505,656 marriages, up two years running — but only the count came back

According to the Ministry of Health, Labour and Welfare’s Vital Statistics (preliminary, December 2025), marriages in 2025 numbered 505,656, an increase of 5,657 (+1.1%) from the prior year. Because 2024 came in at 499,999, just short of 500,000, this is the first time in three years the figure has cleared that mark since 2022 (519,823). The increase is the second consecutive one, and statistically the bottom in marriage counts is clear.

The level itself, however, has not recovered. Against 615,652 in 2019 it is 17.9% lower, and against 650,519 a decade earlier in 2015 it is down 22.3%. In other words the decline has stopped; it has not reversed. For investment decisions on banquet floor, that distinction is decisive.

Dig one layer down and what matters for banquet floor is not the marriage count itself, but how many of those couples hold a reception and how much floor area each one needs. Recruit Bridal Souken’s “Marriage Market Survey 2025” (published 22 January 2026; married-couples module n=821, trends module n=817) puts the share holding a reception or wedding party at 43.1%, average total spend for couples holding both a ceremony and a reception at ¥2.986 million, and average guest count at 57.2.

Multiplying those three against the marriage count puts 2025 receptions at roughly 218,000 and the market at roughly ¥651.0 billion. Note that this survey covers a different population from the “Zexy Wedding Trend Survey” used through the 2024 edition, and the publisher explicitly states it does not compare with earlier years — so spend and guest counts cannot be read as a time series. What to take from it is the level, not the trajectory.

The number that matters as a level is the 57.2 guests. Fifty-seven people seated need roughly 200–250m² of function space. Floor designed around a grand ballroom seating 300 is plainly oversized against that demand. The same structure shows up in disclosures by listed full-service hotels. Imperial Hotel (9708) wrote in its FY March 2026 earnings release that on banquets, “although total attendance for general banquets declined, unit prices rose, improving profitability per event and increasing revenue. Weddings also increased revenue through efforts to win volume centred on small-scale weddings.” Headcount is down; revenue is being built on price and small-format events.

The development question, therefore, is not “keep or scrap the banquet floor.” It is a question of allocation: how much of the large box to cut, and how much of the mid-size box to keep.

Occupancy has already flipped — city 74.1%, business 75.3%

Before turning to floor area, it is worth establishing where the full-service room product itself stands. Room occupancy by accommodation type in the Japan Tourism Agency’s Accommodation Survey shows city hotels at 79.5% and business hotels at 75.8% in 2019, with city hotels 3.7pt ahead. That flipped in the pandemic year of 2020 and has not flipped back: for 2025 (final figures) city hotels are at 74.1% against business hotels at 75.3%, 1.2pt behind. In May 2026 (first preliminary) the gap persists, at 72.6% versus 74.1%.

This reversal is not uniform nationally. Taking the gap between city hotel and business hotel room occupancy across the 12 largest markets in the April 2026 prefecture-level data (second preliminary), only Kyoto is positive (+5.6pt). Osaka is −12.9pt, Hyogo −11.0pt, Tokyo −7.7pt and Kanagawa −7.1pt: the larger the metropolitan market, the wider the gap.

The meaning is plain. In the metropolitan markets, limited-service rooms fill better. Full-service hotels retain an edge in the room market only where the stay itself is the purpose of the trip, as in Kyoto. This is less a story about full-service rooms being unusually weak than about limited-service products maturing and gaining pricing power, narrowing the relative gap.

From an investor’s standpoint, the next question follows. If the relative advantage in the room market is thinning, what exactly is the “floor other than guestrooms” that full-service hotels carry actually earning?

The structure of floor area — 85.7m² per room at city hotels, 27.6m² at limited service

Across the hotels for which MetroEngines Research & Consulting can confirm gross floor area, we aggregated “gross floor area ÷ number of rooms.” City hotels come to a median of 85.7m² per room (N=188 properties), business hotels 27.6m² per room (N=841 properties). The gap is 3.1x.

From here we can define a “floor-efficiency parity hurdle.” Operated at the same occupancy, the difference between the room revenue generated by one square metre of city hotel gross floor area and that generated by one square metre at a limited-service hotel is the amount the non-guestroom floor has to fill. Converting to revenue at a departmental gross operating margin of 70% for rooms and 25% for banquet and F&B (near the midpoint of the industry rule of thumb: 50–60% for limited service, 10–30% for full service), the required revenue works out as follows.

In Kyoto, 1,000m² of non-guestroom floor has to generate revenue equivalent to 60 weddings a year just to match limited-service floor efficiency. In Nagasaki, 25 will do. The same “1,000m² of banquet hall” faces a requirement that varies by a factor of 2.4 depending on the market.

Location analysis — divided by plot ratio, land cost per m² of floor differs 7.5x between Tokyo and Sendai

Reallocating floor is ultimately a question of how much land cost that floor is carrying. Taking the median published land price for commercial land in three markets from the MLIT Real Estate Information Library’s 2026 official land price survey, and dividing by the median designated plot ratio for the same set of points, gives land cost per square metre of gross floor area.

Holding an 85.7m² per room floor configuration on commercial land in Tokyo means each guestroom carries land cost of roughly ¥143 million. On the same land, a limited-service configuration carries roughly ¥46 million. In Sendai the figures are ¥19.1 million and ¥6.2 million, an absolute gap of only ¥12.9 million. The impact of the floor configuration choice on required investment scales with land prices, so it is largest in central urban locations. All three sites are in commercial zones with an 80% building coverage ratio and fire-prevention district designation, and no regulatory factor obstructing retention of banquet floor is identified. The constraint is economic, not regulatory.

The break-even point — an estimated settled ADR of about ¥9,600, clearing in 20 of 41 prefectures

The next step is to put these numbers into the form of an actual investment decision: whether the room-department gross operating profit gained by converting banquet-related floor into guestrooms exceeds the banquet-department gross profit lost plus the required return (8%) on the capital committed to the conversion works. Converted rooms are assumed at 35m² per room (gross) and construction cost at ¥2.15 million per tsubo.

On these terms we solved, prefecture by prefecture, for how much banquet-related floor must earn annually before “not converting” becomes the better outcome. The higher the value, the higher the bar imposed on banquet floor — that is, the more favourable conversion is. A negative value means conversion fails to reach the required yield even if banquet floor revenue is zero.

Conversion clears the hurdle in 20 of the 41 prefectures. Tokyo (¥395,486/m²) and Kyoto (¥309,002/m²) stand well clear of the rest, at wedding-equivalent levels of 132 and 104 couples per 1,000m² per year. Those are figures you could not reach even by filling every weekend with weddings, and the natural reading is that in these two markets most banquet floor is a candidate for conversion. Okinawa (68.0 couples) and Fukuoka (59.9 couples) follow.

Conversely, in the 21 prefectures from Miyagi downward the break-even point is negative. Translated into estimated settled ADR, roughly ¥9,600 is the dividing line: in markets where the 12-month average ADR for city hotels sits below it, converting banquet floor to guestrooms cannot repay construction at ¥2.15 million per tsubo at 8%. The question to debate in these markets is not “keep the banquet floor or not” but whether to invest in large-scale reconfiguration of floor at all.

What matters most is this: of the 78,010 city hotel rooms opened in or before 1990, 74.5% — 58,107 rooms — sit in the 20 prefectures where conversion works. Ageing stock and markets with conversion economics overlap to a considerable degree.

Development scenarios — “drop only the large box” beats a full conversion

With the break-even established, we apply it to an actual asset. As a model asset for a mid-sized regional city, we posit a city hotel with 200 rooms, 17,140m² of gross floor area (85.7m² per room) and 3,000m² of banquet-related floor, assuming an estimated settled ADR of ¥12,787 (the 12-month average level in Aichi) and room occupancy of 74.1% (Japan Tourism Agency “Accommodation Survey”, final figures for January–December 2025, city hotels nationwide). RevPAR is ¥9,475.

The 3,000m² of banquet floor is split into a grand ballroom zone of 1,500m² (annual revenue of ¥70,000/m², equivalent to 23.4 weddings) and a mid-and-small function zone of 1,500m² (¥130,000/m²). The large box only turns over on weekend weddings and a handful of large banquets a year, while the smaller rooms also move on weekdays with corporate meetings, small weddings and private dining — a structure commonly seen in practice.

Partial conversion (B) yields more than full conversion (A). The reason is simple: A also demolishes the mid-and-small function rooms that are earning. The numbers show that reallocating floor is not a question of total volume but of whether you can identify the low-utilisation zones.

Scenario C, retaining banquet floor and refurbishing interiors, returns only 2.6% on capital even with banquet revenue lifted 25%. Because the banquet and F&B gross margin is structurally low, moving unit prices alone struggles to justify the capital committed. Reinvestment in banquet floor works when it is paired not with price increases but with more operating days — building weekday demand from corporate MICE, civic ceremonies, corporate training and retail events beyond weddings. There is clear headroom here.

On sensitivity, the variable that moves this project most is ADR. A ±10% swing takes return on capital to a range of 6.9–9.0%. Because a 10% ADR decline (−1.0pt) bites at least as hard as a 15% construction cost increase (−1.0pt), the conversion decision starts with testing in the market what rate the converted rooms can actually command. Conversely, a 30% recovery in banquet revenue still leaves a 7.1% return, only 0.8pt below the base case. Even if the recovery in marriages continues, at this asset scale it is unlikely to reverse the conversion conclusion.

Where the stock is — 78,010 rooms opened in or before 1990, 58% with a renovation record

Japan has 1,174 city hotels with 203,151 rooms (within our coverage, excluding closed and suspended properties). Of these, 465 properties and 78,010 rooms opened in or before 1990, accounting for 38.4% of rooms. That cohort is now 36 years old or more, and major repairs and equipment replacement fall due in sequence.

Cohorts that opened in the 1980s or earlier show high renovation-record rates of 54–65%, whereas the 1990s cohort sits at just 38.2%. The 48,390 rooms built in the 1990s are now 26 to 36 years old and reaching their first major renewal cycle, yet the share for which a renewal history can be established is lower than for the generations on either side. The greatest scope for reconsidering the allocation of floor is in this 1990s cohort. Picking up the properties whose good locations are held back by dated guestrooms is a related question that turns on identifying renovation candidates from guest review data.

The supply side is also worth noting. New city hotel rooms fell 65% from a peak of 13,791 in 2017 to 4,809 in 2025. Business hotels fell 62% as well, from 64,272 in 2018 to 24,222, so the decline is not specific to the full-service format. What deserves more attention is that because new supply is thinning across the board, the scarcity value of existing banquet floor as “large-volume space with no substitute in the local market” may rise over time. Almost no operator is building a new mid-size or larger banquet hall in a regional city. Deciding to convert is also deciding to give up that scarcity. The reasons behind thinning new supply are tracked year by year in Construction +41% & Labor Crunch Erase Japan’s 2027-2029 Hotel Supply.

What listed REIT monthly results say about full-service versus limited-service today

Differences in floor configuration also show up in the monthly disclosures of listed hotel REITs. Lining up the figures published for June 2026 gives the following.

Japan Hotel REIT, which includes full-service properties, grew RevPAR +1.7% year on year with occupancy of 81.2% (June 2026) and ADR of ¥18,509. Invincible, centred on limited-service properties, posted high occupancy of 82.2% (June 2026) but ADR down 5.0% year on year, taking RevPAR to −6.0%. Limited-service leads on occupancy, full-service on rate — the same pattern holds here. That said, scope differs by company (property count, operating structure, regional mix), so these four rows cannot be read as a ranking between formats. They are a reference point for where the market stands.

The implication for investors is that as long as the full-service room department can compete on rate, the decision to replace banquet floor with guestrooms comes down to whether you can add rooms that command a rate. That lands on the same conclusion as the sensitivity analysis in the previous section, where ADR mattered most.

Investment summary — think in three bands

Three things are needed as next steps in practice. First, break the banquet floor down by zone (grand ballroom / mid-size function rooms / small function rooms and meeting rooms) and measure annual revenue and operating days for each. A whole-property banquet revenue figure cannot distinguish the zones to cut from the zones to keep. Second, verify what rate the converted rooms can command in the market, benchmarked against same-grade properties in the same catchment. Third, confirm conversion feasibility at schematic design level for floor-to-floor height, floor loading, shaft positions and escape routes.

Conclusion

Marriages in 2025 rose for a second consecutive year to 505,656, clearing 500,000 for the first time in three years. The premise that “weddings are in permanent decline” is no longer accurate. At the same time the figure is 17.9% below 2019, and guest counts have shrunk to an average of 57.2. The count has bottomed, but the floor area each couple needs has fallen — that is the real position banquet floor finds itself in.

Measuring the economics of floor, city hotels carry 85.7m² of gross floor area per room against 27.6m² for limited service (medians), a gap of 3.1x. The revenue non-guestroom floor must earn to close that gap ranges from ¥180,375/m² a year in Kyoto to ¥74,536/m² in Nagasaki, a spread of 2.4x across markets. Setting conversion works at ¥2.15 million per tsubo and the required yield at 8%, conversion clears in the 20 prefectures where estimated settled ADR exceeds roughly ¥9,600 — and 74.5% of the 78,010 city hotel rooms opened in or before 1990 are located there.

What the model asset showed, however, is that partial conversion works better than full conversion. Demolishing the mid-size and small function rooms that are earning costs 1.1pt of yield. Reallocating floor is not a decision to abandon banquet space; it is the work of identifying, zone by zone, the large volumes that are under-utilised. That begins with measuring revenue and operating days by zone rather than for the property as a whole. The recovery in marriages has, if anything, bought a little time to get that work started.

Related Reading

• Corporate Retreat Conversions: 8.27% GOP Yield Across 5 Onsen Areas
• Hotels at Year 10: The Renovation Gap Is +13.6% Below 100 Rooms
• Old Malls and Warehouses as Next Hotel Rooms: Conversion Investment Boundaries
• Hotel Category Mix Explains Half of City ADR: 34 Municipalities
• Kanagawa ADR Splits: City Hotels +20.9%, Business −1.7% in June 2026
• New Hotel ADR Ramp-Up: 91 Japan Openings Split Into 4 Pricing Types
• Construction +41% & Labor Crunch Erase Japan’s 2027-2029 Hotel Supply
• Policy Rate 1.00%: Break-Even ADR Across Japan’s 8 Hotel Markets