US Business Formation Statistics 2026
By Axis Intelligence Research
Co-author: Mia Scarlett (Business & Entrepreneurship) | Last updated: August 8, 2026 | License: CC BY 4.0
Americans filed 6,020,322 business applications in the twelve months ending June 2026, the first time the trailing-twelve-month count has crossed six million since the Census Bureau began the series in 2004. Applications indicating planned wages fell 16.9% over the same period. According to Axis Intelligence Research, the two numbers describe one economy moving in opposite directions.
Quick Answer
The U.S. recorded 6,020,322 business applications in the twelve months through June 2026 (U.S. Census Bureau, Business Formation Statistics, July 9, 2026), a record. But only 7.45% of those applications indicated planned wages, against 13.68% in 2019. According to Axis Intelligence Research, the Entrepreneurial Formation Signal Index (EFSI) reads 68.6, down from a 2019 baseline of 100.
Key Findings
- According to Axis Intelligence Research, U.S. business applications reached 6,020,322 in the twelve months ending June 2026, the highest trailing-twelve-month total in the twenty-two-year history of the Census Bureau’s Business Formation Statistics series.
- According to Axis Intelligence Research, applications with planned wages fell to 7.45% of all applications in the year through June 2026, down from 13.68% in calendar 2019 — the share has been cut nearly in half while total volume rose 62%.
- The Axis Intelligence Research Entrepreneurial Formation Signal Index (EFSI) reads 68.6 as of June 2026, its seventh consecutive annual decline from a 2019 baseline of 100.
- According to Axis Intelligence Research’s combination of Census and Bureau of Labor Statistics data, it took 4.37 business applications to produce one employer establishment birth in 2025, against 3.37 in 2019 — a 29.6% increase in application volume per real employer.
- Business applications for June 2026, seasonally adjusted, were 531,423, up 1.1% from May 2026, with 149,714 classified as high-propensity (U.S. Census Bureau, July 9, 2026).
How Many Businesses Are Started in the US Each Year?
The honest answer starts with a definition problem, and the definition problem is the whole story.
The Census Bureau’s Business Formation Statistics program counts applications for an Employer Identification Number filed on IRS Form SS-4. That is not the same thing as counting businesses. An EIN application is a legal act that costs nothing and commits the filer to nothing. It might precede a restaurant with twelve employees. It might be a single-member LLC opened to hold a rental property, or to receive 1099 income, or because a marketplace platform asked for a tax ID.
The Census Bureau built four application series precisely because it knew this. Business Applications (BA) is the raw count. High-Propensity Business Applications (HBA) filters for the characteristics that historically predict payroll. Applications with Planned Wages (WBA) is the narrowest and most demanding filter: the filer entered a date on which wages would first be paid. Applications from Corporations (CBA) covers filings by corporate entities.
Here is what those four series show at the annual level, all figures not seasonally adjusted and drawn from the current data vintage.
| Year | Business applications | High-propensity | With planned wages | Wage-intent share | Source |
|---|---|---|---|---|---|
| 2019 | 3,498,990 | 1,316,191 | 478,640 | 13.68% | U.S. Census Bureau BFS |
| 2020 | 4,356,498 | 1,518,956 | 524,651 | 12.04% | U.S. Census Bureau BFS |
| 2021 | 5,390,816 | 1,841,954 | 638,140 | 11.84% | U.S. Census Bureau BFS |
| 2022 | 5,062,563 | 1,722,399 | 578,669 | 11.43% | U.S. Census Bureau BFS |
| 2023 | 5,469,302 | 1,848,540 | 613,708 | 11.22% | U.S. Census Bureau BFS |
| 2024 | 5,224,176 | 1,715,458 | 550,423 | 10.54% | U.S. Census Bureau BFS |
| 2025 | 5,671,836 | 1,708,842 | 501,223 | 8.84% | U.S. Census Bureau BFS |
Sources: U.S. Census Bureau, Business Formation Statistics, series BABATOTALNSAUS, BAHBATOTALNSAUS, BAWBATOTALNSAUS, retrieved August 8, 2026. Wage-intent share is calculated by Axis Intelligence Research as WBA divided by BA.
Calendar 2025 was the record year at 5,671,836 applications. The twelve months through June 2026 beat it: 6,020,322, computed by Axis Intelligence Research by summing the six monthly observations from July through December 2025 with the six from January through June 2026.
Mia Scarlett on the denominator problem: Every headline written about this series treats the total as a proxy for entrepreneurial energy, which is the same error as reading gross bookings and calling it revenue. The wage-intent column is the one that matters. It fell from 638,140 in 2021 to 501,223 in 2025 while the top line rose. A filing that names a first payroll date is a filer who has thought about employees. Everything else is optionality — cheap to acquire, frequently abandoned, and increasingly the bulk of the series.
The Entrepreneurial Formation Signal Index (EFSI)
Axis Intelligence Research built the Entrepreneurial Formation Signal Index (EFSI) to answer one question the raw counts cannot: how much employer-business signal does each month of application volume actually carry?
What EFSI measures
EFSI is a composite of three ratios, each expressing a quality subseries as a share of total applications, each indexed to calendar 2019 and weighted.
Formula
EFSI = 100 x [ 0.40 x (WBA_t/BA_t) / (WBA_2019/BA_2019)
+ 0.25 x (HBA_t/BA_t) / (HBA_2019/BA_2019)
+ 0.35 x (PBF4Q_t/BA_t) / (PBF4Q_2019/BA_2019) ]
Inputs
| Component | Series | Weight | 2019 baseline share | Rationale |
|---|---|---|---|---|
| Wage intent | WBA / BA | 0.40 | 13.6794% | Filer named a first wages-paid date; strongest declared commitment to payroll |
| High propensity | HBA / BA | 0.25 | 37.6163% | Census Bureau’s own predictive classification |
| Projected formation yield | PBF4Q / BA | 0.35 | 7.0548% | Census econometric projection of employer formation within four quarters |
Baseline: calendar 2019, the last full pre-pandemic year, set to 100.
EFSI readings
| Period | Wage intent | High propensity | Formation yield | EFSI |
|---|---|---|---|---|
| 2019 | 13.68% | 37.62% | 7.05% | 100.0 |
| 2020 | 12.04% | 34.86% | 5.99% | 88.1 |
| 2021 | 11.84% | 34.17% | 5.87% | 86.4 |
| 2022 | 11.43% | 34.02% | 6.00% | 85.8 |
| 2023 | 11.22% | 33.80% | 5.76% | 83.8 |
| 2024 | 10.54% | 32.84% | 5.71% | 80.9 |
| 2025 | 8.84% | 30.13% | 5.52% | 73.2 |
| Jul 2025 – Jun 2026 | 7.45% | 28.64% | 5.60% | 68.6 |
Source: Axis Intelligence Research, computed from U.S. Census Bureau Business Formation Statistics, current vintage, retrieved August 8, 2026. All inputs are in the accompanying CSV.
The current reading of 68.6 is the baseline reading for this index. Seven consecutive periods of decline, no reversal in any single year, and the steepest single-year drop in the most recent one. The wage-intent sub-index alone stands at 54.5 — the component doing most of the damage.
What moved, and what would reverse it
The decline is not a pandemic artifact that is slowly unwinding. If it were, the index would have troughed in 2021 and recovered as the labor market normalized. Instead 2025 and the trailing year through June 2026 produced the two largest annual declines in the series.
EFSI would reverse on a rising wage-intent share, not on rising application volume. A reading back above 80 would require roughly 315,000 additional planned-wage filings per year at current application volumes. Axis Intelligence Research will publish an updated reading after each Census Bureau annual revision and flag any component whose definition changes.
Are Business Applications Actually Turning Into Employers?
This is where two federal datasets have to be put side by side, because neither answers the question alone.
The Census Bureau’s projected business formations series estimates how many applications will become payroll-paying employers within four quarters. The Bureau of Labor Statistics counts something different and harder: establishment births, drawn from the Business Employment Dynamics program’s longitudinal linkage of unemployment insurance records. A BED birth is an establishment appearing with positive employment for the first time. It is administrative fact, not projection.
Axis Intelligence Research combined the two.
| Year | Business applications (Census) | Establishment births (BLS BED) | Applications per birth |
|---|---|---|---|
| 2019 | 3,498,990 | 1,038,000 | 3.37 |
| 2020 | 4,356,498 | 1,069,000 | 4.08 |
| 2021 | 5,390,816 | 1,397,000 | 3.86 |
| 2022 | 5,062,563 | 1,408,000 | 3.60 |
| 2023 | 5,469,302 | 1,307,000 | 4.18 |
| 2024 | 5,224,176 | 1,292,000 | 4.04 |
| 2025 | 5,671,836 | 1,298,000 | 4.37 |
Sources: U.S. Census Bureau, Business Formation Statistics (applications); U.S. Bureau of Labor Statistics, Business Employment Dynamics, Fourth Quarter 2025, released July 29, 2026, Table 8 (establishment births, seasonally adjusted quarterly levels summed to annual). Ratio calculated by Axis Intelligence Research.
According to Axis Intelligence Research, the United States needed 4.37 business applications in 2025 to produce one employer establishment birth, against 3.37 in 2019 — 29.6% more paperwork per real employer.
Two caveats belong with that number, and stating them is the difference between a metric and a talking point. BED births include establishments opened by firms that already exist, so the denominator is not purely new-firm activity. And an application filed in December 2025 may produce a birth in 2026, so the annual pairing is approximate rather than a cohort match. Neither problem moves with time, which is why the trend is informative even where the level is not exact.
BLS recorded 338,000 establishment births in the fourth quarter of 2025, accounting for 1,006,000 jobs. That is the highest quarterly birth count since 2022 and a real signal of strength — which is exactly why the divergence matters. Employer formation is holding up. Application volume is running far ahead of it.
Mia Scarlett on the two datasets: Analysts who cover this space tend to pick one series and defend it. The useful move is to hold both and read the gap. BED says roughly 1.3 million establishments opened with payroll last year, a number that has been remarkably stable since 2022. Census says applications went from 5.06 million to 5.67 million over the same stretch. Stable output, rising input. In any other operating context that ratio would trigger a review.
Business Formation Statistics by Region
The June 2026 release is the first Census publication carrying a full regional cut on the current methodology vintage, which makes cross-region comparison clean.
| Region | Applications (Jun 2026, SA) | Share of U.S. | High-propensity share | Wage-intent share | EFSI |
|---|---|---|---|---|---|
| Northeast | 74,558 | 14.0% | 31.7% | 7.00% | 72.6 |
| West | 132,134 | 24.9% | 29.4% | 7.10% | 72.2 |
| Midwest | 88,533 | 16.7% | 26.7% | 7.16% | 65.2 |
| South | 236,198 | 44.4% | 26.9% | 6.24% | 61.1 |
| United States | 531,423 | 100% | 28.2% | 6.72% | 66.1 |
Source: U.S. Census Bureau, Business Formation Statistics, June 2026 (Release CB26-115, July 9, 2026), seasonally adjusted. Shares and regional EFSI calculated by Axis Intelligence Research against the 2019 national baseline.
The South files 44.4% of all U.S. business applications and posts the lowest formation signal of any region. The Northeast files 14.0% and posts the highest. According to Axis Intelligence Research, that inversion is the single most under-reported fact in the state and regional formation data: the regions generating the most filings are generating the least employer intent per filing.
Regional application growth in June 2026 ran +3.6% in the Midwest, +2.3% in the West, +0.4% in the Northeast and -0.2% in the South, month over month.
Which Industries Are Generating New Business Applications?
| NAICS sector | Applications (Jun 2026, SA) | Share | MoM change | Source |
|---|---|---|---|---|
| 44-45 Retail Trade | 97,096 | 18.3% | +2.0% | U.S. Census Bureau BFS |
| 54 Professional Services | 82,990 | 15.6% | +1.3% | U.S. Census Bureau BFS |
| 23 Construction | 48,933 | 9.2% | +0.8% | U.S. Census Bureau BFS |
| 81 Other Services | 44,314 | 8.3% | +0.9% | U.S. Census Bureau BFS |
| 56 Administrative and Support | 36,909 | 6.9% | 0.0% | U.S. Census Bureau BFS |
| 48-49 Transportation and Warehousing | 34,512 | 6.5% | +3.1% | U.S. Census Bureau BFS |
| 62 Health Care and Social Assistance | 33,436 | 6.3% | -0.7% | U.S. Census Bureau BFS |
| 72 Accommodation and Food Services | 28,777 | 5.4% | +2.6% | U.S. Census Bureau BFS |
| 53 Real Estate | 26,565 | 5.0% | +1.1% | U.S. Census Bureau BFS |
| 52 Finance and Insurance | 19,512 | 3.7% | +0.9% | U.S. Census Bureau BFS |
| 71 Arts and Entertainment | 15,204 | 2.9% | -0.8% | U.S. Census Bureau BFS |
| 51 Information | 12,999 | 2.4% | +4.7% | U.S. Census Bureau BFS |
| 42 Wholesale Trade | 9,061 | 1.7% | -3.1% | U.S. Census Bureau BFS |
| 61 Educational Services | 8,345 | 1.6% | +0.2% | U.S. Census Bureau BFS |
| 31-33 Manufacturing | 7,424 | 1.4% | +3.1% | U.S. Census Bureau BFS |
| 55 Management of Companies | 6,517 | 1.2% | +1.1% | U.S. Census Bureau BFS |
| 11 Agriculture | 3,794 | 0.7% | -1.3% | U.S. Census Bureau BFS |
| 22 Utilities | 610 | 0.1% | +8.2% | U.S. Census Bureau BFS |
| 21 Mining | 554 | 0.1% | +6.7% | U.S. Census Bureau BFS |
| Unclassified | 13,872 | 2.6% | -2.7% | U.S. Census Bureau BFS |
Source: U.S. Census Bureau, Business Formation Statistics, June 2026, Release CB26-115. Shares calculated by Axis Intelligence Research. Utilities are not seasonally adjusted. Details may not sum to totals because of rounding.
Retail trade, professional services and construction account for 43.1% of all June 2026 applications. Retail alone is 18.3% — nearly one in five filings, in a sector where the marginal new entrant is far more likely to be a sole proprietor selling on a marketplace than a store with a lease.
That composition is the mechanical explanation for the EFSI decline. The sectors gaining share are the sectors with the lowest historical conversion to payroll. Our small business statistics research covers the resulting stock of firms, and the US small business failure rate statistics analysis tracks what happens to the ones that do reach payroll.
The Data Trap: Why Year-Over-Year Comparisons Are Broken Right Now
This section exists because a specific, avoidable error is about to propagate through every article written on this topic in 2026, and Axis Intelligence Research would rather publish the warning than the error.
As of the January 2026 monthly release, the Census Bureau excluded applications associated with internet sales from the high-propensity and corporation series. The Bureau restated the entire historical series to match. Separately, the January 2026 release was the first published on the 2022 NAICS vintage, with the full time series restated from the 2017 vintage.
The consequence is arithmetic. Compare the archived May 2025 release against the current June 2026 release and high-propensity applications appear to fall from 159,009 to 149,714 — a decline of 5.8%. Compare June 2025 to June 2026 within the current vintage and high-propensity applications rise from 134,376 to 141,641, up 5.4%.
Same series. Same question. Opposite sign.
According to Axis Intelligence Research, any 2026 comparison of high-propensity or corporation applications that draws its earlier figure from an archived press release is reporting a definitional change as an economic trend. The wage-intent series was not touched by the exclusion, which is one reason EFSI weights it most heavily.
Mia Scarlett on restated series: This is the filing-footnote problem in a federal dataset. The headline moved, the definition moved underneath it, and only one of those movements is news. Anyone who has restated a segment after a divestiture recognizes the pattern immediately: you compare like to like or you compare nothing at all.
Business Formation Statistics: June 2026 in Detail
| Series | June 2026 (SA) | Change vs May 2026 | Source |
|---|---|---|---|
| Business applications | 531,423 | +1.1% | U.S. Census Bureau, CB26-115 |
| High-propensity applications | 149,714 | +1.9% | U.S. Census Bureau, CB26-115 |
| Applications with planned wages | 35,695 | +0.3% | U.S. Census Bureau, CB26-115 |
| Applications from corporations | 43,346 | +5.5% | U.S. Census Bureau, CB26-115 |
| Projected formations within 4 quarters | 29,741 | +0.7% | U.S. Census Bureau, CB26-115 |
| Projected formations within 8 quarters | 41,042 | +1.3% | U.S. Census Bureau, CB26-115 |
The June-to-June comparison on the current vintage, not seasonally adjusted, is the cleanest year-over-year read available:
| June, NSA | 2019 | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|---|
| Business applications | 270,168 | 435,825 | 403,642 | 428,838 | 496,832 |
| High-propensity | 104,840 | 146,344 | 133,423 | 134,376 | 141,641 |
| With planned wages | 38,582 | 45,610 | 42,549 | 39,745 | 33,031 |
| Projected formations (4Q) | 19,779 | 24,938 | 23,331 | 23,709 | 27,808 |
| EFSI | 103.9 | 81.3 | 81.5 | 75.4 | 66.2 |
Source: U.S. Census Bureau, Business Formation Statistics, current vintage; EFSI calculated by Axis Intelligence Research.
Applications in June 2026 ran 83.9% above June 2019. Planned-wage applications ran 14.4% below June 2019. According to Axis Intelligence Research, that is the entire finding in two sentences.
Applying the 2019 wage-intent structure to the first half of 2026 implies 451,030 planned-wage applications. The actual figure was 230,872 — a shortfall of 220,158 filings that the raw application count conceals entirely.
Methodology
Collection. All application and formation figures come from the U.S. Census Bureau’s Business Formation Statistics program, retrieved August 8, 2026 from the June 2026 monthly release (CB26-115, published July 9, 2026) and from the full monthly time series distributed through the Federal Reserve Bank of St. Louis: BABATOTALNSAUS, BAHBATOTALNSAUS, BAWBATOTALNSAUS and BFPBF4QTOTALNSAUS. Establishment birth counts come from the BLS Business Employment Dynamics release for the fourth quarter of 2025, published July 29, 2026, Table 8.
Vintage discipline. Every historical figure in this article is drawn from the current data vintage as published on or after February 11, 2026. No figure is taken from an archived press release except where an archived figure is explicitly identified as such in the section on broken year-over-year comparisons.
Seasonal adjustment. Monthly release figures are seasonally adjusted, as published. Annual totals, June-to-June comparisons and all EFSI readings use not seasonally adjusted data, because summing twelve months of seasonally adjusted values reintroduces adjustment residuals and because June-to-June comparison controls for seasonality directly. Regional EFSI uses the seasonally adjusted regional figures published in CB26-115, since no regional NSA cut was available in the same release.
EFSI construction. Components, weights and baseline shares are disclosed in the index section above. Weights reflect the declared strength of each signal: a named first-wages date is the most demanding commitment on IRS Form SS-4, the Census projection model is the most statistically grounded, and the high-propensity classification is the broadest. A competent analyst can recompute every reading from the CSV.
Cross-source ratio. Applications per establishment birth divides annual Census BFS business applications by the sum of four seasonally adjusted quarterly BLS BED establishment birth levels. The two programs draw on different administrative universes — IRS EIN filings and state unemployment insurance records — so the ratio is an indicator of relative movement, not a conversion rate.
Arithmetic. All shares, growth rates, sums and index values were computed programmatically from the source series and independently re-run before publication.
Scope and caveats
EFSI measures declared intent at the point of application, not outcomes. A filer who does not name a wages date may still hire; a filer who does may never open. The index tracks the composition of stated intent, which is what the source data supports.
The high-propensity component reflects the Census Bureau’s classification as it stands today, including the internet-sales exclusion applied in January 2026. Because the Bureau restated the full history, the component is internally consistent across the series, but it is not identical to the classification in use before 2026.
The applications-per-birth ratio pairs calendar-year applications with calendar-year births rather than tracking application cohorts through to formation. Cohort-level matching is not possible from published data.
Census reports actual, non-projected business formations only through the fourth quarter of 2023, with formations within four quarters stated through December 2022. Everything after that is projection.
About This Dataset
business-formation-statistics-2026.csv contains every figure appearing in this article, one observation per row, with source organization, source document, URL, retrieval date, primary-source flag and an Axis-calculated flag. Rows marked axis_calculated = yes carry a method note pointing to the formula disclosed above.
Coverage: United States, national and Census region; calendar 2019 through June 2026; monthly and annual frequency.
License: CC BY 4.0
Format: UTF-8, comma-separated, one header row, ISO 8601 dates, raw numeric values with units in a separate column.
Distribution: Hugging Face, Kaggle and GitHub, in addition to direct download.
Cite as:
APA — Axis Intelligence Research. (2026). Business formation statistics 2026: Entrepreneurial Formation Signal Index. https://axis-intelligence.com/business-formation-statistics/
MLA — Axis Intelligence Research. “Business Formation Statistics 2026: Entrepreneurial Formation Signal Index.” Axis Intelligence, 8 Aug. 2026, axis-intelligence.com/business-formation-statistics/.
Chicago — Axis Intelligence Research. “Business Formation Statistics 2026: Entrepreneurial Formation Signal Index.” Axis Intelligence, August 8, 2026. https://axis-intelligence.com/business-formation-statistics/.
Frequently Asked Questions
Does a business application in the Census BFS data mean a business actually opened?
No. The series counts applications for an Employer Identification Number filed on IRS Form SS-4. The Census Bureau publishes a separate formation series precisely because the two differ. In 2025 there were 5,671,836 applications and, per BLS, 1,298,000 employer establishment births — a ratio of 4.37 to one.
What is the difference between business applications and high-propensity business applications?
High-propensity applications are the subset whose characteristics historically predict payroll: filings from corporate entities, filings indicating hiring, filings naming a first wages-paid date, and filings in specific NAICS codes. In June 2026, 149,714 of 531,423 applications were classified high-propensity, or 28.2%.
Why did high-propensity application counts change between 2025 and 2026 releases?
As of the January 2026 release the Census Bureau excluded internet-sales applications from the high-propensity and corporation series and restated the full history. A separate restatement moved the series from 2017 NAICS to 2022 NAICS. Comparing a 2025 archived release against a 2026 release therefore mixes vintages and can reverse the sign of the year-over-year change.
What does the Entrepreneurial Formation Signal Index measure?
EFSI measures how much employer-business intent each unit of application volume carries, combining wage-intent share, high-propensity share and projected formation yield against a 2019 baseline of 100. The reading for the year through June 2026 is 68.6.
Which US region files the most business applications?
The South, at 236,198 seasonally adjusted applications in June 2026 — 44.4% of the national total. It also posts the lowest regional EFSI at 61.1, meaning it generates the most filings and the least employer intent per filing.
Which industries generate the most new business applications?
Retail trade led June 2026 with 97,096 applications (18.3% of the total), followed by professional services at 82,990 (15.6%) and construction at 48,933 (9.2%). Those three sectors account for 43.1% of all applications.
How far ahead can Census project business formations?
The projected formations series estimates employer formations within four and eight quarters of application. Actual, non-projected formation data currently runs only through the fourth quarter of 2023, which means formations within four quarters are confirmed through December 2022 and within eight quarters through December 2021.
Are business formation statistics a leading indicator for employment?
Applications lead employer formations by roughly four quarters by construction, and BLS establishment births confirm the outcome with a further lag. The predictive value sits in the high-propensity and wage-intent subseries rather than the headline count, which is why EFSI weights those components rather than total volume.
How often is BFS data released?
Monthly, roughly five weeks after month end. The June 2026 data was published July 9, 2026; July 2026 data is scheduled for August 12, 2026. Weekly application data is also published, approximately five days after each week ends.
Where can I download the underlying data?
The CSV accompanying this article carries every figure cited here with full source provenance, released under CC BY 4.0. The Census Bureau publishes the complete time series through its own data tables and through the Federal Reserve Bank of St. Louis.
Related Axis Intelligence Research: Small Business Statistics 2026 · Startup Statistics 2026 · US Small Business Failure Rate Statistics 2026 · Venture Capital Statistics 2026 · Fintech Funding Statistics 2026 · US Business Operating Costs Statistics · Startup Failure Rate Statistics