The Mandatory Line — The Law Requires Power for “Time to Escape”, Sized at 30–60 Minutes

Let us first confirm where the regulatory line is drawn. The basis for requiring emergency power at lodging facilities falls into two systems: the “emergency power source” under the Fire Service Act and the “standby power source” under the Building Standards Act. Both cover equipment for firefighting, alarm and evacuation; operating loads such as guest room lighting, air conditioning, hot water and kitchens are not included.

Under the Fire Service Act, hotels and ryokan are specified fire-prevention properties under item (5)(a) of Appended Table 1 of the Order for Enforcement. Indoor fire hydrant equipment is required at a total floor area of 700㎡ or more (relaxed to two or three times that figure where fire-resistive construction and interior finish conditions are met), and sprinkler equipment is required on the 11th floor and above, or where total floor area is 6,000㎡ or more in buildings other than single-story ones. This equipment must be fitted with an emergency power source, and Article 12 of the Enforcement Regulations sets the capacity for indoor fire hydrant equipment as being able to “operate effectively for thirty minutes or more.” There is also a line on the type of power source: specified fire-prevention properties with a total floor area of 1,000㎡ or more cannot choose dedicated emergency power receiving equipment and must use private generating equipment, storage battery equipment or fuel cell equipment. Since a 100-room hotel usually has a total floor area of several thousand square meters, in practice almost all of them can be considered to fall inside this line.

The Building Standards Act, meanwhile, requires emergency elevators in buildings more than 31m high (Article 34, paragraph 2 of the Act) and requires those emergency elevators to have a standby power source (Article 129-13-3 of the Order for Enforcement). Emergency lighting is required in the habitable rooms and evacuation routes of hotels and ryokan (Article 126-4 of the Order), and smoke exhaust equipment that needs power must also have a standby power source (Article 126-3 of the Order). The required duration is 30 minutes for emergency lighting and smoke exhaust equipment and 60 minutes for emergency elevators. According to commentary by the Nippon Engine Generator Association (日本内燃力発電設備協会), the building standards legislation sets no specific standards for the structure and performance of standby power sources other than operating time, and they are administered in line with the standards of the fire service legislation.

The maintenance obligation is also worth confirming. According to the Tokyo Fire Department, the inspection standards for private generating equipment were revised in June 2018: internal observation and similar methods can now be chosen in place of load operation, and where preventive maintenance measures to maintain operating performance are in place, the cycle for load operation or internal observation can be extended to once every six years. Equipment that uses a gas turbine as its prime mover no longer requires load operation. In other words, hotels inside the mandatory line already have generators and storage batteries and already bear the cost of inspection. The question for voluntary investment is not “installing power where there is none” but the increment: “how far to extend 30–60 minutes of fire-safety power toward 72 hours of operating power.”

The Gateway to Voluntary Investment — A Shelter Agreement Opens Routes to Subsidies and Payment

The 72-hour benchmark comes from the government’s shelter schemes. The “Guidelines for Securing and Operating Temporary Shelter Facilities” issued by the liaison council on measures for commuters stranded by a Tokyo inland earthquake assume that stranded commuters are accommodated for up to three days after a disaster, and the Tokyo Metropolitan Ordinance on Measures for Stranded Persons also asks businesses to make efforts to stockpile three days of supplies. Whether a hotel becomes a temporary shelter facility or an evacuation shelter, however, is a voluntary decision based on an agreement with the municipality. Precisely because it is not mandatory, agreements are paired with payment and support.

There are three main routes. The first is subsidies for physical improvements: under the Tokyo Metropolitan Government’s program for temporary shelter facilities, the incremental cost that private operators incur for the shelter portion is subsidized 2/3 by the national government and 1/3 by Tokyo. The second is subsidies for fuel stockpiling. The Ministry of Economy, Trade and Industry (METI) subsidy for promoting self-reliant fuel stockpiling at socially important infrastructure in preparation for disasters covers private facilities such as lodging facilities recognized by their municipality as temporary evacuation shelters, as “facilities that can serve as temporary evacuation shelters,” and in the call under the FY2025 supplementary budget and FY2026, applications are accepted under the FY2026 program. The subsidy rate is up to 1/2 of eligible costs, with a cap of ¥10 million for LP gas containers and supply equipment only, rising to ¥50 million when combined with generators and air-conditioning equipment. For LP gas, the requirement is that 50% of the maximum storage volume can cover operation for at least three and no more than seven days; here too, “three days” appears as a design condition. In FY2026, following the first call (May 27–June 16) and the second (July 1–21), a third call ran from August 5 to October 5, and an additional (fourth) call for the FY2026 program is planned from early October onward.

The third is payment for the accommodation itself. According to Cabinet Office guidelines (December 2025), when hotels and ryokan are used as evacuation shelters in a disaster to which the Disaster Relief Act applies, the cost that can be paid is the actual cost up to ¥10,000 per person per night including tax (not including meals) under the standard amount as of FY2025, and materials from a Cabinet Office study group introduce examples such as Kumamoto Prefecture and Gero City, where agreements are concluded in advance with ryokan and hotel associations or individual properties. A shelter agreement can help fill rooms at a time when ordinary demand stops. Since sheltering itself is impossible in a building where electricity and water have stopped, the effectiveness of an agreement can be said to depend on whether the property has independent power and independent water supply.

Capex by Equipment — The Basic Set Costs ¥10.88–14.21 Million for 100 Rooms, ¥109,000–142,000 per Room

Next, we build up the content of voluntary investment equipment by equipment. On the power side we cover emergency generators (portable, permanently installed diesel, and LP gas) and storage batteries; on the water and heat side, a larger receiving water tank, a well, well-water filtration and an LP gas bulk tank. Unit prices are taken from published list prices and market-rate articles and applied to two models of 100 and 200 rooms. Because installation costs vary widely with the installation site and delivery conditions, the figures should be read strictly as rough ranges.

The basic set of three items — a portable generator, a storage battery and a larger receiving water tank — comes to ¥10.88–14.21 million for the 100-room model, or ¥109,000–142,000 per room. The 200-room model comes to ¥16.35–23.94 million, which falls to ¥82,000–120,000 per room. Because generators and wells do not increase in proportion to room count, the per-room burden becomes lighter the larger the property. The extended set, which adds a well and an LP gas bulk tank to a permanent generator, has a wide range: ¥19.44–39.22 million for 100 rooms and ¥34.98–56.40 million for 200 rooms. Most of that range comes from the unit price of the permanent generator, because there is a fourfold gap between the market-rate lower bound and the list-price upper bound.

Regulation also affects the choice of fuel. Diesel oil is a hazardous material under the Fire Service Act (Class II petroleum), with a designated quantity of 1,000 liters. Running a generator that consumes, say, 15 liters per hour for 72 hours requires 1,080 liters, which exceeds the designated quantity. The more that is stockpiled, the heavier the storage permit and equipment requirements become; this is the background to why 72-hour diesel capacity does not advance easily in city-center buildings. LP gas, by contrast, keeps its quality for about 20 years according to Iwatani’s commentary, whereas diesel oil needs replacing once every five years. The unit price is about 1.5 times that of a diesel unit and the maximum output is small at 54kVA, but combined with a bulk tank, a single fuel can cover power generation, hot water and the kitchen heat source. For storage batteries, materials from a METI study group put the equipment cost in subsidized projects at an average of ¥92,000/kWh (FY2023, 11 projects) and the prevailing price outside subsidized projects at around ¥200,000/kWh; batteries suit the role of keeping communications and front-desk functions running until the generator starts up.

We estimate maintenance at ¥370,000–970,000 per year per generator. However, hotels inside the lines in Table 1 already carry out statutory inspections, so if the investment takes the form of upgrading an existing generator, the increment is limited to around ¥30,000 a year for fuel renewal. When a generator is added voluntarily, the inspection cost is added in full, so whether the unit can double as mandatory equipment is the fork that determines how heavy maintenance costs become.

Location Analysis — 59.2% of the 332,483 Rooms in 16 Trade Areas Are Inside at Least One of the Four Risk Areas

We now turn to location. For 16 trade areas nationwide from Sapporo to Naha, we took the 2,711 properties and 332,483 rooms with 10 or more rooms that lie within a 1.5km radius of the center point and for which selling prices were observed between September 2025 and August 2026, and determined whether each property’s location falls inside the four areas: flood (inundation depth of 0.5m or more), tsunami, landslide and liquefaction. Note that we count at the points where rooms actually stand, not by the land area of the trade area.

Across the 16 trade areas in total, 196,684 rooms (59.2%) are inside at least one of the four risk areas, and 116,126 rooms (34.9%) are where two or more overlap. By type, flood of 0.5m or more accounts for 45.7%, liquefaction 40.5% and tsunami 17.5%, while landslide is only 0.4%. What characterizes the location of urban hotels, it can be said, is the overlap of flood and liquefaction. Alluvial lowlands and reclaimed land are also where railways and commerce cluster: in Fukuoka (Hakata/Tenjin) 70.7% of rooms, and at Nagoya Station 56.4%, fall under both flood and liquefaction.

The differences between trade areas are large. In Hiroshima (Hatchobori/Station) and Kochi (Harimayabashi), every room is inside one of the areas, directly reflecting a landform in which the city spreads across the deltas of the Ota and Kagami rivers. Osaka Umeda stands at 94.8% and Yokohama (Kannai/Minato Mirai) at 92.9%, both with tsunami and liquefaction added. In contrast, Sendai Station is at 0.0%, Shinjuku at 0.1% and Osaka Namba at 14.7%. These are trade areas where lodging is clustered on uplands and terraces. However, there is no visible correspondence between being inside or outside an area and the level of ADR. Median ADR is ¥11,900 in Umeda, where the share inside any area exceeds 90%; ¥15,400 in Shinjuku, which is almost entirely outside; and ¥19,200 around Tokyo Station, at 66.1%. Rates are determined by the depth of demand. Risk areas are best read not as an indicator of whether a location is good or bad, but as a design condition that determines the order in which equipment is added.

Priorities by Risk Category — “Placement” in Inundation Areas, “Water” in Liquefaction Areas, “Power” Outside the Areas

So which equipment is most effective to add first? Even the same generator works differently depending on where it is located. If the generator and fuel tank are placed in the basement or on the first floor in a place where inundation is expected, they may be unusable at the critical moment, and on liquefaction-prone ground, restoring buried water, sewer and gas pipes tends to take time. We have set out the equipment to tackle first for each of the four risk categories.

In flood and tsunami areas, deciding where to put equipment comes before buying it. We have separately estimated the cost of moving substation equipment to upper floors at ¥30,000–250,000 per room, and this should be considered together with the basic set in this article. Because portable generators can be stored on upper floors or the roof, they can be an option that takes effect before permanent units in inundation areas. In liquefaction areas, it is worth assuming situations where water stays off longer than electricity. Additional cost for a receiving water tank can be contained by adding capacity at the regular replacement cycle, while a well presupposes confirmation of ground conditions. Our analysis of liquefaction and room stock in Niigata’s Chuo Ward found that even on liquefaction-prone ground, no difference appears in ADR.

On the other hand, the 135,799 rooms (40.8%) that fall under none of the four risks are the segment where the effect of resilience investment shows up most directly. Because the building and equipment are less likely to be damaged, the likelihood of continuing to operate is high as long as power and water are self-reliant, and the property’s role as a shelter base stands out when surrounding facilities have stopped. In trade areas such as Sendai Station and Shinjuku, the combination of 72-hour power and a shelter agreement can be said to be effective.

How to Think About Payback — The Basic Set Balances at 1.2–4.7 Events, While Years Range From 5.5 to More Than 80

Finally, we consider payback on the investment through three routes. The first is revenue preserved when the hotel can keep operating during power and water outages. Assuming the 16-trade-area median ADR of ¥13,200 (N=1,967 properties) and occupancy of 75.3% (full year; the business hotel average in the Japan Tourism Agency’s 2025 final figures), room revenue for the 100-room model is about ¥1 million per day, and about ¥2 million for 200 rooms. If independent power and water supply avoid a three-day closure, about ¥2.99 million of revenue is preserved for 100 rooms; for seven days, about ¥6.98 million. Dividing the Capex of the basic set by these amounts, the investment balances at 1.2–4.7 disaster events without subsidies, and at 0.6–2.4 events after applying a 1/2 subsidy rate.

The second is insurance. Flood premium rates for fire insurance have been subdivided according to regional risk since the October 2024 revision. The annual cash effect, however, is small. If property and casualty insurance premiums are put at ¥4,000 per room per year (an editorial rule of thumb obtained by dividing the premiums stated in listed hotel REITs’ securities reports by the number of rooms held), then even if equipment measures produce a 5–10% difference, the amount is ¥20,000–40,000 a year for 100 rooms. PML (an indicator expressing the scale of damage a building suffers in an earthquake as a ratio of replacement cost) measures damage to the building, so adding emergency power does not move PML itself. The insurance and PML route is thought to work less through rates than through the qualitative side: how business continuity is described in engineering reports and loan screening.

The third is payment for sheltering under an agreement. Use of hotels and ryokan as evacuation shelters under the Disaster Relief Act is capped at ¥10,000 per person per night including tax (not including meals), so accommodating two people each in half of 100 rooms yields up to ¥1 million per day in room charges alone. The Cabinet Office guidelines call for amounts below normal room rates, so this will not exceed ordinary room revenue (about ¥1 million), but it serves as a floor that secures occupancy even when tourism and business travel demand stops. Because it uses the same rooms as the first route, the two cannot be added together, but it means having two exits: normal sales if demand remains, sheltering if it stops.

Once an event frequency is assumed and the result converted to years, the outcomes spread widely. For the basic set, Scenario A — using subsidies and assuming one event every 10 years with a 7-day closure avoided — gives 7.4 years for 100 rooms and 5.5 years for 200 rooms, while Scenario C — no subsidy, one event every 20 years and 3 days — exceeds 70 years. The extended set takes 12–13 years even in Scenario A. To sum up so far, expected value alone can explain payback only up to “the basic set with subsidies”; investment beyond that has to be measured by value other than preserved revenue. With construction costs continuing to rise, the fact that building the equipment in at the time of new construction or major renovation makes it easier to contain costs than adding it later should also be factored into the decision.

That value is the valuation of the property as an asset. A hotel equipped with independent power and water supply that has concluded a shelter agreement with its municipality can demonstrate business continuity in numbers and in writing during due diligence at the time of sale. As Table 5 shows, 60% of rooms in the 16 trade areas are inside one of the risk areas, and being located inside an area has itself become a standard condition for urban hotels. What makes the difference is likely to be less whether a property is inside or outside an area than how much preparation it has built up against that condition. Resilience investment is positioned less as recovering a cost than as upside that makes an asset more likely to be chosen within the same trade area.

Sensitivity Analysis — What Drives the Payback Period Is “Closure Days Avoided” More Than Event Frequency

Moving the assumptions along two axes shows that what drives the payback period is the number of closure days avoided per event, more than event frequency. Even assuming one event every 20 years, extending the days avoided from 3 to 10 shortens payback from 69.8 years to 23.7 years without subsidies, and to 11.9 years when combined with a 1/2 subsidy rate. Water and fuel reserves that allow operations to continue beyond 72 hours directly raise the efficiency of payback. Conversely, if event frequency is set at once every 50 years, payback takes more than 27 years even with subsidies, so the cases where expected value alone can justify the investment are limited to locations where power and water outages are expected about once every 10 years, or equipment configurations that allow a long period of closure to be avoided.

Conclusion — Add the Increment Beyond the Mandatory Line in an Order That Fits the Location

The power that the law requires of hotels covers 30–60 minutes; continuing operations for 72 hours and sheltering people belong to the domain of voluntary investment. That investment came to ¥10.88–14.21 million for 100 rooms, or ¥109,000–142,000 per room, for the basic set of a portable generator, a storage battery and a larger receiving water tank. In the 16 trade areas, 59.2% of rooms are inside a flood, tsunami, landslide or liquefaction area, and it can be said to be effective to start with placement in inundation areas, water in liquefaction areas, and power and agreements outside the areas. Payback is at a level that balances at 1.2–4.7 disaster events, and becomes shorter the more subsidies and shelter agreements are combined. Properties with generators or receiving water tanks approaching replacement would do well to estimate the increment for 72-hour capacity before replacing them at the same capacity.