Atholl House v HMRC: Winning At The Third Stage

HMRC's status decision—typically a Regulation 80 PAYE determination, a section 8 National Insurance decision or a closure notice—leans hard on two phrases: mutuality of obligation and control. You have read that they are the "irreducible minimum" of employment, the features without which no contract can be employment at all; you have been told you satisfy both; and you are wondering whether the case is already over.

It is not. Most contested status appeals are decided at a third stage: a weighing of everything else about the arrangement. The case that tells tribunals how to run that stage is HMRC v Atholl House Productions Ltd [2022] EWCA Civ 501—the Kaye Adams case.

Adams presented "The Kaye Adams Programme" on BBC Radio Scotland through her personal service company, Atholl House Productions Ltd. HMRC decided a direct contract with the BBC would have made her its employee, and pursued £124,441.58 in PAYE and National Insurance. Proving otherwise took four hearings and the better part of a decade. Mutuality and control were both resolved against her, and stayed that way. She won anyway.

This analysis is the final panel of a four-case arc: Ready Mixed Concrete gives the three conditions of employment, Tanton ends cases at the first, PGMOL settled the first two, and Atholl House governs the third—and the rule that decides which facts count there at all.

The Case In Brief

Atholl House signed annual written contracts with the BBC. HMRC's determinations covered 2013/14 to 2016/17, but it dropped the first two years to save resources, leaving only 2015/16 and 2016/17 in dispute.

Two Sets Of Three Stages, And Why It Matters

In Adams's years the IR35 rules were Chapter 8 of Part 2 of ITEPA 2003, mirrored for National Insurance by the Social Security Contributions (Intermediaries) Regulations 2000 (SI 2000/727): would you have been an employee if you had contracted directly with the client instead of through your company? Since 6 April 2021 most medium and large private-sector engagements fall under Chapter 10 instead—our IR35 and off-payroll hub maps which regime covers you—but the status question at the centre is the same.

Two nested frameworks answer it, and Atholl House is easier to follow once they are separated. The IR35 exercise has three stages: find the terms of the actual contracts and the surrounding circumstances; build from them a hypothetical contract directly between worker and client; and decide whether that hypothetical contract would be one of employment.

Inside that last stage sits the Ready Mixed Concrete test and its three conditions: personal service and mutuality of obligation (each side owes the other something—work provided, pay in return); a sufficient right of control (the client can direct what is done and how); and a final check that everything else is consistent with employment. The remitted First-tier Tribunal labelled these Stage 3A, Stage 3B and Stage 3C; this article uses those labels throughout.

By the time the Court of Appeal sat, 3A and 3B were settled against Adams: mutuality was common ground, the Upper Tribunal had found a sufficient framework of control, and the Court of Appeal left both findings standing ([163]). The appeal—and ultimately the entire case—was in substance about Stage 3C.

What The Court Of Appeal Settled

Sir David Richards gave the lead judgment; Arnold LJ concurred, emphasising four points, and Peter Jackson LJ agreed. Four holdings matter for your appeal.

There Is One Test, Not Two

HMRC argued that the Upper Tribunal had swapped Ready Mixed Concrete for a rival test: whether the worker was in business on her own account, the line running through Market Investigations and Hall v Lorimer. The court refused to choose:

"It would be intolerable if an individual's employment rights or tax position could depend on the choice of one out of two or more different, and potentially conflicting, tests." ([60])

On either formulation, mutuality and control are "necessary pre-conditions to a finding that a contract is one of employment. Once those necessary, but not necessarily sufficient, conditions are satisfied, both approaches require the identification and overall assessment of all the relevant factors present in the particular case" ([122]).

Passing Stages One And Two Does Not Put You Behind

An earlier Upper Tribunal gloss had treated mutuality plus control as creating a prima facie case of employment for the remaining factors to displace. The Court of Appeal rejected it:

"Nor do I think it appropriate to add a gloss of a "prima facie affirmative conclusion" of a contract of employment if the pre-conditions of mutuality and control are satisfied." ([113])

Stage 3C therefore starts level. Losing 3A and 3B does not put you behind; it means the weighing begins.

One caution before that reassurance settles: "level" describes the legal test, not the hearing. On the appeal itself the burden is on you to displace HMRC's determination—under section 50(6) TMA 1970 the figures stand good unless you show they are wrong—so the evidence work later in this article is your job, not HMRC's. Our guide to writing grounds of appeal covers the burden in detail.

Control Is Weighed Twice

HMRC also argued the opposite simplification: that mutuality and control, once established, drop out of the picture. The court rejected that too. Contracts display those features to very different degrees, and:

"I can think of no good reason why account should not be taken of these differences in what all agree is a multi-factorial process addressing all the relevant factors." ([76])

So a skeletal framework of control that scrapes past Stage 3B is worth less at Stage 3C than day-to-day command over how you work. The Supreme Court adopted exactly this point in PGMOL, citing paragraph 76.

Running Your Own Business Is A Factor, Not A Test

Then the holding appellants quote most:

"If the person providing the services is known to carry on a business, profession or vocation on their own account as a self-employed person, it would in my judgment be myopic to ignore it, when considering whether or not the parties intended to create a relationship of employment." ([124])

The very next words keep it in proportion: "The weight to be attached to it is a matter for the decision-making court or tribunal." The ceiling is equally clear: other self-employed work is a relevant fact "but it goes no further than that" ([128]). The same person can be an employee in one engagement and an independent contractor in another, in the same tax year.

The judgment also repeated Nolan LJ's warning from Hall v Lorimer that the in-business question is of limited use for a profession or vocation: a self-employed author, actor or singer "may earn his living without any of the normal trappings of a business" ([96]). Having no premises, staff or capital does not disqualify you.

The Filter: What The Engager Knew

Now the heart of the case. If Stage 3C weighs all the relevant factors, what makes a factor relevant? The court's answer is a filter, and it is the reason the Upper Tribunal's decision fell.

The question, judged objectively, is whether the parties intended to create a relationship of employment when they made their agreement. So the court drew the boundary from ordinary contract-law principles: the admissible circumstances are the "facts or circumstances which existed at the time that the contract was made, and which were known or reasonably available to the parties" ([123]). Arnold LJ made the same point in his concurrence: the hypothetical contract "should not be construed in a vacuum, but in the light of the admissible factual matrix" ([170]).

Two hard edges follow.

First, timing. Earlier dealings between the same parties are part of the picture; later ones are not. On Adams's own facts the court was blunt: her "activities in later years cannot be used to assess whether she was employed in earlier years" ([131]).

Second, knowledge. Your other engagements are capable of counting—and count only if the engager knew of them, or could reasonably have found out, when the contract was signed. Arnold LJ gave the example: the precise terms on which you worked for other parties cannot be weighed if they were never disclosed to the alleged employer ([170]). Real work the engager knew nothing about does not move the needle.

Measured against that filter, the Upper Tribunal had gone wrong in four ways—the standard failure modes of stage-three reasoning:

• It came close to treating Adams's wider freelance career as the whole answer, when it is one admissible factor among many.
• Beyond the time her commitment absorbed, there was "no consideration at all by the UT of the terms of the hypothetical contracts" ([130]). The enquiry is not confined to the contract's terms, but the terms remain central.
• It drew on her career across surrounding years, when the critical periods are the years under appeal ([131]).
• It misread HMRC's decision to drop 2013/14 and 2014/15 as a concession that she was self-employed in those years. HMRC had dropped them to save resources, and the misunderstanding was itself a further reason to set the decision aside ([132]-[136]).

An appellate court cannot re-weigh an evaluative decision merely because it disagrees with the answer; it interferes only for errors of approach like these—the Edwards v Bairstow discipline. That is why the list is worth keeping: it maps the mistakes that get stage-three decisions set aside.

You Cannot Ask A Tax Tribunal To Rewrite The Contract

The 2019 tribunal had reached Adams's answer by a different route. It used Autoclenz Ltd v Belcher [2011] UKSC 41—the employment-rights doctrine that lets a tribunal find the "true agreement" behind written terms that do not reflect reality—to read two restrictive clauses out of the contracts.

The Court of Appeal closed that route in tax ([156]-[160]). Uber BV v Aslam [2021] UKSC 5 had explained Autoclenz as purposive interpretation of protective employment statutes; section 49 ITEPA asks the common-law employee question and protects no one, so "it is not legitimate to apply the Autoclenz approach". The clauses would have survived even under Autoclenz anyway, because a right that is never exercised is still a term of the agreement ([159]).

The consequence for a tax appeal: the paper matters more. HMRC's attack narrows to whether the written terms are genuine; yours narrows to what the document actually says.

How Stage Three Was Actually Won

The Court of Appeal could not safely re-make the decision, so it remitted the case to the Upper Tribunal, which sent it on to the First-tier Tribunal, the tribunal of fact, before the original panel. That panel noted the "considerable uncertainty and debate" the Court of Appeal's decision and order had caused, held Stages 3A and 3B finally settled against Adams, and fought the case out at Stage 3C alone.

The scoresheet was close. Pointing to employment (the decision's paragraph 162): the sheer scale of the commitment, 160 programmes for at least £155,000; the length of her BBC relationship; and the mutuality and control carried forward from 3A and 3B.

What won it (paragraph 169):

• Industry custom and practice. The industry treated presenters engaged as Adams was as self-employed—a custom recorded in the Radio Industry Guidelines that HMRC itself published in 2008 and the BBC followed—and both parties knew it or could readily have known it.
• Her own brand. The BBC did not merely tolerate her outside career; it positively wanted to exploit and facilitate it, and did not curtail her other engagements.
• No employee treatment. No holiday pay, sick pay, maternity leave or pension; no appraisals; no internal-vacancy rights; her own equipment. Facts both sides knew because they lived them.
• The stated intention. There was "an unequivocal statement in each written agreement to the effect that the parties did not intend to create a relationship of employment"—"by no means determinative", but "capable of carrying weight in a case which is finely-balanced".

One honest negative: ordinary financial risk—a client defaulting, or too few engagements coming in—carried no weight. What distinguishes a business is "the ability to make a profit or loss from the engagements", for instance through sub-contracting.

Notice the filter doing the work throughout: factor after factor is tested against whether the matter was "known or reasonably available to each of the relevant individuals" when the agreements were signed. The BBC's knowledge of Adams's wider career was proved by what it had historically paid, her company's filed accounts, and the running dialogue about her other engagements.

The tribunal did not pretend the answer was obvious: "the position is finely-balanced and we have reached our conclusion with some diffidence". But:

"when one stands back and considers the various factors as a whole, as we are required to do - see Nolan LJ in Hall at 216 - the hypothetical contracts appear to us to be contracts for services and not contracts of service"

(A contract of service means employment; a contract for services means self-employment.)

What To Gather, And How To Prove They Knew

The remitted decision reads like an evidence list. No factor list produces an answer—this was a finely balanced case decided on its own facts—but these are the pairings that persuaded the tribunal, each a fact plus proof of the engager's knowledge of it, and the questions to ask about your own arrangement. Nothing here is special to broadcasting: the same exercise decides the appeals of IT contractors, locums, engineers and consultants.

• Other clients in the disputed years, plus whatever put the engager on notice: filed accounts, invoices, scheduling correspondence, the negotiation trail, the public profile of the other work.
• Industry custom, plus the documents that record it: trade-body guidance, published guidance the engager followed (here, HMRC's own 2008 Radio Industry Guidelines, applied by the BBC), how comparable workers were engaged.
• Absence of employee treatment, plus the absences themselves: no holiday or sick pay, pension, appraisals or internal-vacancy rights; no access to internal systems; your own equipment. Both sides know these facts by definition.
• The contract's statement of intention, if it has one. Weight only in a close case—but yours may be one.
• The whole picture, presented the way our Hall v Lorimer analysis describes: a portrait, not a checklist.

Equally useful is what not to lead with: the mere risk of a client not paying, which carried no weight; evidence about later years, which is inadmissible; and paper rights the evidence suggests were never genuine, because genuineness is the one attack on written terms that survives in tax.

The timing rule doubles as a filing instruction. This evidence belongs in your witness statement and hearing bundle at the First-tier Tribunal, the tribunal of fact—prepared before the hearing, not saved for an appeal that can only review errors of law.

The Same Court, The Same Day, The Opposite Answer

The same three judges gave judgment the same day in Kickabout Productions Ltd v HMRC [2022] EWCA Civ 502, the appeal of Talksport presenter Paul Hawksbee's company—and dismissed it. Hawksbee had presented his show for 18 years; his Talksport income was "approximately 90% of his total income"; and he "did not work as a radio presenter, in those tax years, for anyone other than Talksport". His hypothetical contract was employment.

Side by side, the contrast is dependence on a single paymaster. Adams had a portfolio and the BBC knew it; Hawksbee had Talksport. But there is no percentage threshold at which self-employment switches on. The BBC's fees were between 50% and 70% of Adams's own gross income, and she still won: both decisions rested on the overall picture, not on any one number.

Four Hearings, Ten Years, And What An FTT Win Is Worth

Press reports traced HMRC's enquiries back to 2014; the end came in January 2024. Adams reportedly put her legal fees at close to £300,000, more than double the tax at stake. Those are press-reported figures, not tribunal findings—and they are not the price of a First-tier Tribunal appeal, either. They bought four hearings, including two appellate courts. There is no fee for bringing an FTT appeal, most first-instance cases take 6-12 months, and around 45% of appellants represent themselves.

Cost still needs clear eyes, in three places. In the standard case categories each side bears its own costs win or lose, as our costs guide explains, but a dispute of this size may be allocated to the Complex category, where costs shifting applies unless you opt out within 28 days. Interest is the clock you cannot stop: it runs on the disputed tax from the original due dates until payment, so a long fight means a long accrual if you lose—see interest on unpaid tax. And if HMRC alleges a careless or deliberate inaccuracy, a penalty may sit on top, with its own mitigation rules.

After the remitted decision, HMRC had 56 days to seek permission for another appeal to the Upper Tribunal. It let the window close, telling reporters, as Scottish Financial News reported: "Given this litigation has been ongoing for a number of years and the FTT does not set binding legal precedents, we don't think it would be proportionate to appeal in this case."

That is a resourcing decision reported in the press, not a concession on the law. But the point inside it is real and cuts both ways: First-tier Tribunal decisions do not bind other tribunals. A win in your case is a win in your case, and an HMRC win against someone else does not decide yours.

The framework itself is not so fragile. In PGMOL the Supreme Court applied the Atholl House approach with both parties' agreement, recording that no one challenged the approach and analysis in Atholl House—a decision that had in material respects rejected submissions made on HMRC's behalf. The lead judgment in PGMOL was given by Lord Richards—the same judge who, as Sir David Richards, wrote Atholl House.

What This Means For Your Appeal

Losing on mutuality and control is not losing. Those are Stages 3A and 3B; contested status appeals are usually decided at Stage 3C, where the weighing starts level, with no presumption of employment to displace, and where the strength of the control HMRC established is weighed again.

Your business, brand and portfolio are admissible there and can carry real weight, but only through the filter: the engager must have known the facts, or been reasonably able to know them, when the contract was made. Later years are out. The written contract cannot be rewritten in a tax appeal, which makes its actual terms, and their genuineness, the battlefield. And no list of factors produces an answer: the tribunal that finally decided for Kaye Adams called the case finely balanced and said so with diffidence.

If a CEST result is being held against you, keep it in proportion: the tribunal applies Atholl House and PGMOL, not the calculator, and the contextual factors that decided Adams's case—a portfolio the engager knew about, industry custom, how she was actually treated—are exactly what a checkbox tool cannot weigh. Our IR35 hub covers CEST's status in detail.

One practical footnote. A status decision arrives as an appealable document—typically a Regulation 80 PAYE determination, a section 8 National Insurance decision or a closure notice—and you normally have 30 days from its date to appeal. Appealing does not by itself pause collection, so ask for the disputed tax to be postponed while the appeal runs. You can also request a free statutory review first, explore settlement or ADR at any stage, and if the deadline has already passed, a late appeal is sometimes still possible.

Under Chapter 10—applying to public bodies since 6 April 2017 and to medium and large private-sector clients since 6 April 2021—the decision often goes to the deemed employer or fee-payer rather than the worker, so check whose appeal right it actually is. The evidence in this article still has a home there: it goes into your representations under the section 61T ITEPA status disagreement process, which the client must answer within 45 days, and into the appeal of whoever holds the right. Our guide to appealing to the tax tribunal covers the steps.

Key Legislation And Resources

The Judgment

• HMRC v Atholl House Productions Ltd [2022] EWCA Civ 501 — Court of Appeal (Civil Division), 26 April 2022; lead judgment by Sir David Richards, Arnold LJ concurring with four emphasised points, Peter Jackson LJ agreeing
• Atholl House Productions Ltd v HMRC [2024] UKFTT 37 (TC) — the remitted First-tier Tribunal decision (judgment 29 November 2023) that finally decided Stage 3C in Adams's favour
• HMRC v Atholl House Productions Ltd [2021] UKUT 37 (TCC) — the Upper Tribunal decision the Court of Appeal set aside
• Atholl House Productions Ltd v HMRC [2019] UKFTT 242 (TC) — the original First-tier Tribunal decision of April 2019, the first of Adams's wins

Legislation

• Chapter 8, Part 2, ITEPA 2003 — the intermediaries (IR35) rules that applied to all of Adams's years
• section 49 ITEPA 2003 — the gateway: would the worker have been an employee under a direct contract?
• Social Security Contributions (Intermediaries) Regulations 2000 (SI 2000/727) — the National Insurance mirror of Chapter 8

Key Cases

• HMRC v Atholl House Productions Ltd [2022] EWCA Civ 501 — one test, no presumption, control weighed twice, in-business a factor not a test; admissible facts limited to those known or reasonably available to both parties at contract date
• Kickabout Productions Ltd v HMRC [2022] EWCA Civ 502 — same panel, same day, opposite outcome: 18 years and roughly 90% income dependence on one paymaster
• HMRC v Professional Game Match Officials Ltd [2024] UKSC 29 — the Supreme Court applies the Atholl House framework, unchallenged by either side
• Ready Mixed Concrete (South East) Ltd v Minister of Pensions and National Insurance [1968] 2 QB 497 — the three conditions that structure the employment question
• Hall v Lorimer [1993] EWCA Civ 25 — the whole-picture evaluation Atholl House integrates with Ready Mixed Concrete
• Express & Echo Publications Ltd v Tanton [1999] EWCA Civ 949 — the Stage 3A knockout: a genuine substitution right is inconsistent with employment
• Autoclenz Ltd v Belcher [2011] UKSC 41 — the employment-rights "true agreement" doctrine that Atholl House holds unavailable in tax
• Uber BV v Aslam [2021] UKSC 5 — recast Autoclenz as purposive statutory interpretation, the ground for the tax carve-out

HMRC Guidance

• ESM7335 — HMRC v Atholl House Productions Ltd — HMRC's own case page, recording the saga and the Court of Appeal's propositions
• ESM0560 — the evaluative exercise at the third stage — HMRC's stage-three guidance following Atholl House and PGMOL

On This Site

• IR35 and off-payroll appeals — the hub: which regime covers your years, CEST, and who holds the appeal right
• PGMOL v HMRC: employment status — Stages 3A and 3B after the Supreme Court, and its endorsement of Atholl House
• Ready Mixed Concrete: the three conditions — the framework Stage 3C sits inside
• Express & Echo v Tanton — when a substitution clause ends the case at Stage 3A
• Hall v Lorimer: painting the whole picture — how the final weighing is meant to feel
• Edwards v Bairstow: error of law — why appellate courts interfere with evaluations only for errors of approach
• Employment income appeals — the wider employment-tax landscape around status disputes
• Upper Tribunal appeals — the permission window, and what "set aside and remit" means
• After your tax tribunal decision — written reasons, remittals and permission to appeal
• Tribunal tracks and costs — who pays for a decade of litigation
• How to appeal to the tax tribunal — the deadline and filing steps once a status decision lands
• Preparing for your tax tribunal hearing — witness statements and bundles: getting the knowledge evidence before the tribunal of fact
• Writing grounds of appeal — the burden of displacing an assessment, and structuring grounds around 3A/3B/3C
• Postponing payment during an appeal — stopping collection of the disputed tax while the case runs
• Interest on unpaid tax — the clock that keeps running through a long dispute
• Reducing HMRC penalties — mitigation if a penalty sits on top of the tax
• HMRC internal review — the free second look before the tribunal
• Settling your tax tribunal case — agreements and ADR as the alternative to years of hearings
• Late appeal to the tax tribunal — if the deadline has already passed

This article is for informational purposes only and does not constitute legal or tax advice. For advice specific to your situation, consult a qualified tax adviser, accountant, or solicitor.