China Inbound -60.4%, Yet Tokyo & Kyoto ADR Rose: Labour Day 2026

Posted: 2026.09.03

Chinese visitors to Japan numbered 313,000 in May 2026, down 60.4% year on year. Taken at face value, that figure suggests the lodging market built on Greater China inbound demand should have collapsed. Yet in that same month, estimated settled ADR rose in Tokyo (+6.7% YoY), Kyoto (+3.5%) and Hokkaido (+3.5%). Osaka was the only area to fall sharply (−26.1%), and once the drivers are separated out, that decline is not the result of weaker Chinese demand either. This article takes China’s Labour Day holiday (1–5 May 2026) as its axis and cross-checks estimated settled ADR, daily occupancy and room-level inventory sell-through across four areas — Osaka, Kyoto, Hokkaido and Tokyo — to test how far the “China −60%” headline actually reached the real economy.

Metric Definitions Used in This Article

ADR (average daily rate): An estimated settled rate (tax-exclusive equivalent) calculated by applying a property-type correction factor to the lowest published plan level each property lists on OTAs and similar channels (two guests per room, per-room rate, tax-inclusive). Cross-checked against property-level results disclosed by listed hotel REITs (91 properties, most recent three months), the median error is approximately 7%. These are estimates and differ from each property’s actual transacted rates and accounting figures. ADR at the area (prefecture) level is the median across the properties covered — the level of a typical property in that area.
OCC (occupancy): The share of sold rooms against total rooms in the area, estimated from OTA sales inventory. Used only for macro aggregates such as prefectures; not applied to individual properties.
Listed price: The average across all plans published on OTAs and similar channels (two guests per room, per-room rate, tax-inclusive). Because its basis differs from estimated settled ADR, it is used only for observing day-to-day movement.
LT (lead time): Days remaining until the check-in date. LT0 = same day. Early sell-out LT = the lead time at which remaining rooms first reached zero (a larger value means an earlier sell-out).
Data sources: MetroEngines Research / Japan National Tourism Organization (JNTO) / monthly disclosures from individual REITs

Key Takeaways

— China fell 60.4%, yet total inbound arrivals fell only 3.6%. Led by South Korea (+15.2%) and Taiwan (+14.6%), non-China markets added 344,000 visitors, absorbing 72.1% of the 477,000-visitor decline from China.
— Three of the four areas rose in May 2026. Estimated settled ADR was up in Tokyo (+6.7%), Kyoto (+3.5%) and Hokkaido (+3.5%); only Osaka fell, at −26.1%.
— Osaka’s decline began in April, not January. The drop in Chinese arrivals has applied uniformly since the start of the year, while the onset of the decline coincides with the opening month of Expo 2025 Osaka, Kansai. Against May 2024, Osaka is still +7.6%.
— Labour Day is no longer May’s demand peak. Kyoto at 90.1% and Tokyo at 92.6% both came in below their full-month May averages (92.0% and 94.7%); inventory was tighter on late-May weekdays.
— The direction is consistent with reported results. Two listed hotel REITs closed May 2026 above the prior year on occupancy, ADR and RevPAR alike.

How Labour Day 2026 Landed — Osaka Was the Only Area to Fall

Start with May 2026 estimated settled ADR in the four areas set against the same month a year earlier. Osaka moved from ¥13,600 to ¥10,100, down 26.1% (N=650 properties). Kyoto, by contrast, went from ¥16,400 to ¥17,000, up 3.5% (N=607 properties); Hokkaido from ¥9,200 to ¥9,500, up 3.5% (N=949 properties); and Tokyo from ¥14,100 to ¥15,100, up 6.7% (N=1,145 properties). Three of the four areas beat the prior year.

Measured by reliance on Greater China inbound demand, Kyoto and Hokkaido are no less exposed than Osaka. If evaporating Chinese demand fed straight through to room rates, those areas should have sunk at the same time for the story to hold together. In practice, Osaka was the only area that fell.

Estimated settled ADR, YoY change across four areas (January–July 2026)
2026	Jan	Feb	Mar	Apr	May	Jun	Jul
Osaka	+5.5%	-9.9%	-7.6%	-22.1%	-26.1%	-32.3%	-30.7%
Kyoto	+1.1%	+2.2%	+7.2%	+0.6%	+3.5%	-8.6%	-12.8%
Hokkaido	+26.5%	+11.9%	+8.9%	+2.8%	+3.5%	+0.1%	+1.4%
Tokyo	+9.7%	+4.0%	+4.8%	+5.2%	+6.7%	-3.9%	-4.8%

Year-on-year change in estimated settled ADR. Blue = increase, red = decrease. Source: MetroEngines Research; compiled by the HotelBank Editorial Team (N = Osaka 650, Kyoto 607, Hokkaido 949, Tokyo 1,145 properties)

Laid out month by month, a clearer shape emerges. Osaka was in positive territory in January at +5.5%, then dropped abruptly to −22.1% in April, widening to −26.1% in May, −32.3% in June and −30.7% in July. The decline began in April, not January. That single point governs everything that follows. Before the holiday began, forward booking conditions had pointed the same way: pace for the Labour Day window was building even with inbound arrivals from China down about 60%.

Osaka’s −26% Is Not China-Driven — Isolating the Expo Base Effect

The decline in Chinese visitors has been continuous since the start of 2026. JNTO estimates put January at −60.7%, February at −45.2%, March at −55.9%, May at −60.4% and June at −57.3% — every month of the first half well below the prior year. In other words, the China factor has applied uniformly since January.

So why did Osaka alone break down from April? The answer lies on the comparison side. Expo 2025 Osaka, Kansai ran from 13 April to 13 October 2025. From April 2026 onward, in other words, the “same month a year earlier” is Osaka during the Expo.

Source: MetroEngines Research; compiled by the HotelBank Editorial Team (estimated settled ADR for May of each year)

Line up three years and the picture is unmistakable. Osaka’s May estimated settled ADR traced ¥9,300 in 2024 → ¥13,600 in 2025 (+45.7%) → ¥10,100 in 2026 (−26.1%). Over the two years combined, that is a net +7.6%. It is hard to read more into this than that a one-off Expo premium carried in 2025 came off, leaving the rate slightly above its pre-Expo level.

For comparison, over the same two years Tokyo built up +22.1%, Kyoto +14.1% and Hokkaido +9.1%. Osaka’s +7.6% is slower than the others, but it points the same way. Had evaporating Chinese demand pushed rates down, the two-year figure should be negative as well — and it is not.

The reading that follows is simple. Osaka’s year-on-year shortfall reflects a comparison against a one-off demand peak in 2025, not a decline in Chinese inbound demand. If the China factor were the driver, the place to look is January to March, when it was already applying uniformly — and over that stretch Osaka ran between −9.9% and +5.5%, Kyoto between +1.1% and +7.2%, Hokkaido between +8.9% and +26.5%, and Tokyo between +4.0% and +9.7%.

Reading Real Demand from Inventory Sell-Through — Labour Day Is No Longer the Peak

Price embeds the seller’s expectations, so confirming real demand requires looking at how inventory actually cleared. The following traces daily occupancy for the four areas in May 2026 (estimated on an OTA-listed inventory basis).

Source: MetroEngines Research; compiled by the HotelBank Editorial Team (Osaka 705 properties / 111,000 rooms; Kyoto 1,026 properties / 54,000 rooms; Hokkaido 1,289 properties / 92,000 rooms; Tokyo 1,331 properties / 182,000 rooms)

Across 1–5 May, the Labour Day window, all four areas peaked on 3 May (Osaka 94.7%, Kyoto 96.4%, Hokkaido 95.1%, Tokyo 96.1%). That window, however, overlaps entirely with Japan’s Golden Week, so it cannot be separated from domestic demand.

Compare instead the five-day Labour Day average against the full-month average. Kyoto ran 90.1% over Labour Day versus 92.0% for May as a whole, and Tokyo 92.6% versus 94.7% — in both cases the Labour Day window came in below the monthly average. The Greater China holiday no longer forms May’s demand peak for Kyoto or Tokyo. Osaka, by contrast, ran 89.2% against 88.0% and Hokkaido 88.9% against 84.5%, with the Labour Day window slightly above the monthly average.

Lining up May’s Fridays makes the point concrete: on Friday 1 May, the first day of Labour Day, occupancy was 89.5% in Tokyo and 82.4% in Kyoto, whereas Friday 22 May reached 98.8% in Tokyo and 98.1% in Kyoto. Inventory was tighter on late-May weekdays than over Labour Day.

Source: MetroEngines Research; compiled by the HotelBank Editorial Team (average remaining rooms and average occupancy for Labour Day, 1–5 May, and the weekdays immediately following, 6–10 May)

The drop-off after the holiday is worth noting as well. Comparing the five Labour Day dates with the five weekdays immediately after (6–10 May), occupancy falls steeply in Osaka, from 89.2% to 78.3% (−10.9pt), and Hokkaido, from 88.9% to 76.1% (−12.8pt), while the trough is shallower in Kyoto, 90.1% to 83.1% (−6.9pt), and Tokyo, 92.6% to 85.5% (−7.2pt). Kyoto and Tokyo depend comparatively less on holiday peaks; their inventory clears in ordinary periods too.

Early Sell-Outs Measured at the Room Level

Beyond area aggregates, it is worth checking when individual properties ran their remaining rooms down to zero. The table below covers the five Labour Day dates, limited to properties whose OTA-published allotment is at least 30% of total rooms.

Room-level early sell-out results over the five Labour Day dates (1–5 May 2026)
Area	Property	Type	Total rooms	Avg. early sell-out LT	Days sold out	Published allotment
Osaka	Hotel Consort Shin-Osaka (ホテルコンソルト新大阪)	City hotel	319 rooms	LT36	1 of 5	57%
	Toyoko Inn Osaka Namba Nippombashi (東横INN大阪なんば日本橋)	Business hotel	238 rooms	LT34	4 of 5	31%
	Toyoko Inn Shin-Osaka Chuo-guchi Honkan (東横INN新大阪中央口本館)	Business hotel	202 rooms	LT34	3 of 5	37%
Kyoto	Kyoto Travelers Inn (京都トラベラーズイン)	Business hotel	78 rooms	LT33	3 of 5	38%
	Kyoto Daiichi Hotel (京都第一ホテル)	Business hotel	165 rooms	LT28	3 of 5	93%
	Hotel Binario Saga Arashiyama (ホテルビナリオ嵯峨嵐山)	City hotel	28 rooms	LT27	3 of 5	79%
Hokkaido	Hotel Edel Warme (ホテルエーデルヴェルメ)	Resort hotel	74 rooms	LT34	3 of 5	45%
	Dormy Inn PREMIUM Sapporo (ドーミーインPREMIUM札幌)	Business hotel	168 rooms	LT34	1 of 5	64%
	JR Inn Asahikawa (JRイン旭川)	Business hotel	198 rooms	LT34	1 of 5	34%
Tokyo	Super Hotel Shinagawa Aomono-yokocho (スーパーホテル品川・青物横丁)	Business hotel	104 rooms	LT34	3 of 5	57%
	Super Hotel Shinjuku Kabukicho (スーパーホテル新宿歌舞伎町)	Business hotel	117 rooms	LT34	3 of 5	46%
	Sauna & Capsule Hotel Hokuou (サウナ&カプセルホテル北欧)	Capsule hotel	198 rooms	LT34	2 of 5	50%

Room-level aggregation over the five Labour Day dates (1–5 May 2026). Properties whose OTA-published allotment is at least 30% of total rooms only. Coverage: Osaka 169 properties / Kyoto 154 / Hokkaido 292 / Tokyo 237. Source: MetroEngines Research; compiled by the HotelBank Editorial Team

Across all four areas, average early sell-out LT fell within a range of 23 to 36 days. Because inventory observation began in mid-March 2026, the maximum observable lead time for early-May check-in dates is roughly 50 days, and allotments that sold out before that point are not captured here. The table’s figures should therefore be read as lower bounds indicating that remaining rooms were zero at least by the stated LT.

Even so, sell-outs around LT34 in Tokyo and Hokkaido and around LT36 in Osaka are observable — not the profile of areas where demand has vanished. Properties with strong drawing power cleared their holiday allotments early regardless of what happened in Greater China.

Substitute Markets Filled 72% of the China Gap

Where, then, did the real demand come from? JNTO’s May 2026 estimates by country and region show South Korea at 951,300 visitors (+15.2% YoY), Taiwan at 616,800 (+14.6%) and the United States at 333,700 (+7.0%) — each an all-time high for the month of May. In total, 19 markets set May records.

Source: Japan National Tourism Organization (JNTO), “Foreign Visitors to Japan (May 2026 estimates)”; compiled by the HotelBank Editorial Team. Prior-year figures by country and region are approximations back-calculated from the year-on-year rates JNTO published

Converting the changes into visitor counts reveals the structure. China fell by roughly 477,000. Against that, South Korea added 125,500, Taiwan 78,600, the United States 21,800, and all other markets a combined 118,100. Excluding China, arrivals grew from 2.903 million to 3.247 million, up 11.9% (+344,000) — absorbing 72.1% of China’s decline. The total consequently held at −3.6%, from 3,693,300 to 3,559,900. Whether the same pattern holds in regional Japan is examined against April lodging statistics in Did Non-China Inbound Fill China’s −58% Gap?.

The “Chinese arrivals −60%” figure is therefore real as a number. JNTO attributes the decline to travel advisories regarding Japan and reduced flight capacity, plus the Dragon Boat Festival — which normally generates inbound demand in late May — falling in mid-June in 2026. The net effect on total volume in the lodging market, however, compressed to −3.6%. A −60% result in one market cost the whole only 3.6% precisely because substitute markets filled the gap in volume terms.

Consistency with Monthly REIT Disclosures

It is worth checking these OTA-based estimates against reported results. Invincible Investment Corporation (インヴィンシブル投資法人) disclosed monthly operating results for its 101 domestic hotels showing May 2026 occupancy of 85.5% (+0.6pt YoY), ADR of ¥15,288 (+3.9%) and RevPAR of ¥13,066 (+4.7%). Japan Hotel REIT Investment Corporation (ジャパン・ホテル・リート投資法人), covering 29 hotels on variable-rent and similar structures, likewise posted May 2026 occupancy of 84.0% (+1.0pt YoY), ADR of ¥21,395 (+6.6%) and RevPAR of ¥17,961 (+7.9%) — all above the prior year.

In a month when travellers from China fell by 60%, listed hotel REIT portfolios closed above the prior year on both occupancy and rate. That is consistent with the direction the estimated settled ADR figures here point to: Tokyo +6.7%, Kyoto +3.5%, Hokkaido +3.5%.

Conclusion — How to Reframe the “China −60%” Headline

The conclusions drawn from the measured Labour Day 2026 data are as follows.

First, “Chinese arrivals −60.4%” is a fact. Its effect on total inbound arrivals, however, stopped at −3.6%. Nineteen markets, led by South Korea, Taiwan and the United States, set May records, and arrivals excluding China grew 11.9% (+344,000), absorbing 72.1% of the decline.

Second, explaining Osaka’s −26.1% through the China factor does not hold up. The Chinese decline has applied uniformly since January 2026, yet Osaka’s fall began in April — a point that lines up precisely with the opening of Expo 2025 Osaka, Kansai on 13 April 2025. Against May 2024, Osaka’s estimated settled ADR remains in positive territory at +7.6%. What came off was the Expo premium, not underlying demand.

Third, the Greater China holiday is no longer May’s demand peak. In Kyoto and Tokyo, average occupancy over the five Labour Day dates fell below the monthly average, and inventory was tighter on late-May weekdays. That structural shift suggests the assumptions behind the pricing calendar need rebuilding. The conventional design — stacking rate peaks onto the Greater China holiday schedule — is out of step with actual conditions, at least in Tokyo and Kyoto in 2026.

For operators, the implication comes down to how to spread sensitivity away from any single market. South Korea and Taiwan are already larger sources than China in volume terms, and their patterns of length of stay, room-type preference and F&B demand are not the same as the Chinese market’s. There is considerable room to rebuild inventory and pricing designs that were optimised around Labour Day and Lunar New Year around the school-holiday and long-weekend calendars of South Korea and Taiwan.