What the ONS Series Captures and What It Misses

The Survey Behind the Number

The Office for National Statistics measures self-employment through the Labour Force Survey, a rolling quarterly household survey covering tens of thousands of UK households. Within that count, the ONS distinguishes two types: employers — self-employed people who hire others — and own-account workers, those who operate entirely alone. By the mid-2020s, own-account workers made up the substantial majority of the self-employed total, a proportion that had grown steadily since the 2008 financial crisis.

The distinction matters because the two groups behave differently. Employers accumulate capital, carry payroll obligations, and are structurally closer to small businesses. Own-account workers — sole traders, freelancers, contractors working through personal service companies — represent something closer to the contested legal territory that IR35 and the UK's off-payroll rules were designed to address. Collapsing both into one headline figure obscures that difference entirely.

What Eurostat Counts Differently

Eurostat uses the same own-account worker category, drawn from the European Union Labour Force Survey, and the definition is broadly comparable. But the operational detail diverges. Eurostat includes contributing family workers in its self-employment aggregate; the ONS excludes them from its primary self-employment headline. Eurostat also applies a harmonised reference-week approach across member states, while the LFS uses a four-week averaging method for some sub-series — producing figures that are directionally similar but not directly equivalent.

The result is a familiar problem: Eurostat's own-account worker data and the ONS series describe overlapping populations with enough methodological friction to prevent clean comparison. A researcher tracking UK self-employment against, say, Germany or the Netherlands is comparing instruments as much as workers.

The Post-2020 Blind Spot

The ONS added remote-work breakdowns to its employment data after 2020, but these cross-cuts do not map cleanly onto self-employment status. A self-employed graphic designer working from home appears in the remote-work count and the self-employment count but never in a cell that holds both attributes reliably. That gap limits what any analyst can say with precision about where self-employed work happens — a limitation the ONS has acknowledged without yet resolving.

The count exists. Its edges are soft, and the softness is load-bearing.