In hotel acquisition work, the seismic risk assessment in the engineering report and the fire insurance quotation usually sit in two different folders. The first gets read as a question about construction cost; the second gets glanced at once as a small line in the expense schedule. Yet these two documents, without a single yen of disaster-mitigation investment, quietly move both the annual P/L and the very structure of the debt from the moment the site is chosen. This article sets Capex aside entirely and breaks down how much the same hazard differential is worth to NOI and equity returns through two channels — the insurance premium paid, and the LTV a lender is willing to extend — using actual data from 10 hotel markets across Japan.

The Rule Itself — Five Flood-Rate Tiers Split Premiums by Municipality

On 21 June 2023, the General Insurance Rating Organization of Japan (GIROJ) filed a revision to the fire insurance reference loss cost rates with the Commissioner of the Financial Services Agency, receiving notice of the conformity review result on 28 June of the same year. The revision had two pillars. One was a 13.0% nationwide average increase in the reference loss cost rates for comprehensive residential insurance. The other — the subject of this article — was subdividing the flood rate, previously uniform nationwide, into five tiers according to regional risk. Insurers have been rolling this into their products in stages since October 2024.

The design of the subdivision is set out in GIROJ’s published materials. The geographic unit is the municipality. There are five tiers, from Zone 1 (cheapest premium) to Zone 5 (most expensive). Relative to no subdivision at all, Zone 1 comes in about 6% below and Zone 5 about 9% above, so the most expensive area runs roughly 1.2 times the cheapest. Both figures are on total premium — fire, wind, snow and flood cover combined — and are stated after the transitional smoothing measures designed to avoid sharp increases.

The data behind the zone assignments is also public. Within flood risk, river (fluvial) flooding accounts for 56.2% of flood damage, with inland (pluvial) flooding, landslides and other causes making up the remaining 43.8%. Fluvial flooding is assessed using flood inundation hazard maps; pluvial flooding and landslides draw on flood damage statistics and topographic data. What matters in practice is a caveat GIROJ itself states explicitly: because the zones reflect total flood risk including pluvial flooding and landslides — not fluvial flooding alone — they do not necessarily align with general risk information such as flood hazard maps. A site left white on the hazard map is not guaranteed a low zone.

Note that as of August 2026, GIROJ’s June 2023 filing remains the most recent published revision to the fire insurance reference loss cost rates; no subsequent filing is available on the organization’s website. This article does not incorporate figures reflecting any future revision.

Ten Markets Nationwide — Zones Split Even Within the Same City

We selected 10 markets where hotels actually cluster and looked up the flood-risk zone for each municipality in GIROJ’s flood zone search system. Alongside that, we pulled from the National Research Institute for Earth Science and Disaster Resilience (NIED) National Seismic Hazard Maps (J-SHIS), for the centre point of each market, the peak ground velocity for a 475-year return period (a 10% probability of exceedance over a 50-year service life) and the probability of experiencing JMA seismic intensity 6-lower or greater within 30 years. The former is the ground-motion input side of a PML calculation.

The results show how risky it is to talk at the level of a city. Within Tokyo’s 23 wards, Shinjuku, Shibuya and Toshima are Zone 1, while Sumida, Taito, Koto and Edogawa are Zone 5. The minimum and the maximum coexist inside the same 23 wards. Nagoya is even sharper: Nakamura Ward in front of Nagoya Station is Zone 5, while Sakae (Naka Ward), about 2km due east, is Zone 1. In Osaka, Kita Ward (Umeda) and Chuo Ward (Shinsaibashi) are both Zone 1, while the bayside wards of Minato and Konohana are Zone 3. At the granularity of “it’s Tokyo” or “it’s Nagoya,” this difference disappears.

The spread on the seismic side is far wider than on flood zones. Peak ground velocity for a 475-year return period ranges from 35.7cm/s in Nagano City to 166.0cm/s in Suruga Ward, Shizuoka — a factor of about 4.7. On the probability of intensity 6-lower or greater within 30 years, Nagano City and Aoba Ward in Sendai sit at 0.8%, while Suruga Ward reaches 44.3% and Nagoya Station 32.3%. Because PML is set by “this ground motion × that building’s seismic performance,” placing an identical building of identical vintage in each location produces PMLs that differ by multiples on location alone. For an analysis pairing seismic risk with ADR and risk-adjusted yields across Pacific-coast destinations, Nankai Trough 80% Era: Pacific Coast Hotel Investment works through 10 tourist destinations.

Plotted on these two axes, the 10 markets do not line up neatly. It becomes clear that flood and earthquake are independent axes. Hakata in Fukuoka is flood Zone 3 but its ground motion of 47.7cm/s is second-lowest of the 10. Aoba in Sendai is flood Zone 1 with ground motion of 46.9cm/s — low on both. Nagoya Station, conversely, is flood Zone 5 with ground motion of 141.0cm/s, high on both. Rolling “disaster risk” into a single score erases this distinction. Since insurance quotations and financing terms are decided on separate axes, keeping them separate in due diligence matches how the work is actually done. The same hazard information can also be read as the return on resilience Capex — the trade-off between mitigation spend and payback is a distinct exercise that sits outside this article’s scope.

Channel 1: P/L — Insurance Premium Moves 0.25–0.61% of GOP

Start by counting the recurring expense. To isolate the effect of market-level conditions, building specification, insured amount and GOP margin are held identical across all markets; the only variables are that market’s ADR and OCC, and that municipality’s flood-risk zone. The model assumptions are as follows.

The annual premium gap between Zone 1 and Zone 5 is about ¥1.01M. As a share of GOP that runs from 0.25% in Shinjuku, where GOP is thick, to 0.61% in Suruga, where it is thin. Converted to RevPAR terms, at 80% occupancy across 200 rooms, that is roughly ¥17 per room sold — about ¥14 on a total-room-nights basis. Less than 0.1% of a night’s rate.

Put plainly, the result means there is no need to walk away from a location on the flood-zone premium differential alone. Two qualifications apply, however.

The first qualification. The −6% / +9% on reference rates applies to total premium across fire, wind, snow and flood cover, and includes transitional smoothing. If a commercial underwriter applies a specific regional coefficient to the flood portion alone, the gap widens. If flood accounts for 25% of premium and the coefficient runs 0.7x to 1.5x, the gap becomes ¥1.35M a year; at 0.7x to 2.0x, ¥2.19M; at 0.6x to 2.5x, ¥3.21M. That puts it in a range of 0.3–1.9% of GOP. At that point it is hard to keep calling it a rounding error.

The second qualification. In commercial lines the difference often shows up in coverage terms rather than premium. A flood deductible set at tens of millions of yen per occurrence, or a flood sub-limit capped at a fixed percentage of the insured amount, will not appear in the quoted premium. In acquisition work, the comparison to run is not premium alone but a terms table covering three items: deductible, sub-limit and coinsurance. Two policies that both say “flood cover included” can carry very different retained exposure once the property sits in Zone 5.

Read the other way, a property in Zone 1 or Zone 2 has more room to negotiate on those same terms. The advantage of a low-risk location shows up not as a few hundred thousand yen of premium but as “wider cover for the same premium.” Whether that difference in terms makes it into the NOI underwriting at acquisition is what separates practice from theory.

Channel 2: B/S — PML Moves IRR 1.9pt via Earthquake Cover and LTV

Now the second channel. PML (probable maximum loss) is the metric produced by the seismic risk assessment in an engineering report, defined as the maximum expected physical loss as a percentage of replacement cost. The reference ground motion corresponds to a 475-year return period — an event with a 10% probability of exceedance over a building’s 50-year service life. The “475-year return-period peak ground velocity” shown in the table above is precisely the input side of that PML calculation.

In practice, this number feeds straight into financing terms. A report published by the risk management arm of Sompo Japan notes that “where PML exceeds a certain level (generally around 15%), earthquake insurance is often placed.” Where cover is required, the premium eats into NOI. On top of that, lenders generally structure terms with lower LTV for properties whose collateral value can swing more widely. PML, in other words, triggers a higher premium and reduced leverage at the same time.

The reading is simple. A property with low PML that can carry 60% LTV without earthquake cover produces an equity IRR of 6.36%. The same NOI, where PML forces LTV down to 50% and requires ¥30M a year of earthquake cover, yields 4.42%. The gap is 1.94pt, and equity required rises from ¥2.20bn to ¥2.70bn — an extra ¥500M. To reach the same return, you either pay less for the asset or grow NOI.

Splitting the effect, the move from 60% to 50% LTV costs 0.79pt and the ¥30M earthquake premium a further 1.15pt. Because leverage works less hard as rates rise, this picture shifts with the rate environment. At 1.8%, the gap between 60% and 50% LTV is 0.98pt; at 3.4% it narrows to 0.24pt. The lower the rate environment, the more valuable it is to be able to take LTV — which also means that in a rising-rate environment, the relative weight of a PML-driven LTV constraint falls. On how far the rate level itself moves the feasibility of development, Policy Rate 1.00%: Break-Even ADR Across Japan’s 8 Hotel Markets runs the numbers market by market.

Set the two channels side by side and the difference in magnitude is obvious. The flood-zone gap (¥1.01M a year) moves IRR by roughly 0.05pt. The PML-driven change in terms moves it by up to 1.94pt — a difference of about 40 times. If you are allocating time in acquisition due diligence, the conclusion is to finish the flood-zone check on the insurance quotation in 30 minutes and spend the rest on PML and the design of the financing.

What should be stressed here is that this is not an argument that high-PML locations are disadvantaged. PML is not determined by location alone. For the same ground motion, structural system, design vintage, seismic grade and ground conditions move PML substantially. The practical implication runs the other way: in markets with high ground motion, a building with low PML can command relatively better financing terms. In markets such as Shizuoka or Nagoya Station, the PML level itself becomes a negotiating lever on price, and the funding premium attached to post-1981 seismic-code buildings is large. In markets with low ground motion, that scope for differentiation is small.

Sorting 298 Listed Hotel REIT Properties Into Flood Zones

How is this expense distributed across the market as a whole? We extracted the domestic properties from the portfolio master of Japan’s seven listed hotel REITs and matched each property’s municipality to GIROJ’s flood-risk zones. 298 properties and 52,106 rooms could be classified (properties with disclosed acquisition prices total approximately ¥1.80 trillion). For designated cities, the lookup was run down to the ward level.

The distribution shows a clear pattern. By property count, Zones 1–3 account for 82.5% and Zones 4–5 for 17.4%. By room count, Zones 4–5 are 14.1%; on an acquisition-price basis, just 10.7%. Zone 1 accounts for 21.8% of properties but 26.1% of acquisition value, which suggests the higher the price per property, the more likely it sits in a lower-risk zone. That is consistent with the fact that the areas where high-price central assets cluster — the upland wards of central Tokyo, Kita and Chuo in Osaka, Shimogyo in Kyoto — all fall in Zones 1–2.

Two things follow from this distribution. First, at the portfolio level, upside in premium driven by flood zone is limited. If roughly 90% of acquisition value sits in Zones 1–3, the effect of the zone differential on total NOI is small enough on the figures in the previous section. Second, and conversely, zone still matters for an individual acquisition decision. The 52 properties in Zones 4–5 exist in the market and are being operated. A high zone is not in itself a problem; the question is whether that difference has been priced into the negotiation and into the insurance programme at acquisition.

Seven Checkpoints for Reading Insurance and PML in Acquisition Due Diligence

Translated into acquisition practice, the breakdown above comes down to the following seven items. None involves a single yen of Capex, and most can be done at the desk before work begins.

Summary

As a line item, insurance premium stays an inconspicuous row in the P/L. Yet the same hazard information sitting behind it moves the equity cheque and the IRR substantially, through the lender’s credit decision. Check zone and PML separately before acquisition, and the two stop being overlooked expenses and become variables you can underwrite in advance. Low-risk locations hold an advantage on insurance terms; in high-ground-motion markets, buildings with low PML can differentiate on financing terms. Either way, there is room for a sharper investment decision.

Related Reading

• Nankai Trough 80% Era: Pacific Coast Hotel Investment — Disaster Risk × ADR × Risk-Adjusted Yield Across 10 Destinations
• Policy Rate 1.00%: Break-Even ADR Across Japan’s 8 Hotel Markets
• 226k Hotel Rooms in 32 Regional Cities: 42.7% Built Before 1996
• Station Premium Ends at an 8-Minute Walk: 8 Stations, 1,012 Hotels