How often new businesses survive, according to the government's own data

What the data actually tracks

BLS measures survival at the establishment level, not the company level. An establishment is a business operating at one physical location. If a chain opens a new branch, that counts as a new establishment being born, even though the parent company already exists. This distinction matters: an establishment can close because a business relocated, merged locations, or was absorbed by another company, not only because the underlying business failed outright.

BLS tracks this through a survival matrix, a table showing how many establishments born in a given year are still operating one year later, two years later, three years later, and so on. The survival rates calculated from that matrix are published as Table 7 in the agency's establishment age data, and are what make year-by-year comparisons across different birth cohorts possible, rather than relying on a single national failure statistic.

The one-year number, across three decades

BLS has published one-year survival rates by Census division for establishments born from 1994 through 2022. Across that span, one-year survival has typically run in the mid-to-high 70s percent, with a low of 71.4% for establishments born in 2008 in the South Atlantic division, the lowest point in the entire series, and a high of 84.6% for establishments born in 2021 in the Pacific division, the highest point in the series.

The national cohort figures in Table 7 tell the same story with even more consistency. Of establishments born in the year ending March 1994, 79.6% were still operating a year later, in March 1995. Of establishments born in the year ending March 2020, 80.9% were still operating a year later, in March 2021. Those two cohorts are separated by twenty-six years and a global pandemic, yet their one-year survival rates differ by little more than a single percentage point.

Even at the weakest point across twenty-eight years of division-level data, more than seven in ten new establishments were still open a year after opening. That is the opposite of a majority failing in year one.

The five-year and ten-year curve

Survival keeps falling after year one, but at a slower pace each year. For the 1994 cohort, 68.1% of establishments were still open two years after birth, 60.6% at three years, 54.3% at four years, and 49.6% at five years, in March 1999. For the 2020 cohort, the equivalent figures were 72.3% at two years, 63.6% at three years, 57.2% at four years, and 51.4% at five years, in March 2025.

That puts five-year survival close to half for both cohorts, a round number that holds up across twenty-six years of data. Beyond five years, the decline continues but slows further. Of establishments born in the year ending March 2013, 34.7% were still operating in March 2023 — the most recent cohort for which BLS has published a detailed industry breakdown. (BLS's underlying data table shows a newer full ten-year cohort too: establishments born in the year ending March 2015 also had a 34.7% survival rate in March 2025.)

That ten-year figure is not the product of steady annual losses. BLS reported that the survival rate for the 2013 cohort "dropped most in first year of operation," falling 20.4 percentage points between 2013 and 2014. That single year accounts for close to one-third of the total ten-year decline for that cohort. Every year after the first one removes a smaller share of the establishments that remain.

Recessions move the number more than the calendar does

One-year survival rates were generally lowest for establishments born in 2001 and 2008, the two years recessions began in that period. In the 2001 cohort, the Mountain division had the lowest one-year survival rate at 74.0%, while the West North Central division had the highest at 79.5%. In the 2008 cohort, the South Atlantic division posted the series low of 71.4%, while the Middle Atlantic division led at 78.0%.

By contrast, BLS reported increases across most regions for the 2020 cohort, and the 2021 cohort's Pacific division figure of 84.6% was the highest one-year rate recorded since the series began. Declines returned for the 2022 cohort, with division rates ranging from 74.4% to 78.6%. The South Atlantic division experienced the most change in one-year survival rate of any division across the 1994-2022 series, moving a cumulative 50.9 percentage points since the start of the series, in addition to recording the series' single lowest year.

Even at the weakest point across twenty-eight years of division-level data, more than seven in ten new establishments were still open a year after opening.

Industry matters more than a fixed timeline

For the 2013 cohort tracked through 2023, agriculture, forestry, fishing and hunting had the highest ten-year survival rate at 50.5%, followed by utilities at 45.7% and manufacturing at 43.6%. Mining, quarrying, and oil and gas extraction had the lowest rate at 24.5%, followed by information at 29.1% and wholesale trade at 30.1%.

Construction, an industry frequently cited as high-failure, posted 40.1% for this cohort, above the 34.7% economy-wide average. Health care and social assistance came in close to the average at 35.7%. That contrasts with BLS's own broader description of survival patterns across 1994-2015 cohorts, where health care and social assistance ranked consistently highest and construction ranked among the lowest. BLS did not publish exact figures for that longer comparison, so the two industry rankings cannot be reconciled precisely; the sources agree only that industry produces large, persistent differences, not that any one industry's rank is fixed across every cohort.

What survival doesn't tell you: shrinking birth size

Surviving is not the same as growing at the pace earlier cohorts did. BLS reported that employment from establishment births peaked at 1.4 million jobs in the first quarter of 1998 and fell to 661,000 jobs in the second quarter of 2020, during the pandemic recession, before recovering to nearly 1.1 million jobs by the second quarter of 2022. That decline happened even as the number of new establishments rose, from 181,000 births in the second quarter of 1993 to a record 379,000 in the fourth quarter of 2021.

The gap is explained by what BLS calls a new establishment's "birth weight," the average number of employees it starts with. That figure fell from 6.4 employees in 1996 to 2.7 in 2021, which BLS said means a new establishment born in the 2020s creates about 40 percent of the jobs it would have created in the 1990s. The pattern persists among survivors: establishments born in the 2010s employed fewer workers than 1990s cohorts at their one-, three- and five-year survival marks alike, at roughly 60% of the earlier levels. BLS also found that establishment births are the only age category of establishment that consistently creates more jobs than it loses.

The gap between the popular failure claim and the BLS data seems to come from two sources: conflating establishment-level closures with business failure, and citing multi-year cumulative decline figures as if they described the first year alone. A ten-year cohort losing two-thirds of its establishments sounds close to the popular claim, but the decline is spread across a decade, with the largest single-year drop concentrated in year one and each subsequent year removing a smaller share of what remains. BLS does not publish a single national one-year survival rate averaged across the whole 1994-2022 series; the reported range across all divisions and years is 71.4% to 84.6%, while Table 7 gives a national rate for each individual birth-year cohort, such as 79.6% for 1994 and 80.9% for 2020. Under that range, or under the national cohort figures in Table 7, a majority of new establishments were still open twelve months after opening in every year the program has measured.