Conspiracy to Cheat HMRC: Knowledge of Precise VAT “Off-Setting” Mechanics Not Required Where a Broad Cheat Is Indicted

Case: Rush & Anor v R Neutral citation: [2026] EWCA Crim 442
Court: Court of Appeal (Criminal Division) (Lady Justice May DBE, Mr Justice Pepperall, HHJ Plaschkes)
Date: 19 March 2026

1) Introduction

The Court of Appeal considered conviction challenges arising from “Operation Barbados”, an HMRC investigation into a sophisticated VAT fraud said to have cheated HMRC of about £20 million. The principal trading company, Winnington Networks Ltd (“WNL”), generated large amounts of output VAT on real UK sales and then sought to neutralise or reduce its VAT liability by claiming input VAT on purported purchases, including alleged (and largely fictitious) “VOIP” trading supported by bogus offshore banking platforms.

The appellants (referred to by surname) were convicted at Southwark Crown Court of a single count of conspiracy to cheat the public revenue. Trial 2 focused on the “metals” strand (copper rod and cathode supply chains) said to generate output VAT that WNL dishonestly retained. The appeal raised two tightly connected issues:

  • Scope of the indicted conspiracy: must each conspirator know the precise mechanism by which WNL’s VAT returns were rendered false (i.e. bogus VOIP “off-setting”)?
  • Proper jury directions and defence forensic choices: could the defence invite the jury to acquit on the basis that the evidence showed only a different, “lesser” conspiracy focused on withholding VAT from metals trading?

2) Summary of the Judgment

  • Rush: appeal dismissed; conviction held safe.
  • Leighton: renewed application for leave refused; conviction held safe.
  • The Court held that, given the way the conspiracy was pleaded (“dishonestly inducing or facilitating the submission of false VAT returns”), the prosecution did not have to prove that each conspirator knew the detailed accounting device (bogus VOIP input VAT) by which WNL falsified returns. It was enough that the defendant joined an agreement whose common purpose was to cheat HMRC by wrongfully retaining VAT through false returns.
  • The trial judge was entitled to prevent the defence (on the facts of this case) from advancing to the jury a positive alternative case that the defendant joined some different conspiracy, where the defence position denied membership of any conspiracy and there was no evidential foundation for such an alternative agreement.
  • For Leighton, the Court confirmed that conspiracy under section 1 of the Criminal Law Act 1977 criminalises the agreement; it is not necessary to prove further “participation” in implementing it, and any suggestion from Anderson [1986] AC 27 of a universal need for intended active participation was qualified by later authority.

3) Analysis

3.1 Precedents cited and their role

(a) R v Ali [2011] 2 Cr App R 22

Ali was central to the Court’s approach to how broadly conspiracies may be framed. The judgment emphasised that prosecutors commonly formulate conspiracy counts in broad terms; the jury decides whether the defendant was party to the broadly formulated conspiracy, while the sentencing judge later differentiates roles and culpability. The Court in Rush applied this to reject the argument that the prosecution had to prove knowledge of a particular “mechanic” (VOIP off-setting) as an essential element for every defendant.

(b) SFO v Papachristos [2014] EWCA Crim 1863

The Court relied on Papachristos for the proposition that conspirators may pursue a shared common aim by different means and that it is unnecessary to prove that all conspirators were aware of the full range of execution methods. What matters is “a shared common purpose or design in conformity with the charge”. This provided direct support for the conclusion that a metals-linked defendant need not be proved to know VOIP off-setting details if the indicted common purpose was to cheat the revenue via false returns and retention of VAT.

(c) Anderson [1986] AC 27 and Siracusa (1990) 90 Cr App R 340

Leighton’s renewed application leaned heavily on Lord Bridge’s statement in Anderson suggesting that mens rea requires an intention “to play some part” in the agreed course of conduct. The Court treated Siracusa as the corrective lens: it held that the cited Anderson passage was not ratio and explained that participation in conspiracy can be “active or passive”, and intention may be inferred from knowledge and adherence to the agreement.

The Court also endorsed the textbook summary (Smith Hogan & Ormerod) that conspiracy liability does not require an agreement that the defendant personally will commit the substantive offence; it is enough that the defendant agrees that a crime will be committed.

(d) Other authorities referenced: Roberts and Taylor [1998] 1 Cr App 441; R v Wells [2010] EWCA Crim 1564

These were cited within the Ali discussion, illustrating prosecutorial choice in charging one conspiracy or multiple conspiracies (particularly where different substantive offences are involved). In Rush, they reinforced the broader theme: the indictment’s framing matters, and a single conspiracy may encompass multiple strands when directed to a single substantive criminal aim.

3.2 Legal reasoning

(a) The statutory anchor: section 1 Criminal Law Act 1977

The Court began from the statutory language: (i) the offence is the agreement; (ii) the agreed course of conduct need not be embarked upon; (iii) if carried out in accordance with the parties’ intentions it must necessarily amount to or involve the commission of an offence by one or more parties (not necessarily the defendant).

(b) Indictment-driven scope: “cheat” vs “mechanics”

Rush argued that because WNL’s VAT returns were accurate as to output VAT on metals, the “falsity” depended on VOIP input-tax fiction; therefore, knowledge of VOIP off-setting was said to be essential. The Court rejected that as a matter of what was charged. The conspiracy was pleaded as cheating HMRC by dishonestly inducing or facilitating false VAT returns—focused on the wrongful retention of funds due to HMRC. The Court treated the VOIP off-setting as an evidential route to proving falsity and dishonest retention, not as an element that each conspirator must understand in detail.

(c) Jury directions: “no other participation is required”

The Court upheld the direction that, once a defendant “joined” the agreement with the requisite mental elements, no additional participation was required. That did not eliminate mens rea: the jury were still required to be sure (via the written directions and route to verdict) that the defendant knew the agreement involved depriving HMRC of VAT and intended that VAT would not be paid, and that the defendant acted dishonestly.

(d) The “alternative conspiracy” submission and forensic fairness

The Court rejected the complaint that the judge should have allowed the defence to suggest to the jury a “lesser/different” conspiracy (e.g. withholding VAT from metals trading alone). Two considerations drove this:

  • Charge fidelity: the jury’s task was to decide whether the defendant was party to the indicted broad conspiracy; the route to verdict properly required the jury to confirm that the conspiracy included the metals strand (Question 2) and then decide whether the defendant joined it (Question 4).
  • No evidential foundation / inconsistent defence posture: the defence statement denied membership of any conspiracy; the defendant gave no evidence and advanced no positive case supporting a distinct alternative agreement. In those circumstances, the judge was entitled to prevent counsel from positively advancing an alternative conspiracy theory to the jury.

(e) Section 74 PACE 1984 (Trial 1 convictions) and the “metals question” safeguard

The trial judge admitted convictions from Trial 1 under section 74 PACE 1984. A key appellate concern was whether this risked pre-determining “scope”. The Court’s answer was practical: the route to verdict explicitly required the Trial 2 jury to be sure that the accepted overarching conspiracy included use of metals trading as a means by which WNL stole VAT. This prevented an impermissible “scope shortcut” from the section 74 evidence.

3.3 Impact

  • Broad conspiracies in complex fraud: the decision supports prosecutorial ability to charge a single broad conspiracy embracing multiple operational “strands” (real trades, fictitious trades, financing/investment, accounting devices) without having to prove that every conspirator understood each strand’s mechanics—provided the indictment is framed around a shared criminal aim consistent with section 1.
  • Jury directions in multi-strand conspiracies: trial judges may direct that detailed knowledge of execution mechanics is unnecessary, while still insisting on knowledge of the dishonest objective (here, depriving HMRC of VAT through false returns) and dishonesty. Tailored “scope-check” questions (like the metals question) can be used where necessary to ensure the jury addresses contested strands.
  • Constraints on “lesser conspiracy” arguments: where the defence advances no positive evidential basis for an alternative agreement and the defence case is outright denial of any conspiracy, trial judges may properly prevent closing submissions that invite acquittal by hypothesising a different conspiracy.
  • Executory intent and passive participation: the judgment consolidates the post-Siracusa approach: conspiracy liability does not invariably require proof of intended active steps by the defendant; agreement plus the requisite mental elements suffices.

4) Complex concepts simplified

  • Output VAT vs input VAT: output VAT is collected from customers on sales; input VAT is VAT paid on business purchases. VAT returns typically pay HMRC the difference (output minus input).
  • “Off-setting” fraud (as used here): a trader reduces VAT payable by inventing purchases (fake input VAT) to cancel out real output VAT collected on sales, allowing the trader to keep money that should be paid to HMRC.
  • Conspiracy under section 1 Criminal Law Act 1977: the crime is the agreement that a criminal course of conduct will be pursued. The prosecution need not prove the plan was carried out, nor that the defendant personally would commit the substantive offence.
  • “Executory intent” (in conspiracy): the defendant must intend that the criminal objective be carried out (in the sense captured by joining/adhering to the agreement with the requisite mental elements). It does not necessarily require proof that the defendant intended to take hands-on operational steps.
  • Section 74 PACE 1984: allows proof of a fact by relying on a prior conviction, subject to limits. In a conspiracy context it can prove existence of a conspiracy, but careful directions may be needed to ensure the jury still decides disputed features (e.g. whether a particular “strand” was part of it).

5) Conclusion

[2026] EWCA Crim 442 confirms an indictment-led approach to conspiracy scope in complex tax fraud: where a broad conspiracy to cheat HMRC is charged, juries need not be satisfied that each conspirator knew the precise accounting mechanics (such as VOIP-based off-setting) by which false returns were achieved, provided they are sure the defendant joined the dishonest agreement to deprive HMRC of VAT and acted dishonestly. The Court also endorsed robust trial management of closing submissions: absent an evidential foundation and where the defence position is complete denial, it is permissible to prevent the jury being invited to speculate about a different “lesser” conspiracy.