An open cream storefront, blue clock and sand-coloured key symbolise business continuity
Contents

Korea’s provisional 2025 Business Demography Statistics, released by the Ministry of Data and Statistics on October 22, reports a one-year survival rate of 63.9% and a five-year rate of 36.5%. The first follows businesses born in 2023 through 2024; the second follows the 2019 birth cohort through 2024. Because the starting cohorts differ, their gap is not an attrition rate for one group of firms.

Active businesses in 2025
7.748 million · up 106,000 (+1.4%)
Business births in 2025
908,000 · birth rate 11.7%
Different survival cohorts
2023 births: 63.9% at one year · 2019 births: 36.5% at five years
Fact-checked
October 25, 2026, 09:09 KST · MODS release and official PDF

Release year and measurement year are mixed

There were 7.748 million active businesses with sales or a regular employee in 2025, up 106,000 or 1.4% from a year earlier. Business births numbered 908,000, down 14,000 or 1.5%; the birth rate fell 0.3 percentage points to 11.7%.

Deaths are reported for 2024: 769,000 businesses, down 21,000 or 2.7%, with a death rate of 10.1%, 0.4 percentage points lower. A later year of administrative data is needed to establish that a 2024 active business disappeared, which is why the death series lags.

Provisional 2025 Business Demography Statistics · periods shown explicitly
MeasureOfficial figureChangeReference
Active businesses7.748m+106,000 · +1.4%2025
Business births908,000−14,000 · −1.5%2025
Birth rate11.7%−0.3 percentage pointsShare of 2025 active businesses
Business deaths769,000−21,000 · −2.7%2024
Death rate10.1%−0.4 percentage pointsShare of 2024 active businesses

63.9% and 36.5% start with different groups

The one-year survival rate of 63.9% is the share of businesses born in 2023 that had not died through 2024, down 0.5 percentage points from the comparable prior measure. The five-year rate of 36.5% follows businesses born in 2019 through 2024, up 0.1 percentage points.

Both the starting year and observation length differ. It would therefore be wrong to say one cohort fell by 27.4 percentage points from 63.9% to 36.5%, or to apply 36.5% as a guaranteed probability for an individual new venture.

Separate blue and sand trays hold storefront models with a clock and blank calendar cards
A conceptual illustration of one-year and five-year statistics using different starting cohorts. The number and colours of stores do not represent actual counts or survival shares.

An industry average is not your venture’s probability

Wholesale and retail was among the lower one-year survival rates at 54.1%, while accommodation and food services was among the lower five-year rates at 27.8%. Electricity, gas and steam stood at 87.5% for one year and 78.3% for five years. Entry structures, business composition and firm size differ across industries, so rankings are not a like-for-like risk score.

Use the same survival horizon and the closest relevant industry, then separately model rent, payroll, inputs, debt service and months of available cash. These statistics do not forecast an individual firm’s success or investment return.

Business registrations are not the same as business owners

An active business is a for-profit enterprise with sales or a regular employee during the year. A case in which at least two of representative, address and industry change together may be classified as a birth when the administrative data indicate genuinely new economic activity.

Death is assessed from administrative data when both sales and regular employment are absent for at least a year; it is not determined solely by a closure filing. If one sole proprietor holds several registrations, each can be counted as an individual enterprise, so the firm count is not a count of founders.

The release is provisional and uses administrative records received through September 2026. It may differ from survey statistics and can be revised as further data arrive.

Add your own cash-flow line to the cohort statistic

After finding a survival rate for a comparable industry and horizon, write down monthly fixed costs, a conservative sales case, break-even revenue and the number of months your cash can last. The population average cannot replace those calculations.

A live business in this measure is not necessarily profitable, and a closure filing does not immediately determine the statistical death date. Check the current loan, guarantee, tax and lease rules that apply to your own decision.

We checked the provisional result and cohort definitions together

The official release was posted on October 22, 2026. Active firms and births refer to 2025, while death and survival are observed through 2024. We have not rewritten the publication date as a common measurement year.

The MODS webpage, the official PDF’s main results, definitions and user notes were cross-checked on October 25, 2026 at 09:09 KST. This article does not estimate an unpublished outlook, market reaction or an individual venture’s probability of success.

Frequently asked questions

Does 63.9% at one year fall to 36.5% at five years?

Not as one observed cohort in this release. The one-year figure begins with 2023 births; the five-year figure begins with 2019 births. Their difference is not one cohort’s decline.

Does keeping a business registration mean the enterprise survived?

Not necessarily. The statistic uses sales or regular employment in administrative data, and death means both are absent for at least a year rather than relying only on a closure filing.

Summary

The 63.9% one-year rate follows businesses born in 2023; the 36.5% five-year rate follows 2019 births. Check the cohort and administrative definitions, and do not turn an industry average into an individual venture’s probability.

Sources

Ministry of Data and Statistics · Provisional 2025 Business Demography Statistics ↗Posted October 22, 2026. Active businesses, births, deaths, birth/death rates, survival rates and industry highlights checked.MODS · Official release PDF ↗Definitions of active, birth, death and survival; administrative-data basis; provisional status and user cautions checked October 25, 2026 at 09:09 KST.