Constructive trusteeship: trap or tool?

In Paragon Finance v Thakerar & Co, Millett LJ warned that “the expressions ‘constructive trust’ and ‘constructive trustee’ create a trap.” The warning has stuck. A loose consensus has formed (across Lord Millett’s later judgments, Lord Neuberger’s reasoning in Williams v Central Bank of Nigeria, and the “fashionable” critiques of Birks and Swadling) that the language of constructive trusteeship, when used to describe the liability of dishonest assisters and knowing recipients, is misleading at best and incoherent at worst.

I think this consensus is half right. The phrase is genuinely strained when applied to dishonest assistance, which is better analysed simply as an accessorial version of a wrong. But in the context of knowing receipt, the idea of a trust being breached retains a meaningful and useful role, particularly when examined through the lens of proprietary arithmetic. Removing the language from dishonest assistance is overdue. Stripping it from knowing receipt would obscure something important about what beneficiaries actually have.

The rise and fall of the language

The modern recognition of constructive trusteeship traces back to Barnes v Addy, where the court proclaimed the now-canonical demarcation of “third party liability”: “strangers are not to be made constructive trustees… unless [they] receive… some part of the trust property, or unless they assist with knowledge in a dishonest and fraudulent design”. Although each limb is no longer fully accurate, this roughly corresponds to the realms of “knowing receipt” and “dishonest assistance” respectively. In Agip (Africa) v Jackson, the then Millett J borrowed the phrase while describing the “two main classes” of knowing receipt: the first, one who “receives for his own benefit trust property transferred to him in breach of trust”, and the second, one who receives the property lawfully but “deals with it in a manner which is inconsistent with the trust”. Millett J then reiterated the phrase when describing dishonest assistance: one who “knowingly assists in the furtherance of a fraudulent and dishonest breach of trust”.

Recent cases have cast doubt on the role of this language. In Paragon, Millett J accused “equity lawyers” of using “constructive trusteeship” in “two entirely different situations”; where one was liable through being “implicated in a fraud”, one “is not in fact a trustee at all” and “never assumes the position of a trustee”. The language is thus “misleading” for its proprietary connotations. Such concerns re-surfaced in Dubai Aluminium v Salaam, where Lord Millett reiterated how implicated third-parties never claimed to assume the position of trustee. More egregiously, the language of constructive trustee implies a proprietary claim, but claims in knowing receipt and dishonest assistance are personal claims, operating precisely where specific claims by the beneficiary are unavailable.

Williams v Central Bank of Nigeria explicitly considered the issue. In a dispute over whether an action was time-barred by s 21(3) of the Limitation Act 1980, a key contention was whether the third-party purportedly liable was a trustee, which included a “constructive trustee”. The majority of the UKSC held that there was no “true trust”; Lord Neuberger described reference to knowing recipients as trustees as “unreal” since no one had “placed trust in the recipient”, and that “it must follow that dishonest assisters are not either”. Lord Sumption, with the support of Lord Neuberger, argued that such third-parties had merely exposed themselves to equitable remedies by participating in the unlawful application of trust assets. A distinction had to be drawn between trustees with powers and duties, and the sole obligation of said third-party to restore the assets immediately, in some senses, a “self-defeating trust”.

Lord Mance, dissenting, focused on the potential inconsistency from the varied application of the Limitation Act: he attempted to offer a principled rebuttal, arguing that the “resulting position is still not incoherent”, and that equating one who “dishonestly joins with a dishonest trustee to defraud the trust, with the dishonest trustee for limitation purposes is entirely natural”. However, this sat in stark contrast with the prevailing scepticism in Lord Neuberger and Lord Sumption’s judgments, and the argument that constructive trusteeship is “entirely natural” was barely substantiated.

Williams is certainly good authority for the statutory meaning of “trustee” under the Limitation Act, but it does not resolve the central debate. It is entirely possible, for instance, for the language of “like a constructive trustee” to be used, averting the stricter construction of “constructive trustee” under the Limitation Act. Whether the courts’ broader aversion to the language is justified for both heads of liability is a separate, and I think more interesting, question.

Dishonest assistance: the language can go

The overwhelming majority of commentators have argued for the dismissal of constructive trusteeship in the context of dishonest assistance, and rightly so. Birks regarded the language as “surplusage”, while Lord Nicholls has extra-judicially remarked that it is “otiose and indeed, confusing”. The chief problem, as Lord Neuberger observes in Williams, is that “dishonest assisters do not take possession of any of the funds at issue”; it seems odd to speak of a “trust” if the “trustee” does not have the legal title to the property.

Mitchell and Watterson acknowledge this, admitting that the lack of legal title means the liability of an accessory cannot be derived from being a “trustee in his own right”. They propose instead the observation of an analogy, a “duplicative liability” that “mirrors” the trustee whom they assisted. They justify this by noting that accessories are often “liable to pay the same sums as the trustee”.

There are two problems with this. First, Davies rightly observes that the liability may be “parasitic”, but is not always “duplicative”: breach of trust and dishonest participation in breach of trust are, after all, “two species of equitable wrongs” (Lord Nicholls). A breach of trust is generally treated as a wrong incurring strict remedies; dishonest assistance is fault-based. In Novoship (UK) v Mikhaylyuk, the court explicitly rejected a gain-based remedy for a dishonest assister, arguing that the third-party did not “breach a fiduciary duty” but “committed an equitable wrong”. “There is no reason why the common law rules of causation… should not be applied by analogy”.

Second, given the non-duplicative nature of accessorial liability, calling the liability of a dishonest assister a “mirror” of the trustee’s liability is akin to describing accessorial liability in criminal law as a “mirror” of the liability of the primary wrongdoer. It would be a stretch to regard, for instance, someone who intentionally gave the address of a known victim to a gang as having “constructive[ly]” committed Grievous Bodily Harm. Such a stretch renders “trustee” meaningless, and distorts “constructive” to the point of excessive judicial discretion. Even Mitchell recognises the challenge, admitting that “it is harder to characterise dishonest assistants as ‘trustees’ than it is knowing recipients”.

So far, then, the courts and commentators are on solid ground. The language of constructive trusteeship can be dismissed for questions of liability for dishonest assisters.

Knowing receipt: the language earns its keep

Knowing receipt is where the consensus goes wrong.

The most forceful argument for retaining like-constructive-trusteeship in knowing receipt is offered by Mitchell and Watterson. Liability for knowing receipt, they argue, is a “distinctive, primary, custodial liability” that “closely resembles” the liability of express trustees. The main flaw of contrary views, in their account, lies in a misplaced “desire to rationalise equitable principles… [per] common law models of liability”, with Birks’ proposal that the personal claim for knowing receipt is grounded in unjust enrichment being the obvious example.

Treated as a “distinctive” liability instead, “[e]quity fixes [recipients] with [similar] custodial duties… [and similar] accounting mechanisms… [to] express trustees”. Where the third-party fulfils their “custodial duty” to return the rights to the trustee, the beneficiary may then falsify the account and enforce the original trustee’s duty to reinstate the trust fund. Mitchell and Watterson expound the precedent for their vision, and the benefit of a juristic foundation for the requirement of knowledge; according to them, this turns individuals into trustees.

Their explanations are cogent, but they have generated a host of responses. These can be grouped into two objections. Each deserves a closer look.

”Knowing recipients have only one duty, unlike other kinds of trustees”

There is a real difficulty in comparing knowing recipients to trustees of an express trust. Knowing recipients “have no trust powers or duties; they cannot invest, sell or deal with the trust property” (Paragon Finance). Instead, their “sole obligation… is to restore the assets immediately” (Williams). On the face of it, this casts doubt on the strength of the analogy.

In a response to Williams, Watterson issues a quick rebuttal: trusts are “diverse phenomena, with diverse purposes”. This sounds like a convenient excuse, but a quick look at the genuine variety of trusts (from the seeming simplicity of “automatic” resulting trusts where trusts have failed, as in Vandervell v IRC, to the potential complexity of express trusts in commercial scenarios) reinforces Watterson’s point. Not all trusts have the “same extensive array of powers and duties”. In particular, trusts arising by law are typically “bare” arrangements with “limited” duties.

Going further than Watterson, Georgiou argues against this granularity altogether, declaring it a “mistake to focus on this level of specificity”. Recipients, he posits, “owe the same duties at the abstract level”. He draws a useful analogy with the law on delegation: trustees are barred from delegating the exercise of their powers (subject to the Trustee Act 2000, ss 11–15 and other statutory inroads), and just as a purported “delegatee” does not perform duties “in accordance with the terms of the trust”, recipients do not deal with rights “in accordance with the terms of the trust”. The existence of a sole duty is not a problem; it is a natural conclusion of the unique form of “like-constructive trusteeship”.

Together, Watterson and Georgiou sufficiently discredit the courts’ concern about the discrepancy between the duties and powers of the original trustee and those of the knowing recipient.

”Why would ‘knowledge’ turn one into a trustee?”

A more interesting gap in Mitchell and Watterson’s account is why a defendant with “knowledge” may, by operation of law, be transformed into a trustee. Swadling rightly observes that most trusts “arise consensually”, where trustees are bound “precisely because [the obligations] have been voluntarily assumed”. The burden, he argues, still lies with Mitchell and Watterson to explain why such duties “ought” to be imposed.

The observation is salient, but it threatens to reduce the debate to a question-begging contention. The vast majority of commentators agree that “knowledge” (or some form of dishonesty) leads to liability. Some believe that this is sufficient for a trust to arise. Others do not.

Georgiou offers a refreshingly different move. The trust arises, he argues, simply on the disposition of trust property which failed to extinguish the trust. This stands in stark contrast to Swadling’s position, and ends up reinforcing the idea of constructive trusteeship. Georgiou’s vision is somewhat scattered in his case note on Byers v Saudi National Bank, but he reaches his thesis in two stages.

First, he rejects Lord Browne-Wilkinson’s holding in Westdeutsche Landesbank Girozentrale v Islington LBC that knowledge is a necessary condition for trusteeship. Unlike Swadling, Georgiou believes that knowledge is not at all needed for a recipient to become a trustee. He observes how Lord Browne-Wilkinson’s remarks (which serve as Swadling’s starting point) were dicta, and that Lord Browne-Wilkinson admitted in the very same case that whether one is a trustee “may only be a question of semantics”.

Second, Georgiou asserts that knowledge is nonetheless necessary for third-party liability. This requirement has always existed; in most “express trust” cases, however, a trustee already knows this, and so “there will be no issue”. This brings Georgiou’s vision in line with the outcomes of existing case law on knowing receipt.

Georgiou’s analysis is admittedly somewhat lacking at the second step: after a forceful rejection of the need for knowledge for trusteeship, he turns around and argues that knowledge is needed for liability stemming from trusteeship. On first glance, this might appear a convenient device to retain the language.

But when one understands that Georgiou’s second step is a necessary extension of his first, the “proprietary arithmetic” does the work. Where the rights in property were already encumbered by a trust, and where the disposition (from trustee to knowing recipient) did not destroy any trusts, the eventual right held by the knowing recipient must remain encumbered by a trust. The most damning attack on Swadling’s argument, therefore, is this: under Swadling’s understanding of knowing receipt, “it is not obvious how the encumbrance has somehow changed character from ‘trust’ to ‘not-trust’.”

Viewed fully, this leads to a startling conclusion about the nature of knowing receipt: a knowing recipient cannot be personally liable while still holding the trust rights, because the recipient would not be in breach of trust. While the trust rights are held by the recipient, the beneficiary can recover rights in specie. Only subsequent dissipations of the rights generate a right in personam by the beneficiary against the knowing recipient.

Beyond a single remark by Millett LJ in Paragon that recipients are “accountable… only by parting with the trust property”, there has yet to be detailed analysis of the precise moment at which in specie rights end and in personam rights start. That gap is itself, I think, evidence of how thinly the alternative accounts have been worked out. Georgiou’s analysis holds, and it is built on a deep understanding of the nature of trusts, an understanding that can only be elucidated through a recognition of the “constructive” trusteeship that knowing recipients undertake.

A trap worth keeping in part

Millett LJ was right that the language of constructive trusteeship creates a trap. He was right, too, that it should be discarded where it does not earn its keep; and in dishonest assistance, it does not. But the wholesale exile that Paragon, Dubai Aluminium, and Williams point towards goes further than the doctrinal terrain warrants. In knowing receipt, the language is not a trap. It is the lens that lets us see the proprietary arithmetic at work, and the route by which a beneficiary’s in specie and in personam rights can be properly distinguished. Concerns about it obscuring the nature of liability can be answered by a diverse and generous understanding of trusts, and a careful examination of what is actually happening to the rights in question.

The phrase is most useful when it is used most precisely. Dismiss it from dishonest assistance. Keep it for knowing receipt.