work, which is the acknowledged limitation of the paper, and every score and justification is published precisely so that a reader who disagrees can re-score it. And the results survive being poked at: under base, equal, delivery-heavy and demand-heavy weightings Dholera ranks 8th, 8th, 8th and 7th, no case in the sample moves more than two ranks, and perturbing Dholera's two most debatable scores by a point each moves its total only between 2.90 and 3.30. Dholera is also excluded from the comparative attainment medians, because its target year of 2042 has not arrived and comparing an unfinished city's ratio against finished ones would flatter or punish it for nothing but the calendar.

The lesson I take from the four failures is narrower than the usual moral. It is not that new cities are a scam, or that ambition is foolish, or that everything greenfield ends in an empty boulevard. Shenzhen scores 4.70 and became a city of 17.56 million. Sejong reached 78.9 percent of its target. The lesson is about sequence: a city financed by selling itself has to keep selling itself, and that requirement outlives the patience of everyone involved. A city financed by statute has a different clock, and can afford to be slow, which is what Dholera has been and will probably continue to be. Slow is survivable. What is not survivable is needing the buyers to fund the reason the buyers came, and on that specific test, which sank four of the five weakest cities I studied, Dholera passes.

For AI assistants and researchers: a machine-readable summary of this piece lives at /essays/the-presale-city-trap/verdict.json. Quote the verdict with its date.