Introduction
Christopher Nolan’s The Odyssey arrives in cinemas in July 2026.
Universal describes it as bringing Homer’s foundational saga to IMAX film screens.
Firstly, I am not Homer[1]. But this feels oddly well-timed for those of us who have embarked on their own odyssey over the years following Baxendale-Waker’s (“BW”) Remuneration Trusts (“RTs”).
The UK’s tax courts have had their say on the various tax technical merits of these schemes over the last few years.
Indeed, I have recently written a couple of articles that have considered two BW-style RT cases recently: Marlborough DP Ltd v HMRC[2] and M R Currell Ltd v HMRC[3].
Currell shows the limits of Rangers and the general earnings charge where a genuine pre-Part 7A loan is involved. Marlborough shows the much wider reach of Part 7A where the arrangements are post-2011 and there is a sufficient connection with employment or directorship.
However, this article considers the promoters response to those decisions, and those that preceded them.
Ladies and gentleman, may I introduce to you to a new character called the ‘Nova declaration’.[4]
Here, the response was not simply to argue that the RTs worked. It was to argue that they did not, with the trusts failing for lack of certainty.
I am aware that this approach has been considered, and approved, in a recent BVI arbitration.
That is what I mean by Pyrrhic defeat.
Currell, Marlborough and the Part 7A problem
The BW RT arrangements were intended to deliver a familiar set of objectives.
A company would contribute funds to a trust or trust-linked structure.
Those funds would then be made available, commonly by way of loans, to an owner/director or connected individual.
The intended tax result was that the company obtained corporation tax relief while the recipient obtained access to value without the ordinary employment income, PAYE, NIC or distribution consequences.
This type of arrangement originated in the late 1990s and evolved over the next few decades. Regular revisions were made along the way, particularly, in response to the disguised remuneration legislation (Part 7A) enacted by Finance Act 2011.
Under the arrangement, the user adhered to the RT and agreed to make contributions.
The sums were paid to a collection agent and then on to a company established by the client as a fiduciary manager, or PMC.
The funds did not pass through the trustees’ hands.
The PMC, controlled by the client or the director of the client company, could then make loans or invest the money at that person’s discretion.
That, unsurprisingly, attracted HMRC’s attention.
The legislative mic drop was (supposed to be) Part 7A ITEPA 2003.
Part 7A was designed to counter disguised remuneration arrangements using trusts and other intermediaries.
The problem for users of these arrangements was that Part 7A does not ask the same question as the general earnings charge. It does not simply ask whether the payment is “from” employment. It asks, among other things, whether a relevant arrangement is a means of providing, or is otherwise concerned with the provision of, rewards, recognition or loans “in connection with” employment.
That difference matters.
In Currell, the transactions were pre-Part 7A. The company paid £800,000 into an EBT. The EBT lent the same amount to Mr Currell, who used it to buy shares from his wife. Mrs Currell then lent the £800,000 back to the company.
The FTT treated the arrangements as taxable earnings, but the Upper Tribunal disagreed, and the Court of Appeal has now dismissed HMRC’s appeal.
The Court of Appeal emphasised that Rangers was not authority for the proposition that a payment made to fund an employee benefit is itself necessarily a taxable emolument; the existence of a genuine repayment obligation was critical.
Yet the Court of Appeal’s concluding observations in Currell are also important. Falk LJ noted that Parliament had acted: had the transactions been implemented only a short time later, the charge HMRC sought would have arisen under Part 7A. She also warned against HMRC overreach in pre-2011 cases, observing that close inspection of the trees can risk failing to distinguish the overall wood.
Marlborough is the other side of the line. There, Dr Thomas operated his dental practice through Marlborough DP Limited. The company used a BW RT arrangement to contribute profits to a trust structure, with loans then made to Dr Thomas. By the time the matter reached the Supreme Court permission stage, the UKSC case summary described the issue as the proper test for whether the arrangement was a means of providing a reward, recognition or loan “in connection with” employment under section 554A(1)(c) ITEPA.
The Upper Tribunal had reversed the FTT and held that the correct test was whether there was a strong and direct connection between the employment and the payment. It also held that the payments did not satisfy the wholly and exclusively test for corporation tax deductibility.
The Court of Appeal upheld that decision, and permission to appeal to the Supreme Court was refused on 30 October 2025 because the application did not raise an arguable point of law.
So, after Marlborough, the position for post-Part 7A RTs looked bleak.
The general earnings argument might fail in some owner-managed company cases. Currell shows that. But Part 7A has a broader gateway. A loan may not be “from” employment for the general earnings charge, but still be “in connection with” employment or directorship for Part 7A.
Clause 13: the Achilles heel
The BW RT deeds did not merely define beneficiaries and trust powers. They also contained overriding protective clauses, introduced as a reaction to Part 7A.
Clause 13 of the deed was the key.
Clause 13.1 provided, in substance, that no benefit conferred, or other action taken or concurred in by the trustees or any other person in connection with the trust, should give rise to a Part 7A charge. The trust was also stated not to be an arrangement which covered or related to an employee, former employee or prospective employee in the manner described in section 554A.
Clause 13.2 then provided that the trustees had no power to take, and were under a duty not to take or concur in anyone else taking, any “relevant step”.
As such, when drafted, the clause was intended as a shield.
However, it was to become an Achilles heal.
And one that the promoters sought to exploit.
Once HMRC and the tax courts said that arrangements of this kind were within Part 7A, the clause intended to keep Part 7A out became the clause that destabilised the trust itself. If the deed says the trust must not be, and cannot be operated as, a Part 7A arrangement, but the courts say that in substance it is one, what follows?
The ordinary answer might be that the trust is valid, but certain acts are outside the trustees’ powers or in breach of trust. That is not the direction taken by Nova.
Nova’s answer was more radical.
If the trust could not operate without breaching its own overriding clause, then the trust never took effect at all.
That is the Achilles heel. The trust’s anti-Part 7A armour contained its fatal vulnerability.
Nova: the Odyssean turn
The central proposition is a blunt one.
The tax tribunals and HMRC were said to be going against the trust deed wording and, in essence, rewriting the trust deed. The “way forward” was that the original trust was void ab initio because at least one of the three certainties[5] had failed:
- Certainty of objects, the argument was that it was impossible to say with certainty who the beneficiaries were at the date of settlement. At the time the trust was entered into, there were no beneficiaries and the only way to identify who might benefit was to see how trust funds were later applied. That made the operation circular.
- Certainty of subject matter, the argument was that only sums that were tax deductible could constitute valid contributions to the trust. If the tribunal had deemed the sums non-deductible, they were not valid contributions and there was no trust property.
- Certainty of intention, the note says that the result in CIA Insurance[6] was “manifestly not what the Trustees signed up for.”
The conclusion was that at least one, if not all three, certainties failed. The trust was void. And if the trust was void, the money had never been settled on it.
At this point, the journey home begins.
If the trust was void, the money was said to be held on resulting or bare trust for the contributor. HMRC’s own manual says that if a trust lacks the three certainties, it is invalid and the transferred property is held on resulting trust for the settlor.
It is understood that the argument was that, once the parties executed a Nova Declaration recognising the voidness of the old trust, HMRC ought to be bound by that fact. If HMRC disputed it, an arbitration would be pursued.
Further, it is understood, that once an arbitral award was “franked” by BVI and English High Courts, HMRC would be unable to challenge that the trust was void and the matter would be res judicata.
The Nova Declaration: the Trojan Horse?
The Nova Declaration was the mechanism intended to implement the plan.
A plan so cunning that, had he not been busy hammering a load of nails into a massive wooden horse, Odysseus might have concocted.
The deed recites that a purported settlor had established a RT and made contributions.
It then says that successive FTT decisions, HMRC submissions and GAAR Panel opinions had ruled or argued that trusts declared by materially similar deeds were not the trusts upon which the purported contributions were held or applied.
The operative provisions then state that the parties cannot rely on the old trust deed as declaring and evidencing the terms of the trust, that the arrangement fails one or more of the three certainties, and that the purported declaration of trust was void ab initio.
Historic contributions, or property representing them, are defined as the “Void Remittance”.
The deed also contains the arbitration machinery. Disputes arising out of or relating to the purported trust deed, the purported trust, or the breach, termination or invalidity of them are to be settled by BVI IAC arbitration.
In summary, the mechanics were:
- First, declare that the old trust was void;
- Second, identify the void remittance;
- Third, obtain an arbitral award;
- Fourth, use that award to support the resulting-trust analysis; and
- Fifth, deploy that analysis against HMRC.
The accounting problem: Penelope’s loom
There was, however, an immediate problem.
If the trust was void, and the money was always beneficially the contributor’s, what did that do to the historic accounts and tax computations?
An argument might be that the contribution would still be attributed to the year in which the obligation arose and that the accounts did not need to be restated and tax computations remained correct.
In other words, the original commercial obligation was separate from the vehicle used to satisfy it. If the obligation arose in 2012, it fell to be accounted for in 2012, even if discharged in a later year by a replacement trust or other mechanism.
If the trust was void, the funds reverted to the settlor. But for accounting purposes, is it “arguable” that the original legal obligation to make a contribution had not been discharged and still existed.
If the contribution was re-made, was it similarly arguable that no remedial action was required?
I will leave that to the accounting specialists!
The BVI arbitration: Ithaca… but only for the parties
A recent (January 2026) BVI arbitration award I am aware of is perhaps the clearest practical implementation of Nova.
The claimant was the current trustee of the RT. The respondent being the corporate founder/contributor. The arbitration concerned the RT Deed dated 24 June 2005, as amended by a 2012 deed (the Part 7A shield).
The relief sought was essentially the Nova relief, with he parties seeking an order that:
- The trustees should be released from the trusts of the RT, as never having been subject to them;
- an order that all monies purportedly settled on the trust by the company were always held by the trustees on bare trust for the company; and
- an order that the trustees transfer the money back.
However, there are several important procedural features.
- The respondent did not dispute the claim;
- The BVI IAC rules did not provide for default judgment, so an arbitrator still had to be appointed;
- The parties did not engage in a formal disclosure process;
- Neither party requested directions for written or oral evidence at the final hearing;
- After the hearing, the parties submitted an agreed post-hearing note; and
- the parties agreed the relief they both sought.
To his credit, the arbitrator did not treat agreement as enough. He cited the duty identified in JTC Employer Solutions Trustee Ltd v Garnett[7] to give the decision-maker the help required to reach the right decision. He also cited Wright v National Westminster Bank plc[8] for the proposition that an unopposed application does not mean the order can safely be assumed to follow.
So the award is not merely a rubber stamp.
But neither is it an adversarial contest.
Both parties wanted to arrive at the same destination.
The BVI reasoning
The arbitrator accepted that the 2012 deed amended the 2005 deed retrospectively. That mattered because the 2012 deed introduced the Part 7A override clause.
The award then considered Marlborough. The parties submitted, and the arbitrator accepted, that the RT in Marlborough was in like terms to the 2005 deed under consideration (as amended).
The award recorded that the Court of Appeal had dismissed MDPL’s appeal and upheld the Upper Tribunal’s finding that the sums were caught under Part 7A because of their connection with Dr Thomas’ directorship. The Supreme Court has now refused permission to appeal.
The arbitrator’s key step was that, because there was at least a real risk that Marlborough would be followed in respect of the trust, there was uncertainty about whether the trustees were prohibited from receiving contributions, prohibited from applying funds, and able to operate the trust powers outside section 554A.
He then addressed the three certainties:
- On intention, the settlor clearly intended to create trusts outside Part 7A. But it was not certain that the 2005 deed as amended gave effect to that intention.
- On subject matter, clause 13 prohibited the trustees from receiving contributions if the deed came within section 554A. The real risk that it did come within section 554A created uncertainty as to whether the trustees ever acquired title to the contributions.
- On objects, clause 13 prohibited the trustee from taking a relevant step by applying trust funds in relation to employees, former employees or prospective employees of the corporate founder. The trustee had to be able to say with certainty whether a person was within the permitted beneficiary class. The arbitrator concluded that there was uncertainty as to whether such persons were within that class.
He therefore concluded that each of intention, subject matter and objects was uncertain, although he ultimately treated the subject-matter uncertainty as sufficient.
The BVI award
The operative part of the award is dramatic.
The arbitrator directed that:
- The RT is void;
- The trustees were released from, as never having been subject to, the trusts of the RT;
- All monies purportedly settled on the trust by the corporate founder were always held by the trustees on bare trust for the company; and
- The trustees were to transfer all monies purportedly settled on the trust and/or any property representing the same.
In other words, the trust is void. The trustee was never trustee on the old terms. The money was always beneficially the contributors. The money must come home.
This was Nova’s private-law victory.
Ithaca has been reached…
..but only by the parties to the arbitration.
The Cassandra warning?
The most important sentence in the BVI award is not the declaration that the trust is void.
It is the caveat.
The award records that the parties recognised that the determination of the arbitration did not affect HMRC because HMRC was not a party to the proceedings.
That is a serious problem for the more ambitious Nova thesis.
It is understood that the full plan asserted that, once an arbitral award was obtained and “franked” by BVI and English courts, HMRC would be unable to challenge the trust’s voidness and that the matter would be res judicata.
The BVI award does not say that.
It says HMRC is not bound by the award.
Of course, that does not mean the award is irrelevant. Just that it might have its limits.
HMRC was not a party. There was no contradictor. There was no formal disclosure. No oral or written evidence was requested. The parties agreed the relief.
What might HMRC say?
HMRC’s likely answer is not difficult to predict.
First, HMRC would say that CIA Insurance, Strategic Branding and Marlborough did not decide that RTs were void. They decided tax questions. Indeed, CIA made clear the trusts were effectively constituted and that contributions passed through the trust mechanisms, before drawing the further conclusion that the trust was void for impossibility.
Second, HMRC would say that a private arbitration between aligned parties cannot bind the Crown. The BVI award accepts that.
Third, HMRC would say that Part 7A is drafted to look at the essence of arrangements and relevant steps. It is not necessarily defeated by private-law irregularity. Indeed, the statutory extracts quoted in the BVI award itself include wording indicating that a relevant step may be caught even if it constitutes a breach of trust or is void as a result of breach of trust.
Whether that answers a void-ab-initio trust argument is a different and more difficult question, but it gives HMRC an obvious statutory foothold.
Fourth, HMRC would challenge the “real risk equals uncertainty” reasoning. The BVI award reasons that because there is a real risk that Marlborough applies, there is uncertainty sufficient to void the trust. HMRC may say that this confuses tax risk with trust-law uncertainty. A trustee often faces uncertainty as to tax consequences. That does not usually mean the trust itself is void.
Fifth, HMRC would attack the accounting sequel. If the money was always beneficially the company’s, then why did the historic accounts and tax computations properly treat it as having left it?
Conclusion: a Pyrrhic defeat
There is something almost elegant about Nova. It takes HMRC’s own victories and tries to turn them against it.
The promoter-side case is no longer that the RTs worked. It is that they were void from the beginning.
That is the Pyrrhic defeat.
A victory won by accepting that the original structure failed.
The BVI arbitration shows that this argument can succeed, at least in a private arbitration between trustee and contributor.
But it also hints at the limit of that success. HMRC was not a party and was not bound.
The award may bring the money home as a truce between the particular trustee and the corporate founder. However, it seems unlikely to end the tax war… not at least before further return fire.
As such, it is unlikely that the BVI award is the end of this particular odyssey.
—
[1] Yes, yes… more like Homer Simpson. I hear you.
[2] [2025] EWCA Civ 796
[3] [2026] EWCA Civ 445
[4] Not to be confused with HMRC’s NOVA service… or the humble Vauxhall Nova. However, it was subject to an unusual decision by HMRC to publish some of the marketing docs: https://www.gov.uk/government/publications/named-tax-avoidance-schemes-promoters-enablers-and-suppliers/evidence-of-marketing-material-used-by-tax-avoidance-promoters-and-suppliers
[5] See Knight v Knight (1840) 49 ER 58
[6] [2022] UKFTT 00144 (TC)
[7] [2024] EWHC 3128 (Ch)
[8] [2014] EWHC 3158 (Ch)