The industry-level figures in the Bank of Japan’s Tankan (September 2026 survey, the 210th), published on October 1, 2026, show the business conditions DI for “accommodation and food services” at +38 among large firms, down 8 points from +46 in the June survey. Mid-sized firms rose 3 points, from +33 to +36, while small firms fell 6 points, from +18 to +12, so the direction differs by firm size. The December forecast is +34 for large firms, +25 for mid-sized firms and +18 for small firms, which is −4, −11 and +6 points from the September levels, and is just as uneven. In a report dated the same day, Daiwa Institute of Research (大和総研) cited, as factors behind the decline among large non-manufacturers (+35, −2 points), a deterioration in consumer-related industries such as “accommodation and food services” and “retail” against a backdrop of rising costs including fuel and labor. This article sets out the four DIs for this industry (business conditions, employment, output prices and input prices) by firm size exactly as in the BOJ’s original tables, places them in the trend since the March 2019 survey, and compares them with MetroEngines Research’s estimated settled ADR and estimated occupancy for July–September 2026 and with the accommodation CPI from e-Stat. A DI is the share of firms answering “favorable” minus the share answering “unfavorable”; it is not a level of sales or room rates. With that in mind, we read “the DI fell” and “how room rates moved” as separate facts.

Accommodation and food services business conditions DI — large +38, mid-sized +36, small +12, moving in different directions

We begin with the original September survey figures by firm size. The table below takes the four DIs for accommodation and food services from Table 1 of the Tankan summary and from the judgment items in the full results, showing “June survey actual”, “September survey actual”, the change, and the December forecast. The business conditions DI for non-manufacturing (large and small firms) is included for comparison.

Source: Compiled by the HotelBank Editorial Team from the Bank of Japan’s “Tankan (September 2026 survey)” summary and full results (published October 1, 2026) and its Time-Series Statistical Data Search Site. Unit: percentage points

The first point to note is that the September actuals exceeded the forecasts made in June at every firm size. In the June survey, firms expected September conditions of +34 (large), +22 (mid-sized) and +7 (small), while the actuals were +38, +36 and +12, beating forecasts by 4, 14 and 5 points respectively. In other words, “down from June” and “better than we expected three months ago” are both true at once. The large-firm drop of 8 points is bigger than the −2 points for non-manufacturing overall, but the level remains 3 points above non-manufacturing’s +35, and it holds sixth place among the 12 non-manufacturing industries, behind construction (+56), real estate (+52), information services (+51), goods rental and leasing (+46) and services for businesses (+39).

Next, we place this +38 in the trend since 2019. The line chart below covers the 31 quarters from the March 2019 survey to the September 2026 survey, taken from the BOJ’s Time-Series Statistical Data Search Site. In pre-pandemic 2019, the range was +9 to +17 for large firms and −8 to +1 across all sizes. From a trough of −91 (large firms) in the June 2020 survey, the DI recovered to +36 in the June 2023 survey, peaked at +52 in March and September 2024, then fell to +26 and +25 in September and December 2025. So far in 2026 it has gone back and forth, at +34 in March, +46 in June and +38 in September, and the latest decline is the second correction from the 2024 plateau. For small firms, the high was +27 in December 2024, and the DI has since moved in a +10 to +19 range, with September’s +12 toward the lower end.

Source: Compiled by the HotelBank Editorial Team from the BOJ’s Time-Series Statistical Data Search Site (Tankan judgment-item D.I., accommodation and food services, actual). March 2019 survey to September 2026 survey, percentage points. Figures from the December 2025 survey onward are on the new basis following the revision of surveyed firms

Employment conditions DI at −58 across all sizes — narrower than −75 in December 2023, but 14 points deeper than the non-manufacturing average

Looking at the employment conditions DI on the same timeline as business conditions shows where sentiment in accommodation and food services is supported and where it is constrained. The DI is “excessive” minus “insufficient”, and the all-size total was −58 in the September 2026 survey (June: −57), with a forecast of −61. By firm size, large firms were at −54 (0 from June), mid-sized at −58 (−4) and small at −59 (−1), meaning the share of firms reporting a shortage rose slightly among mid-sized and small firms. The all-size figure is −44 for non-manufacturing overall and −38 for all industries, so accommodation and food services sits 14 to 20 points deeper.

Over time, however, the net shortage has passed its peak. In the June 2020 survey the DI stood at +27, as closures and lost demand made “excessive” the majority answer. From 2022 it swung rapidly negative, reaching its most recent low of −75 in the December 2023 survey. Since then it has been −65 to −70 through 2024, −57 to −62 in 2025, and −57 to −58 in 2026, a narrowing of about 17 points. This does not mean that firms in accommodation and food services now have staff to spare; the plainer reading is that the distribution of answers has moved from “we cannot hire” toward “we are starting to see a way to secure staff”. For how far the sense of labor shortage in the lodging industry has eased compared with other industries, see Japan Hotel Productivity at 79.6% of All Industries; 2.02 Rooms/Worker, which breaks it down across the 47 prefectures. The −61 forecast suggests this trend will stall, and it hints at firms expecting the hiring picture to tighten again ahead of the busy winter season.

Source: Compiled by the HotelBank Editorial Team from the BOJ’s Time-Series Statistical Data Search Site (Tankan employment conditions D.I., accommodation and food services, actual). March 2019 survey to September 2026 survey, “excessive” minus “insufficient”, percentage points

Output price DI 43 and input price DI 82 are 39 points apart — the forecast narrows the gap to 27 points

To read the earnings structure of accommodation and food services through the Tankan, the most informative measure is the gap between the output price DI and the input price DI, in effect a firm-level terms of trade. Across all sizes, the September survey shows an output price DI of +43 (June: +46) and an input price DI of +82 (June: +86), a gap of 39 points. In the March 2019 survey, output was +11 against input of +44, a gap of 33 points. During the pandemic the output price DI sank as low as −28 (June 2020), and in the March 2022 survey the gap widened to 58 points, with input at +62 against output at +4. That was the most input-heavy phase of the past seven years.

The output price DI then jumped from +28 in June 2022 to +48 in June 2023, and the gap narrowed to 27 points in the March 2024 survey. Since 2025, output has moved between +39 and +53 and input between +80 and +86, with the gap going back and forth in a 31 to 42 point range. The September survey’s 39 points sits in the middle of this range and is almost unchanged from 40 points in June. The forecast is what deserves attention: the output price DI is expected at +56 (+13 from actual) and the input price DI at +83 (+1), narrowing the gap to 27 points. This would be the narrowest since March 2024 (27 points), and it means firms are saying that “input costs will stay high, but we will raise selling prices from here”. By firm size, small firms have the most bullish output price forecast at +59 (+15 from actual), followed by large firms at +54 (+12) and mid-sized firms at +48 (+9).

Source: Compiled by the HotelBank Editorial Team from the BOJ’s Time-Series Statistical Data Search Site (Tankan output price D.I. / input price D.I., accommodation and food services, all sizes, actual). Bars show input price DI minus output price DI (points)

For the lodging industry, this phase of “the output price DI catching up with the input price DI” can be framed as room for rate growth. In the first wave, when the output price DI rose from +28 to +48 between June 2022 and June 2023, the accommodation CPI went on to show double-digit year-on-year growth in summer 2023 (+15.1% in July and +18.1% in August 2023). Whether the +56 forecast materializes will be confirmed in the “actual” figure of the December survey (scheduled for publication in mid-December).

Comparing with our data — estimated settled ADR rose year on year in 11 of 12 prefectures in September, within July–September 2026

For July–September, which overlaps the Tankan response period (August 26 to September 30), we lined up MetroEngines Research’s estimated settled ADR (comparing finalized months) for 12 major prefectures by property type. On the overall measure (median across all property types), 9 of the 12 prefectures in July and 10 in August were down or flat year on year, while in September 11 were up. So the year-on-year change in room rates was switching from negative to positive during the period when Tankan responses were being collected.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research. Year-on-year change in estimated settled ADR (each month of 2026 vs the same month of 2025, all finalized months). By property type, combinations with fewer than 20 properties in the sample are shown as “—”. September overall N total = 7,219 properties

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research. Year-on-year change in estimated settled ADR for September 2026 (median across all property types), N=7,219 properties (total for 12 prefectures)

Ranked by September year-on-year change, the order is Aichi +45.9% (N=505), Hiroshima +14.6% (N=301), Ishikawa +13.6% (N=242), Okinawa +10.0% (N=515), Fukuoka +9.8% (N=477), Chiba +8.8% (N=424), Kanagawa +5.6% (N=579), Hokkaido +4.5% (N=960), Nagano +3.7% (N=818), Tokyo +1.9% (N=1,139) and Kyoto +1.1% (N=604), with Osaka −32.4% (N=655) the only decline. Aichi’s outlier is most naturally attributed to the 20th Asian Games, held in Aichi and Nagoya from September 19 to October 4, and rates rose regardless of property type (business +53.2%, city +53.7%). Osaka is lapping the Osaka-Kansai Expo of April–October 2025, with declines of about 30% across all three months: −30.7% in July, −32.7% in August and −32.4% in September. Rates are low against the Expo-period level, but they need to be read with 2025 treated as an exceptional year.

By property type, ryokan were up year on year in many regions in September: Tokyo +9.3% (N=55), Chiba +9.7% (N=130), Hiroshima +12.1% (N=69), Kyoto +6.1% (N=197) and Nagano +4.7% (N=510). City hotels were down in the big cities, at Tokyo −10.3% (N=110), Kyoto −14.6% (N=61) and Osaka −21.8% (N=97), and up elsewhere, at Fukuoka +21.9% (N=43), Hiroshima +18.9% (N=31) and Kanagawa +15.8% (N=42). Big-city city hotels raised rates sharply at the 2025 inbound demand peak, which makes for a tough comparison in 2026. The Tankan’s large firms (capital of ¥1 billion or more) likely include many big-city city hotel operators, so the large-firm DI’s −8 points and the year-on-year decline at big-city city hotels point in the same direction.

We also show estimated occupancy (based on OTA listed inventory). The table below gives estimated occupancy for all property types, business hotels and ryokan by prefecture for each month from July to September, all based on remaining-room observations from a lead time of 45 days to 1 day. We began collecting this indicator in March 2026, so year-on-year comparisons are not available. It is an estimate based on how inventory sold on OTAs is being absorbed, and it differs from the actual occupancy of a whole property.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research. Occupancy (estimated, based on OTA listed inventory), monthly average. Property and room counts are based on listings on major booking sites. Combinations with fewer than 10 properties are shown as “—”

For September overall, Hokkaido at 90.3% (679 covered properties / 53,149 rooms, based on listings on major booking sites), Aichi at 90.4% (328 properties / 33,635 rooms) and Tokyo at 89.8% (805 properties / 104,583 rooms) were around 90%, followed by Osaka at 80.4% (410 properties / 59,647 rooms), Chiba at 83.5% (314 properties / 23,165 rooms) and Kyoto at 84.0% (566 properties / 31,989 rooms). Aichi rose nearly 8 points in September from 81.7% in July and 82.5% in August, with the Asian Games period showing up in both rates and occupancy. Osaka’s rates fell about 30% while occupancy held at around 80%, which can be read as a region where pricing aimed at volume is working.

Accommodation CPI at 172.1 in September 2026, +5.7% year on year — reaccelerating from −0.2% in August

We now check room rates from the public statistics side. The “accommodation charges” item in the Ministry of Internal Affairs and Communications’ Consumer Price Index (national, 2020 = 100) was 172.1 in September 2026, up 5.7% from 162.9 a year earlier. Growth was sluggish over the two summer months, at 169.1 in July (+2.0% year on year) and 181.1 in August (−0.2%), and returned to the 5% range in September. The overlaid chart below shows monthly indexes for 2024, 2025 and 2026; the August peak rises each year, and only August 2026 sits nearly level with the previous year.

Source: Compiled by the HotelBank Editorial Team from the Ministry of Internal Affairs and Communications’ “Consumer Price Index” (e-Stat table ID 0003427113, national, accommodation charges, 2020 = 100). Through September 2026

The Tankan output price DI (all sizes) of +43 in the September survey and +56 in the forecast is consistent with the accommodation CPI’s +5.7% in September. The CPI is an index of prices actually paid, while the DI is the difference in the share of firms saying prices rose. For how far accommodation CPI growth and estimated settled ADR growth have diverged over the long term, see Lodging CPI +25.6% in 3 Years vs Settled ADR +10.9% — 14.7pt Gap. Set side by side, summer 2026 was a phase in which firms saying they raised prices were still the majority, but growth against the previous year’s high level temporarily narrowed, and that growth returned in September.

How to read it — a DI is a share of firms, combines accommodation and food services, and covers firms with capital of ¥20 million or more

Taken together, these figures leave an apparent contradiction: the Tankan DI fell, yet our ADR was up year on year in most prefectures in September. The question is not which is right; we break the reasons for the gap into three.

First, a DI is a difference in the share of firms, not a level. A business conditions DI falling from +46 to +38 means that the share answering “favorable” minus the share answering “unfavorable” narrowed by 8 points; it does not mean sales or rates fell 8%. The DI falls even if firms merely shift from “favorable” to “not so favorable”. Even when rates are up year on year, more firms may hesitate to answer “favorable” if growth is smaller than the year before. The fact that the sluggish summer 2026 accommodation CPI readings of +2.0% and −0.2% came just before the response period is consistent with the DI’s decline.

Second, the Tankan’s “accommodation and food services” combines the lodging industry and the food service industry, and food service carries the larger weight by number of firms. Rising food, utility and labor costs push up the input price DI for food service more strongly. The input price DI of +82 strongly reflects raw material costs on the food service side, whereas our estimated settled ADR looks only at lodging room rates. Daiwa Institute of Research’s citing of “rising costs including fuel and labor” as a factor behind the decline is likewise a reading of the combined industry.

Third, the populations differ. The Tankan covers private firms with capital of ¥20 million or more, and respondents in accommodation and food services are mainly chain operators and ryokan and hotel businesses of mid-size or larger. Our estimated settled ADR uses properties listed on major booking sites as its population, covering 7,219 properties in 12 prefectures for September 2026 overall, and it includes small independently run ryokan and minshuku and single-site business hotels with little capital. The Tankan’s small firm DI (capital of ¥20 million to under ¥100 million) is the lowest at +12, while in our data ryokan are up year on year in many regions; this reflects the difference in population rather than a contradiction. Small properties raised rates by less than big-city city hotels the year before, which makes it easier for them to exceed the previous year in 2026.

In the Japan Tourism Agency’s “Overnight Travel Statistics Survey” for August 2026 (First Preliminary Report), guest-nights were 62.68 million, down 6.7% year on year (Japanese −5.8%, foreign −10.3%), and the room occupancy rate was 64.4%. JNTO’s visitor arrivals for September were 3,103,700 (−5.0% year on year), with typhoon-related flight cancellations and higher airfares linked to the Middle East situation cited as reasons. That rates recovered in September while volume kept falling year on year suggests that properties are starting to prioritize rate over occupancy. The Tankan’s output price forecast of +56 can be seen as firms putting that direction into numbers themselves.

Summary — three things to check by the December survey

In the September Tankan, accommodation and food services split by firm size, at +38 (−8 points) for large firms, +36 (+3 points) for mid-sized firms and +12 (−6 points) for small firms, and all three exceeded the forecasts made in June. The employment conditions DI stood at −58 across all sizes, narrower than −75 in December 2023 but 14 points deeper than the non-manufacturing average. The gap between the output and input price DIs was 39 points, with the forecast of 27 points expected to be the narrowest since March 2024. Our estimated settled ADR rose year on year in 11 of 12 prefectures in September, and the accommodation CPI also recovered to +5.7%.

Operators and investors should watch three things by the December survey (scheduled for publication in mid-December). The first is whether the output price forecast of +56 materializes as an “actual”. If it does, the industry returns, for the first time since the first half of 2024, to a phase in which room rates absorb persistently high input prices. The second is when the year-on-year change at big-city city hotels bottoms out. Tokyo −10.3%, Kyoto −14.6% and Osaka −21.8% are a reaction against the high 2025 base, and the comparison level falls from October 2026 onward. The third is the winter hiring environment indicated by the employment forecast of −61, since setting the number of rooms to sell on the basis of staffing plans is a precondition for any rate strategy. The best use of this September survey is not to read the DI decline as a verdict on the lodging industry, but as the entrance to a phase in which more firms say they will raise rates.

Frequently Asked Questions