Two of the most widely cited sources on self-employment in the United States — the Current Population Survey (CPS) and the American Community Survey (ACS) — both capture data on workers who produce goods independently. But their approaches differ in ways that directly affect how craft-based self-employment is counted, and the resulting numbers do not always agree.
The CPS approach
The CPS, conducted monthly by the Bureau of Labor Statistics and the Census Bureau, classifies workers as self-employed based on their response to a question about "class of worker" in their main job during the survey reference week. As economist Steven Hipple documented in a 2010 BLS Monthly Labor Review article, the CPS distinguishes between incorporated and unincorporated self-employment — a distinction that matters because the two categories are reported separately in official labor-force statistics. An unincorporated craft seller at a weekend market and a sole proprietor with a registered LLC fall into different bins, even if the economic activity is similar.
The CPS reference period is a single week. If a respondent made pottery on Thursday and sold it on Saturday but also worked a wage job on Monday through Wednesday, the CPS classification depends on which job the respondent identifies as primary. A person who considers their craft work secondary — even if it produces meaningful income — would not appear in the self-employment count for that month. Hipple noted that this single-week window tends to produce lower self-employment counts than surveys using longer reference periods.
The ACS approach
The ACS, also administered by the Census Bureau, uses a 12-month reference period. Respondents report their primary activity over the past year, which means seasonal craft producers — those who sell primarily during holiday markets or summer fairs, for instance — are more likely to be captured. However, the ACS shares the CPS limitation of asking about the respondent's primary job. Someone who earns most of their annual income from wage employment but supplements it with craft sales would typically not be classified as self-employed in either survey.
Labor economists Shane Mund and Brian Chansky, in research presented through the BLS, have noted that these surveys were designed in an era when self-employment was more clearly separable from wage work. The rise of hybrid arrangements — part-time platform sales alongside part-time wage employment — creates classification challenges that the original survey instruments did not anticipate.
The Nonemployer Statistics alternative
A third data source, the Census Bureau's Nonemployer Statistics (NES), counts businesses with no paid employees that file at least $1,000 in annual receipts with the IRS. Because the NES is derived from administrative tax records rather than household surveys, it avoids the reference-period problem entirely. A craft business that earned $3,000 over six months of weekend sales would appear in the NES regardless of whether the owner considered it a primary or secondary activity.
However, the NES has its own limitations. As researchers at the Census Bureau have acknowledged, the dataset does not capture businesses operating below the $1,000 threshold or those operating informally without filing. For craft producers who sell small volumes at local markets or through social media without formal business registration, the NES is blind.
What the discrepancies mean
The practical consequence is that no single federal data source provides a complete count of craft-based self-employment. The CPS undercounts because of its short reference window and primary-job filter. The ACS captures more seasonal activity but still misses secondary earners. The NES captures tax-filing businesses regardless of hours worked but misses informal producers entirely.
Researchers who cite self-employment statistics without specifying the source and its methodology risk comparing numbers that were never designed to measure the same thing. The differences are not errors — they are artifacts of instruments built for different purposes, applied to a labor market that has changed since their designs were finalized.