Between Milan and Palermo the price of a new home varies by a factor of 2.3, while the cost of building it varies by twenty per cent. The gap lands entirely on the residual land value, which reaches a six-to-one ratio. Behind that residual sits a long and largely discretionary administrative process, and the landowner monetises all of it.
In the first half of 2026 the average price of a new home in Milan reached 7,300 euros per square metre, according to the Abitare Co. observatory, rising above 15,000 in central districts. Rome follows at 6,600, Florence at 5,850, Genoa at 4,800, Turin at 4,780, Bologna at 4,700, Naples at 4,300. Palermo closes the list at 3,200. A factor of 2.3 separates the first market from the last.
Building those homes, however, costs roughly the same everywhere.
A construction site costs the same from Bolzano to Ragusa
Istat publishes one index for residential construction costs, not one per province. It is a national series, at 120.9 in May 2026 on the 2021 base, tracking direct costs: labour, materials, transport, equipment hire. Land and design are explicitly excluded. A manufacturer's price list does not change because the site sits in Sesto San Giovanni rather than Bagheria. Labour does carry a real territorial component, since the national agreement is layered with provincial bargaining, but the gaps in hourly cost are measured in percentage points rather than multiples. Adding site logistics, seismic zoning and special foundations, a realistic range runs between 1,500 and 1,850 euros per square metre.
The price gap between Milan and Palermo is 128 per cent. The cost gap stops at twenty.
Planning charges explain little
Construction contributions vary between municipalities, and Milan bites harder than most with its extraordinary charge on planning uplift. Even so, the range runs from a few tens to two or three hundred euros per square metre, three or four per cent of a seven-thousand-euro price. They decide whether a marginal scheme stands up, not the hierarchy of prices between cities.
Land value is not a cost, it is what remains
Feasibility works backwards. Start from expected revenues, subtract technical costs, charges, design, marketing, financing and the margin that justifies the developer's risk. What is left is the maximum the scheme can pay for the site. Land value comes out of the calculation as a result, not into it as an input.
The consequence is that the residual moves with leverage against price. Applying the same framework to the eight cities, with an eighteen per cent margin and ancillary costs proportional to revenues, the residual sits around 2,840 euros per square metre in Milan, 2,470 in Rome, 1,130 in Naples and just under 500 in Palermo. Six to one, against 2.3 to one on sale prices.
This also explains why weak markets stop building. Once the price falls below replacement cost plus charges plus margin, the residual turns to zero or negative. Naples recorded twenty-three new-home transactions in the first quarter, down fifty-two per cent, with prices up four. Rising prices and vanishing new supply are the same phenomenon seen from two sides.
Value is created by the process
The residual exists because the right to build is scarce, and the scarcity is produced by an administrative process. Value does not appear when the first column goes up. It appears the day the site moves from agricultural to developable, or from disused industrial to residential, or when a variation raises the density index. It is the only moment in the cycle when wealth is generated without capital or labour, and it coincides with a public decision.
Inside the same use class, too
It would be a mistake to read this only as rezoning. A plot already classified residential under the current plan, with a density index assigned and no variation to request, is still a long way from being a consent.
Building permits carry deadlines written plainly into the consolidated building code, on the order of ninety days from case work to decision. Those deadlines are interrupted every time further documentation is requested and restart on filing, and the sequence can repeat. Deemed consent exists on paper, but nobody finances a scheme on that basis, because the power of administrative annulment survives and no bank treats it as full title.
A ring of secondary procedures then surrounds the main one. Landscape authorisation with a heritage opinion, which is by definition a technical-discretionary assessment rather than a conformity check. Hydrogeological constraint, seismic verification, environmental screening, health opinion, fire clearance, utility connections. Each body has its own timescales and its own threshold of documentary completeness, and each can stop the clock. The service conference is meant to compress the sequence into a single procedure, and sometimes it does.
Then there is the case where the use class is already correct but the plan subordinates the scheme to an implementation plan. At that point the ground shifts from technical to political: council adoption, publication, objections, replies, approval, agreement. Two or three years at best, with an election possibly falling in the middle and sending the file back to the start.
The buildable quantum is itself a matter of interpretation
Within the same use class and the same index, what can actually be built depends on definitions that vary between municipalities and, within one municipality, on how the office applies them. What counts towards gross floor area and what falls outside it. How wall thicknesses, loggias, shafts, plant rooms, roof spaces and underground parking are computed. How height is measured on sloping ground. Which standards must be transferred in kind and which can be monetised, at what rate. Which infrastructure works can be offset against charges, and on what measured basis.
These items shift the saleable product by ten or fifteen per cent without a single rule having changed. On a Milan site the difference between the optimistic reading and the prudent one is worth more than the whole construction cost of a floor. Whoever knows those practices, and has the structure and the time to argue them with the office, obtains a different outcome from whoever does not. This is not a pathology to denounce. It is how the system works, and it should be priced accordingly.
The risk nobody can discount
Permitting risk is the hardest risk in the whole scheme to price, because it has no usable statistical distribution. There is no time series of approval times on which to build scenarios, the way one builds on a yield curve or a price series. The developer estimates by analogy with other files, in the same municipality, with the same office, over recent years. A method that holds until the administration changes or a new official arrives with a different reading.
The cost of that uncertainty is measured in time, and time converts into money at a rate. A complex regeneration scheme takes six to eight years between site acquisition and first completion. At an eight per cent cost of capital, every year of slippage erodes margin that no design revision recovers. Delay does not transfer wealth from one party to another. It destroys it.
The asymmetry between seller and builder
It follows that the scarce factor is not capital, nor the technical competence to build, which the market supplies in abundance. It is the capacity to wait. Whoever can tie up resources for eight years without reporting quarterly reaches schemes closed to everyone else, and few parties are in that position. Hence the concentration of complex projects in a small number of hands.
The landowner occupies the best position in this structure. They have already waited, often for decades, at a holding cost modest against the uplift at stake. They choose when to sell and, above all, at what stage of administrative maturity: raw land, land with a settled use class, land with an approved implementation plan, land with consent granted. Each step removes a portion of uncertainty and is paid for. The differential between one step and the next is the market price of the procedure, and bears no relation to the value of the soil as a physical resource.
In practice the gap is often closed with preliminary contracts conditional on consent, which push the waiting back onto the seller in exchange for a higher price. What changes is who carries the risk, not who collects the value. The land price still absorbs almost all the expected value of the permit.
Whoever sells monetises at completion, in a single moment, an uplift created by the administrative act, by infrastructure paid for collectively, and by demand that thickened at that point in the city. The developer commits that capital before any uncertainty has resolved. Permits that slip, price lists that move, rates that change, absorption slower than forecast: six to eight years of risk for a nominal fifteen to twenty per cent, eroded in practice with a frequency no serious developer denies. The seller earns a certain return, the developer a stochastic one. That the first is on average higher is the structural anomaly of the sector.
Where the leverage is compressing
The Istat direct-cost index has risen roughly twenty-one per cent since 2021, and a shock of that size does not hit all markets equally: an extra three hundred and fifty euros per square metre removes a tenth of the Milan residual and wipes out two thirds of Palermo's. The land market itself has no public index and estimates deserve caution. Operator reports across 2025 and 2026 converge on one point: many negotiations collapsed because owners kept asking for values calibrated on revenue assumptions that predate the cost increase. Bargaining power stays with the seller as long as demand holds, but it has stopped being unconditional.
Read through the residual rather than through headline values per square metre, the geography of Italian property prices measures something other than the wealth of markets. It measures how much of each transaction rewards the permission to build rather than the act of building. The soil itself, as a physical resource, counts for very little in that arithmetic.