kr-company-registry

Read the full write-up →

Five Korean government agencies assign different IDs to the same company. This table links them — for every company listed or delisted on Korean exchanges since DART records began.

Stat	Value
Companies covered	3,949 (active + delisted)
Active listed	2,768 — KOSPI 841 · KOSDAQ 1,817 · KONEX 110
Delisted (included)	1,181
BRN coverage	100% of domestic companies
CRN coverage	~99% (foreign-incorporated companies have none)
Refresh cadence	Weekly, automated (GitHub Actions)

---

The Problem

Korean capital markets data is fragmented across agencies that each use a different company identifier:

Identifier	Format	Assigned by	Used in
corp_code	8-digit, zero-padded	FSS / DART	All DART financial filings
ticker (종목코드)	6-digit alphanumeric	KRX	Stock prices, trading data
BRN (사업자등록번호)	10-digit	NTS / 국세청	Procurement (KONEPS), customs, tax
CRN (법인등록번호)	13-digit	Ministry of Justice	Court records, 법인등기
ISIN	KR + 10 alphanumeric	KSD / SEIBRO	Bond markets, CB/BW issuance

These agencies were built with different mandates and have never been officially linked. The result: connecting a company's DART financial filings to its KONEPS procurement contracts — or tracing it through the corporate registry — requires a lookup table that until now only commercial vendors (FnGuide, KOSCOM, Bloomberg) held as proprietary IP. The same structural gap exists in Taiwan, Japan, and Hong Kong. This project provides that link for Korea as open data.

---

The Data

Two files are published in data/dist/:

File	Format	Opens in
kr_corp_ids.csv	CSV (UTF-8 BOM)	Excel, R, any spreadsheet tool
kr_corp_ids.parquet	Parquet (columnar)	Python / pandas / DuckDB

Schema

Column	Description
corp_code	8-digit DART identifier. Permanent — survives relisting and SPAC mergers.
corp_name	Korean legal company name, as registered in DART.
ticker	6-character KRX ticker. Empty string for delisted companies.
market	KOSPI · KOSDAQ · KONEX · "" (delisted)
bizr_no	BRN (사업자등록번호): 10 digits, no hyphens. Stable unless the company restructures.
jurir_no	CRN (법인등록번호): 13 digits, no hyphens. Permanent — assigned at incorporation.
is_listed	True if currently listed; False for delisted companies.
corp_cls	Raw DART code: Y = KOSPI · K = KOSDAQ · N = KONEX · E = delisted
extracted_at	Date of the most recent extraction run (YYYY-MM-DD).

Sample rows

corp_code	corp_name	ticker	market	is_listed	bizr_no	jurir_no
00126380	삼성전자	005930	KOSPI	True	1248100998	1301110006246
00185301	제이피아이헬스케어	0010V0	KOSDAQ	True	2038148897	1101110285787
00599230	평산차업집단유한공사	950010		False	CT114980
00100258	에스마크	030270		False	3038103040	1511110001366

Row 3 is a foreign-incorporated company (ticker starting with 9): non-standard BRN, no CRN. Exclude from BRN-based joins — see Limitations below.

Identifier stability

Identifier	Stable?	Changes on
corp_code	Permanent	Never
CRN (jurir_no)	Permanent	Never
BRN (bizr_no)	Stable	Corporate restructuring (분할, 합병, 폐업+재설립)
ticker	Changes	Relisting, SPAC merger, 우회상장

---

Limitations

Read these before using the data in joins or published analysis.

• Point-in-time snapshot, not a history. Ticker changes from relisting or SPAC mergers are not tracked. For historical ticker histories, check KIND (한국거래소 상장법인목록) manually.

• Delisted companies are included. DART retains records for delisted companies indefinitely. 1,181 of the 3,949 rows are delisted. Use is_listed = True to filter to active listings only.

• Foreign-listed companies have non-standard identifiers. KRX lists ~39 foreign-incorporated companies in the ticker range 900xxx–950xxx. These companies have no Korean BRN — DART stores their foreign registration number in the bizr_no field, which is the wrong format for any KONEPS or customs join. They also have no CRN. Exclude them with df[~df["ticker"].str.startswith("9")] before any identifier-based join.

• ~36 tickers are alphanumeric. KRX assigns 6-character alphanumeric codes (e.g. 0004V0, 0015G0) to SPACs and some recently listed companies. These are valid KRX identifiers, not data errors. Count updates with each weekly refresh.

• Bond ISINs are not included. ISINs for convertible bonds and bonds with warrants are maintained separately via the FSC API. See kr-dart-pipeline/bond_isin_map.parquet.

• Court registry access is fee-based. CRN enables lookup in the 법원 법인등기 system, but that registry has no public API. Access is manual and fee-based.

---

Get the Data

Both files are committed to this repository. A git clone is all you need — no API key, no pipeline to run.

git clone https://github.com/pon00050/kr-company-registry.git
# data/dist/kr_corp_ids.csv  — open in Excel
# data/dist/kr_corp_ids.parquet  — load in Python or DuckDB

Check extracted_at to confirm when the data was last refreshed.

---

Update Cadence

The crosswalk refreshes automatically every week via GitHub Actions (Sunday 21:00 UTC = Monday 06:00 KST). Each run captures new listings (신규 상장), delistings (상장 폐지), and corporate restructurings that change BRN.

Existing DART responses are cached, so weekly runs complete in seconds unless new companies have appeared. extracted_at records the extraction date.

---

Who Uses This

Who	What they use it for
Investigative journalists	Connect a company's DART financial filings to its KONEPS procurement contracts; trace company networks through the corporate registry using CRN
Academic researchers	Build panel datasets linking DART disclosures, KRX price and volume data, and procurement patterns — identifiers that previously required separate lookup tables to join
Regulators (FSS / NTS)	Cross-system identity resolution across financial, tax, and procurement records for the same legal entity
Foreign institutional investors	Map DART corp_codes to KRX tickers for companies without English-language Bloomberg coverage
Compliance / KYC teams	BRN-based matching against KONEPS exclusion lists or sanctions databases

---

For Developers

Programmatic access

Fetch directly from GitHub without cloning:

import pandas as pd

df = pd.read_csv("https://raw.githubusercontent.com/pon00050/kr-company-registry/main/data/dist/kr_corp_ids.csv")
df = pd.read_parquet("https://raw.githubusercontent.com/pon00050/kr-company-registry/main/data/dist/kr_corp_ids.parquet")

Common lookups

import pandas as pd

df = pd.read_parquet("data/dist/kr_corp_ids.parquet")

# By ticker
df[df["ticker"] == "005930"]                          # Samsung Electronics

# By corp_code
df[df["corp_code"] == "00126380"]

# All KOSDAQ companies with BRN (excludes foreign-listed)
df[(df["market"] == "KOSDAQ") & (~df["ticker"].str.startswith("9"))]

-- DuckDB: join DART disclosures to KONEPS procurement records
SELECT c.corp_name, c.ticker, c.bizr_no, k.contract_amount
FROM 'data/dist/kr_corp_ids.parquet' c
JOIN 'koneps_contracts.parquet' k ON c.bizr_no = k.bizr_no
WHERE c.market = 'KOSDAQ'
ORDER BY k.contract_amount DESC;

Running the extractor

git clone https://github.com/pon00050/kr-company-registry.git
cd kr-company-registry
cp .env.example .env        # add your DART_API_KEY
uv sync

uv run python src/kr_company_registry/build_crosswalk.py --sample 10  # smoke test
uv run python src/kr_company_registry/build_crosswalk.py               # full run (~90 min first time, ~10s from cache)
uv run python src/kr_company_registry/validate.py
uv run pytest tests/

Flag	Default	Description
--sample N	None	Process only the first N companies
--sleep S	0.5	Seconds between DART API calls
--force	False	Re-fetch from DART even if cache exists

DART API responses are cached under data/raw/dart/<corp_code>.json (gitignored). A free DART API key is required to run the extractor — instant registration at opendart.fss.or.kr.

Data sources

Identifier	Source	Endpoint
corp_code, corp_name, bizr_no, jurir_no, corp_cls	DART	company.json via OpenDartReader
ticker, market listing status	DART corpCode.xml	Same library

company.json returns 20 fields per company; no field-selection parameter exists. Raw responses are cached locally and never committed. Published output contains only the 9 columns above.

Relationship to krff-shell

This project is the infrastructure layer consumed by krff-shell at Phase 5, when the forensic pipeline joins DART anomaly signals to KONEPS procurement and customs records via BRN and CRN. It is maintained separately so it is useful to researchers and journalists who have no interest in forensic scoring, and citable as a standalone open data release.

---

License

MIT. Data extracted from DART is subject to FSS terms of service (public disclosure data; generally permissive for research and information purposes).

---

Contributing

Issues and pull requests welcome. If you find a BRN or CRN that looks wrong — wrong length, clearly a different company — open an issue with the corp_code and what you found in the DART portal.