Executive Summary — Replacing a 6.19% Annual Land Cost Ratio with a 4.5% Ground Rent Ratio

The skeleton of the calculation is simple. Whatever remains of the hotel’s GOP after deducting the building recovery amount, building property tax and the capital expenditure reserve is the ceiling on what can be paid for the land. Under ownership that residual has to cover “land price × (cost of capital + property taxes)”; under a leasehold it only has to cover “land price × ground rent ratio”. If the former burden is 6.19% a year and the latter 4.5%, the leasehold route reaches land priced 6.19 ÷ 4.5 = 1.38 times higher.

This 1.38x factor is constant across all nine markets. What varies by market is the absolute amount available for land in the first place — and that is where ADR, occupancy, floor area ratio and construction cost come in. The discussion below starts with the legal frameworks.

Six Frameworks for Building a Hotel on Leased Land — What Separates Park-PFI Is Term and Collateral

“Building a hotel without owning the land” covers several distinct regimes, differing in governing law, term and exit. The three points to sort out first in practice are: (1) the contract term, (2) how the building is treated at expiry, and (3) whether a lender can take security. The third directly determines whether the project can be financed at all.

A distinction frequently muddled — alongside the new funding routes HotelBank has covered previously, such as real estate security tokens — is the one between Park-PFI and a fixed-term leasehold. Park-PFI, created by the 2017 amendment to the Urban Park Act, certifies the installation plan of an operator selected through public solicitation and allows the normal 10-year installation and management permit to be extended to a maximum of 20 years. It also comes with a special provision relaxing the building coverage ratio for park facilities from the usual 2% to 12%. As of March 2026 it is in use at 165 parks nationwide.

The regime is powerful, but from a hotel development standpoint its character is entirely different. An installation and management permit is an administrative disposition; it does not vest the operator with a leasehold as a property right. That makes it hard for lenders to take security over the right itself, and the 20-year cap is short against the physical life of the building (several decades for reinforced concrete). The earning potential inherent in a park location and the constraints the regime imposes are examined case by case in our analysis of the investment opportunity in in-park hotel development under Park-PFI. A paid lease under a business-use fixed-term leasehold, by contrast, is a private-law right established by notarial deed: the term can be designed contractually between 10 and 50 years, and a mortgage can be created over both the leasehold and the building. Term, the nature of the rent, the restoration obligation and the availability of security all differ — so even within the same conversation about “building a hotel on public land”, the feasibility case is assembled in a completely different way.

The Structure of Feasibility — A Leasehold Is a Device for Lowering the Annual Cost Ratio of Land

The essence of the leasehold route is not that you avoid paying for land. It is that the annual cost ratio applied to the land is swapped from the ownership level down to the ground rent level.

Under ownership, the capital committed to the land carries a cost of capital, and fixed asset tax and city planning tax accrue every year on top. Assessed values for commercial land are set at roughly 70% of the official land price, and a combined 1.7% (1.4% fixed asset tax plus 0.3% city planning tax) is levied on that. Restated on an official land price basis, the effective rate is 1.19% a year. At a 5% cost of capital, the annual cost ratio of land under ownership is 5% + 1.19% = 6.19%.

Ground rent under a business-use fixed-term leasehold, meanwhile, is in practice most often set at a benchmark of “4–5% of land value per year”. Fixed asset tax on the land is borne by the landowner, so the tenant’s land-related cost is the rent alone. This is where ownership and leasehold diverge.

Two things follow from this formula. First, whether a leasehold is advantageous is not determined by whether land is expensive or cheap. It is determined by the relationship between the ground rent ratio and the project yield. Second, the higher the land share in a market, the more strongly leasehold leverage works. The corollary is that the moment the ground rent ratio exceeds the project yield, that same high land share goes into reverse.

What matters in practice, then, is the back-calculation: “in this location, for this format, how much ground rent can be paid?” The formula is as follows, and every term can be filled in from observable figures.

Having to recover the building investment in full within the contract term is the defining feature of the leasehold route. Under ownership the land remains at the end of the period, so securing a yield covering the cost of capital is enough. Under a leasehold the building is demolished and the site returned cleared at expiry, so the entire building investment must be recovered within the term. The weight of that recovery is expressed by the capital recovery factor, which spikes as the contract term shortens. As shown below, this is the watershed of feasibility.

Setting the Assumptions — Accommodation Buildings at ¥1.95M per Tsubo, +76% vs 2017

The starting point of the back-calculation is the building investment. Dividing planned construction cost by floor area in the use-classified data of MLIT’s Building Construction Starts Statistics yields a per-tsubo cost series for accommodation-use buildings. From ¥1.11 million per tsubo in 2017 the figure held broadly between ¥1.10 and ¥1.25 million through 2021, jumped to ¥1.39 million in 2022, and reached ¥1.954 million per tsubo in 2024 — +76% over seven years and +41% even against 2022 (¥1.39 million per tsubo). This construction cost level is itself a force compressing new supply volumes over the 2027–2029 window.

The calculations here take that statistic as an upper reference point and set ¥1.8 million per tsubo as the base case, assuming a limited-service format with a tightened specification. With 18㎡ guest rooms and a 72% guest-room ratio, gross floor area is 25.0㎡ per room (7.56 tsubo); adding 18% for design supervision, FF&E and pre-opening costs gives a building investment of ¥16.06 million per room. Sensitivity to construction cost assumptions of ¥1.5 million and ¥2.1 million per tsubo is shown later.

It is worth checking this level against transaction prices. Dividing the property-level acquisition prices disclosed by six listed hotel REITs by room count gives a median of ¥22.6 million per room across 282 properties of 30 rooms or more (first quartile ¥11.7 million, third quartile ¥39.63 million). Restricted to the 48 properties acquired from 2023 onwards, the median rises to ¥28.02 million. Acquisition prices include the land, so an assumption of ¥16.06 million for the building alone is consistent with observed market levels.

There is an implication here. The gap between transaction prices (median ¥22.6 million including land) and the cost of the building alone if built new today (¥16.06 million) is only ¥6.54 million. Urban land cannot possibly be bought with that difference. That distance between cost and market price is part of the background to the wave of cancelled and postponed projects. The leasehold route attacks that distance from the side of the land’s annual cost ratio.

Back-Calculating Nine Markets — Three Work Only on a Leasehold

For nine hotel clusters nationwide, official land price points for commercial land within an 800m radius of each centre point (101 points in total) were aggregated and the median taken as the prevailing land price. The floor area ratio is the median of the same set of points, and with an 85% utilisation rate the required site area per room is derived. On the revenue side, the estimated settled ADR for the ward or city each market sits in (trailing 12-month average) is multiplied by the prefecture’s business-hotel room occupancy (calendar 2025 average) to give RevPAR; other revenue is added at 10% and a 50% GOP margin applied.

Land share ranges from 62.5% at Tokyo Shinjuku Station East to 13.5% at Kanazawa Station-front — a spread of roughly five times. Build the same hotel to the same specification and the weight of land in total investment still differs completely by market. Leasehold leverage bites hardest in high land-share markets such as Shinjuku, Hakata and Namba; in a market like Kanazawa, where the land share is in the 13% range, switching to a leasehold does little to lighten the investment.

Next, the affordable land price derived from these economics is set against the prevailing land price as a ratio — the coverage ratio. Above 100% means the operating economics can carry the standard commercial land price level of that market.

The picture splits into three groups.

The first group works under either ownership or a leasehold. Kyoto Shijo-Karasuma (freehold coverage 230.4%), Tokyo Asakusa (224.9%), Sapporo Susukino (210.8%) and Kanazawa Station-front (124.8%) all have paying capacity well above the prevailing land price even on a freehold basis. Kyoto and Asakusa combine high estimated settled ADR of ¥15,479 and ¥14,398 with prevailing land prices of ¥1.38 million and ¥1.48 million per ㎡ — below the three central Tokyo wards — giving a good balance between earnings and land price. In these four markets the leasehold is not a necessity but an option for improving capital efficiency.

The second group works only on a leasehold. Fukuoka Hakata Station-front, Osaka Namba and Tokyo Shinjuku Station East have freehold coverage ratios of 87.0%, 79.8% and 76.5% — short of 100%. Switch to a 30-year lease and they flip to 119.7%, 109.8% and 105.3%. Simply lowering the annual cost ratio of land from 6.19% to 4.5% moves projects that did not work onto the feasible side. In these markets, where the land share runs 47–63%, the 1.38x factor bites decisively. Which is also to say that the more a development chases the premium of a station-front location, the more a leasehold structure is worth examining.

The third group flips with the term. Hiroshima Kamiyacho has a coverage ratio of 67.4% on a 30-year lease but reaches 112.0% at 49 years. Sendai Station-front sits almost exactly on the watershed at 32.9% for 30 years and 98.5% for 49. In these two markets, extending the contract term does more than haggling over the rent. The next section confirms why, in numbers.

The Contract Term Is the Watershed — 30 Years vs 49 Years Differs by 18% in Recovery Burden

A building on fixed-term leased land is, as a rule, demolished and the site returned cleared at expiry. The building investment therefore has to be recovered in full over the contract term. At a 5% discount rate, the amount required each year for that recovery is 6.51% of the building investment over 30 years, 5.83% over 40 years and 5.50% over 49 years. A 30-year contract imposes a burden 1.00 percentage point — 18% in relative terms — heavier on the building than a 49-year contract.

Applied to a building investment of ¥16.06 million per room, that 1.00 point is ¥161,000 a year. Capitalised at a 4.5% ground rent ratio it is equivalent to ¥3.58 million per room of land-paying capacity. Extending the contract term from 30 to 49 years alone puts land that much more expensive within reach.

As the shape of the chart shows, the capital recovery factor curve rises steeply once the term falls below about 20 years. At a 5% discount rate, a 20-year contract requires 8.02% a year and a 10-year contract 12.95%. That single point explains why the 20-year ceiling on an installation and management permit, and the 10-year floor of a short business-use fixed-term leasehold, are heavy constraints for a capital-intensive use such as a hotel.

Conversely, beyond 40 years the curve all but flattens: the difference between 40 and 49 years is just 0.32 points. The value worth chasing in a term negotiation is concentrated in the 20-to-40-year band. Wanting 50 years or more switches the regime to a general fixed-term leasehold (Art. 22), which means redesigning the contract itself.

It is also worth looking at sensitivity to the construction cost assumption. For Sendai Station-front, which sits on the watershed, varying construction cost across ¥1.5 million, ¥1.8 million and ¥2.1 million per tsubo and the contract term across 30, 40 and 49 years gives the following.

At ¥1.5 million per tsubo the project works even on a 30-year contract; at ¥2.1 million it stalls at a coverage ratio of 30.7% even over 49 years. The higher construction costs go, the longer the term required. While construction costs stay elevated, the term offered by the landowner or municipality directly decides whether the project happens.

Why Public Land Is Different — Backing Out the Assumed Terms for Block 15, Sumiyoshi-cho, Kagoshima

Private landowners and municipalities think about setting ground rent in fundamentally different ways. Private owners derive rent from the expected yield on the land as an asset. Municipalities are trying to satisfy policy objectives — putting idle land to use, creating jobs, generating footfall — alongside securing non-tax revenue. The resulting ground rent ratio lands at a level different from the private market.

As a live example, Kagoshima Prefecture has announced the selected operator for its prefecture-owned “Block 15, Sumiyoshi-cho” site. According to the reported project terms, 21,549㎡ — most of the block — will be leased under a fixed-term leasehold at an assumed ground rent of ¥256 per ㎡ per month, with a project period of 55 years including construction and removal. The original schedule was a July 2028 start with operations opening in April 2031, but the operator asked the prefecture in November 2025 to push this back by about a year; the current assumption is 55 years from August 2029 to July 2084, with operations opening in April 2032. The plan is for a nine-storey, 281-room property with banquet facilities including a grand banquet hall seating up to 2,000.

Viewed as a ratio of annual ground rent to land value, the figure is roughly 0.85% against the median official land price for surrounding commercial land, and stays within a 0.5–1.4% range even when compared against individual nearest points. Against the 4–5% a year used as a benchmark in private business-use fixed-term leaseholds, that is an order of magnitude lower. It is not a strict comparison, given reclaimed port land with no official land price point inside the block, but even discounting for that the difference in level is clear.

The effect on feasibility is large. In the Sendai Station-front calculation above, the supportable ground rent on a 49-year lease was ¥241,000 per room per year. The rent backed out of the Kagoshima project is ¥235,600 per room per year — almost exactly the same level. One reading is that the lease terms for public land happen to line up with what regional-city lodging demand can actually carry. On top of that, the 55-year project period puts the building recovery burden into the region where the capital recovery factor curve has flattened out. On both term and ground rent ratio, the structure makes it easier for the development side to underwrite the investment.

The implication for municipal and public-private partnership teams is clear. Rather than stacking up the absolute amount of rent, designing a longer term does more to raise the odds of attracting an operator. In capital recovery factor terms a term beyond 40 years works in the same direction as cutting the rent — without reducing the municipality’s income. Offer a short term of around 20 years, by contrast, and no amount of rent reduction will easily make a capital-intensive use work.

Where the Nine Markets and the Public-Land Case Sit

On the map, the markets clearing 100% coverage are not simply ordered from cheapest land upwards. Shinjuku Station East carries the heaviest land value per room of the nine at ¥26.82 million, yet earning power — an estimated settled ADR of ¥16,131 and occupancy of 81.5% (calendar year average January–December 2025; Japan Tourism Agency Overnight Travel Statistics Survey, business hotels in Tokyo) — absorbs the ground rent and takes coverage to 118.6% on a 49-year lease. Sendai Station-front, by contrast, carries a light ¥5.44 million of land value per room but only just reaches 98.5% on a 49-year contract on RevPAR of ¥6,702. It is not the level of land prices but the ratio of land price to RevPAR that decides feasibility.

Financing, Accounting and Restoration — Four Issues That Never Appear in the P&L

The back-calculation so far is about economics and yields, but the leasehold route carries several issues outside the P&L that shape the deal. Here are the four points invariably raised in regional and shinkin bank credit assessments and in fund acquisition reviews.

What bites hardest in practice is the interaction between security and term. Lenders customarily set a loan tenor shorter than the contract term, so a 30-year contract often comes with a loan of around 25 years. The capital recovery factor above is calculated on the term itself; where the loan tenor is shorter than the term, principal amortises faster still and cash flow during the period tightens further. Negotiating a longer contract term is simultaneously a rent negotiation and a financing-terms negotiation.

Practical Implications — Checkpoints by Role

Calculation Assumptions

Frequently Asked Questions