Four plants under turnaround simultaneously in March–April 2026 — against zero a year earlier

The turnaround cycle is tied to the continuous-operation period permitted by law. Under the certified operator system of the High Pressure Gas Safety Act, a certified plant may run continuously for four years or such other period as approved by the minister, and facilities holding “super certification” may set their own interval of up to eight years. In other words, “once every four years” is not an industry custom but a periodicity that falls out of how the rules are designed. Crasus Chemical (Oita) states as much on its own website: “In 2026 we plan to carry out the major scheduled maintenance of our ethylene cracker, which takes place once every four years.”

The degree of concentration can be counted mechanically from the JPCA’s monthly performance memos. The chart below sets the number of crackers under turnaround in each month side by side for 2025 and 2026. In 2025, January through April were zero and one or two crackers were sporadically down between May and August. In 2026 there was one in January, two in February, and then four shut down simultaneously in March and April. Ethylene output fell to 272,600 tonnes in March, down 38.8% year on year.

One caveat is in order. The 2026 output decline was not caused by turnarounds alone. For June 2026 the JPCA broke the decline down as “turnaround-related factors -5.8%, utilisation-rate variation -13.3%”, with difficulty sourcing naphtha following the effective closure of the Strait of Hormuz at the end of February contributing at the same time. The size of the production drop cannot be read directly as the scale of turnaround activity. What matters for lodging demand is not output volume but the number of plants shut down and the length of the work window. Turnarounds go ahead as scheduled even when feedstock is not arriving; if anything, a longer shutdown widens the window for the work. At Tosoh’s Yokkaichi Complex (Yokkaichi, Mie), the major turnaround scheduled for March–April was followed by a restart that was postponed because of the naphtha shortage, with the plant eventually coming back online on 25 April.

Lodging capacity differs by up to 8.3x — 364 rooms within 5km at Oita against 3,023 at Yokkaichi

Turnaround technicians want to be as close to the site as possible, but petrochemical complexes sit almost without exception in exclusive industrial districts, and there are no hotels next door. In practice they stay within “car or bus commuting range”. We therefore counted the properties and rooms captured by MetroEngines Research in 5km, 10km and 15km distance bands from each complex’s centre point.

The structural differences are stark. Yokkaichi has 3,023 rooms within 5km, and the cluster of business hotels in front of Kintetsu Yokkaichi Station serves directly as the base for turnaround work. Ichihara is thin within 5km at 1,550 rooms, but thickens rapidly to 3,530 at 10km and 10,779 at 15km, because central Chiba City comes into range. Mizushima reaches 5,128 rooms within 10km, most of them around Kurashiki Station. And Oita has just 9 properties and 364 rooms within 5km, the thinnest of the four, rising only to 1,636 rooms at 10km. The main hotel cluster in front of Oita Station sits 13–15km from the complex. Compared within 5km, Yokkaichi holds 8.3 times as many rooms as Oita.

Yokkaichi’s estimated settled ADR rose 45.9% year on year in March 2026 — Mizushima shows nothing

So did the turnarounds actually move room rates? We compared estimated settled ADR at the municipal level against the same month a year earlier, restricted to months on a consistent actuals basis (through July 2026). Future months are estimates based on currently listed rates and sit on a different basis from actuals months, so they are excluded from the year-on-year comparison.

Yokkaichi was up 45.9% year on year in March (estimated settled ADR ¥6,800 → ¥9,900), clearly separated from the other three cities. That is exactly the window in which Tosoh’s Yokkaichi Complex ran its March–April turnaround. April reversed to -10.0%, suggesting demand peaks early, in the ramp-up phase of the work. Oita had two peaks — +21.4% in January and +11.1% in March — while Ichihara clustered at the start of the year, +16.1% in January and +25.0% in February. Kurashiki, by contrast, had almost no positive months across 2026, running below the prior year at -13.4% in May and -25.9% in June.

Because municipal aggregates also sweep in leisure demand and other business demand, we ran the same calculation on a compset (comparison group) limited to properties within 10km of a complex with 30 or more rooms. The results point in much the same direction as the municipal figures.

Where the difference comes from is readable once you set this against the room stock in Table 1. Yokkaichi has 3,023 rooms within 5km, and the 19 business-oriented properties in front of Kintetsu Yokkaichi Station form the core of the compset. Turnaround demand lands squarely on that narrow trade area, so it showed up in rates without much interference. Oita has only 364 rooms within 5km to begin with, making it easy for demand to concentrate on scarce inventory. Mizushima, by contrast, has most of its 5,128 rooms within 10km around Kurashiki Station, mixed in with leisure demand for the Bikan Historical Quarter. The larger the population of properties, and the heavier the weight of leisure demand, the more the lift from repeat demand is diluted until it becomes invisible. This is the same structure we found in the semiconductor-cluster rate analysis in “TSMC Fab 2 & Kumamoto’s Chip Corridor: Kikuyo–Koshi–Ozu ADR 2024–2026”: whether a demand event can be detected in rates is determined not by the size of the demand but by the dilution rate on the supply side.

Land prices up 9.2% at Ichihara and 4.7% at Oita against 0.5% around Mizushima — trade areas are diverging

As background to the investment case, we checked land prices and building regulations at each location from the MLIT Real Estate Information Library. For commercial land in the 2026 Published Land Prices, we aggregated the standard sites within the tiles covering each location’s lodging cluster.

Positioning within 10km of Yokkaichi — the business tier above ¥16,000 is empty

For Yokkaichi, where turnaround demand showed up most clearly in rates, we mapped the 19 properties within 10km with 30 or more rooms by room count and estimated settled ADR. The horizontal axis is room count, the vertical axis is average estimated settled ADR for January–July 2026, and circle size is proportional to room count.

The supply pipeline is limited — 7 projects, 2,654 rooms across the four prefectures on a building-application basis

We checked how the supply side is responding to this repeat demand using two sources: new openings on an OTA-listing-confirmed basis (establishment dates in the property master), and building plans from MLIT’s Building Construction Dynamics Statistics Survey.

New openings across the four prefectures are numerous in property terms but small in scale — 17 rooms per property in Mie in 2026 and 8 in Oita. These are mostly vacation rentals and simple lodgings, not at a scale that can absorb group demand from a turnaround. On a building-application basis, 5 projects totalling 2,164 rooms are confirmed in Chiba and 2 projects totalling 490 rooms in Okayama, while Mie and Oita currently have none. No new supply aimed at this demand has entered the trade areas where repeat demand is observable.

The investment question — how to evaluate NOI volatility on a four-year cycle

The investment significance of turnaround demand lies not in the single-year revenue impact but in how to underwrite an NOI that swings on a regular four-year cycle. We modelled the turnaround-year uplift for a typical business-oriented hotel within 10km of the Yokkaichi complex.

■ Modelling assumptions

• Subject: a 150-room limited-service hotel (around the middle of the size band in the Yokkaichi 10km compset)
• Normal-year ADR: ¥7,455 (average of the 12 monthly medians for 2025 in that compset, N=18–19)
• Turnaround-year ADR lift: +¥2,859 (March 2026 ¥10,535 less March 2025 ¥7,676)
• Duration of the lift: 61 days (roughly two months, the typical turnaround work window)
• Assumed occupancy: 75% (identical in normal and turnaround years; an assumption for the model, not an observed figure)
• GOP margin: 40% (the lower end of the 40–60% industry guide for limited-service business hotels)
• Variable cost ratio on the lift: 20% (incremental variable cost against incremental revenue)

■ Sensitivity analysis

■ Two-way sensitivity (lift x duration)

Restating the same figures from a development perspective makes the position clear. At 150 rooms, 25 sqm per room and a guestroom-area ratio of 62% (all our own assumptions), gross floor area is roughly 6,048 sqm (about 1,830 tsubo). Applying a guide construction cost of ¥2.2m per tsubo for this hotel type — between the 2025 all-structure average of ¥1.952m and the ¥2.405m for steel-frame construction in MLIT’s Building Construction Starts Statistics, a mid-tier level — gives roughly ¥4.03bn. Assuming a 2028 opening and 24 months of construction, and factoring in inflation during the build (2026 +5.3%, 2027 +5.0% — our own assumptions rather than sourced fixed values, set in light of the upward trend in the construction cost deflator), the figure is about ¥4.45bn. The annualised +¥3.9m from a turnaround year is just 0.09% of that investment. Turnaround demand on its own will never justify a new development.

What does matter is the revenue design of existing stock. If you know in advance that demand will spike for two months once every four years, you can build a staged pricing plan for that window and prepare products aimed at multi-night stays, extracting additional GOP from the same asset. The reversal to -10.0% in Yokkaichi in April 2026 suggests the peak sits in the first half of the work window and that there is headroom left in selling inventory in the second half. Turnaround schedules are disclosed in each operator’s business plans, so unlike event-driven demand they can be read years ahead. That predictability is precisely what separates repeat demand from one-off demand.

Listed REITs have already invested in these industrial trade areas. Invincible Investment Corporation holds HOTEL MYSTAYS Soga (Chiba City, Chiba; 112 rooms; acquired August 2023 for ¥2,039m), ART HOTEL Oita (Oita City, Oita; 228 rooms; acquired July 2024 for ¥5,484m) and HOTEL MYSTAYS Oita (145 rooms; acquired February 2018 for ¥1,604m). Ichigo Hotel REIT Investment Corporation holds THE ONE FIVE GARDEN Kurashiki (Kurashiki, Okayama; 107 rooms; acquired August 2018 for ¥1,725m). Hoshino Resorts REIT Investment Corporation acquired Comfort Inn Kurashiki Mizushima (97 rooms) and Comfort Inn Chiba Hamano R16 (93 rooms) in May 2014, making it an early investor in lodging demand from industrial trade areas.

Looking at acquisition price per room, Comfort Inn Kurashiki Mizushima was ¥7.5m per room (2014), HOTEL MYSTAYS Oita ¥11.1m (2018), HOTEL MYSTAYS Soga ¥18.2m (2023) and ART HOTEL Oita ¥24.1m (2024) — the more recent the acquisition, the higher the unit price. Business hotels in industrial trade areas have been revalued to roughly three times the acquisition unit price over a decade. Set against the new-build cost modelled above (¥4.45bn for 150 rooms, or ¥29.7m per room), acquiring existing properties still sits in a rational range relative to building new.

Risk factors — consolidation at Mizushima gives the repeat demand an end date

Conclusion — demand you can foresee can be captured through pricing design

The concentration of turnarounds in 2026 is confirmed numerically in the JPCA’s monthly memos: four crackers in March and April, against zero in the same months a year earlier. That demand showed through in room rates at Yokkaichi, which has a business-hotel cluster right by the complex (+45.9% year on year in March 2026), and at Oita, where only 364 rooms sit within 5km (+21.4% in January). Ichihara produced only a small peak at the start of the year, and Kurashiki was undetectable, mixed in with leisure demand.

Unlike one-off construction demand, repeat demand tells you when the next one is coming. That is precisely why capturing it through the pricing and product design of existing stock is more rational than building to meet it. In the model, a 150-room hotel sees turnaround-year GOP rise 12.8% (¥122.4m → ¥138.1m), with an annualised contribution to the four-year average of +¥3.9m. The contribution relative to the cost of an acquisition is small, but as an operational improvement on an existing asset it is meaningful.

At the same time, at locations such as Mizushima where production consolidation is targeted for around fiscal 2030, the repeat demand itself has an end date. If you are going to invest on the premise of cyclicality, you also need to count how many cycles are left.

Frequently asked questions

A. Under the certified operator system of the High Pressure Gas Safety Act, a certified plant may operate continuously for four years or such other period as approved by the minister, and facilities holding “super certification” may set their own interval of up to eight years. In practice a cycle of two to four years is common, and Crasus Chemical (Oita) states on its own website that “in 2026 we plan to carry out the major scheduled maintenance of our ethylene cracker, which takes place once every four years.”

A. According to the Japan Petrochemical Industry Association’s performance overview memos, the number of naphtha crackers under turnaround was 1 in January 2026, 2 in February, 4 in March, 4 in April, 2 in May, 3 in June and 2 in July. In the same months a year earlier, January through April were all zero. Ethylene output fell to 272,600 tonnes in March (-38.8% year on year), though the association notes that the decline reflects not only turnarounds but also utilisation-rate variation caused by difficulty sourcing naphtha feedstock.

A. In March 2026, when Tosoh’s Yokkaichi Complex ran its turnaround, estimated settled ADR in Yokkaichi was up 45.9% year on year (¥6,800 → ¥9,900, N=26 properties). Narrowing to a compset of the 19 properties within 10km of the complex with 30 or more rooms, the median was still up 37.2% (¥7,676 → ¥10,535). April reversed to -10.0%, however, so the lift tends to sit in the first half of the work window.

A. Oita. Within 5km of the centre point of the Oita complex there are only 9 properties and 364 rooms, and even within 10km the figure is 46 properties and 1,636 rooms. Against Yokkaichi’s 50 properties and 3,023 rooms within 5km, that is an 8.3x difference in room count. The main hotel cluster in front of Oita Station sits 13–15km from the complex.

A. In our model, a 150-room limited-service hotel sees a turnaround-year GOP uplift of ¥15.7m, with an annualised contribution to the four-year average of ¥3.9m. Setting the new-build cost for the same scale at roughly ¥4.45bn (after factoring in inflation during construction), the contribution amounts to just 0.09% of the investment. Building a development case on turnaround demand alone is difficult; capturing it through the pricing and product design of existing stock is the more realistic route.

Related reading

• Chitose & Tomakomai Worker Lodging — 7 New Hotels, 738 Rooms in 2026
• Fukushima Hamadori: 4,434 Rooms, Three-Layer Demand, Thin Center
• TSMC Fab 2 & Kumamoto’s Chip Corridor: Kikuyo–Koshi–Ozu ADR 2024–2026
• Hotel Supply Pressure Ranking 2026-2027: Ozu Kumamoto Tops at 30.8%
• Mie August Booking Curves: Business Hotels Swing 25.5pt, Ryokan 9.7pt