Are Japan's shopping arcades half empty? What the counts show

In Japan's FY2024 national arcade survey, 4.1 per cent of shopping arcades reported a vacancy rate of 50 per cent or more, while 24.6 per cent reported no vacant shop at all (中小企業庁商業課, 今後の中小商業支援策等について, 令和7年8月, checked 11 September 2026). The average vacancy rate across all responding arcades was about 11 per cent — lower than the 14.62 per cent recorded in the FY2012 edition of the same survey, twelve years earlier.

So the arcade you walked through is not statistically half empty, and the emptiness is not accelerating. Something else is: the number of arcades that exist at all fell from 14,271 in 1994 to 11,280 in 2024, on the Small and Medium Enterprise Agency's own chart.

This guide covers the national survey figures and how to read an arcade on the ground, using Osaka's Tenjinbashisuji as the worked counter-example. It does not cover individual Osaka markets, shopping-mall retail, or department stores.

How empty is the average arcade?

Japan runs a recurring national survey of shopping arcades, and the vacancy rate it produces has sat in a narrow band for fifteen years. The 商店街実態調査 is commissioned by the Small and Medium Enterprise Agency, surveys arcade associations rather than individual shops, and publishes one average vacancy rate (空き店舗率) and one average shop count per arcade for each edition (中小企業庁, 商店街実態調査, checked 11 September 2026).

The table below compares survey editions on three attributes: the average vacancy rate, the average number of shops in an arcade, and the response base behind each figure.

Figures for FY2009 to FY2021 are read from the agency's own edition reports: FY2009, FY2012, FY2015, FY2018 and FY2021, all checked 11 September 2026. The FY2024 row is the agency's commerce division stating "商店街の空き店舗率は約11％" in an August 2025 briefing document, on a chart base of 2,071 arcades.

Two cautions belong with that table rather than after it. The FY2024 figure is published as an approximation on a smaller chart base than the survey's other tables, so this guide does not treat it as directly comparable to the earlier decimal figures. And the FY2009 report states that 10.82 per cent was the first time the series had passed 10 per cent, which means the band this article describes is a recent one, not a permanent condition.

The distribution matters more than the mean, because a mean of 11 per cent hides both extremes. In FY2024 the arcades split as follows: 24.6 per cent with no vacant shop, 39.0 per cent above zero but under 10 per cent, 26.6 per cent between 10 and 30 per cent, 5.8 per cent between 30 and 50 per cent, and 4.1 per cent at 50 per cent or more. The same document notes that arcades above the 10 per cent line — the point it describes as the threshold at which a street begins to feel run down — reach 35 per cent of the total.

Are arcades closing or shrinking?

The arcades themselves are disappearing faster than their shops are emptying, and that is the finding the vacancy rate hides. The Small and Medium Enterprise Agency's August 2025 briefing charts the number of arcades in Japan at 14,271 in 1994, 14,070 in 1997, 12,597 in 2002, 12,407 in 2004, 12,568 in 2007, 12,681 in 2014, 12,096 in 2018, 12,210 in 2021 and 11,280 in 2024, and states that more than 11,000 arcades exist nationally when merchants' associations, cooperatives and informal tenant groups are all counted.

That series carries a footnote you should carry with it: the same page states that from 2018 the figure is the number of arcades to which the survey was posted. The pre-2018 and post-2018 halves are therefore not the same measurement, and a drop from 12,210 to 11,280 between 2021 and 2024 is partly a change in the mailing frame. This guide does not treat the series as a clean census of arcades.

A second, independently produced series points the same way without the frame problem. The national economic census counted 990,246 retail establishments in Japan in 2016, down 4.2 per cent on 2012 (平成28年経済センサス‐活動調査, checked 11 September 2026), and 880,031 retail establishments as of 1 June 2021 (令和3年経済センサス‐活動調査, checked 11 September 2026). That is a fall of about 11 per cent in five years, in a period when the arcade vacancy rate barely moved.

Put the two together and the mechanism changes shape. Small shops are closing throughout Japan, but an arcade that loses its merchants does not usually sit at 50 per cent vacancy for long — it stops being an arcade, its association dissolves, and it leaves the survey. What you see as a visitor is a survivor population, which is why the average arcade still reports around fifty shops.

What do arcade associations blame?

The survey asks arcades to name their own problems, and the same answer has led every edition for fifteen years. Respondents pick up to three, so the percentages do not sum to 100. The table below lists the leading items from the two most recent editions with their figures as printed.

FY2021 figures are from the 概要版 and FY2024 figures from the agency's August 2025 briefing, both checked 11 September 2026. The two editions do not print an identical item list, so the blank cells mean "not quoted in the source read", not "zero".

The ranking is stable much further back. Succession led at 51.3 per cent in FY2009, 63.0 per cent in FY2012, 64.6 per cent in FY2015 and 64.5 per cent in FY2018, with ageing premises and a shortage of draw-card shops rotating through second and third place. Vacancy itself has never been the top answer in any edition read for this article.

What makes the succession answer more than a complaint is a second FY2021 figure sitting beside it. Asked what they were doing about succession, 96.0 per cent of arcades answered 対策は講じていない — no measures taken; 2.2 per cent ran training and 1.8 per cent were recruiting a successor from outside. The age profile of the people answering explains some of that: among arcade association chairs, 37.5 per cent were in their seventies or older and 33.5 per cent in their sixties, against 0.7 per cent aged under forty.

And when the FY2021 survey asked why shopkeepers had actually left, 68.1 per cent of arcades gave the proprietor's age or the absence of a successor as the reason. The vacancy in a Japanese arcade is, on the survey's own evidence, more often a retirement than a bankruptcy.

Did the 2000 law change this?

Japan deregulated large-store openings in 2000, and that date is the standard structural explanation for suburban retail displacing arcades. The 大規模小売店舗立地法 (Large-Scale Retail Store Location Act), promulgated on 3 June 1998 as Act No. 91 of 1998, came into force on 1 June 2000 and replaced the earlier 大規模小売店舗法, the act on adjusting the business activities of retailers in large-scale stores, which was repealed at the same time (Tokyo Metropolitan Government and Hokkaido Government, both checked 11 September 2026).

The change was in what the state was allowed to weigh. The old act adjusted the commercial activity of large stores, which in practice let existing local retailers slow a newcomer down. The replacement applies to buildings with more than 1,000 square metres of retail floor area and asks the operator to have regard for the surrounding living environment — traffic congestion, noise, waste — with no commercial adjustment mechanism at all.

On the same day in 1998, Japan promulgated the 中心市街地の活性化に関する法律 as Act No. 92 of 1998, originally titled the act on integrated promotion of urban improvement and the revitalisation of commerce in city centres (National Diet Library, Japanese Law Index, checked 11 September 2026). The pairing is deliberate: one law removed the brake on suburban stores, the other funded town-centre revitalisation.

This article stops at the dates. No source read here demonstrates that the 2000 deregulation caused the arcade vacancy figures above, and the survey does not offer a "large suburban store opened nearby" answer option in the problem list quoted. The laws are confirmed; the causal chain from them to a specific shuttered storefront is not, and you should be sceptical of any guide that presents it as settled.

Why is Tenjinbashisuji still full?

Osaka has the arcade most often used as proof that the form still works, and what distinguishes it is length, transport and catchment rather than any policy. The Tenjinbashisuji Shopping Street runs roughly 2.6 km north to south, and its own 2-chome and 3-chome promotion association states that length and describes the street as Japan's longest shopping arcade (天神橋筋商店街2丁目3丁目, checked 11 September 2026).

The Osaka Convention & Tourism Bureau, the city's official tourism body, gives the same 2.6 km figure, puts the number of shops along it at about 800, and says walking the whole thing takes around 40 minutes. It also lists the stations that feed it: Minamimorimachi and Ogimachi and Tenjinbashisuji 6-chome on the Osaka Metro, and Temma on the JR Osaka Loop Line (osaka-info.jp, checked 11 September 2026).

Hold that against the survey's average arcade of 51.2 shops in FY2021. Tenjinbashisuji is not a slightly better arcade; on the tourism bureau's shop count it is roughly sixteen average arcades laid end to end, with four rail stations distributed along it rather than one at a single entrance. An arcade with one station serves whoever lives at that station. An arcade with four is a corridor that people cross rather than a destination they must choose.

That is the honest version of "what officially distinguishes it": published length, published shop count, and multiple rail access points, all from the arcade's association and the city tourism bureau. No government source consulted here publishes a vacancy rate for Tenjinbashisuji specifically, and this guide does not claim one. Treat it as the visible upper end of the distribution — part of the 24.6 per cent of arcades reporting zero vacancies is what a working arcade looks like, and this is the largest example of it.

What do these numbers not show?

The statistics on this page are narrower than their decimal places suggest, and four gaps matter. First, the survey samples arcade associations, not shops: an arcade whose association has dissolved stops being surveyed, so the population being measured is a survivor population and the vacancy average is biased toward arcades organised enough to answer a questionnaire. Response rates ran between 35.0 and 45.1 per cent in the editions read here.

Second, nothing in the survey tells you what a vacant unit is doing now. There is no official breakdown of whether a shuttered storefront is for rent, held by an owner who lives upstairs, converted to storage, or awaiting demolition. The FY2021 edition records that 58.4 per cent of arcades said they were not involved at all in what happens to vacant units in their street, which is a statement about the association, not about the building.

Third, no government source read here publishes a vacancy rate, shop count or opening hours for any named individual arcade, in Osaka or anywhere else. The prefecture-level and national aggregates are what exists. Any figure you see for a specific street is either that street's own claim or an estimate.

Fourth, the survey does not measure the thing a visitor actually cares about. There is no official statistic for how many arcade tenants are independent rather than chains, how many still trade in a single category, what an arcade meal costs, or how many shops close on which weekday. This guide gives no prices for that reason. The word シャッター通り, shutter street, appears nowhere in the survey as a defined category, so no official count of shutter streets exists to quote.

How do you read an arcade?

An arcade is worth walking because it holds retail categories that a shopping mall has no leasing model for, and you can tell whether a given one still holds them within the first hundred metres. Malls lease to covenant-bearing chains on standard terms; an arcade unit is often owned by the person behind the counter, which is exactly why the survey's leading problem is succession rather than rent. That ownership is what keeps a knife grinder, a single-item pickle shop, a dried-fish seller or a tatami maker in business next to a bank branch.

Read the street in this order before you commit an hour to it.

1. Count shutters on one side for one block. Under ten per cent closed puts the arcade in the healthiest two-fifths of the national distribution; consistently a third or more closed puts it in the bottom tenth, where the FY2024 survey places 9.9 per cent of arcades.
2. Look at what time the shutters come down rather than whether they are down. Many independent shops keep short daytime hours, so an arcade at 18:00 can read as dead when it was trading at 11:00. Judge on a weekday late morning.
3. Check for a fresh tenant among the old ones. A new café or a bakery in a converted unit means the association or a landlord is letting vacancies rather than holding them, which the FY2021 figures show is not the norm.
4. Look for the categories a mall cannot carry — knives, seals and stamps, futons and tatami, single-ingredient food shops, repairs. Their presence is the reason to be there; their absence means you are in a chain street with a roof.
5. Note how many rail entrances the arcade has. Tenjinbashisuji's four stations across 2.6 km are the structural reason it stays busy, and an arcade with one distant station is relying on residents alone.
6. Ask about payment before you choose a shop. No official source consulted here publishes card acceptance rates for arcade tenants, so carry yen and treat cashless as a bonus.
7. Go now rather than on a future trip, in the specific sense the data supports: with 71.0 per cent of arcade association chairs in their sixties or older and 96.0 per cent of arcades taking no succession measures, the shops most worth seeing are the ones with the shortest confirmed horizon.

The number to keep is not 50 per cent. It is 13.59 per cent average vacancy in FY2021 and about 11 per cent in FY2024, against 14,271 arcades in 1994 and 11,280 in 2024. Japan's arcades are not filling with empty shops. They are being outlived by their shopkeepers, one association at a time, and the ones still standing are fuller than the photographs suggest.