Japan Construction & Real Estate License Registry Stats

How many licensed construction companies are there in Osaka? How many of them do demolition? What share are nationally licensed? How old are the real-estate brokerages in Fukuoka? One call answers it — with the exact population count, not an estimate.

This Actor queries Japan's official MLIT registry of licensed businesses (建設業者・宅建業者等企業情報検索システム), the legally authoritative source for:

• 建設業許可業者 / licensed construction companies — ~480,000 nationwide, every one of them licensed in 1 or more of 29 regulated trades
• 宅地建物取引業者 / licensed real-estate brokers — the only public list of who may legally broker property in Japan

There is no official API and no bulk download for this registry, and the search UI is Japanese-only, session-based and Shift_JIS encoded. This Actor turns it into one clean, English-labelled JSON record per prefecture.

What you get per prefecture (one record)

Field	Meaning
totalLicensedCompanies	Exact registry count for the prefecture — the size of the market
byBusinessType	Exact count for each of the 29 construction trades (civil engineering, building, electrical, plumbing, demolition…) — the industrial structure of the area
topBusinessTypes	The five largest trades, ranked
ministerLicensedCount / ministerLicenseRatio	Exact count (not an estimate) of companies holding a national (minister) license = they operate in more than one prefecture. The single best proxy for how consolidated the local market is — e.g. 2,746 of Tokyo's 44,827 construction licensees (6.1%) vs 24 of Kochi's 2,907 (0.8%)
cityDistribution	Top 10 municipalities by number of licensees
generalLicensedCount / specificLicensedCount	一般 vs 特定 license mix (optional)
licenseRenewalDistribution, medianRenewalCount, newEntrantsShare	Real estate only: Japanese broker licenses are renewed every 5 years and the renewal number is printed in every license number — so this is a business-age profile of the local brokerage market, including the share of brand-new entrants

Optionally, every sampled company as its own record (name, license authority, license number, office, address, city).

Privacy: the registry displays each company's representative (a natural person). This Actor parses that column away and never outputs personal names.

Pricing — from $0.02 per area, no subscription

Event	Price	When
Area registry summary	$0.02	Per prefecture (× registry) analyzed
Individual company	$0.002	Only if you enable Include individual companies

A default run (1 prefecture, summary + all 29 trade counts) costs $0.02. Areas that return zero companies are never charged.

Input

Field	Example	Notes
registry	"construction"	construction (建設業許可業者) or realestate (宅地建物取引業者)
prefectures	["13"], ["Tokyo","Kochi"], ["東京都"]	English names, Japanese names or JIS codes — all accepted. Matched on head-office location. $0.02 each
businessTypes	["Electrical works","解体"]	Construction only. Codes 1–29, English or Japanese names. Empty = all 29 trades
breakdownByBusinessType	true	Construction only. Adds the per-trade counts (~30s)
includeIndividualCompanies	false	Enable to also get each sampled company as a record
maxCompaniesPerArea	150	Sample size per prefecture (50–600)
licenseTypeBreakdown	false	Construction only. Adds general/specific license counts

Exact vs sampled — which is which

Most of this Actor's numbers are population counts, not estimates: the registry returns a total for any query in a single request, so each one costs one request rather than a full crawl.

Field	Basis
totalLicensedCompanies	Exact — registry count for the prefecture
byBusinessType (29 trades)	Exact — one registry count per trade
ministerLicensedCount, prefecturalLicensedCount, ministerLicenseRatio	Exact — dedicated national-license query (ministerLicenseSource: "exact")
generalLicensedCount / specificLicensedCount	Exact (optional)
cityDistribution	Sampled
licenseRenewalDistribution, medianRenewalCount, newEntrantsShare	Sampled

Sampled fields are always listed in estimatedFrom.sampledFields together with sampledCompanies and sampledPages, so you never have to guess.

The registry's default sort groups nationally licensed companies at the top, so sampling the first page only would badly skew a sample. This Actor sorts by company name and jumps to evenly spaced pages across the whole result set (e.g. pages 1 / 449 / 897 of 897).

Output example (abridged; totals and license split are real measured values, per-trade counts are illustrative)

{
  "type": "registry_summary",
  "registry": "construction",
  "prefecture": "Tokyo",
  "prefectureNameJa": "東京都",
  "prefectureCode": "13",
  "totalLicensedCompanies": 44827,
  "resultPages": 897,
  "sampledCompanies": 150,
  "sampledPages": [1, 449, 897],
  "byBusinessType": [
    {"code": "1", "name": "Civil engineering works", "nameJa": "土木一式", "companies": 6120},
    {"code": "2", "name": "Building construction", "nameJa": "建築一式", "companies": 14980},
    {"code": "8", "name": "Electrical works", "nameJa": "電気", "companies": 7340},
    {"code": "29", "name": "Demolition", "nameJa": "解体", "companies": 3980}
  ],
  "topBusinessTypes": [{"code": "2", "name": "Building construction", "companies": 14980}],
  "ministerLicensedCount": 2746,
  "prefecturalLicensedCount": 42081,
  "ministerLicenseRatio": 0.0613,
  "ministerLicenseSource": "exact",
  "cityDistribution": [{"city": "世田谷区", "companies": 11}, {"city": "大田区", "companies": 9}],
  "estimatedFrom": {"sampledCompanies": 150, "sampledPages": [1, 449, 897], "sampledFields": ["cityDistribution"]},
  "checkedAt": "2026-07-25T09:00:00+00:00"
}

Real-estate runs add instead:

{
  "licenseRenewalDistribution": [{"renewalCount": 1, "companies": 34}, {"renewalCount": 16, "companies": 8}],
  "medianRenewalCount": 4,
  "newEntrantsShare": 0.2267
}

Use cases

• B2B lead scoping & territory planning — building materials, tools, heavy equipment and rental firms sizing a prefecture before hiring a rep
• Insurance & guarantee products — the licensee population and its trade mix is the addressable market for construction liability and workers' cover
• Construction SaaS / PropTech go-to-market — TAM per prefecture, split by trade and by minister vs prefectural license
• Supply-chain & subcontractor due diligence — verify that a trade is actually licensed in the area, and how deep the local pool is
• Market research & policy — 47-prefecture comparison of construction density, consolidation and brokerage churn (run one call per prefecture)

Notes & limits

• Companies are matched on their head-office location, so a Tokyo-headquartered firm working in Chiba counts under Tokyo
• One company is usually licensed for several trades, so byBusinessType counts overlap and sum to far more than totalLicensedCompanies — read them as "how many licensees can legally do X", not as a partition
• 一般 (general) and 特定 (specific) licenses can both be held by one company, so those two counts may overlap
• The registry is refreshed roughly monthly by MLIT — treat it as a monthly-fresh source, not a daily feed
• Renewal-count → years-in-business is an approximation (the cycle was 3 years before 1996); use the distribution, not a hard age
• Read-only, throttled to ≥1 request/second out of respect for a government system. No login, no personal data, nothing stored between runs
• Keep it to a few prefectures per run; a full per-trade breakdown is ~34 requests (~40 s) per prefecture, a real-estate area 5 requests (~6 s)

---

日本語ガイド

国土交通省「建設業者・宅建業者等企業情報検索システム」の許認可データを、1コールで都道府県単位の市場統計にして返す Actor です。

• 概要: 指定した都道府県の 建設業許可業者数(29業種すべての内訳つき) または 宅地建物取引業者数 を、システムが返す実数で取得します。あわせて、大臣許可・大臣免許の業者数と比率も推定ではなく実数(専用クエリ1回で取得。例: 東京の建設業は 44,827 社中 2,746 社 = 6.1%、高知は 2,907 社中 24 社 = 0.8%)、市区町村別の分布、一般/特定の内訳(オプション)、宅建業では免許証番号の更新回数分布(5年更新のため業歴の指標。新規参入=更新回数1の比率も算出)を返します。公式APIも一括ダウンロードも存在しないデータです。
• 使い方: registry で建設業(construction)か宅建業(realestate)を選び、prefectures に都道府県を指定します(JISコード 13 / 日本語 東京都 / 英語 Tokyo のいずれでも可)。業種を絞るときは businessTypes にコード(1〜29)・日本語名(電気、解体)・英語名を指定してください。個別業者の一覧が必要な場合のみ includeIndividualCompanies を ON にします。
• 料金: 都道府県サマリー1件 $0.02、個別業者レコードは $0.002/件(既定は OFF なので、既定実行は $0.02 固定)。0件の地域には課金されません。サブスクリプション不要です。
• 実数とサンプルの区別: 件数(総数・業種別・大臣許可/知事許可の内訳・一般/特定)はすべて全数の実数です(ministerLicenseSource: "exact")。標本から算出するのは 市区町村分布 と 宅建の更新回数分布 のみで、対象フィールド名・標本ページ・標本数は estimatedFrom.sampledFields に明記されます。標本は商号順に並べた検索結果を全ページ範囲へ等間隔にジャンプして層化サンプリングします(先頭ページだけを見ると大臣許可業者に偏るための対策)。
• 個人情報: 検索結果に表示される代表者名は出力しません(取得後に破棄)。商号・許可番号・所在地など法人情報のみを返します。