The off-payroll headline is real. It is also a count of one channel — and HMRC tells you, in its own report, which channels it left out.

My first piece in this series argued that the 2021 off-payroll reform was an own goal: it scattered a skilled onshore workforce, handed the work to consultancies, and left the Treasury worse off. The obvious objection landed almost immediately, and it is a fair one. In February 2025, HMRC published an updated assessment estimating the reform generated around £4.2 billion in additional tax, National Insurance and Apprenticeship Levy up to March 2023. If the policy raised £4.2 billion, how can it be a fiscal failure?

The answer is not to dispute the figure. It is to read the report closely enough to see what the figure counts — and, more revealingly, what HMRC openly states it does not. The £4.2 billion is not a lie. It is a measurement of the one channel HMRC chose to look at, accompanied by explicit admissions that the two channels where the work actually went were left out of the count.

Generated is not collected

Start with the word HMRC chose. The reform generated £4.2 billion — and the report is explicit that this is an estimate on a National Accounts basis, which it defines as aligning to the period when the activity giving rise to the tax takes place, not when the tax is collected. This is a modelled figure of tax that should arise from changed behaviour, reconciled against a counterfactual, not a reconciled receipt. That distinction is HMRC’s own, stated in its methodology — not a sceptic’s gloss.

The timing compounds the point. The rules took legal effect in April 2021. Yet the £4.2 billion is claimed from October 2019 onward, on the basis that behaviour changed ahead of the law. Strip the figure back to its own annual breakdown and the shape is telling:

Source: HMRC, Update to the impacts of the 2021 off-payroll working rules reform, Table 3 (February 2025).

Two things fall out of HMRC’s own table. First, roughly £0.75 billion of the headline — the £50 million plus the £0.7 billion — is attributed to periods before the rules even took legal effect. Second, and more important, the revenue peaks and then falls: £1.9 billion in the first full post-reform year, £1.6 billion in the second. A genuine, durable closing of a structural tax gap should hold or compound. A figure that spikes once and then declines is the signature of a one-off reclassification event — the immobile population moved onto payroll, the uplift was banked once, and the following year already shows the gain eroding.

The channel HMRC counted — and the two it didn’t

An aggregate tax estimate can only measure activity that stayed inside the base HMRC chose to model. And HMRC was admirably candid about that base: it is built almost entirely on workers who left a personal service company payroll and moved onto another organisation’s payroll — overwhelmingly as employees, agency workers, or umbrella-company staff. That is the channel where reclassification produces visible extra PAYE. It is also the channel for the worker who could not move.

The work that could move had two exits. HMRC names both in its own report — and excludes both from the count.

Exit one: the partnership channel, excluded by HMRC’s own admission.

In a methodology note headed “Partnerships and other intermediaries,” HMRC states plainly that the off-payroll rules apply to intermediaries beyond personal service companies, but that it did not include partnerships in the analysis, on the expectation that the impact on those populations would be small. That is not a minor footnote. The consultancy channel — the Big Four and the major actuarial and advisory firms — is precisely a partnership structure. HMRC’s scorecard cannot see the channel that absorbed the displaced work, because it assumed that channel away at the outset.

And the assumption is not idle. IR35 was never a tax on a business form — a partnership can be a caught intermediary just as a company can. It is a tax on a relationship: the individual who performs services personally for a client through an intermediary they control, and who would be that client’s employee if you stripped the intermediary away. Apply that counterfactual test to a consultancy engagement and it finds nothing. Remove the firm and the consultant is not the client’s employee — they are the firm’s employee, already on PAYE, or the firm’s partner, already taxed on profit share. The client is a customer of a business, not a temporary employer; the firm can substitute who does the work, so the personal-service hinge of the whole regime is absent. IR35 has, quite literally, nothing to reach. The consultancy door was never a door the legislation could close — and HMRC’s analysis then declined to look through it.

Exit two: offshore delivery, examined and then not measured.

The rules do not apply where the end client is wholly overseas. The report acknowledges the overseas dimension directly — it lists workers with an overseas home address among the exit routes it examined, and its own VAT methodology spells out the asymmetry for financial services. Because financial and insurance services are VAT-exempt, a firm buying services from a contractor pays unrecoverable VAT; once that worker moves onto an organisation’s payroll, the VAT is no longer chargeable, and HMRC dutifully counts the lost VAT — but only for onshore moves. There is no equivalent line for work that left the country.

The fiscal mechanism for that offshore work is doubly adverse, and certain in direction. When a UK insurer’s modelling is delivered from Bangalore, Warsaw or Kuala Lumpur, the labour generates tax in that jurisdiction and none here — and the UK buyer deducts the cost against its own UK corporation tax. The Exchequer loses on both sides of the same transaction. No public dataset isolates the volume of insurance or actuarial modelling that has gone offshore since 2021, so the honest claim is not “HMRC lost £X” but “this is a structurally guaranteed leak of unmeasured size, and the body that published the inflow chose not to estimate the outflow.”

Why the headline looks like a win

Assemble the pieces and the £4.2 billion stops looking like a verdict and starts looking like an accounting artefact. HMRC modelled the channel that produces visible PAYE — the immobile worker who reclassified and stayed. It excluded the partnership channel by stated assumption. It examined the overseas channel and left it unquantified. The number is positive because it is, by construction, blind to the two exits the mobile work actually took.

HMRC’s own trajectory is consistent with exactly that reading. A one-off reclassification gain in 2021–22, already eroding to £1.6 billion the next year, is what you would observe if the durable effect at the high-skill, mobile margin were neutral or negative — the immobile banked their uplift once, while the mobile work quietly found the open doors. The report even concedes it cannot isolate the reform’s effect from the pandemic, and that those affected are around 1% of the workforce. These are not the hedges of a measure that has confidently closed a structural gap.

The second column

None of this is a call for repeal, and none of it requires accusing anyone of bad faith. It is a measurement objection, and a precise one. A policy cannot be declared a fiscal success on a number that, by its own author’s description, counts only the activity that stayed on a UK payroll, excludes the partnership structures that captured the displaced work, and leaves the offshore migration unmeasured.

The £4.2 billion is real. It is also one column of a two-column ledger. The missing column holds the labour tax forgone where work went wholly overseas, the corporation-tax deductions those offshore contracts generate against UK profit, and the changed tax profile of work routed through a partnership channel HMRC told us it did not analyse. Until someone fills that column in, “£4.2 billion” is not the answer to my first piece. It is a precise measurement of the part of the story that was easy to count — published by the only institution with the data to count the rest, and choosing not to.

That, in the end, is the quiet theme running through this whole series: the channels that are winning are, almost without exception, the channels the UK tax system is least equipped to see.

Sources: HMRC, “Update to the impacts of the 2021 off-payroll working rules reform in the private and voluntary sectors,” 27 February 2025 (Executive Summary; Findings; Tables 1–3; methodology notes on Partnerships and other intermediaries, and on VAT). Figures quoted are HMRC’s own estimates.