Korea Company Data

Data guide

What is in here, where each value came from, and how complete it is — including the parts that are thin.

Scope

Companies152,367
Who is includedKorean for-profit corporate headquarters with 10 or more insured employees, identified from employment-insurance workplace records
Who is notSole proprietors, branches, non-profits, and companies under 10 insured employees. A registration number outside this set is not necessarily invalid
Employment history36 months, monthly

Coverage by field

Measured across all 152,367 companies, not a sample.

FieldPresentSource
Employees (Employment Insurance)100.0%Workers' Compensation & Welfare Service
Address in English100.0%Rendered by rule
Established date100.0%FSC, else insurance inception
Worksite count100.0%Workers' Compensation & Welfare Service
Monthly employment series83.3%National Pension Service
Corporate registration number70.5%Financial Services Commission
Phone69.3%Financial Services Commission
Listing market · DART code25.5%Financial Services Commission
Homepage18.3%Financial Services Commission

10,284 companies (6.7%) have neither a Financial Services Commission match nor a pension link. For those, the record is employees, address, an insurance-inception date and a worksite count. It is thin, but no record is empty.

Sources and how often they change

SourceGivesAs ofRefresh
Korea Workers' Compensation & Welfare Service Company and worksite master, employment-insurance headcount, industry code 2025-12-31Annual
National Pension ServiceMonthly subscriber counts, hires, separations 2026-07Monthly
National Tax ServiceRegistration status (active, suspended, closed) Per recordWeekly
Financial Services Commission Corporate registration number, incorporation date, contact, listing Per recordDaily
Statistics KoreaKSIC 11th industry names in official English Rare

Fields that mean less than they look like

Three fields carry a companion field saying what kind of value it is. Read them together.

name_en_class official — the English name registered with the Financial Services Commission.
generated — romanized from Korean by rule. It is a reading aid, not a registered name. Do not use it as a legal name.
established_date_source fsc_estb — the incorporation date.
insurance_inception — the date the company first enrolled in employment insurance. These are different facts and we do not blur them into one "founded" field.
nps_quality A company can run several worksites, and pension records are per worksite. When we sum them we say how confident the sum is: clean, review, contaminated, unknown. clean means no sign of double counting — it is not a verification.

Two employee numbers, on purpose

ei_workersEmployment Insurance enrollees. Annual, from the master file.
nps_latestNational Pension subscribers in nps_period. Monthly. Excludes subscribers aged 60 and over and includes registered directors, so it differs from the insurance figure by design.

We do not reconcile them into a single "headcount". They are two government counts of two different things, and the gap between them is information.

What this dataset does not do

No assessmentNo score, no grade, no ranking, no opinion about any company. We state facts and where they came from.
No negative recordsNo sanctions, penalties, litigation or industrial-accident history. Not "coming later" — out of scope by design.
No peopleNo representative names, no personal contact details. Company records only.
No inferenceNothing is modelled, imputed or estimated. If a value is absent, it is absent, and the record says so.

Known limits

API & MCP docs Get a key — 50 free credits