Docket Article
Census Annual Business Survey: Owner Data, Tables and API
What the Census Annual Business Survey says about owners, including owner age by sector and state, how to pull ABS tables and API data, and its limits.
Summary
- 01Record the table ID and reference year for every pull. The API and published tables use the reference year, while survey naming conventions change between cycles.
- 02Owner age is a coded question in abscbo, using QDESC and OWNCHAR. The figures count responding owners of employer firms, so they do not establish how many companies have an older owner.
- 03ABS employer data and NES-D nonemployer data cover different populations and use different methods and geographic codes. Keep them separate in employer-firm analysis.
- 04Keep error fields and suppression flags beside each value. Smaller geographic and industry slices can have large errors, and the summed older-owner share has no published sampling error.
- 05Use aggregate owner age to choose and size a sector and geography for outreach. Company-level evidence must establish ownership and succession signals; ABS does not measure intent to sell.
Inside this article
- 01What the ABS is and which vintage you are reading
- 02What it says about owners, including owner age
- 03Employer vs nonemployer coverage
- 04Tables vs API: which dataset answers which question
- 05Suppression, noise and sampling
- 06Building an ownership view by NAICS and state
- 07Getting owner age and succession signals at company level
- 08Worksheet: ABS pull checklist
- 09Sources
What the ABS is and which vintage you are reading
The Annual Business Survey (ABS) is the Census Bureau's main source of statistics on who owns U.S. employer businesses. It is a joint project of the Census Bureau and the National Center for Science and Engineering Statistics (NCSES) within the National Science Foundation. It covers owner demographics, research and development, innovation and other business characteristics [1]. It replaced the five-year Survey of Business Owners (SBO) for employer businesses. It also superseded the Annual Survey of Entrepreneurs and the BRDI-M survey [2].
Check the naming convention before you quote any figure. Through the 2024 cycle, a survey's name was the year it was fielded, not the year it measures: "the 2024 ABS ... covers reference year 2023" [3]. The API and the data.census.gov tables are filed under the reference year. So the most recent owner tables, labelled 2023, come from the survey fielded as the 2024 ABS. The convention changes from the next cycle on. The survey fielded in 2026 is called the 2025 ABS, after its reference year, and earlier releases keep their original names [2]. A citation should therefore record the table ID and the reference year, not just "the 2024 ABS". As of 2026-10-28, reference year 2023 is still the newest vintage: the Census Data API catalog lists no ABS dataset later than 2023, and data.census.gov has no 2024 or 2025 owner table [4]. The 2025 ABS was scheduled to be mailed in May 2026 [2], so its tables are not yet out.
The 2024 cycle went to about 330,000 employer businesses, and about 63.3% of firms still operating responded [3]. The sample was drawn from the Business Register, which the Census Bureau built from business tax returns and its other economic surveys. Several industries are out of scope, including parts of agriculture, rail, the postal service, private households and public administration (same source).
For the ABS row in the reference catalog, see the Docket data-source page for the Census Annual Business Survey. This article covers the owner tables only.
What it says about owners, including owner age
The 2023 owner dataset (abscbo) has five dimensions with their own variables: sex, ethnicity, race, veteran status, and number of owners (OWNPDEMP). Every other owner trait sits inside two coded variables. QDESC is the question code and OWNCHAR is the answer category [5].
Owner age is one of those coded questions. No variable is named "age", so a scan of the variable list misses it. In the published 2023 owner table, AB2300CSCBO, owner age is question O09 (label OWNRAGE), with six bands: under 25, 25 to 34, 35 to 44, 45 to 54, 55 to 64, and 65 or over [6]. The same table carries these other owner questions under QDESC:
| QDESC | Label | What it records |
|---|---|---|
| O02 | YRACQBUS | Year the owner acquired the business (banded: before 1980 through 2023) |
| O05 | PRIMINC | Whether the business was the owner's primary source of income |
| O07 | EDUC | Highest education level |
| O08 | FLDEDUC | Field of degree |
| O09 | OWNRAGE | Owner age band |
| O10 | REASONOWN | Reasons for owning the business |
| O11 / O12 | USBORN / USCITIZEN | Nativity and citizenship |
| O15 | PRIORBUSTAT | Whether the owner owned an earlier business, and what happened to it |
| O16 | DISABLED | Disability status |
Source: question codes read from table AB2300CSCBO, national, all sectors, 2023. Other codes exist; the list above is not exhaustive.
The headline figures, read correctly
In the 2023 table, 4,208,231 owners of respondent employer firms reported an age. Of these, 28.2% were 55 to 64 and 23.1% were 65 or over, so 51.3% were 55 or older [6]. The 2022 table gives 28.9% and 22.8%, or 51.7% [7]. The Census Bureau's graphic for reference year 2018 gave 51% aged 55 and older [8].
The unit matters. These are counts of owners, not firms. The ABS collects demographics for "up to four persons owning the largest percentages in the business" [3], so a firm with three owners can add three people to the table. The Census Bureau's own note on the 2018 graphic says the percentages "do not represent all U.S. business owners" and cover only responding owners of employer businesses [8].
"Half of business owners are over 55" is therefore accurate as a statement about owners. "Half of companies have an owner over 55" is a different claim, and the ABS owner table does not support it.
Owner age by sector
Owner age differs by sector more than the national figure suggests. National, 2023, all owners reporting:
| NAICS sector | Owners reporting | 55 to 64 | 65 or over | 55+ (sum) |
|---|---|---|---|---|
| Wholesale trade (42) | 223,182 | 30.7% | 32.7% | 63.4% |
| Manufacturing (31-33) | 206,182 | 30.9% | 30.7% | 61.6% |
| Real estate and rental and leasing (53) | 256,602 | 27.2% | 30.0% | 57.2% |
| Retail trade (44-45) | 418,614 | 28.1% | 26.1% | 54.2% |
| Professional, scientific and technical services (54) | 704,052 | 28.9% | 23.9% | 52.8% |
| Construction (23) | 552,579 | 29.2% | 19.9% | 49.1% |
| Transportation and warehousing (48-49) | 130,862 | 28.7% | 20.2% | 48.9% |
| Health care and social assistance (62) | 431,969 | 25.7% | 19.3% | 45.0% |
| Accommodation and food services (72) | 373,761 | 26.7% | 17.9% | 44.6% |
| All sectors | 4,208,231 | 28.2% | 23.1% | 51.3% |
Source: AB2300CSCBO, reference year 2023. The 55+ column is our sum of the two published bands, so its sampling error is not published.
They describe the population of owners in a sector, not the age of any particular owner.
Employer vs nonemployer coverage
The ABS covers employer firms only, defined as firms "with payroll at any time during the reference year" [3]. Owners of businesses with no employees are in a separate product, the Nonemployer Statistics by Demographics series (NES-D). The Census Bureau says NES-D "is not a survey". Instead it assigns owner traits to the nonemployer universe from individual-level administrative records, and those traits include owner age, place of birth and citizenship [9].
The NES-D owner table for 2023 (AB2300NESD04) uses the same O09 age question. It counts about 32.7 million nonemployer owners reporting an age, of whom 18.5% were 55 to 64 and 15.9% were 65 or over [10]. NES-D counts are published rounded to thousands.
Most acquisition mandates have an EBITDA floor that rules out nonemployer businesses. Keep NES-D out of employer-firm analysis. The two products differ in method (survey versus administrative records) and in their geographic codes, so adding them together produces a number that neither source supports.
- Covers firms with payroll at any time during the reference year.
- Owner-table figures count owners, rather than firms.
- Covers owners of businesses with no employees.
- Assigns owner traits from individual-level administrative records, including age, place of birth and citizenship.
Keep NES-D out of employer-firm analysis. Adding the products together produces a number that neither source supports.
Nothing in the table identifies a company: it says where to point outreach, not whom to contact.
Tables vs API: which dataset answers which question
| Question | Dataset (API name) | Published table | Key fields |
|---|---|---|---|
| How old are owners in sector X, state Y? | abscbo | AB2300CSCBO | QDESC=O09, OWNCHAR, OWNPDEMP, OWNPDEMP_PCT, OWNPDEMP_PCT_S |
| How long have firms been in business? | abscs | AB2300CSA02 | YIBSZFI, FIRMPDEMP, EMP, RCPPDEMP |
| Firm counts, receipts, payroll by owner demographic group | abscs | Company summary tables | SEX, ETH_GROUP, RACE_GROUP, VET_GROUP |
| Business practices, financing, technology | abscb, absmcb | Characteristics tables | QDESC, BUSCHAR |
| Nonemployer owners' age, nativity | absnesdo | AB2300NESD04 | QDESC=O09, OWNNOPD |
Dataset list from the Census Data API catalog for the 2023 vintage. The YIBSZFI "Years in business code" variable is listed in the 2023 abscs variables.
There are two ways to retrieve the data.
- data.census.gov tables. Open a table ID such as AB2300CSCBO, filter geography and NAICS, and download. The national owner table is too large to display unfiltered (the viewer reports roughly 285 MB), so filter to one geography first.
- Census Data API. The ABS developer page says "All data queries to the Census Data API now require an API key" [2]. Keys are free from the Census Bureau's key signup page. The variable lists and the official example calls can be read without a key.
The official 2023 owner examples follow this pattern [11]:
https://api.census.gov/data/2023/abscbo?get=NAME,GEO_ID,NAICS2022_LABEL,OWNER_SEX,OWNER_ETH,OWNER_RACE,OWNER_VET,OWNPDEMP&for=state:*&OWNCHAR=CG&NAICS2022=00&QDESC=O02&key=YOUR_KEY_GOES_HERE
To pull owner age for wholesale trade in every state, change the question and industry codes and select all age bands:
https://api.census.gov/data/2023/abscbo?get=NAME,NAICS2022_LABEL,OWNCHAR,OWNCHAR_LABEL,OWNPDEMP,OWNPDEMP_PCT,OWNPDEMP_PCT_S&for=state:*&NAICS2022=42&QDESC=O09&OWNER_SEX=001&OWNER_ETH=001&OWNER_RACE=00&OWNER_VET=001&key=YOUR_KEY_GOES_HERE
The second call follows the documented pattern. We did not execute it, because doing so requires a key; the same rows were read from the data.census.gov table service, which serves the published table without one. The owner-demographic filters set to 001/00 return the "all owners" rows, and the age bands come back as OWNCHAR codes (ED = 55 to 64, EE = 65 or over in the 2023 table, as returned by that service on 2026-10-28). Check the codes against the vintage you pull, because labels and codes have changed between vintages.
Suppression, noise and sampling
- Noise infusion. This is the ABS's primary disclosure-avoidance method. A random multiplier is applied before tabulation, which causes "the vast majority of cell values to be perturbed by, at most, a few percentage points" [3].
- Suppression. The methodology states that "estimates with high errors of 50 percent or more are suppressed" (same source). Suppressed cells carry a flag such as
S. In the Ohio manufacturing slice below, the under-25 band is suppressed. - Sampling and imputation. About 14.4% of the firm-count estimates were imputed. The owner-characteristic tables have "no weighting adjustment to account for nonresponse" (same source). Every cell is an estimate with a published error. In the owner table,
OWNPDEMP_Sis the relative standard error of the count, andOWNPDEMP_PCT_Sis the standard error of the percentage [5].
Errors grow quickly as cells get smaller. The national wholesale 65+ share has a standard error of 0.5 points. In a single state the same cell can carry a standard error of several points, and at county level a cell may not be usable at all.
Building an ownership view by NAICS and state
Worked example: Ohio wholesale and manufacturing
The example is illustrative: a sponsor tests a succession-driven thesis in Ohio wholesale distribution and wants to know whether owner age in the state differs from the national figure. The figures are real, from AB2300CSCBO, reference year 2023, all owners.
| Slice | Owners reporting (RSE) | 55 to 64 (SE) | 65 or over (SE) |
|---|---|---|---|
| Ohio, all sectors | 129,108 (1.5%) | 29.2% (0.8) | 26.3% (1.1) |
| Ohio, wholesale trade (42) | 8,649 (4.8%) | 26.9% (2.6) | 40.0% (3.7) |
| Ohio, manufacturing (31-33) | 9,547 (6.6%) | 37.7% (4.3) | 29.7% (4.0) |
| U.S., wholesale trade (42) | 223,182 | 30.7% (0.4) | 32.7% (0.5) |
How to read it:
- The Ohio wholesale 65+ share (40.0%) is above the national wholesale share (32.7%). The difference of about 7 points is roughly twice the Ohio standard error. That is suggestive, not decisive, and a formal comparison would also need the national error.
- The survey counts about 8,600 owners in Ohio wholesale. That is the population behind the percentage. It is not a count of firms, and it is not a target list.
- Nothing in the table identifies a company: it says where to point outreach, not whom to contact.
Add business age from the same survey
Firm age is a different variable from owner age, and the ABS has it as years in business (YIBSZFI) in the company summary. In the 2023 table, 1,921,775 of the roughly 5.93 million employer firms across the six years-in-business classes had been in business 16 years or more. That is about 32% of firms, but they accounted for about 75% of the employment in those classes. In wholesale trade, firms 16 years or older were about 48% of firms (Census Bureau, table AB2300CSA02; shares are our arithmetic on the published counts).
For a longer time series, the Business Dynamics Statistics (BDS) gives annual measures of firm startups, shutdowns and job flows by firm characteristics. Its tables run from 1978 to 2023 [12]. BDS defines firm age by "the age of the oldest establishment in the firm" [13]. A firm that is old by that measure may still have a young owner who bought it recently. The owner table's O02 (year the owner acquired the business) is the closer measure of how long the current owner has held it.
For firm counts by size class to use as a denominator, the comparison of CBP, QCEW and SUSB explains which source answers which sizing question.
- 01Record the vintage
Record the table ID and reference year before quoting any figure.
- 02Filter geography and industry
Open the owner table, filter geography and NAICS, and download the selected rows.
- 03Verify question and category codes
Check age-band codes against the selected vintage because labels and codes have changed.
- 04Retain uncertainty and flags
Store count and percentage error fields beside every value, and list suppressed and other flagged cells.
- 05Preserve the retrieval record
Record the full query URL with the key removed, retrieval date and file hash.
- 06Qualify owners company by company
Use the aggregate view to choose and size outreach, then gather company-level evidence to qualify owners.
Getting owner age and succession signals at company level
The ABS tells you the age mix of owners in a sector and geography. It cannot rank targets, because microdata are confidential and every published cell is an aggregate. Company-level succession evidence has to come from other places:
- the company's own statements (founding date, leadership pages, announced transitions)
- state registry filings that name officers and record changes
- ownership-structure records such as employee benefit plan filings
- direct conversation with the owner
The succession and ownership guide sets out how to gather that evidence and grade it. Use the ABS to choose and size the hunting ground, and company-level research to find and qualify owners within it.
There are also things the ABS does not measure. It has no question on owner intent to sell, succession plans or retirement timing. "Older owner" stands in for "likely seller" only as a hypothesis, which outreach then tests.
Worksheet: ABS pull checklist
Copy this into the research log for each pull (illustrative template):
| Field | Entry |
|---|---|
| Product | ABS employer (abscbo / abscs) or NES-D (absnesdo) |
| Table ID and reference year | e.g. AB2300CSCBO, reference year 2023 |
| Survey name as published | e.g. 2024 ABS (fielded) = reference year 2023 |
| NAICS vintage | 2022 NAICS for the 2022 and 2023 tables; 2017 NAICS in earlier vintages |
| Geography | US / state / metro / county; note which levels the vintage publishes |
| Question and category codes | e.g. QDESC=O09; OWNCHAR codes checked against this vintage's labels |
| Owner-demographic filters | "All owners" codes recorded (001/00), not assumed |
| Unit | Owners (not firms) for abscbo; firms for abscs |
| Error fields kept | OWNPDEMP_S, OWNPDEMP_PCT_S stored next to every value |
| Flags | Suppressed (S) and other flagged cells listed, not dropped silently |
| Derived figures | Each sum or ratio marked as analyst arithmetic, with no published error |
| Retrieval record | Full query URL with the key removed, retrieval date, file hash |
The last row makes the pull reproducible: a reviewer can rerun the request and confirm the figure, as the Docket evidence standards require for any cited number.
Before a memo goes out, also check that 2017-NAICS tables (2018 and earlier) are not compared with 2022-NAICS tables at detailed industry levels without checking the NAICS changes.
Docket's agents can take a mandate's criteria, including a succession hypothesis, and turn them into questions answered company by company, each answer quoting a retained page.
Sources
- Annual Business Survey (ABS) — U.S. Census Bureau (page revised 2026), https://www.census.gov/programs-surveys/abs.html, accessed 2026-10-28.
- Annual Business Survey (ABS) APIs, developer data set page — U.S. Census Bureau (2026), https://www.census.gov/data/developers/data-sets/abs.html, accessed 2026-10-28.
- Annual Business Survey Methodology (2024 ABS, reference year 2023) — U.S. Census Bureau, https://www.census.gov/programs-surveys/abs/technical-documentation/methodology.html, accessed 2026-10-28.
- Census Data API: 2023 ABS Characteristics of Business Owners, variables — U.S. Census Bureau, https://api.census.gov/data/2023/abscbo/variables.html, accessed 2026-10-28.
- Census Data API: 2023 ABS Characteristics of Business Owners, example calls — U.S. Census Bureau, https://api.census.gov/data/2023/abscbo/examples.html, accessed 2026-10-28.
- Table AB2300CSCBO, Owner Characteristics of Respondent Employer Firms by Sector, Sex, Ethnicity, Race, and Veteran Status: 2023 — U.S. Census Bureau, https://data.census.gov/table/ABSCBO2023.AB2300CSCBO, accessed 2026-10-28.
- Table AB2200CSCBO, Owner Characteristics of Respondent Employer Firms: 2022 — U.S. Census Bureau, https://data.census.gov/table/ABSCBO2022.AB2200CSCBO, accessed 2026-10-28.
- Business Owners' Ages (graphic, 2019 ABS, data year 2018) — U.S. Census Bureau (2020), https://www.census.gov/content/dam/Census/library/visualizations/2020/comm/business-owners-ages.pdf, accessed 2026-10-28.
- Nonemployer Statistics by Demographics (NES-D) — U.S. Census Bureau (page revised 2026), https://www.census.gov/programs-surveys/abs/data/nesd.html, accessed 2026-10-28.
- Table AB2300NESD04, Owner Characteristics of Nonemployer Firms: 2023 — U.S. Census Bureau, https://data.census.gov/table/ABSNESDO2023.AB2300NESD04, accessed 2026-10-28.
- Census Data API dataset catalog — U.S. Census Bureau, https://api.census.gov/data.json, accessed 2026-10-28.
- Census Data API: 2023 ABS Company Summary, variables — U.S. Census Bureau, https://api.census.gov/data/2023/abscs/variables.html, accessed 2026-10-28.
- Table AB2300CSA02, Years in Business Statistics for Employer Firms: 2023 — U.S. Census Bureau, https://data.census.gov/table/ABSCS2023.AB2300CSA02, accessed 2026-10-28.
- Business Dynamics Statistics (BDS) — U.S. Census Bureau (page revised 2025), https://www.census.gov/programs-surveys/bds.html, accessed 2026-10-28.
- BDS Methodology — U.S. Census Bureau, https://www.census.gov/programs-surveys/bds/documentation/methodology.html, accessed 2026-10-28.
External Sources (13)
Published by
Docket Research
Docket provides deal-origination research software and managed research for private equity teams, screening companies against a mandate and showing the evidence behind each finding.
Editorial note
Researched and written by Docket Research with AI-assisted research tools; sources are cited inline and checked before publication. Report an error: https://docket.capital/evidence-standards#corrections.