• Nonemployer Universe
• Industry Classification of Businesses
• Geographic Classification
• Protecting Confidentiality
• Reliability of Data
• Special Note for 2019 and 2020 Reference Years
• Historical Comparability

For Nonemployer Statistics (NES), the U.S. Census Bureau annually creates the universe of nonemployer businesses in conjunction with identifying the employer business universe. For data on businesses with paid employees, refer to the County Business Patterns (CBP) website. When the U.S. Census Bureau receives information through administrative records that a business has no paid employees, the business becomes part of the potential nonemployer universe. The source of this data consists primarily of the annual or quarterly business income tax returns filed with the Internal Revenue Service (IRS) and maintained in the U.S. Census Bureau's Business Register. The U.S. Census Bureau does not collect survey data for the identification of businesses in the nonemployer universe.

The U.S. Census Bureau reviews the potential nonemployer universe to detect and remove businesses incorrectly identified as nonemployers. Business establishments associated with large, multi-unit companies not properly linked to the parent company account for the largest group of cases removed (for a description of how multi-unit companies are handled on the Business Register, refer to the County Business Patterns Methodology). See subsequent sections for other various exclusions from the nonemployer universe.

The legal form of organization (LFO) for nonemployer business establishments is derived from administrative record sources. NES publishes U.S.-level data (starting with 2008) and state-level data (starting with 2010) by the following LFOs:

• Sole Proprietorships
• Partnerships
• Corporations
  
• S-Corporations

Note: Starting in reference year 2014, NES began breaking out S-Corporations from Corporation estimates.

Since nonemployer data are, by definition, subject to federal income tax, other forms of LFO  such as government and non-profit are not included in the NES tabulations.

Back to Top

The U.S. Census Bureau excludes businesses with receipts above a predetermined cutoff from nonemployer tabulations. The logic for limiting receipts to a maximum amount assumes that receipts above the cutoff constitute part of the employer universe and should not get duplicated in nonemployer tabulations. Maximum receipts cutoffs vary by industry classification. For example, some nonemployers, such as those engaged in investment or entertainment, may have higher cutoffs than other industries. Conversely, it is unlikely that a nonemployer restaurant would have a similarly large amount in receipts without also having paid employees.

In addition, the nonemployer universe excludes businesses with less than $1,000 in receipts for all industries except construction. The small receipts of these non-construction businesses indicate that they may represent hobbies as opposed to normal business activities.

For reference year 2022, the U.S. Census Bureau carefully researched and implemented new receipts cutoffs to make the nonemployer universe more precise. Therefore, when comparing estimates for 2022 and subsequent reference years to estimates for prior years, the changes in receipts cutoffs must be taken into consideration. The maximum receipts cutoffs were previously updated with the release of the 2009 NES. See the Historical Comparability section for more information.

Back to Top

Historically, the permanent on-site workforce at a business location consisted of paid employees working at that location. While this traditional practice of companies directly hiring employees remains the dominant employer/employee relationship in the U.S., other work force arrangements exist. Companies commonly known as employee leasing companies or Professional Employer Organizations (PEO) operate in a co-employment relationship with client businesses. Employee leasing establishments typically acquire and lease back some or all employees of their clients and serve as the employer of record of the leased employees for payroll, benefits, and related purposes.

In cases where all employees are leased, the payroll for the established business is zero, placing it in the potential nonemployer universe. In some cases, these employers get tabulated as nonemployers due to this lack of payroll. The U.S. Census Bureau currently does not have a reliable method to identify the universe of companies that lease the entirety of their employees. The U.S. Census Bureau uses the Receipts Cutoffs methodology mentioned above in an attempt to remove these likely employers.

Back to Top

Since its inception in 1997, Nonemployer Statistics have been tabulated based on the North American Industry Classification System (NAICS). To see a detailed description of the relationship between industry classification systems, visit the Concordances section of the NAICS website.

Industry classification of nonemployer businesses currently uses the 2022 North American Industry Classification System (NAICS). The NAICS vintages for corresponding nonemployer reference years are provided below:

• 2022 through 2026 nonemployer data uses 2022 NAICS vintage
• 2017 through 2021 nonemployer data uses 2017 NAICS vintage
• 2012 through 2016 nonemployer data used 2012 NAICS vintage
• 2007 through 2011 nonemployer data used 2007 NAICS vintage
• 2002 through 2006 nonemployer data used 2002 NAICS vintage
• 1997 through 2001 nonemployer data used 1997 NAICS vintage

In general, Nonemployer Statistics adopt the classification system employed in the most recent Economic Census.

Nonemployer tabulations are typically provided at broader levels of industry detail than data for employers.  For the 2022 NAICS vintage, there are approximately 450 industry codes available through administrative record sources and common to the four Legal Forms of Organization (Sole Proprietorships, Partnerships, Corporations, and S-Corporations). While these industry codes are primarily self-reported by tax filers through IRS forms, other sources for assigning industry classifications include the Social Security Administration (SSA) and the Bureau of Labor Statistics (BLS).

The following table lists the NAICS sectors included in Nonemployer Statistics:

While Nonemployer Statistics cover most NAICS industries within the above listed sectors, the following industries are excluded from nonemployer tabulations and publication:

• Crop Production (NAICS 111)
• Animal Production and Aquaculture (NAICS 112)
• Postal Service (NAICS 491)
• Funds, Trusts, and Other Financial Vehicles (NAICS 525)
• Management of Companies and Enterprises (NAICS 55)
• Private Households (NAICS 814)
• Public Administration (NAICS 92)

Back to Top

The U.S. Census Bureau directly reclassifies businesses in certain industries when they appear in the nonemployer universe because these industries are assumed to require employees. These include

• New Car Dealers (NAICS 44111) -- reclassified to Used Car Dealers (NAICS 44112)
• Rail Transportation (NAICS 482) -- reclassified to Support Activities for Transportation (NAICS 488)
• Hospitals (NAICS 622) -- reclassified to Other Ambulatory Health Care Services (NAICS 6219)
• Oil and Gas Extraction (NAICS 2111) only in areas without oil or gas production -- reclassified to Other Financial Investment Activities (NAICS 5239)

In instances where IRS, SSA, and BLS do not provide a valid industry code, the U.S. Census Bureau uses an automated coding system based on Form 1040 Schedule C write-in descriptions of the Principal Business Activity (PBA). Prior to reference year 2007, this description coding process only took place in Economic Census years (years ending in 2 and 7), but it has occurred annually since reference year 2007.

Back to Top

A small percentage of nonemployers are unclassified. These businesses receive an imputed industry classification from a classified business within the same county and LFO that has a similar value of receipts. This imputation procedure preserves the distribution of businesses and receipts by industry while also eliminating an unclassified component varying in size each year. 

To control the contribution of imputed classifications, if nonemployer businesses with imputed classifications account for more than 40 percent of either receipts or number of businesses in a published row, then both items are suppressed. At the national level, less than 5 percent of nonemployer businesses have an imputed classification accounting for 1-2 percent of receipts.

Back to Top

Nonemployer Statistics show data for the U.S. by National, State, County (or county equivalent), Metropolitan/Micropolitan Statistical Area (MSA), and Combined Statistical Area (CSA) levels.

Most geography codes come from the business mailing address identified from administrative records. Because this address may not match the physical location of the business, the resulting geography codes do not always indicate where business operations occur. However, this represents the best information available regarding business location.

Back to Top

Nonemployer Statistics publish data for Louisiana (parishes), Alaska (organized boroughs, city and boroughs, municipalities, and census areas), and Connecticut (planning regions) as the equivalent of a county. The independent cities in Virginia, and the cities of Baltimore, MD; Carson City, NV; and St. Louis, MO are treated as separate counties. Nonemployer tabulations exclude records with invalid, foreign, or military geographic locations as well as Puerto Rico and the Island Areas (American Samoa, Guam, The Commonwealth of the Northern Mariana Islands, and the U.S. Virgin Islands).

Unlike the Decennial Census, Nonemployer Statistics are not available by sub-place geographies such as blocks, block groups, or tracts. Nonemployer Statistics also excludes many of the sub-county and sub-state geographies (such as school, congressional, voting, and state legislative districts) published by the Decennial Census.

Back to Top

While County boundary changes may occur between each publication year, these changes are only implemented once every 5 years. Such changes are detailed in the reports on the following page: Substantial Changes to Counties and County Equivalent Entities: 1970-Present . This is primarily done to maintain data consistency and comparability over time.

MSAs are redefined after each Decennial Census and may be revised based on population estimates. Additional details about MSA changes can be found on the following page: Metropolitan and Micropolitan Guidance for Data Users. New areas will qualify and the boundaries and/or titles of many existing area will change. Like counties, these changes are implemented in batch once every 5 years. For questions regarding specific years, please see the MSA Geography Reference page.

Back to Top

Nonemployer Statistics follow U.S. Census Bureau and IRS disclosure guidelines, as the majority of the data originates with the IRS. In Nonemployer Statistics tables from reference year 2004 or earlier, cell suppression was used to protect data confidentiality for cells that had a value based largely on contributions from a small number of businesses. These cell values were replaced with a “D” indicating they were withheld to protect the confidentiality of the businesses contributing to the cell value.

Starting in reference year 2005, the Noise Infusion disclosure avoidance method was incorporated to protect receipt totals. Noise Infusion is a method of disclosure avoidance in which values are perturbed prior to table creation by applying a random noise multiplier to the magnitude data (in this case, receipts) independently, for each business. This method of disclosure protection results in a relatively small change for the majority of published receipts totals. For reference years 2005 through 2016, some noisy cell values were also suppressed with a “D.” However, after further review by the U.S. Census Bureau’s Disclosure Review Board (DRB) it was determined that the “D” suppression symbol was reserved for the cell suppression method of disclosure avoidance, and its use was not recommended for estimates released with noise infusion. The DRB further recommended that suppression of these additional noisy cell values was unnecessary. As a result, during reference years 2017 and 2018, the use of the “D” suppression symbol was phased out.

Starting in reference year 2007, each published cell value has an associated noise flag indicating the relative amount of distortion in the cell value resulting from the perturbation of the data. The flag for ‘low noise’ (G) indicates a change to a cell value by less than 2 percent with the application of noise. The flag for ‘moderate noise’ (H) indicates a change to a cell value by 2 percent or more but less than 5 percent. The flag for ‘high noise’ (J) indicates a change to a cell value was 5 percent or more.

For further explanation of the noise infusion method, see Using Noise for Disclosure Limitation of Establishment Tabular Data by Timothy Evans, Laura Zayatz, John Slanta in the Journal of Official Statistics (1998).

Starting in reference year 2018, the number of establishments and receipts in a data cell must contain three or more businesses to be published. If this condition does not hold, the U.S. Census Bureau removes the cell from publication. Cells may also be suppressed because the quality of the data does not meet publication standards. In accordance with U.S. Code, Title 13, Section 9, and Title 26 no data are published that would disclose the operations of an individual business.

The Census Bureau has reviewed this data product to ensure appropriate access, use, and disclosure avoidance protection of the confidential source data (Project No. 7503950, Disclosure Review Board (DRB) approval number: CBDRB‑FY25‑0157).

Back to Top

Nonemployer tabulations are not based on a sample; therefore, they are not subject to sampling error. However, the nonemployer tabulations are subject to nonsampling error. The nonemployer data undergo complex processing, editing, and analytical review prior to release. Despite extensive review, nonsampling errors may occur for the following reasons:

• Inability to identify all cases that should be in the nonemployer universe
• Industry misclassification and definition changes
• Errors in recording or coding the data obtained
• Other errors of coverage, processing, and estimation for missing or misreported data.

The improper inclusion of possible employer establishments in the nonemployer universe serves as the primary source of nonsampling error. The U.S. Census Bureau takes steps to identify and remove these establishments from the nonemployer universe, but because of the size of the universe, limited resources, and limited data on which to make a determination, not all employer establishments get removed.

The assignment or imputation of an industry classification to nonemployer businesses with missing classification provides another source of nonsampling error. Imputation procedures are designed to preserve the overall distribution of totals by industry at the county and higher geographic levels, though this may not be possible for counties or industries with a small number of businesses.

Additionally, a small amount of undercoverage in the universe occurs because of processing deadlines. In a typical year, if the U.S. Census Bureau receives information on a potential nonemployer business more than one year after the end of the reference year, it is excluded from the nonemployer tabulations. The percentage of nonemployer Corporation and Partnership businesses not included in the published totals because the U.S. Census Bureau did not receive the tax return information in time to include them is small, with an even smaller percentage of Sole Proprietorship businesses not included in the published totals.

A lag in filing with the IRS may also cause exclusions from the reference year data. For instance, the IRS Tax Reporting Relief program allowed taxpayers impacted by Hurricane Katrina additional time in filing tax returns, resulting in lower totals for 2004 and 2005 Nonemployer Statistics in hurricane impacted counties/regions. Another example of this involved delayed processing of 2019 tax returns due to the COVID-19 pandemic. For additional information, see the Special Note for 2019 Reference Year.

The accuracy of tabulated data is determined by the joint effects of all nonsampling errors. No direct measurement of these effects has been obtained except for the estimation of missing or misreported industry classifications. However, the U.S. Census Bureau takes precautionary steps in all phases of the nonemployer processing to minimize the effects of nonsampling errors. Analysts use the "S" symbol to suppress estimates affected by large nonsampling errors which should not be published. Though some of the suppressed values may be obtained by subtraction, the results are not official and should be used with caution as they may differ substantially from the true unknown population values.

Back to Top

The 2019 and 2020 Nonemployer Statistics were released on June 20, 2022 and August 17, 2023, respectively. For both years, the percentage of undercoverage may be slightly higher than a typical year because the tax return filing and processing period took place during a global pandemic. To provide tax filing and payment relief to individuals and businesses in response to the COVID-19 outbreak, the IRS adjusted the tax filing deadlines for each year for all Federal income tax returns from:

• April 15, 2020 to July 15, 2020, and,
• April 15, 2021 to May 17, 2021

The change in deadlines delayed the processing of 2020 and 2021 tax returns. Additionally, many IRS offices were also closed for part of 2020. Once the IRS processing offices re-opened, staffing levels were below normal and processing of returns required additional time.

For both reference years 2019 and 2020, Census Bureau delayed processing of Nonemployer Statistics to mitigate potential undercoverage. The late processed IRS returns were eventually received by the Census Bureau and these additional records were included in the 2019 and 2020 Nonemployer Statistics tabulations if they met all other criteria to be considered in-scope.

Back to Top

Changes in industry classifications, geographic definitions, and methodology may impact the comparability of data over time, as the Census Bureau does not restate previously released NES estimates to reflect changes in the industry classifications, geographic definitions, or methodology used to create the NES estimates. For this reason, it is not recommended to analyze multiple years of NES data as a time series. Significant changes are listed below:

• Years Ending in 2 and 7: Changes in NAICS industries. See Industry Classification of Business section for more information.
• Every 5 years: County and MSA geography changes. See Geographic Comparability section for more information.
• For reference year 2009: The U.S. Census Bureau implemented new receipts cutoffs to make the nonemployer universe more precise, specifically improving the accuracy of the Sole Proprietorships included in the nonemployer tabulations. Additional improvements included filtering out likely employers from the Corporation and Partnership data. Therefore, when comparing 2009 and subsequent reference years data to prior years, these changes in methodology must be taken into consideration. A Revised 2008 Nonemployer Statistics Table is available to gauge the effects of these changes.
• For reference year 2016: The U.S. Census Bureau began using business descriptions to reclassify Sole Proprietors in specific industries after identifying significant self-misclassification by tax filers. This methodology corrected Urban Transit (NAICS 4851) records, reclassifying the majority to Taxi and Limousine Service (NAICS 4853) records.
• In reference year 2017: The U.S. Census Bureau applied the same industry reclassification methodology from 2016 to the Utilities sector (NAICS 22). As a result, the Utilities sector data for reference year 2017 are not comparable to prior years.
• In reference year 2018: The same industry reclassification methodology from 2016 and 2017 was applied to the Other Services (except Public Administration) sector (NAICS 81) which impacted all industries. All reference year 2018 data are not comparable to prior years Refer to the 2018 NES: Change in Nonemployer Establishment Counts Due to Sector 81 Recode and 2018 NES: Change in Nonemployer Receipts Due to Sector 81 Recode graphs to see the number of establishments and receipts data that moved by NAICS sector due to the recoding operation.
• For reference year 2022: The U.S. Census Bureau carefully researched and implemented new receipts cutoffs to make the nonemployer universe more precise. Therefore, when comparing estimates for 2022 and subsequent reference years to estimates for prior years, the changes in receipts cutoffs must be taken into consideration.

Definitions available on the NES Glossary page.

Back to Top