Regional Distribution of 'Vacant Houses' and 'Public Facilities' Debate in Japanese Local Assemblies — A Seven-Year Structural Analysis Across 1,320 Municipalities

This note reads year-by-year and prefecture-by-prefecture counts of Japanese local assembly speeches containing terms related to public-asset utilisation — akiya / kuya (vacant houses), kokyo-shisetsu / shisetsu-management / FM (public facilities), and haiko / haiko-katsuyou / tohaigo (school closure and consolidation) — from 2018 to 2024, using the machikarte corpus (about 125 million assembly speech records, with up to 1,320 municipalities ingested in 2024). The aim is to examine the time structure of statutory revisions in public-asset policy and the regional distribution of assembly discussion. Naming individual assemblies or councillors in evaluative ranking terms lies outside the article's purpose. The scope of observation is held at national-trend and prefecture-aggregate granularity; no top-or-bottom ranking by municipality name is published here.

Across the seven-year window from 2018 to 2024, mentions of vacant-house-related terms (akiya / kuya) in Japanese local assemblies range from 24,100 to 34,847 speeches per year. The annual counts are 34,573 (2018), 26,126 (2019), 24,100 (2020), 25,791 (2021), 28,538 (2022), 34,551 (2023), and 34,847 (2024) — a U-shape that drops below the 2018 peak between 2019 and 2021 and recovers to the 2018 level by 2023-2024.

Public-facility mentions move along a different curve. From 64,205 in 2018 they decline to the 55,000-61,000 range in 2019-2021, then recover to 67,014 in 2024 — the highest level in the seven-year window. School-closure / consolidation mentions also reach their seven-year high in 2024 (13,749).

Absolute counts must be read with care. Machikarte's municipal ingestion grew from 1,150 (2018) to 1,320 (2024), so part of the rise reflects expanding coverage. That said, the 2019-2021 dip in akiya mentions (34,573 to 24,100, about a 30% decline) moves opposite to the monotonic rise in covered municipalities, suggesting a time structure not explainable by coverage alone.

Prefecture-level variation in discussion density

The 2024 per-municipality discussion density for akiya, across the 47 prefectures with at least 6 ingested municipalities, ranges from 9.4 to 92.4 mentions per municipality — roughly a tenfold spread. The prefectures that lead by absolute count are not the same as those that lead by per-municipality density.

By absolute count, large prefectures dominate: Saitama (2,405), Tokyo (1,917), Chiba (1,582), Nagano (1,532), Hokkaido (1,414), Osaka (1,392), Hyogo (1,257).

By per-municipality density, prefectures with 11-34 ingested municipalities sit at the top — the top five span roughly 92, 45, 41, 39, and 38 mentions per municipality. Prefectures with more than 100 ingested municipalities sit at the top of absolute counts but lower in density rankings.

Co-mentions with small-concession-related terms

Co-mentions of PPP framework terms within speeches that already contain akiya or public-facility terms reveal a distinct shift. Designated-manager (shitei-kanri) co-mentions stay in the 4,251-5,132 range across the seven years, while Park-PFI / Park PFI co-mentions rise from 1 (2018) to 16 (2024), and small-concession (sumokon / small concession) co-mentions rise from 2 (2023) to 36 (2024).

The designated-manager framework is well established in assembly discourse, persistently appearing in the akiya and public-facility contexts. Park-PFI and small concession remain in the low double digits in absolute terms, but the direction of change from 2023 to 2024 is unambiguously upward.

Policy timeline for public-asset utilisation

The decade from 2014 to the mid-2020s contains several major milestones in public-asset policy.

• November 2014: Enactment of the Act on Special Measures concerning the Promotion of Countermeasures against Vacant Houses, etc. (Act No. 127 of 2014). This law organises municipal vacant-house plans, the designation of "specified vacant houses", and authority for inspection and demolition orders.
• 2017: Creation of the Park-PFI (Designated Setting and Management System) under the amended Urban Parks Act. Private operators are selected by public solicitation to install revenue-generating facilities inside parks, with the resulting revenue redirected to redeveloping park infrastructure.
• Early 2020s: Revision cycles for municipal "Comprehensive Management Plans for Public Facilities". Under MIC guidance, first-generation plans were drafted across municipalities around 2014-2017, and many enter their revision phase between 2022 and 2025.
• December 2023: Enforcement of the Amended Vacant-House Act (Act No. 50 of 2023). A new "management-deficient vacant house" classification is added, intervention is strengthened in the pre-designation phase, and conditions for removing the residential-land tax break are clarified.
• May 2026: MLIT PPP/PFI Promotion Office publication of "A Recommendation for Small Concession". This guide is aimed at municipal staff hesitant to start projects on idle public facilities, organising the process from project conception through public solicitation and operator selection.

Time synchrony with assembly discourse

Overlaying these milestones on the annual trend shows that the 2018 peak and the 2023-2024 recovery in akiya mentions sit in time alignment with the 2014 enactment and the 2023 amendment of the Vacant-House Act respectively. The seven-year high for public-facility mentions in 2024 overlaps with the revision phase for municipal comprehensive management plans.

The shift in small-concession-related terms — 2 in 2023 to 36 in 2024 — predates the MLIT guide's May 2026 publication. Assembly-level mentions appear before the official guide was published, indicating that the policy guide is not the only entry point for assembly-level discussion. Disentangling the effects of pre-guide ministerial documents, accumulated public-solicitation case practice, and pre-publication outreach by the PPP Promotion Office requires monthly resolution and municipality-level timing data, both beyond what surface aggregation can provide.

What "discussion density" captures

The per-municipality mention count (avg_per_muni) reads as a quantitative indicator of how intensively akiya is debated in assembly responses. Treating low density as "low interest" would, however, miss at least four structures that can each contribute to low density:

• The topic has been institutionalised in prior cycles and current debate has moved to newer issues.
• Depopulation has advanced to a stage that precedes formalising vacant houses as an agenda item.
• The municipality is small and total assembly speech volume is low.
• The agenda portfolio is dominated by other priority items.

Treating density as a single-axis evaluation risks overlooking these structures.

What the keyword set captures

The regex aggregation will inevitably pick up uses of akiya, public-facility, and school-closure terms across multiple contexts — vacant-house policy, public-facility management, school reorganisation, tourism, and industrial revitalisation. This article treats the matches collectively as "public-asset-utilisation-related mentions"; context-level classification is left to a later iteration.

Time-lag structure between policy and assembly speech

Overlaying the policy milestones (2014 Vacant-House Act, 2017 Park-PFI, 2023 amended Vacant-House Act, 2026 small-concession guide) on the speech time series shows peaks lining up with each milestone, although their lead time and amplitude differ.

The akiya trend follows a diffusion phase after the 2014 enactment (peak in 2018), a dip (2019-2021), and a recovery in 2023 aligned with the amended Act. The lag between enactment and assembly-level diffusion is several years, and the curve shows a re-activation around amendment timing.

Park-PFI-related terms show a long lag from 2017 creation to assembly-level penetration. From 1 mention in 2018 to 16 in 2024, the six-year trajectory toward the low double digits suggests how slowly the framework was absorbed into assembly-level vocabulary.

The shift in small-concession-related terms in 2023-2024 follows a different pattern. Assembly-level mentions start before the MLIT guide's May 2026 publication, indicating that the guide is not the trigger point for the discussion. Which phase of the broader promotion activity around small concession produced assembly-level uptake cannot be settled by surface aggregation here.

A parallel structure appears in the care mentions seven-year trend, where assembly discussion synchronises with the three-year Long-Term Care Insurance Act revision cycle. The difference here is that public-asset utilisation has multiple statutory and guidance milestones running in parallel (Vacant-House Act, Park-PFI, comprehensive management plans, small-concession guide), so the visible peaks reflect the overlap of several policy nodes rather than a single revision cycle.

Geographic variation in density and the absolute-vs-density divergence

The 2024 per-municipality density for akiya spans roughly ninefold across prefectures, and the absolute-count leaders differ from the density leaders. Absolute counts are dominated by prefectures with many ingested municipalities (Saitama, Tokyo, Chiba, Hokkaido, Osaka, Hyogo), while the density ranking is led by mid-sized prefectures with 11-34 ingested municipalities.

No single explanation accounts for this divergence. The following candidate factors stand in parallel:

• Differences in vacant-house rate: How prefecture-level vacant-house rates from the 2023 Housing and Land Survey map onto density.
• Timing of stock ageing: Overlap between high-growth-era housing supply and current ageing cycles.
• Progress of facility-specific plans: Implementation pace of individual facility plans and the resulting agenda load.
• Assembly size and agenda mix: Variation in session length, speech volume, and agenda portfolio inside each prefecture.

Testing these factors is beyond what the current aggregation can do. Cross-referencing with vacant-house-rate data (MIC), public-facility stock data (MIC / MLIT), and municipal fiscal-strength indices is the next priority.

Skew in policy-framework usage

There is a clear skew in which PPP frameworks actually appear in public-asset-utilisation discussions. Designated-manager terms sit in the 4,200-5,200 range each year and are firmly embedded in assembly vocabulary. Park-PFI (1-16 mentions per year) and small concession (1-36 mentions per year) remain limited at the assembly level.

This skew likely reflects a combination of framework utilisation conditions, administrative familiarity, and private-sector entry incentives. The designated-manager framework has accumulated more than 20 years of practice since the 2003 Local Autonomy Act revision, and its vocabulary is well diffused among municipal staff and councillors. Park-PFI and small concession received their institutional or guidance anchor in 2017 and 2026 respectively, and assembly-level diffusion will need more time.

The low count alone, however, cannot be read as "the framework is dormant". The pre-publication appearance of small-concession terms in assembly speeches suggests that the framework's diffusion channels run in parallel to assembly debate — through municipal-staff briefings, PPP-promotion platforms, and similar venues.

Intersection with the school-closure debate

School-closure and consolidation discussion (1.0-1.4 million range per year) sits at the intersection of vacant-house policy, public-facility management, and small-concession promotion. It is counted as a separate axis in assembly speech, but structurally it is part of the same problem set.

Closed school sites tend to satisfy several conditions that ease PPP entry: single ownership (municipal), large built footprint, and existing utility infrastructure. They lie outside the Park-PFI scope (not parks) but sit at the centre of small-concession case studies — MLIT's "A Recommendation for Small Concession" includes multiple closed-school cases. The overlap between assembly-level school-closure debate and public-facility management debate can serve as an indicator of how far PPP frameworks have moved into implementation.

Room for cross-lab citation

Public-asset utilisation lies at the intersection of fiscal policy, industrial policy, regional economy, and education. One connection with other articles in this lab is the distribution of deferral phrasing in assembly responses (case-sakiokuri-rate); combining the two would let us measure how often "we will consider" hedging appears within public-asset-utilisation discussion. A second link is the question of how foreign-workforce intake debate (foreign-workers-regional-2018-2024) and closed-school / community-infrastructure reorganisation debate co-occur within the same assemblies.

• No semantic classification: Contextual classification of speeches matching the akiya / public-facility / school-closure terms (vacant-house policy / facility management / school reorganisation / tourism and industrial revitalisation, etc.) is not applied; the aggregation stays at surface keyword level.
• No agenda-portfolio normalisation: The aggregation is not normalised for the agenda portfolio of public-asset-utilisation-related items, so the effect of agenda composition on mention counts is not separated.
• No separation of speakers: Councillor questions and administrative responses are not distinguished; analysing the leading / following structure of any time lag requires speaker-level aggregation.
• Coverage bias: The number of ingested municipalities varies by prefecture, so avg_per_muni cannot be equated with "prefecture-wide debate temperature". This article restricts to prefectures with at least 6 ingested municipalities to preserve basic comparability, but does not correct fully for coverage variation.
• No population weighting: avg_per_muni is a simple per-municipality average without weighting by municipal population size. It does not carry the same meaning across prefectures dominated by designated cities and those dominated by small towns and villages.
• No individual ranking: The article stays at the structural level and does not publish municipality-level or councillor-level top-or-bottom rankings.
• No alignment with implementation data: The aggregation does not measure the correspondence between assembly-level discussion density and the actual count of PPP / small-concession projects implemented.

These limits will be addressed in later iterations. Priorities are: (1) cross-reference with prefecture-level vacant-house-rate data from the 2023 Housing and Land Survey (MIC), (2) correlation with the revision status of municipal Comprehensive Management Plans for Public Facilities, (3) monthly resolution to measure time lag around statutory milestones, and (4) cross-checking against PPP / small-concession project databases.

The query specification (spec_version v1-akiya-public-2026-06) is documented at the end of this article, and the BigQuery aggregation queries will be placed in the machikarte GitHub repository so that third parties can reproduce the results independently.

A corrections contact is provided separately to receive notices about aggregation errors and notation drift. The lab's editorial procedure — including the three-tier rule on councillor-level data publication — is documented in the lab overview note (hypothesis-overview).

Aggregation queries (spec_version v1-akiya-public-2026-06)

Q1: Annual trend (2018-2024)

SELECT
  year,
  COUNTIF(REGEXP_CONTAINS(body, r'空き家|空家')) AS akiya_mentions,
  COUNTIF(REGEXP_CONTAINS(body, r'公共施設|施設マネジメント|FM')) AS facility_mentions,
  COUNTIF(REGEXP_CONTAINS(body, r'廃校|廃校活用|統廃合')) AS school_close_mentions,
  COUNT(DISTINCT municipality_code) AS municipalities
FROM `correlate-workspace.isvd_machikarte.speeches`
WHERE year BETWEEN 2018 AND 2024
GROUP BY year
ORDER BY year

Q2: Prefecture aggregation (2024 only, prefectures with at least 6 ingested municipalities)

SELECT
  r.prefecture,
  COUNTIF(REGEXP_CONTAINS(s.body, r'空き家|空家')) AS akiya_mentions,
  COUNT(DISTINCT s.municipality_code) AS municipalities,
  ROUND(
    COUNTIF(REGEXP_CONTAINS(s.body, r'空き家|空家'))
    / COUNT(DISTINCT s.municipality_code), 2
  ) AS avg_per_muni
FROM `correlate-workspace.isvd_machikarte.speeches` s
JOIN `correlate-workspace.isvd_machikarte.municipality_registry` r
  ON s.municipality_code = r.code
WHERE s.year = 2024
GROUP BY r.prefecture
HAVING COUNT(DISTINCT s.municipality_code) >= 6
ORDER BY avg_per_muni DESC

Q3: Framework co-mentions (annual trend)

SELECT
  year,
  COUNTIF(REGEXP_CONTAINS(body, r'Park-PFI|パークPFI')) AS parkpfi_co,
  COUNTIF(REGEXP_CONTAINS(body, r'指定管理')) AS shitei_co,
  COUNTIF(REGEXP_CONTAINS(body, r'スモコン|スモールコンセッション')) AS sumokon_co
FROM `correlate-workspace.isvd_machikarte.speeches`
WHERE year BETWEEN 2018 AND 2024
  AND REGEXP_CONTAINS(body, r'空き家|空家|公共施設')
GROUP BY year
ORDER BY year

Query execution date: 3 June 2026.

This article is published in the public-interest research stream (labs/machikarte) operated by Institute for Social Vision Design (ISVD). ISVD acts as the data operator of machikarte, taking on the role of curating local-assembly speech data and providing verifiable aggregations. Evaluating or holding to account individual municipalities or councillors lies outside this lab's role; that work is left to the press. Use of this article with the citation note "Data collaboration: Institute for Social Vision Design (ISVD)" is welcomed.

machikarte — Nationwide Local Assembly Speech Search Platform (Beta) — Institute for Social Vision Design (ISVD). ISVD

machikarte (GitHub) — schema, aggregation queries, licenses (MIT + CC BY 4.0) — Institute for Social Vision Design (ISVD). GitHub

Act on Special Measures concerning the Promotion of Countermeasures against Vacant Houses, etc. — MLIT Housing Bureau. MLIT

Park-PFI (Designated Setting and Management System) — MLIT City Bureau. MLIT

A Recommendation for Small Concession — A Guide for Utilising Idle Public Facilities — MLIT PPP/PFI Promotion Office. MLIT

Housing and Land Survey (Reiwa 5, 2023) — Statistics Bureau, Ministry of Internal Affairs and Communications. MIC Statistics Bureau

Verification of Role Classification Methods for Diet Records Using BERT-Based Classifiers (in Japanese) — Miyaki, Y. and Uchida, Y.. Journal of Japan Society for Fuzzy Theory and Intelligent Informatics, Vol. 37, No. 1, pp. 530-534