Skatteforvaltningen v Solo Capital Partners LLP & Ors

Introduction [1]

Invalidity [43]

Main Narrative [101]

Trial Witnesses [376]

Sham Trading? [414]

Was SKAT Misled? [424]

Result (except SKAT vs. Syntax) [610]

SKAT vs. Syntax [615]

Introduction

This litigation concerned share trading on cum-ex terms, Danish dividend tax, and the approach of the Danish customs and tax administration (‘SKAT’) to the processing and paying of claims for dividend tax refunds between mid-2012 and mid-2015. The final trial in these proceedings, on which this is the judgment, was the trial of claims pursued by SKAT concerning 4,170 of the dividend tax refund claims that were submitted to it and paid by it during that period. The aggregate amount paid by SKAT in response to those 4,170 claims was just under DKK12.1 billion (c.£1.4 billion at today’s exchange rate).

An equity trade is on cum-ex terms if it is traded, i.e. entered into, on or before a dividend declaration date, for settlement, i.e. performance, after the record date for that dividend. In October 2021, in an interview for a German television documentary, Sanjay Shah said of the Danish cum-ex scandal and his part in it: “… just going back to what happened in Denmark, why would they pay out for years and years and then, after four years of payments, they say, “Oh, we made a mistake, or we were cheated”? If there’s a big sign on the street saying, “please help yourself”, then me or somebody else would go and help themselves”.

This judgment determines whether SKAT was cheated in the sense that is relevant to this litigation. That is to say, it determines whether Sanjay Shah and others practised upon SKAT the deceit it alleged by its pleadings in this court. On the evidence put before the court at trial, and without forgetting that SKAT bore the burden of proof, one alternative to emerge was that Sanjay Shah and others were able to and did help themselves to a fortune because SKAT’s processes were so limited as to fit Mr Shah’s street sign analogy, not because they needed to or did practise deceit upon SKAT as it alleged.

Sanjay Shah is one of several trial defendants convicted in Denmark on criminal charges arising out of the activities considered in this judgment. A New York federal jury has awarded SKAT US$500 million in damages against a number of parties involved in some of those activities. There are also some judgments in favour of SKAT in Dubai. None of those decisions is relevant to whether any of the trial defendants in these proceedings is liable as alleged by SKAT. That falls to be judged by reference to the claims pursued by SKAT at the trial in this court, all of which are governed by English law, and upon the evidence and argument put before this court at that trial. Those claims, and that evidence and argument, may or may not be the same as or similar to the criminal charges or civil claims brought in other jurisdictions under systems of law other than English law, or the evidence and argument put to those other courts; and even if all was the same or materially similar, the decisions of judges or juries in other jurisdictions as to what had or had not been proved are inadmissible opinions about the facts in this court, which must reach its own decision.

SKAT accepted that if it made a financial recovery pursuant to any of the decisions elsewhere, credit would have to be given for that if remedies fell to be considered here, but that is a different point. Any such recovery might reduce the loss ultimately suffered by SKAT at the date of this judgment, or it might involve a transfer or restoration to SKAT of assets in which it claimed to have an ownership interest. Either way, the recovery might need to be taken into account, without the decision of the other court having anything to say as to liability on a claim pursued here or as to the remedies available in this court if such a claim succeeded.

The broad factual claim advanced by SKAT was that market appetite to provide funding to support the cum-ex trading strategies upon which Sanjay Shah’s main business had been focusing dried up by late 2011, and that he and senior individuals then working for him turned to fraud, developing a cum-ex trading model they knew and intended would result in false statements being made to certain tax authorities, including SKAT, in the hope of tricking them into paying tax refund claims, Sanjay Shah and his colleagues realising at the time that they would be invalid claims or at any rate having no honest belief that they were or might be valid claims. A large cast of individuals was assembled, all of whom, SKAT alleged, were made aware or became aware that the business in which they were participating existed to generate the deception of SKAT through false statements made in tax refund claims submitted to it, but participated anyway, through greed in view of the financial rewards on offer.

Greed can be a powerful motive, and I consider there was substantial greed here. However, the evidence at trial did not persuade me to accept SKAT’s claim, and I do not make the findings it sought. I also do not accept, on the other hand, the positive factual case advanced by Sanjay Shah and the other architects of the cum-ex trading models that gave rise to SKAT’s claims in the proceedings, that is Graham Horn, Rajen Shah and Guenther Klar, that during the relevant period they held a reasonable belief that those models generated tax refund entitlements under Danish tax law. In Mr Klar’s case, I accept his evidence that he did think at the time that the resultant tax refund claims were valid in that sense, but I conclude that he had no reasonable basis for that view. In the case of Sanjay Shah, Mr Horn and Rajen Shah, I do not accept evidence they each gave that at the time they thought the tax refund claims they were facilitating were valid claims. That does not mean that the deceit alleged by SKAT occurred, however. My rejection of narratives put forward by many of the trial defendants, with findings I make that some, notably Sanjay Shah, were dishonest in various ways at the time and untrustworthy as witnesses, does not prove the case pleaded by SKAT and pursued by it at trial, although it is capable of lending some indirect support to it. On the evidence as a whole, however, I have concluded that that case was not established.

The money-making strategy devised and implemented by Sanjay Shah, with Mr Horn and Rajen Shah, likewise Mr Klar’s version of it, was pursued, and worked between mid-2012 and mid-2015, not because they identified that it involved falsehoods being communicated to SKAT that might trick it into paying, but went ahead anyway. Rather, it was pursued and worked because they did not identify any such thing, that is to say they did not consider that anything untrue would be or was being stated to SKAT, and when they implemented the strategy, they found that SKAT paid. They engaged in concealment and obfuscation that involved pervasive collateral dishonesty, because of an instinct that if it was known that the tax refund claims resulted from pre-planned and coordinated trading, the sole purpose of which was to create tax refund claims predominantly (by value) for the benefit not of the tax refund claimant, SKAT might challenge the claims rather than pay and/or the Financial Conduct Authority (‘the FCA’) might disapprove (some of the entities central to the trading models being FCA regulated financial services firms). Even bearing that well in mind, I was not persuaded that the strategy would have been pursued if Sanjay Shah, Mr Horn or Rajen Shah, respectively Mr Klar, thought that it would or did involve false statements being made to SKAT to mislead it into paying claims.

More fundamentally still, and the basis upon which I have concluded that the claims pursued by SKAT in this court on which liability was in issue all fail, I was satisfied on the evidence that SKAT was not misled by misrepresentations made to it through the tax refund claims it received, as it alleged. Its controls for assessing and paying dividend tax refund claims were so flimsy as to be almost non-existent. That, it might be thought, came to be exploited somewhat ruthlessly. However, that did not require, nor did it involve in fact, that the misrepresentations alleged by SKAT were made or induced SKAT to pay claims it would otherwise not have paid.

None of the 4,170 tax refund claims with which the trial in these proceedings was concerned was a valid claim under Danish tax law, and SKAT would have been entitled not to pay any of them. SKAT did not suggest, however, that paying a tax refund claim that it was not obliged to pay gave it a cause of action it could pursue in this court. The causes of action that SKAT did pursue in this court all required it to prove that it had been deceived into paying, except for one cause of action against Lindisfarne Partners LLP, claiming damages for negligent misstatement, but that cause of action still required SKAT to prove that it was misled into paying claims, as it alleged, by the misrepresentations it claimed were made to it.

The result is that SKAT failed to establish any of the claims it pursued at trial, where liability was disputed, and all those claims will be dismissed.

Against one trial defendant, Syntax GIS Ltd, SKAT was entitled to treat a default judgment as establishing liability on at least some of the claims pleaded against it, so that what was left was the measuring of the amount to be awarded to SKAT under that judgment. The default judgment against Syntax proved nothing against any other defendant; and my conclusions as to liability in respect of SKAT’s claims against all the other trial defendants are irrelevant to SKAT’s entitlements under the default judgment.

SKAT established at trial that, on tax refund claims for which it pursued remedies against Syntax under the default judgment, it paid in aggregate some DKK2,763,859,045.79, against which credit must be given by SKAT in respect of recoveries it has made to date. I shall ask counsel to assist further as to the amount of that credit and also as to whether, if SKAT asks for judgment for the net amount, after recoveries, by way of restitution to reverse unjust enrichment, there can then be any judgment for damages. A claim was also pleaded against Syntax for a declaration in contingent form, to the effect that if Syntax has retained to date any of the sums paid to it by SKAT, or their traceable proceeds, then it holds the same on trust for SKAT. No attempt was made to establish that Syntax has retained anything traceable to sums paid to it by SKAT, and I shall not grant any declaratory relief.

The claimant is named in the proceedings as “Skatteforvaltningen (the Danish Customs and Tax Administration)”. It is not a separate legal person from the Kingdom of Denmark. Rather, it is an integral part of the Danish state, being that part of the state with responsibility for the areas of administrative activity that have been allocated to it. In other words, and as a matter of identity or substance, the claimant is the Kingdom of Denmark, acting by and suing in the name of Skatteforvaltningen, a ministerial authority going by that name since 1 July 2018, previously (from 2005) having been named “SKAT”.

The nation state of Denmark is a constitutional monarchy. Professor, Dr.Jur Frederik Waage gave unchallenged expert evidence on that, confirming and explaining the view given in his treatise, “Offentligretlige retssager” (2020), at pp.188-189, that (in translation): “The entire state administration is perceived in case law as one unit. This is consistent with the fact that [the monarch] is understood in The Constitutional Act as a collective term for all the ministries under the executive power … . The fact that the state is a single entity has, among other things, the consequence that different state authorities can set off a claim against a citizen across authority boundaries … . Although the State is perceived as an entity, it is the specific administrative authority responsible for the subject matter that acts as plaintiff or defendant on behalf of the State … .”

Prof. Waage explained aspects of the independence of action, and knowledge, of different ministerial authorities as a matter of Danish public law that inform, for example, his observation that such an authority can and will act, in a Danish court, as plaintiff or defendant on behalf of the Danish state. I do not consider that sense in which, as Prof. Waage put it, a ministerial authority “thus possesses an independent status and legal personality”, falsifies the fundamental “single entity” principle stated in his text and confirmed by his report. That principle, however, would not mean that any of the claims pursued in this litigation fell to be dismissed because the claimant was named in the proceedings as it was rather than as the Kingdom of Denmark; and I continue to refer to the claimant, as I have done throughout the litigation, simply as ‘SKAT’.

This case concerned exclusively the taxation of dividends on shares in Danish listed companies. Such shares existed at all material times as a legal construct, in dematerialised and fungible form. Everything I say refers only to such shares and such companies, during the period of interest in the case (2012-2015).

As I explained and found in SKAT (Validity Issues) [2023] EWHC 590 (Comm), Danish dividend tax was one head of the unlimited tax liability of legal persons who are tax domiciled in Denmark, and as a head of limited tax liability imposed on legal persons who are not, on a withholding tax (‘WHT’) basis. Subject to exceptions that are not relevant to this judgment, Danish companies were obliged to pay to SKAT 27% of the dividends they declared and only the balance to the Danish Central Securities Depository for distribution. The Danish Central Securities Depository was VP Securities (‘VPS’).

The payment by a Danish company to SKAT of 27% of its declared dividend (or possibly just the company’s obligation to pay SKAT) discharged the Danish dividend tax liability of those liable to such tax on that dividend. Some such legal persons might be entitled under a double taxation treaty (‘DTT’) between Denmark and their tax domicile not to be taxed on Danish dividends, or not to be taxed at a rate exceeding some rate below 27%. For example, tax-exempt US pension funds (‘USPFs’) were entitled under the Denmark-US DTT not to be taxed on Danish dividends (so long as they were the beneficial owner, for the purpose of the DTT, of the dividend); and tax-exempt Labuan corporations (‘LabCos’) were entitled under the Denmark-Malaysia DTT not to be taxed on Danish dividends (with no express beneficial ownership condition).

A beneficial ownership condition such as was explicit in the Denmark-US DTT was an additional requirement for relief from Danish dividend tax. In Danish tax law, it did not create an entitlement to relief in a party that had not been taxed by Denmark, but it could mean that, for example, a USPF which had been taxed by Denmark was not entitled to relief because it had not been the beneficial owner, in the sense used by the DTT, of the dividend on which it had been taxed. SKAT did not pursue at trial any claim that any of the tax refund claims impugned in these proceedings was a bad claim because the entity on behalf of which the claim was submitted, though entitled to and taxed on a Danish dividend, was not the beneficial owner in the relevant sense. Rather, SKAT’s case was always that the entity on behalf of which a tax refund claim was submitted was not entitled to, or taxed on, any Danish dividend in the first place.

It is possible, but this was not explored fully at trial, that this approach of Danish law, providing an entitlement to a tax refund only where the beneficial owner of the dividend, in the sense used in DTTs, was also the party treated by Danish tax law as liable to tax on that dividend, did not properly discharge Denmark’s treaty obligations. That possibility arises because it may be the Treaty obligation was not an obligation not to tax beneficial owners of dividends, if they qualified for a tax-favoured status recognised by the DTT, but an obligation not to tax dividends, if their beneficial owners so qualified. It is not necessary to make any definite finding about that, nor am I in any position to do so; it arises from my finding in SKAT (Validity Issues) that DTT beneficial ownership of a dividend, and tax law share ownership under Danish law, were cumulative requirements of entitlement to a tax refund in Denmark, and I find, below, that SKAT was not at the time operating its tax refund claim process on that basis.

In any event, the beneficial ownership condition in DTTs is part of the context for the standard form by which SKAT required tax refund claims to be made. It is therefore relevant when construing that form for the purpose of SKAT’s allegation that misrepresentations were made to it. The same form was specified by SKAT for all tax domiciles with which I am concerned. The form current in the period I am considering required the applicant to identify themselves either as, or as acting on behalf of, the “beneficial owner”, without further explanation, and without reference to whether any such concept was mentioned expressly in the pertinent DTT. SKAT’s approach was that the beneficial ownership requirement went without saying, albeit sometimes it was made explicit in the DTT. Where I refer to ‘beneficial owner’, I shall be referring to that concept as used in, or as may be implicit in, a DTT.

Each of the tax refund claims made to SKAT with which I am concerned was made by an agent on behalf of an entity which had been inter alia the buyer under an equity trade on cum-ex terms in respect of Danish shares, i.e. an equity purchase traded on or before a dividend declaration date for settlement after the record date for that dividend.

In SKAT (Validity Issues), I found that any entitlement to a Danish dividend tax refund from SKAT, based on a DTT, was not a matter of unjust enrichment with a claim against SKAT on that basis, but a statutory right under Danish tax law, arising (after 1 July 2012) under s.69B(1) of the Danish Withholding Tax Act, which provided as follows (in translation):

“If a person who is liable to pay tax pursuant to section 2 hereof or section 2 of the Danish Corporation Taxation Act has received dividends, royalties or interest, of which tax at source has been withheld pursuant to sections 65-65D which exceeds the final tax under a double taxation treaty, …, the amount must be repaid within six months from the receipt by [SKAT] of a claim for repayment. …”

This litigation (and related litigation in other jurisdictions, in particular the US, Malaysia and Dubai) concerned Danish dividend tax refund claims presented to SKAT and paid between August 2012 and July 2015. In the proceedings in this court, SKAT said that it was wrongfully induced to pay those claims by misrepresentations made by the documents submitted to it, interpreting those documents sensibly in their context, namely that of making claims for relief and a refund in respect of Danish dividend tax.

As I mentioned at the outset (paragraph 1 above), this judgment concerns 4,170 tax refund claims accepted and paid by SKAT, for a total of just under DKK12.1 billion. Another DKK400 million odd was paid on tax refund claims arising from structured transactions involving MCML Ltd (previously ED&F Man Capital Markets Ltd). Claims against MCML in these proceedings failed, and a subsequent attempt by SKAT to pursue differently formulated claims against MCML was struck out earlier this year (Skatteforvaltningen v MCML Ltd [2025] EWCA Civ 371, a decision from which the Supreme Court has granted permission to appeal, with the appeal listed to be argued in Feburary 2026).

Each of those tax refund claims followed a series of what purported to be share trades and derivatives relating to shares in one of the leading Danish companies making up the OMX Copenhagen 20 (‘C20’) Index. SKAT’s primary case on the facts was that those apparent transactions were not contracts at all, in that, so SKAT alleged, the respective parties to them intended only to create a documented pretence of share trading. That is not a necessary ingredient of any of the causes of action SKAT has pursued, and this judgment would be clumsy to read if it constantly qualified the transactions or their effects by ‘purported’, ‘purportedly’ or the like, so it does not do so.

I introduce the trial defendants and the claims SKAT made against them below, after first outlining the main stages of the litigation that brought it to the ‘Main Trial’ upon which this is now the judgment.

Five Claims were consolidated: CL-2018-000297 (the ‘First Claim’, with 70 defendants); CL-2018-000404 (the ‘Second Claim’, with 25 defendants); CL-2018-000590 (the ‘Third Claim’, with 8 defendants); CL-2019-000487 (the ‘Fourth Claim’, with 9 defendants); and CL-2020-000369 (the ‘Fifth Claim’, with 7 defendants). Allowing for overlap (some defendants were named on more than one Claim Form), in total 106 defendants were named (in error, I said 114 in my Revenue Rule judgment, referred to in paragraph 32(i) below).

At the date of this judgment, there are, on the defendants’ side of the litigation: 4 legal teams representing between them 30 defendants (subject to the point mentioned in paragraph 38 below); 14 individuals litigating in person; a limited liability partnership (Lindisfarne) represented at trial, with my permission, by its partners; 10 unrepresented corporate defendants; and 1 corporate defendant (Syntax) against whom judgment was entered in default, for an amount to be determined. They are the 56 defendants SKAT’s pleaded claims against whom fall to be determined, and are now determined, by this judgment (the ‘trial defendants’), subject again to the point in paragraph 38 below. I note for completeness that the default judgment entered against Syntax with which I am concerned was entered against it as the 22^nd defendant to the Second Claim. Syntax was also named as the 68^th defendant in the First Claim. That is one of the duplications referred to in the previous paragraph. The First Claim stands stayed against Syntax.

The remaining 50 defendants originally in these proceedings, claims against whom are not now before the court for judgment, were:

The first main CMC in these proceedings was in January 2020, before which I was appointed as designated judge by Teare J. At a further CMC in July 2020, I gave a case management ruling, [2020] EWHC 2022 (Comm), explaining my decision to direct three trials:

The Main Trial thus occupied 138 days over 33 weeks between 9 April 2024 and 10 April 2025. It ran generally as scheduled, sitting for 108 days between 15 April 2024 (Day 1) and 10 April 2025 (Day 108), with 30 reading days spread across the trial timetable, plus breaks for court vacations and for the parties to prepare closing argument after the completion of the evidence on (in the event) 2 December 2024 (Day 76). Given the number of parties to whom time needed to be allocated for argument and the number of points to be addressed, oral closing argument took 32 days, between 24 February 2025 (Day 77) and 10 April 2025 (Day 108), supported by written closing materials running to c.5,350 pages.

I am very grateful to all who participated in the preparation and conduct of the trial, helping me to plan its schedule in some detail and to keep substantially to that schedule when the time came. At the risk of saying something that is invidious by omission, I wish to record here my particular thanks for three types of contribution:

For the purpose of reaching my decisions and preparing this judgment, I have reviewed and considered the evidence adduced at trial and all the arguments put forward by the parties. It is neither necessary nor realistic for me to set out and deal individually with every point taken, and I have not attempted to do so. Nor do I descend in this judgment to anything like the level of detail the parties gave me in their written submissions. That level of detail contributed to the very great length of some of those submissions, which will have involved huge effort on the part of the limited number of well-resourced legal teams in the case. It was not as helpful as would have been a focused written presentation of the important findings of fact contended for, identifying for each the case pleaded, and then the evidence said to establish it and/or the facts, inherent probabilities or other matters from which it was said that it should be inferred. A key task of litigation advocacy is to help the court to see the wood said to be constituted by the trees, which is not usually done by devoting hundreds of pages to describing in close detail the leaves or bark patterns of individual saplings. With hindsight, I regret that I was persuaded by the difficulty of identifying how many pages should suffice not to impose page limits on the written closing submissions.

As I described the decision-making process in Kyla Shipping Co Ltd et al v Freight Trading Ltd et al [2022] EWHC 1625 (Comm) at [24], so also here, “My conclusions as to the facts follow from a consideration of all the evidence in the round, and all the parties’ submissions on the evidence, even if I do not mention or summarise all of that evidence or all of those submissions. It is rarely possible to do full justice to the holistic, iterative, self-critical and cross-checking nature of the process of assessing a case on the evidence, in an essentially ‘linear’ written judgment. Thus, for example, my assessment of the factual witnesses was informed by the plausibility of their evidence, and its consistency or inconsistency with the documentary record, as well as by the ability “which cross-examination afford[ed] to subject the documentary record to critical scrutiny and to gauge [the witnesses’] personality, motivations and working practices” (per Leggatt J, as he was then, in Gestmin SGPS S.A. v (1) Credit Suisse (UK) Ltd & (2) Credit Suisse Securities (Europe) Ltd [2013] EWHC 3560 (Comm) at [22]); but at the same time, my final sense of the plausibility of rival accounts on disputed matters, bearing in mind what is or is not in the documentary record, was informed by the personalities involved (and their motivations and working practices), the most important of which I had an opportunity to gauge through the trial process.”

Appendix 1 to this judgment lists the trial defendants. It identifies where each is to be found on one or more of the five consolidated Claim Forms, and it defines the terms I use to refer to them in this judgment. Taking Sanjay Shah, for example: he is identified in Appendix 1 as the 34^th defendant to the First Claim; the ‘Sanjay Shah Defendants’ or ‘SSDs’ are then identified as Sanjay Shah himself plus some 23 corporate entities; and the ‘Shah Defendants’ or ‘Shah Ds’ are then the SSDs plus Sanjay Shah’s wife, Usha Shah, the 36^th defendant to the First Claim.

Having mentioned the SSDs, it is convenient to explain now the point referred to in paragraph 30 above. The SSDs include nine companies that were struck off in their respective places of incorporation prior to the Main Trial, six of which remain struck off at the date of this judgment, all as noted in Appendix 1. The title page to this judgment says that the Shah Ds (which I have defined to include all of the SSDs) were represented by Mr Jones KC and his juniors, instructed by Meaby & Co. Strictly, that is not true for the corporate SSDs that still stand struck off, and it was not true during the Main Trial for those that were struck off at the time but have now been restored to the register (Colbrook, Ganymede and T&S).

What can or should happen to SKAT’s claims against any of the SSDs that still stand struck off, if restored to a register after the date of this judgment, is not something I have been asked to consider or could properly consider until the restored company had been given a reasonable opportunity to be heard; and I was told that, as things stand, there is no plan to seek to have any of those companies restored to its respective register.

As for Colbrook, Ganymede and T&S, after those companies were restored to the register in the Cayman Islands, Meaby & Co confirmed to the court in writing that they, and counsel instructed by them, now represent the companies. By that letter, on behalf of those companies and with the agreement of Pinsent Masons on behalf of SKAT, Meaby & Co confirmed that the companies and SKAT jointly wished the Main Trial as conducted between April 2024 and April 2025 to be treated as the trial of SKAT’s claims against the companies, even though they stood struck out at the time. Meaby & Co proposed, and I am content, that my recording all of that now in this judgment should be sufficient for the result under this judgment to be effective as between SKAT and each of Colbrook, Ganymede and T&S, and for any order made upon this judgment, so far as material, to determine SKAT’s claims against each of those companies.

Appendix 2 to this judgment identifies in very summary form the causes of action pursued by SKAT at trial and maintained in closing argument against the various trial defendants (other than Syntax). All those defendants had a proper opportunity to plead a case for, call evidence at, and otherwise participate generally in the Main Trial (subject to paragraph 38 above). The heading of this judgment identifies which trial defendants in fact participated. I was and remain satisfied that the case management of the litigation provided all defendants with a fair opportunity to participate, even if in some cases their financial or other circumstances were not ideal for full participation. I am satisfied that it is just for the court now to proceed, on the basis of the Main Trial as completed on 10 April 2025, to determine all live claims pursued by SKAT, even where that means for some trial defendants having proceeded in their absence, in their presence but without hearing from them, or where they were unrepresented but would have preferred to have had legal representation.

Pursuant to case management directions concerning how any case was to be raised to the effect that any of SKAT’s claims is not governed by or should not be determined under English law, all of SKAT’s claims fall to be determined under English law, albeit, of course, by reference to Danish tax law as it governs the question whether the tax refund claims impugned by SKAT were claims that SKAT was not obliged to pay. That had become the case by the time of oral opening remarks at trial except for SKAT’s claims against Mr Fletcher, the Jain Ds and the Godson Ds, who at that stage maintained a pleaded case that SKAT’s claims against them were governed by and should be determined under Danish law. As to that:

Invalidity

In this section, I explain why I concluded that SKAT was not liable to pay any of the 4,170 tax refund claims that give rise to the claims pursued in these proceedings. At trial, that was conceded, so far as material to the claims pursued against them, by the Shah Ds, the DWF Ds, Mr Klar and Lindisfarne. It was for SKAT to prove it, if it could, against all other trial defendants, although it was actively contested only by the Jain Ds, Godson Ds and Mr Fletcher.

It is convenient to explain how the trading worked in sufficient detail that what needs to be said about that for the purpose of later sections of this judgment has also been set out. This section of the judgment therefore goes further than would be required just to confirm the invalidity of the tax refund claims.

SKAT made claims arising out of three trading models, the Solo Model, the Maple Point Model and the Klar Model. The scale involved is indicated by the aggregate total paid out by SKAT, to which I have already referred, namely c.DKK12.1 billion (≈ c.£1.4 billion). Another indication of scale is given by the aggregate purchase price of Danish shares on the face of things contracted to be acquired, namely c.DKK1.9 trillion (≈ c.£220 billion), made up as follows:

It should not be thought, however, that those who designed the trading models had anything like that scale in mind when doing so. On 21 March 2012, when the Solo Model was being developed, Sanjay Shah forwarded to the DWF Ds at Solo a spreadsheet that had been prepared indicating the anticipated scope for the business as regards Danish shares. The email with the spreadsheet, from Edo Barac at Solo who had prepared it, expressed the view that “the returns are potentially very good”, and when forwarding it to the DWF Ds, Sanjay Shah endorsed that view: “mmmmmmm”. The spreadsheet envisaged trading 3% of the total issued volume of 13 Danish stocks and estimated a gross revenue from tax refund claims, if paid, of €19.1m (on an assumed WHT rate of 28%; at 27%, that would have been €18.4m). In the event, significantly larger volumes were traded, in aggregate, from the outset, but still nothing like the scale later seen in 2014/2015.

I aim to be precise and consistent in my terminology when describing the trading models. They were designed and implemented so that each trade would generate a payment credit referable to a Danish dividend in the account at a custodian of an entity on behalf of whom a tax reclaim agent would claim a tax refund from SKAT on the basis of inter alia a credit advice note (‘CAN’) issued by the custodian in respect of that payment credit. I shall use that term throughout; ‘dividend credit advice’ or ‘DCA’ was also frequently used, both at the time and at trial.

My aim to be precise and consistent does not mean that participants involved in dividend arbitrage (‘div-arb’) strategies, or cum-ex trading, were as precise, or were consistent, in their use of language. Nor does it mean that how I define or use terms is necessarily how such parties did so at the material time or might do so today. I should also make clear that where I use or explain a use of language that involves legal concepts or effects, they are the ordinary legal concepts and effects of securities trading, not (if different) those of Danish tax law, unless I indicate otherwise.

An immediate example of imprecision or variability of terminology concerns the two terms just mentioned, div-arb and cum-ex. It would be wrong to imagine that if a trading strategy were described as div-arb, or if a person said they had experience, knowledge or awareness of div-arb trading in the market, that would necessarily mean trading focused on the generation of a tax refund claim, or trading on a cum-ex basis. In general, arbitrage trading is the practice of taking advantage of pricing differences in two or more markets relating to the same asset. Economic theory has it that in perfectly efficient trading markets, arbitrage opportunities should not exist; but trading markets are not perfectly efficient. A simple div-arb strategy, then, might use options, acquiring shares and put options for a matching share volume on or before a dividend declaration date, and exercising the put options after the record date, having earned and collected the dividend. That would be an attempt to make an arbitrage profit out of spot market equity prices and options market prices relating to the same shares, with particular reference to pricing around a dividend declaration date.

A div-arb strategy of that kind would be unrelated to tax refund claims in respect of the dividend. It would be a classic arbitrage, looking to create a small margin out of pricing differences across markets. It is conceivable, I think, that there could be some complex relationship between the potential availability of tax refund claims to some equity investors and market prices for spot trades and/or options that in some indirect way connects tax refund claims to that simple type of arbitrage; but that is a very different point.

A structured tax trade under which a tax-advantaged buyer acquires shares so as to earn and receive payment of a dividend from the share issuer, suffering withholding tax on that income, before then selling the shares ex-div and claiming a refund, is not a simple price arbitrage strategy of that kind. However, the evidence at trial indicated that those involved in such structured tax trades might consider them and refer to them as (a species of) div-arb, although another term might be ‘tax arbitrage’.

Div-arb of that kind does not involve trading on cum-ex terms, under which a buyer contracts for a deferred settlement such that, if the trade is performed as per contract, the buyer will not acquire cum-div shares. Again, however, the evidence at trial indicated that those involved in structured trades using cum-ex, in a structure designed to facilitate and profit from a tax refund claim made by or on behalf of the cum-ex buyer, might consider and refer to them also as (a type of) ‘tax arbitrage’, or as another (sub-)species of div-arb.

The trading with which I am concerned always centred on a cum-ex trade by a tax-advantaged buyer with, in the Solo and Maple Point Models, price hedges, stock loans, subsequent unwinds, and settlement procedures, all coordinated so as to facilitate a tax refund claim on behalf of that buyer to a national tax authority like SKAT. As I describe below, the Klar Model had the same focus, but it was simpler, with no price hedges and not always using stock loans.

In equity trading, participants will use ‘sale / purchase’, ‘sell / buy’ and ‘sold / bought’ to refer to an equity trade that, strictly, is only a contract to buy / sell, under which no shareholding is transferred on the trade date. They will use those terms even though the question whether any shareholding is transferred to the buyer will depend on whether and if so how the trade is ‘settled’, i.e. performed.

That use of language is so ubiquitous and efficient that I adopt it, but it has the consequence, and this must be borne in mind throughout, that “V has today sold shares to B”, or similar, does not mean that any shareholding has been created in or transferred to B. It does not mean, without more, that any shares ever will be owned by or transferred to B. There are differences between different types of personal property, but at the basic level of that last proposition, a contract to buy Carlsberg A/S shares is not a shareholding in Carlsberg A/S any more than a contract to buy a pint of Carlsberg is a pint of Carlsberg.

I have laboured that slightly because it was apparent at trial that there may be those, possibly including some of the trial defendants, who would take the use of the language of ‘bought’ and ‘sold’ to the false conclusion that some kind of share ownership passes to the buyer the moment they execute a trade, i.e. enter into a contract to purchase shares. At that moment, however, all the buyer ‘owns’ is the seller’s promise to perform the bargain on the settlement date. Just as a contract to buy a shareholding in a Danish company is not a shareholding in the company, ‘owning’ such a contract is not share ownership. Whether the buyer ever becomes the owner of any shares in the Danish company will depend on whether, and if so how, the bargain is in fact settled (performed).

Messrs Jain, Godson and Fletcher made the point in their closing argument that in a modern system of recording shareholdings through custody chains (as to which, see further paragraph 61ff, below), whether or what true ownership rights can be said to exist is a complex question, because “Intermediation by the custody chain gives rise to the separation of a party who is economically or beneficially entitled to a security (the end investor or ultimate account holder) from a party who, under the statutory or contractual regime (which constitutes the security) is recognised as the holder of it, and therefore as entitled to exercise and enjoy the rights attached or flowing from it.” There is force in that point, but it does not detract from what I said in paragraphs 54 to 56 above. It goes to what may nowadays be a complex question of what precisely is ‘owned’ by a holding of shares constituted by a complete custody chain. Whatever may be the correct answer to that possibly complex question, such a holding is not created merely by entering into a trade under which a seller is obliged to cause the buyer to acquire (receive a transfer of) such a holding some days later.

So, for example, in Sample Trade Solo 1 (see paragraph 101 below for the Sample Trades), on 27 November 2012 Mill River Capital Management Pension Plan sold, at DKK187.5903 per share, 1,000,000 shares in CHR Hansen Holdings, for settlement on 3 December 2012. But that means only that on that trade date (27 November), Mill River contracted to acquire, on 3 December, a holding of 1,000,000 shares in CHR Hansen. The conclusion of that contract on that trade date did not transfer a holding of shares to Mill River, Mill River did not become in any sense a shareholder by concluding it, and whether Mill River became a shareholder as a result of it would depend on how, if at all, the contract was performed.

That will always be true for the trading considered in this judgment. By design, the sellers always held no shares, on the trade date or ever (and any direct or indirect sellers to those sellers likewise never held any shares). If the sales had settled in such a way that the buyers became shareholders, the more complex aspects of Danish tax law that in SKAT (Validity Issues) I found to exist might have been relevant to whether the buyers were liable to Danish dividend tax on a dividend declared after the trade date and prior to settlement, so as potentially to be entitled to a tax refund. Those aspects of Danish tax law do not arise, however, because, again by design, no buyer was ever going to become or ever in fact became a shareholder pursuant to any of the equity purchases, due to the way in which they were (treated as) settled.

For clarity, then, I need to distinguish between contracting to buy shares, on the one hand, and obtaining a holding of shares pursuant to such a contract, on the other hand. Since I use ‘buying’, ‘selling’, etc, for the former, for the latter, as in the preceding few paragraphs, I use ‘acquiring’ or ‘acquisition’ by a buyer, ‘transferring’ or ‘transfer’ by a seller.

Dematerialised, fungible shares in Danish listed companies can only be held as a shareholding recorded with VPS or with a custodian at the end of a custody chain leading to VPS, as I explained and sought to illustrate in SKAT (Validity Issues) at [154]-[163]. The diagram, below, is a revised version of the diagram at [159] in that judgment, adding cells to the right:

Illustrative Shareholding Diagram

The section of the revised diagram that features B, V, C4 and C5 illustrates the effect of vendor V, who holds no shares, selling X% of DanCo to buyer B, where B and V are both custody clients of custodian C4, and C4 is in turn a custody client of custodian C5. C4 holds no shares, since C5 holds no shares for it. C5 therefore also holds no shares so far as might be material to B or V, i.e. as sub-custodian of C4 indirectly through whom any shareholding of B or V would be held. I note for completeness that C5 need not be a different custodian to C1 or C2. For example, if C5 was C2, then it would have shares (a 30% shareholding), but not shares held for C4, and therefore not shares held via C4 as sub-custodian for B or V. That is illustrated by the shaded ‘barrier’ between C2 and C5 in the revised diagram.

The sale, assuming it is priced on the trade date, gives B and V immediate exposure, long and short respectively, to movement in DanCo’s share price, but it gives neither of them any shares. If account entries at C4 recording the sale from V to B purport to show X% of DanCo in a securities account of B’s at C4, prior to any completion of that sale, that does not make B a shareholder in DanCo. Any such record is not connected to DanCo via VPS, whether directly or via a sub-custodian or chain of sub-custodians. B may not realise that, if C4 produces an account statement purportedly recording that it holds X% of DanCo in custody for B, depending on what B knows or understands, if anything, of any wider set of transactions of which its equity purchase may form a part, and on whether it appreciates that concluding a purchase does not make it a shareholder. B’s possible ignorance that it is not immediately a shareholder in DanCo upon entering into its purchase with V does not affect the reality that it is not.

The other party in the new section of the diagram, L, is a stock lender, also a client of C4, that contracts to borrow X% of DanCo from B and to lend X% of DanCo to V, for settlement on the same settlement date as the sale by V to B. If the two stock loans and the sale are settled internally at C4, even though none of V, B or L has any shares to deliver, settlement date account entries at C4 might show X% of DanCo being:

In that way, the equity trade (V selling to B) and the stock loans (B lending to L; L lending to V) might be treated as having settled although none of B, V or L at any time has, acquires or transfers any shareholding in DanCo. It is easy to see that as a self-fulfilling settlement loop: B lends to L, so that L can lend to V, so that V can complete the sale to B. Mr Wade, the market expert instructed by SKAT, illustrated that with this diagram of the settlement transfers prima facie required to settle Sample Trade Indigo 1 (in which DWM Pension Plan bought from Maven Asset Management, Potala was the stock lender, and Indigo was the custodian):

In his oral evidence at trial, Mr Baker of Lindisfarne said that he did not think of this settlement process as a loop, because “I don’t see any start or finish in it, I see things concurrently”, and “there is a certain connotation with a loop that … nothing changes and everybody is the same … [but] this is different. I think people are taking on contractual obligations. I think there is risk there.” I agree with Mr Baker that there is no start or finish to the loop. To spell that out:

I do not think that makes it inappropriate or inaccurate to think of this as a loop. There is then, as Mr Baker said, a certain connotation that nothing changes, but as regards share ownership indeed nothing changes. At no point, before, during (if that is meaningful) or after settlement, does any of V, L or B own, acquire or transfer any shares. I agree with Mr Baker nonetheless that it would not be true to say that nothing at all has changed. The position after settlement is not identical to the position prior to settlement, or prior to trading. That is because if this internalised settlement by the custodian is performance, not cancellation, of the individual transactions, then V and L now owe to L and B, respectively, a stock borrower’s contractual obligation to transfer securities at the end of the stock loan that (other things being equal) will have to be covered in due course, and I come back to that in paragraph 74 below.

For such an internalised, share-less settlement to be a contractual performance, all of B, V and L would need to have agreed to it or to ratify it after the fact. But then, taking account of the agreement to allow that kind of settlement (or the ratification of it after the fact), what might otherwise have been a sale by V to B requiring a share transfer was in fact a trade that V could perform without any share transfer (and likewise for the stock loans).

Mr Baker’s observation that the settlement loop has no beginning is an important one for that issue about the contractual arrangements: prima facie, V’s sale to B cannot settle unless V has shares to transfer; V will have no shares to transfer unless its stock loan from L settles with a transfer of shares; but that cannot happen unless V’s sale to B settles with a transfer of shares. On the face of it, therefore, the equity sale and both stock loans should all fail to settle since no party has any shares to transfer. That is what a contract term or ratification as to settlement methodology would have to overcome.

The simplicity of the basic analysis I have now set out can be clouded by jargon. Talk of ‘traded positions’ in contrast to ‘settled positions’ (or ‘depot positions’) may cloud the difference between a contract to buy shares and a shareholding. Talk of whether a custody record represents a ‘settled balance of shares’ may cloud the reality that a custody record in the books of a custodian with no shares cannot be a shareholding, and cannot make the custody client a shareholder.

In his written evidence in chief, Mr Dhorajiwala noted that the idea at the centre of the transaction structures in this case, on the cash side, was that “it might be possible, as a custodian to create leverage synthetically via [as he put it] netting transactions between buyers and sellers of shares in the same omnibus account of a custodian”. I would prefer to say via the internal settlement of transactions, between buyers, sellers and interposed stock lenders, by equal and opposite account entries at a common custodian; but Mr Dhorajiwala was right to say that the transaction funding was synthetic, that is to say artificial. On the equity side, where all there ever were or would be were book entries in the accounts of a custodian with no shares, likewise any ‘shareholding’ evidenced was synthetic, not a holding of any shares.

To explain my quibble over terminology in the previous paragraph, I would now not say that an ‘omnibus account’ or ‘netting’ was used, and I acknowledge that means my own use of that language in SKAT (Validity Issues) was not as precise as it might have been:

I therefore disagree with Messrs Jain and Fletcher, who in one of their written submissions for closing said that “SKAT proposes that there was no delivery (gross discharge of primary obligation) if the net balance within an omnibus account at the end of the day was zero”. Firstly, that description treats ‘delivery’ (of a shareholding) and ‘gross discharge of primary obligation’ as one and the same, whereas SKAT’s allegation was, and my finding is, that the settlement methodology utilised here treated the primary obligation as discharged even though there was no delivery. Secondly, SKAT’s point was not that on the settlement date there was a nil end-of-day omnibus account balance (although no doubt that was true), rather its point was, and my finding is, that any omnibus account that existed at all was empty (devoid of any shares held by anyone for anyone) at all times.

In the trading models considered in this judgment, there was always a second phase of trading, an ‘unwind’ phase. Where stock lending had been used to settle the initial phase, there had to be an unwind because, as Mr Rabinowitz KC put it on Day 1, “otherwise the short seller is left with an obligation to return shares that it doesn’t own and the buyer is left with an obligation to return cash collateral that it doesn’t have.” It suffices to say, in broad terms, that:

In cross-examination, Mr Horn suggested that there was no requirement, when each set of initial trades was put on, for there to be an unwind set in the future. He was correct about that only in a very narrow sense, namely that the individual trade terms for the initial equity purchase and stock loans did not provide for it. However, in line with Mr Rabinowitz KC’s pithy summary, the practical reality was that stock loans would come to an end and the participating entities would have no means to perform the obligations then arising without a managed unwind of the type just described. Furthermore, it would not have been rational for those entities to enter into their respective initial trades unless they were confident that such an unwind would be arranged. Mr Horn’s attempt to quibble about that was not, in my judgment, because he was trying to be precise, making just the narrow point with which I opened this paragraph. He was trying to lay the ground for a line he intended to spin (and which he did spin the following day), in relation to Solo Model trading in Belgium, that the equity buyer was somehow free to do whatever it wanted with its entitlement to recall its stock loan, whereas in reality, as (I find) Mr Horn knew, the equity buyer was only ever going to recall the loan when told to do so by Solo, on terms determined by Solo, as part of Solo’s management of its trading model.

None of what I have said so far is affected by whether, or as from when, as a matter of market practice, a trading party might think of itself as ‘owning’ ‘shares’, or by what exactly they might think they meant by that if they did. Any market practice of that kind cannot convert book entries at a custodian that holds no shares into shareholdings.

It makes no difference of substance in paragraphs 61 to 76 above if, rather than a direct sale by V to B, there is a matched pair of sales, by V to C4 and by C4 to B (as illustrated by Mr Wade’s diagram reproduced at paragraph 65 above); and it then makes no difference whether, if that is the position, it is created by two trades concluded with C4 (V selling to C4, C4 selling to B), or by trades matched through a broker (V selling, B buying) where the broker gives the trades up to C4 for settlement, or at all events uses C4 as custodian for settlement of the matched trades. It is also no difference of substance if, where a broker is used, a different broker is used on the unwind.

Such additional complexity does not change the fact that, in the settlement loop described above, reversed on the unwind, even if, as a matter of contract:

In that simple description, only a single set of transactions is assumed, in which B buys and agrees to lend X%, V sells and agrees to borrow X%, and L agrees to borrow and lend X%. However, it again makes no difference of substance if, instead, “B”, “L” and “V” are sets of “buyer-lenders”, “borrower-lenders” and “borrower-sellers”, respectively, each of whose own positions is matched as to share volume, so that, for example, each buyer-lender, for its own part, contracts to lend whatever volume it buys, and all transactions are brought together for internalised, share-less, settlement at C4. Similarly, it makes no difference of substance if there are additional links in the transaction chains, putting greater distance between “B” and “V”, for example:

The more complexity of that sort that is introduced, the more challenging it may be to draft the contractual terms required for C4 to be entitled to operate a share-less settlement process equivalent to that which I have described for a simpler structure. But if that were achieved, still the upshot would be that: at C4, where everything balanced out, all transactions could be settled internally without any party ever having, acquiring or transferring any shares; even if under each contract, any seller or stock lender is treated as having delivered rather than defaulted, still the whole operation can and does occur in a world disconnected from VPS and DanCo, as illustrated by the diagram at paragraph 61 above, and thus involving no shares, that is to say, so far as shareholding (share ownership) is concerned, in a world entirely synthetic.

Sanjay Shah illustrated a more complex loop within a set of PowerPoint slides he prepared for use in his criminal trial in Denmark and adopted as part of his evidence in chief before me. The slides were proffered as an encapsulation of how Mr Shah says he understood things at the time. They included the following diagram to illustrate a settlement loop relating to a TDC dividend declared in March 2015:

The numbers in brackets identify the number of parties in each category. Thus:

Mr Shah acknowledged in his explanatory slides that “all depot positions and cash balances being zero at the start and end of trading (ie not “creating” or “destroying” money or shares) … accounts held by [the custodian, C4] with its sub-custodians therefore showed no balances.” Putting it more plainly, this was share trading by parties with neither shares nor money, settled at a custodian with neither shares nor money, by and through which neither shares nor money ever changed hands. (I put to one side the possible view, on the cash side, that in some sense money moved because simultaneous matched debit and credit entries in B’s cash account at C4 constituted or evidenced equal and opposite debts owed between B and C4 that instantly discharged each other (likewise for V and for L). No such argument could affect the conclusion that there were no shareholdings.)

All of the trading models involved activity materially similar, for my purposes, to that of B, V, L and C4 described above, in which (a) the equity purchase by B was cum-ex, (b) none of B, V, L or C4 ever held or transferred any shares, and (c) trades were settled internally at C4 by the share-less settlement loop method explained above. In fact, in every case, the cum-ex purchase was traded on the dividend declaration date for settlement on the dividend payment date, one business day after the record date for the dividend in question.

To be clear, I do not say that internalised settlement at a custodian necessarily means that no shares are ever held or transferred, only that internalised settlement was always used in this case in such a way that that was true. Messrs Jain, Godson and Fletcher argued that if the equity trades settled, there must have been a transfer of shareholdings between the parties. That is a false logic. A buyer who knew only that, according to its custodian, its share purchase had settled, might assume that a holding of shares had been transferred to it, especially if the buyer understood the custodian to be (as Messrs Jain, Godson and Fletcher put it) ‘plumbed in’ to the system for holding dematerialised shares via chains of custody; but that is a different point. Messrs Jain, Godson and Fletcher went on to submit that they never agreed to internalised settlement with no transfer of any shareholding, delivery versus payment. I agree with them, as indeed SKAT also submitted, that none of the written terms of business entered into between custodians and trading participants entitled the custodians to settle trades by the share-less internalised method in fact used. I also accept Messrs Jain, Godson and Fletcher’s case that nor was the share-less settlement method that was used ever agreed to by them in some other way. But whether it was or was not agreed to by all concerned, it was the method used in fact, with the result that no shareholding was ever acquired by the equity buyer.

In those circumstances, none of B, V or L could or did ever acquire a right to a real dividend, or receive a real dividend payment, i.e. a payment coming to them through a custody chain of payments by which a payment of dividend proceeds was made by the company to VPS for onward distribution to the company’s shareholders. If B, the cum-ex buyer, had a contractual right to a dividend compensation payment calculated by reference to the declared dividend, that would necessarily be a manufactured dividend, as I used that term in SKAT (Validity Issues) (see at [53], [181]).

That use of terminology was not fixed in the market. Some used ‘manufactured dividend’ only in a more limited way, to refer to obligations under a stock loan or repo, and may have used a different term for the sort of payments with which I am concerned. That different term, for some in the market, may have been ‘market claim’, although that term was far more commonly reserved for the reallocation of distribution proceeds to a party contractually entitled, i.e. (so far as material) to the passing on of real dividend payments. A cum-ex buyer, by definition, is not contractually entitled to distribution proceeds, because it has contracted not to be put in a position which could have entitled it to such proceeds.

That more common, narrower, meaning of the term ‘market claim’ in relation to European equity markets was influenced particularly by the “Market Standards for Corporate Actions Processing” promulgated by the European Corporate Actions Joint Working Group (the ‘CAJWG’), first published in 2009, re-published in 2012 and again in 2015 in identical terms so far as material (the ‘CAJWG Standards’). They defined a ‘Market Claim’ as a “Process to reallocate the proceeds of a Distribution to the contractually entitled party”, and they defined a ‘Distribution’ as a “Corporate Action whereby the Issuer of a security delivers particular proceeds to the holder of the Underlying Security without affecting the Underlying Security”. The making or processing of a dividend compensation payment of the type with which I am concerned is not the processing of a Corporate Action as the CAJWG defined it, for which the CAJWG publication was designed to set market standards, viz. an “Action initiated upon a Security by the Issuer or an Offeror”. It is therefore logical that the CAJWG defined a Market Claim as it did, such that it did not encompass dividend compensation payments of the kind I am considering.

I adhere to the finding that it was understood by market participants that only record date shareholders would receive a real dividend payment, and that a cum-ex buyer would not be a record date shareholder (see SKAT (Validity Issues) at [137]). I also adhere to what I said in that judgment at [146]-[147] and [182], given the definition of ‘manufactured dividend’ I was using; but I supplement it now by making clear that the use of that label was not fixed in the market. On any view, the sort of payment to which I referred at [147], and in the second half of [182(1)], would not be a real dividend payment. In this judgment generally, as I have just been doing, I refer to a payment from a cum-ex seller to its cum-ex buyer, required by contract to compensate the buyer for the fact that it will not receive any real dividend payment, as a ‘dividend compensation payment’.

It was also plain on the expert evidence, and that of factual witnesses at the Main Trial, in line with conclusions I stated in SKAT (Validity Issues), for example at [146]-[147] again, that anyone in the market would understand that what types of entitlement or payment might or might not incur Danish dividend tax, or entitle a tax-exempt party to a WHT refund in respect of such tax, would be a matter of Danish tax law, not a matter of market practice or market terminology. That would obviously include any question of what might or might not count as a ‘dividend’ for that purpose.

In SKAT (Validity Issues), at [269], I said, on the basis of the expert evidence at the Validity Trial, that it was the general practice of custodians to distinguish in CANs between real dividends and manufactured dividends (which, as I was then using that term, included dividend compensation payments). I do not adhere to that finding, as I explain in paragraphs 401 to 413 below. That does not affect the validity or otherwise of the tax refund claims that SKAT says it should not have paid.

In the descriptions I have given above, I have not mentioned the other major element of most of the trading models in this case, namely matched price hedges using exchange-traded futures or over-the-counter forwards put on by B and V as part of the initial phase, traded on the same day as the initial equity trade, reversed by opposite-direction matched futures or forwards as part of the unwind. Those hedges, where used, served a function in the trading models, but they do not affect anything in what I have said as to whether any of B, V or L ever held, acquired or transferred any shares.

At the heart of the validity issue was an idea stated in Sanjay Shah’s PowerPoint slides referred to in paragraph 81 above. One of those slides said this:

“Explanation of dividend compensation payments

Short sellers who sell shares cum-div and settle ex-div are liable to pay a dividend compensation payment to their custodians. …

Long buyers who buy shares cum-div and settle ex-div must receive a dividend compensation payment from their custodians. Such payments are documented as dividend credit advice notices. Under Danish Tax Law, these payments are considered dividend income for tax purposes.” (underlining for my emphasis);

and the slide before the diagram reproduced at paragraph 81 above asserted that “Even though shares and cash are not created or destroyed, the existence of the “legal loophole” results in long buyers becoming entitled to WHT reclaims due to the presence of short sellers”. The “legal loophole”, as Mr Shah called it, was a proposition he stated that, “Under Danish Tax Law, a buyer of shares becomes the legal and beneficial owner on the contract date, the date when he/she enters an irrevocable contract to buy shares.”

Applying the preliminary issue decisions made upon the Validity Trial, the proposition that a dividend compensation payment by a short seller who sells cum-div for settlement ex-div was considered dividend income for tax purposes under Danish tax law was wrong at all material times. It was wrong, if taken literally, because Danish dividend tax attached to the accrual of a right to a dividend, not to the receipt of a payment. It was still wrong if, allowing for that first point, we say the proposition is that a cum-ex buyer’s contractual right to a dividend compensation payment was dividend income for Danish tax purposes. Danish dividend tax did not apply to such a right. Whether at the time Mr Shah or others thought otherwise, or had any basis for doing so, does not affect the true position under Danish tax law as determined for these proceedings by the Validity Trial.

Under SKAT (Validity Issues), it is possible to describe a cum-ex trade (which may or may not be a trade anyone would ever do), where a seller that is short on the trade date completes the trade on the settlement date by a transfer of shares as a result of arrangements made after the trade was executed, but before the dividend was declared, such that (a) the only dividend-related payment the buyer receives is a dividend compensation payment from the seller, but (b) the buyer was the party liable to Danish dividend tax on the real dividend. That is a possibility because of the complexities I found to exist, to which I referred in paragraph 59 above. On the facts, however, the short seller was always short until after the ex-date, and always did not complete the cum-ex trade by a transfer of shares anyway.

In the diagram at paragraph 61 above, B might receive a CAN issued by C4 to reflect a payment credit on B’s account at C4 derived from B’s contractual right to a dividend compensation payment, although C4 would be under no obligation to issue a CAN to report that credit over and above recording it accurately in periodic account statements issued to B. Or in Mr Shah’s diagram (paragraph 81 above), the 71 Investors might each receive such a CAN. But that would not mean they had received or been entitled to a dividend (or a dividend payment) a right to which attracted Danish dividend tax.

In the actual trading activity, B was always an entity capable in principle of qualifying for relief from Danish dividend tax under a DTT between its tax domicile and Denmark. For the most part, B was either a USPF or a LabCo that could have been entitled to relief in full. In some of the Klar Model trades, B was a UK or Luxembourg entity capable in principle of being entitled to relief against tax above 15%. However, none of them had been taxed by Denmark, so none of them was entitled to any tax refund, and therefore in each case the tax refund claim that was made was an invalid claim that SKAT was not obliged to accept and pay.

The final conclusion to identify at this stage, as it follows from the analysis I have set out above, although the invalidity of the tax refund claims in the case does not turn on it, is that all the short selling activity from 1 November 2012 was in breach of Article 12(1) of the EU Short Selling Regulation (‘SSR’), which applied from that date: see the SSR (Regulation (EU) No 236/2012 of 14 March 2012), Article 48, for the commencement date. That is because:

If the sellers had been covered short sellers by virtue of Article 12(1)(a) or 12(1)(b) of the SSR who settled their sales by transferring shares to their buyers, there would have been scope for the buyers to have been entitled to a dividend tax refund under the rules of Danish tax law that I found to exist in SKAT (Validity Issues), supra. Provisionally, I consider the same would not be true for Article 12(1)(c). Its entire purpose, and effect, is to define limited circumstances in which a short seller will be considered to have sufficient cover for a trade that it ought not to be prohibited even though there is not, at the time of the trade, a final and binding right to acquire (in this case) a sufficient shareholding for the trade to settle. In SKAT (Validity Issues), however, I found that such a right in the seller was required for the contract accruals rule to treat the buyer as having become a shareholder for the purpose of Danish tax law. I do not need to make a final decision about that, however, because whether or not the sales here were covered short sales in terms of Article 12(1) of the SSR, they were not performed by any share transfer to the buyers.

To establish any given cause of action pursued against a trial defendant, SKAT had to prove, if it could, the necessary ingredients, as pleaded, of that cause of action, and any pleaded defence on which the defendant in question bore the burden of proof (e.g. time bar or contributory negligence, where either of those arose) had to fail. Without losing sight of that, a general allegation by SKAT was that defendants (a) must have realised that the equity buyers never acquired any shares, so (b) must have understood that those buyers could not be entitled to a dividend that might attract Danish dividend tax, and therefore (c) could not have thought that those buyers were or might be entitled to a tax refund from SKAT. Therefore, a main thread running through the trial, in submissions on the documentary evidence and in the trial evidence of the factual witnesses called by defendants, was the pursuit by SKAT, and the testing by defendants, of that allegation. Another main thread, in submissions on the documents and in the trial evidence of SKAT’s factual witnesses, was whether SKAT was misled, as it claimed to have been, in making decisions to pay out on the tax refund claims submitted to it. Inevitably with such a large trial and many different defendants, there were also specific points relevant only to some of the defendants and the claims pursued by SKAT against them. For the purpose of considering any and all points arising, however, the underlying invalidity of the tax refund claims submitted to and paid by SKAT was established.

Main Narrative

Under the case management directions for the Main Trial, SKAT’s claims were tried on the basis that sample trades, identified by a process that I was satisfied had been fair and sufficient for the purpose, were representative of the various trading models deployed (‘the Sample Trades’). Schedules were created and appended to the relevant CMC Order (in June 2023) that identified the Sample Trades and listed every claim instance, specifying for each one the category of trading model to which it belonged. The Order, in material part, was in these terms:

“… UPON the Claimant’s claims being claims relating to WHT Applications made by WHT Applicants as a result of trading structures and/or series of transactions referred to in the Particulars of Claim (without prejudice to the Claimant’s allegations that they involved sham trading) as:

•

the Solo Model as used in 2012-2013 (“Solo Model 12/13” cases) or the Solo Model as used in 2014-2015 (“Solo Model 14/15” cases), some Solo Model 14/15 cases involving multiple sellers contracting to sell shares through brokers to multiple buyers (“Sub-Variant 1” cases) and some Solo Model 14/15 cases involving multiple sellers contracting to sell shares through brokers to a single buyer (“Sub-Variant 2” cases);

•

the Maple Point Model as used in 2014 (“Maple Point Model 2014” cases), all involving either Indigo Securities (“Indigo” cases) or North Channel Bank (“NCB” cases); or the Maple Point Model as used in 2015 (“Maple Point Model 2015” cases), all involving either NCB cases, or Lindisfarne Partners (“Lindisfarne” cases) contracting as Custodian; or

•

the Klar Model involving Salgado Capital contracting as Custodian (“Salgado” cases)

…

9.

The issues in dispute in these proceedings shall be determined in the Main Trial on the basis that the sample trades identified in the … Sample Trades List … (the “Sample Trades”) are representative of the trading structures and/or series of transactions in Solo Model 12/13 cases, Solo Model 14/15 cases (including where applicable Sub-Variant 1 or Sub-Variant 2 cases), Maple Point Model 2014 cases (and more specifically within that category, as applicable, Indigo or NCB cases) or Maple Point Model 2015 cases (and more specifically within that category, as applicable, NCB or Lindisfarne cases), and Salgado cases, all as respectively indicated by … the … Full Trades List … .”

I mentioned Sample Trade Solo 1 in paragraph 58 above. It is one of the Sample Trades taken under that Order to be representative of Solo Model 12/13 trading. A fuller and by that Order representative description of the trading models, illustrated by summaries of specific examples from the Sample Trades, is set out in Appendix 3 to this judgment.

The 4,170 claims for a tax refund with which this judgment is concerned were made to SKAT between 20 August 2012 and 22 July 2015 using the ‘Form Scheme’, one of the schemes then in operation by which SKAT granted refunds of Danish dividend tax. Each claim referenced a stated number of shares in one of the companies listed on the OMX C20 Index. The 4,170 claims resulted in payments by SKAT totalling DKK12,090,844,948.33.

The Form Scheme involved the submission, with supporting material, of a standard form published by SKAT, to seek relief from and a refund of Danish dividend tax. Therefore, in each instance, SKAT received:

SKAT alleged that, given some of the content of those documents and the fact that they were submitted as part of a tax refund claim in respect of Danish dividend tax, the making of each of the 4,170 tax refund claims involved particular misrepresentations being made to SKAT, intended to induce and in fact inducing it to pay the refund claimed. The detail as to that is considered later in this judgment (paragraph 424ff).

I need to say a little more about the older Form 06.008, used in 84 of 4,170 instances. By June 2012 it was not available on SKAT’s website, which directed use of the then current Form 06.003. Form 06.008 stated by its n.1 that “The claim must be made in triplicate”, and I accept the evidence from recollection of SKAT’s Mr Nielsen that it comprised three materially identical pages, one for each of SKAT, the foreign tax authority and the tax refund claimant. I was shown single-page examples, each of which identified itself as either “Copy 1 (For the Danish Tax Authorities)” or “Copy 2 (For the Foreign Tax Authorities)”. It seems likely, therefore, that the third page was “Copy 3”, for the refund claimant. I do not accept evidence that Mr Nielsen and SKAT’s Ms Rømer both gave that Form 06.003 replaced some yet different, two-page form. The evidence each gave on that was unclear and confusing, and not consistent with the evidence of the other. Form 06.008 did not call for a CAN or anything similar to be submitted in support. It required information about “dividends from shares or participations” of which the refund claimant was “the owner / usufructuary⁽²⁾” (n.2 being an instruction to “Delete whichever is inapplicable”) and the Danish company was supposed to stamp and sign the Form itself by way of certification “that the dividends stated in column 7 were paid after deduction of dividend tax at the amount of DKK: ” (obviously calling for an amount to have been inserted before the Danish company signed). So that I will have been clear in what I say about the older form:

Goal Taxback Ltd (‘Goal’) submitted refund claims arising out of all three trading models (Solo, Maple Point, and Klar), in response to which SKAT paid DKK4.282bn of the DKK12.091bn with which I am concerned. CANs issued by each of the custodians who feature in the case are to be found among the tax refund claims submitted by Goal. The Goal cover letter stated: “Please find enclosed a tax reclaim form together with evidence of payment and tax deduction paid on the above client’s securities”. All 84 of the claims using Form 06.008 were submitted by Goal, in either the second half of 2012 or the first half of 2013.

Acupay Systems LLC (‘Acupay’) submitted refund claims principally arising out of Solo Model trading, but also a small number arising out of Maple Point Model trading. Acupay came to submit CANs issued by all of the Solo Model custodians (SCP, Old Park Lane, Telesto and West Point), and CANs issued by NCB. DKK3.544bn of the DKK12.091bn with which I am concerned was paid out by SKAT in response to Acupay claims. The Acupay cover letter stated: “Please find attach a reclaim application to obtain a full refund of Danish dividend tax for a [qualifying US pension fund / Malaysian company] within the meaning of the Double Taxation Convention concluded between Denmark and [the United States of America / Malaysia]”. It further noted that it enclosed a “Claim to Relief from Danish Dividend Tax Form”, one or more “Dividend Credit Advices”, a tax residence certificate from the relevant foreign country, and a power of attorney authorising Acupay to make the claim.

Like Acupay, Syntax GIS Ltd (‘Syntax’) submitted refund claims principally from Solo Model trading, but also some from Maple Point Model trading, with CANs from all four Solo Model custodians and some NCB CANs, but also some Indigo CANs. Syntax claims generated DKK3.044bn of the DKK12.091bn paid out by SKAT with which I am concerned. The Syntax cover letter stated: “please find enclosed a reclaim application form from a qualifying [US pension fund / Malaysian corporation] for a complete refund of Danish Dividend Tax that was previously withheld in relation to their investments”. It further noted that it enclosed a “Claim to Relief from Danish Dividend Tax Form”, one or more “Dividend Credit Advices”, a tax residence certificate from the relevant foreign country, and a power of attorney authorising Syntax to make the claim.

Koi Associates Ltd (‘Koi’) submitted only refund claims arising from Maple Point Model trading in 2015, involving NCB and Lindisfarne CANs, in response to which SKAT paid out DKK1.221bn of the DKK12.091bn with which I am concerned. The Koi cover letter stated: “please find enclosed a reclaim application form from a qualifying US pension fund for a complete refund of Danish Dividend Tax that was previously withheld in relation to their investments”. It further noted that it enclosed a “Claim to Relief from Danish Dividend Tax Form”, one or more “Dividend Credit Advices”, a tax residence certificate from the relevant foreign country, and a power of attorney authorising Koi to make the claim.

Examples of Form 06.003 and Form 06.008 appear in Appendix 4 to this judgment. I consider their respective content more closely, below, when judging whether any of the representations alleged by SKAT was made by the refund claims with which I am concerned. (It should be noted, for completeness, that the ‘54-…’ and ‘SKSK…’ reference numbers on the copies reproduced in Appendix 4, top right and bottom right respectively, are not original but come from the litigation.)

As indicated in Appendix 3, several custodians were variously involved in the different trading models: SCP in the Solo Model 2012/2013; SCP, Telesto, West Point and Old Park Lane in the Solo Model 2014/2015; Indigo and NCB in the Maple Point Model 2014; NCB and Lindisfarne in the Maple Point Model 2015; Salgado in the Klar Model. Every refund claim paid by SKAT and now a subject of these proceedings was supported by a CAN issued by one of those custodians. Any one refund claim sent to and paid by SKAT might have been for the aggregate of the amounts shown in several CANs submitted with a single Form. The 4,170 tax refund claims that between them form the subject matter of these proceedings are counted on the basis that one CAN equals one refund claim. The 4,170 tax refund claims, counted in that way, were thus constituted from fewer than 4,170 tax refund Forms submitted by the Tax Agents to SKAT over the period (the exact number does not matter).

CAN formats differed between custodians, but the format used by any one custodian remained constant over time, save that in the case of CANs issued by SCP, initially only a “Pay Date” for the referenced dividend was given, whereas later SCP CANs gave its “Ex Date” and “Record Date” as well. An example of a CAN from each custodian (and in SCP’s case, one example of each type) appears in Appendix 5 to this judgment. Again I do not summarise their content further here because I examine it in more detail below, when judging whether any of the representations alleged by SKAT was made (and again, the ‘54-…’ and ‘SKSK…’ reference numbers are not original). However, I do note now that the form of CAN used by NCB stated that it was “Not a tax certificate” and contained some detailed terms on a second page (or in hard copy, it may have been, the reverse side of a standard form), some of which may have been meaningful only in relation to German securities. I have not included that page of ‘small print’ in Appendix 5.

All trial defendants except Lindisfarne had at least some involvement in Solo Model trading or the use of its proceeds. Solo Model trading generated refund claims submitted to SKAT between August 2012 and July 2015 on which SKAT paid out DKK9.025bn. Each trial defendant was pursued by SKAT only for and by reference to the extent of their individual involvement, as alleged by SKAT. For example, claims were not made against the DWF Ds in relation to the Solo Model 2014/2015, the DWF Ds having left Solo in 2013; nor were claims made against them in relation to Solo Model trading in 2013 after their respective dates of departure from Solo. That had a substantial impact on the aggregate amounts in respect of which relief was sought against some trial defendants. For example, of the 4,170 tax refund claims paid by SKAT, 3,238 came from Solo Model trading, but only 286 of those (8.8% by number, 8% by value) came from Solo Model trading in 2012/2013.

Maple Point Model trading generated refund claims submitted to SKAT between May 2014 and July 2015 in response to which SKAT paid out DKK2.74bn. The trial defendants with at least some involvement in Maple Point Model trading were the DWF Ds, Mr Klar and Lindisfarne.

Klar Model trading, in respect of which Mr Klar was the only trial defendant, involved refund claims submitted to SKAT between December 2012 and May 2015 in response to which SKAT paid out a total of DKK321m.

The Solo custodians (SCP throughout, Old Park Lane from August 2014, Telesto and West Point from February 2015) were FCA regulated entities within Sanjay Shah’s Solo Group. There were three key Solo entities:

A fourth entity that featured heavily is Ganymede, incorporated in the Cayman Islands on 16 June 2010. At all times it formed part of the Elysium Group wholly owned by Sanjay Shah. Ganymede was not party to the Solo Model trading itself, but Sanjay Shah arranged for it to be the principal effective beneficiary of the proceeds generated by it, paid by SKAT. Ganymede was, in substance, a simple corporate incarnation of Sanjay Shah. The business concept was that Sanjay Shah, acting as and for Ganymede, was gatekeeper giving access to the GSS business to USPFs (and, later, LabCos) so they could trade with a view to profiting from tax refund claims made by Tax Agents on their behalf, and to the brokers and trading counterparties (short sellers, stock lenders and (from 2013) forwards counterparties) who would indirectly take a small share of that profit.

In the case of the brokers and trading counterparties, I find, their fees per transaction were not, as agreed by them, expressly conditional upon the success of a tax reclaim generated by the transaction. However, the practical reality was that the only source of funds from which they might be paid would be tax reclaims, if successful, and they were in fact paid as part of a coordinated exercise, under Sanjay Shah’s ultimate direction, of accounting for the proceeds of successful tax reclaims amongst the various parties involved.

SCP charged proportionately modest fees for its role, amounting to a tiny fraction of the amounts paid out by SKAT. The main price for the USPFs’ (and LabCos’) access to the GSS trading platform was agreement that the lion’s share of any successful tax refund claims made on their behalf would go to Ganymede. That gave Sanjay Shah effective control over what was done with that lion’s share of what SKAT paid out, including over who else might take a portion of it or benefit from it in some other way.

Reflecting the elements involved in the Solo Model trading, as described in Appendix 3, there were different categories of individuals playing various roles that, when they were all put together, made the Solo Model possible.

Firstly, there were individuals involved in designing, structuring, or coordinating the execution of Solo Model trading. They included, amongst the trial defendants, Sanjay Shah, the DWF Ds when at Solo, i.e. until March 2013 (Rajen Shah), June 2013 (Mr Horn), and September 2013 (Mr Dhorajiwala), Mr Bains (until August 2014), and Mr Patterson (after he joined Solo in 2013). Members of the GSS team, which was managed initially by Mr Horn and then by Mr Dhorajiwala, until their respective departures, and by Omar Arti from 2014, were necessarily involved in the execution of Solo Model trading. Such individuals included Adam Forsyth, Martin Ward, Jessica Spoto, Nirav Patel, Biljit Johal, Claudia Sidoli and Jason Browne.

Secondly, there were principals behind the USPFs and LabCos that traded as buyers and on behalf of whom tax refund claims were submitted to SKAT. There were many such principals, but they included (from the trial defendants) Messrs Godson, Fletcher, Jain and Preston. Other principals of USPFs and LabCos involved in Solo Model trading have been a major target of SKAT’s related proceedings in the US and Malaysia, respectively.

Thirdly, there were principals behind the short sellers. They included, from the trial defendants, Messrs Oakley, Mitchell, Körner and Murphy. Sanjay Shah introduced all four of them to GSS, as indeed he introduced most of those behind short sellers who participated in Solo Model trading, as ex-Solo employees or friends or associates of his. The exceptions were Paul Warner, who owned short sellers called Ceptorbay, Encorelite, Oakholley and Rhaltall, and the individuals behind short sellers called Aronex, Miralty, Wicklow and CFS Group (namely Rajiv Kumar, Satyendra Singh, Bijaya Swain and Jonathan Walton), although the trader for those short sellers, Richard Mills, was known to Sanjay Shah.

Examples of the connections leading to short seller introductions are as follows:

Fourthly, there were principals behind the stock lenders, including Messrs Smith, Klar and Körner from the trial defendants. All were known to Sanjay Shah and became GSS clients at his suggestion. They were friends/family (Bhupendra Mistry, a friend and IT consultant to the Solo group, married to a cousin of Sanjay Shah; Nailesh Teraiya, a friend of Sanjay Shah who was also Mr Dhorajiwala’s brother in law and later incorporated Indigo, one of the custodians in the Maple Point Model 2014), friends/business contacts (Messrs Smith and Körner), a former senior Solo employee who had been part of the initial discussions that culminated in the Solo Model (Mr Klar), and a relative of a former Solo employee (Pratul Shah, a retired accountant whose daughter worked for Sanjay Shah in Dubai).

Fifthly, there were principals behind the forward counterparties involved in the Solo Model for 2014 and 2015, including Mr Klar (again) and Ms Bhudia from the trial defendants. Again, all such principals were known to Sanjay Shah and were GSS clients at his suggestion. They were former Solo employees (Mr Klar and Rebecca Robson (involved as a forward counterparty only for 2015)), or friends of his (Ms Bhudia and Dai Griffiths) or of a Solo employee (Steven North, introduced to Sanjay Shah by Michael Smyth when at Solo, Mr Smyth later leaving Solo to set up LabCos for Solo Model 2014/2015 trading).

Sixthly, there were individuals who acted as traders for parties in the Solo Model trading, including current or former Solo employees Roger Lehman (who acted as authorised representative for 48 USPFs, including those of Mr Godson and Mr Fletcher), Rebecca Robson (who traded for LabCos connected to Mr Preston and others in 2014), and Jaiganesh Sethuraman (who traded for Mr Klar’s forward counterparties from February 2014).

Seventhly, there were the brokers involved in the Solo Model trades, some of whom were entities of Sanjay Shah’s friends or business contacts (e.g. Novus, Bastion, Mako, Sunrise, and FGC).

Eighthly, and finally, there were introducers, to Solo/GSS, of USPFs who went on to participate in Solo Model trades, including (from the trial defendants) Messrs Godson, Murphy, Fletcher, and Devonshire (although in the case of Mr Devonshire, his only direct introduction was of Mr Fletcher, to Mr Murphy, following which Mr Fletcher, and others introduced by him, became involved in Solo Model trades).

Sanjay Shah originally joined KPMG’s corporate tax department in 1994 as a graduate trainee, but left before obtaining accountancy qualifications to pursue a career in investment banking, joining Merrill Lynch in 1996. He worked there in a financial reporting team, then in a middle office role for a precious metals trading desk where he acquired knowledge of derivatives and financial products. In 1998, he moved to Morgan Stanley for a job with similar functional responsibilities in equity derivatives, which introduced him to div-arb trading. He left Morgan Stanley in 1999 (the year in which he and Mrs Shah married), moving to Donaldson, Lufkin and Jenrette (‘DLJ’) as an assistant trader for equity derivatives.

In late 2000, DLJ was bought by Credit Suisse First Boston which led Sanjay Shah to join a prime brokerage team. Over a period of three years or so, his role developed to that of dedicated div-arb trader, and in 2003 he moved to ING to become a senior trader there in an equity finance team. His work at ING covered equity derivatives, stock lending and hedge fund finance lending, and included dealings with Rabobank, to which he moved in 2007, joining as European Head of Equity Finance with a mandate to grow a European equities div-arb business. That was where he first encountered cum-ex trading, and his job at Rabobank was his last employment before setting up his own business under the Solo Capital name.

Rabobank withdrew from the equity finance business after the global financial crisis hit in late 2008, making Sanjay Shah redundant. To set up in business for himself, he incorporated SCL in 2009. His first recruits at SCL were Mr Smyth (a stock loan broker) and Messrs Rajen Shah and Klar (to act as structurers).

In 2010, SCL explored German cum-ex trading. It acted as investment manager for an Irish fund, Broadgate, which used cum-ex trades to generate a German WHT reclaim based on the Germany-Ireland DTT under which Irish entities were entitled not to be taxed above 10% on German dividends. Some of those involved in the Broadgate transaction were significant to the development and implementation of the Solo Model for Danish shares:

The Argre principals were wealthy, sophisticated, knowledgeable and experienced professional investors. Prior to establishing Argre:

The Broadgate transaction was very profitable for SCL. It received c.50% of the WHT refund net proceeds, from which it paid the other counterparties to the trade, including short sellers and stock lenders.

Argre and SCL collaborated on another German transaction in 2011, with a view to facilitating a WHT refund claim by or on behalf of the US tax-domiciled not-for-profit foundation behind the Ezra Academy in the Borough of Queens, New York. Funding of US$40m was created through a total return swap with Deutsche Bank as custodian. The transaction was loss-making because the WHT refund claim was withdrawn following questions from the German authorities, notwithstanding a tax certificate stating that the Ezra Academy had suffered WHT on dividends received. Deutsche Bank resigned as custodian on reputational grounds. The circumstances were explored in greater detail with some of the witnesses at trial, but they are peripheral.

I mention for completeness two other matters, also peripheral for my purposes, that arose from the German cum-ex trading.

Firstly, Solo had been given supportive legal advice on the German tax law treatment of that trading, by Martin Krause of Norton Rose LLP in Frankfurt. On 21 January 2011, he emailed Mr Klar, then still at Solo, to mention a legal article just published in which the author took the view that cum-ex trades created a substantial risk of prosecution as tax evasion. He said that the focus of the article was on a particular German court decision where a short seller sold and repurchased, never acquiring any shares because the delivery claims were netted on the settlement date, but that the author’s explanations were not limited to that scenario and would extend more generally to cum-ex trades. Mr Krause said that to the best of Norton Rose’s knowledge, “the article is the first publication of a Frankfurt-based lawyer taking a particularly negative view to the effect that cum-ex trades could result in criminal prosecution. In terms of reputation, this is an entirely new quality and cannot be disregarded. Rather, one has to state, that a new view among peer lawyers has been formed that condemns cum-ex trades.”

Mr Klar forwarded Mr Krause’s email to Sanjay Shah and Rajen Shah, adding as comment, “For fuck’s sake”, to which Rajen Shah replied, “No director is going to sign up if he refers to criminal prosecution in the tax opinion.” Mr Graham KC submitted in oral closing argument for SKAT that Rajen Shah was thus angling to get Norton Rose not to disclose the risk of criminal sanctions in any opinion. I do not accept that. As I read him, Rajen Shah was recognising by his response that Mr Krause would be bound to refer to that risk in any opinion and that, even if Mr Krause’s own view remained that German cum-ex trading was lawful, or at least could be depending on the detail, that would probably mean that no further German trading would be possible in practice.

That was an important development. I conclude, below, that when putting the Solo Model together, Sanjay Shah, Mr Horn and Rajen Shah did not govern themselves by whether it would generate tax refund claims that would be valid and not questionable under Danish tax law. However, in my judgment, as evidenced by Rajen Shah’s response to Mr Krause’s update, they had no interest in or appetite for anything that might be considered criminal.

Secondly, three of the Zeta Plans (see paragraph 151 below) engaged in German cum-ex trading in 2011, through Macquarie as custodian. In connection with that trading, each of them entered into an agreement with Acupay and SCL dated 1 June 2011 appointing Acupay as agent to secure tax relief from the German tax authority through the use of SCL’s “DTV” claim filer number, although SCL was not the Zeta Plans’ custodian on the trades. On 10 October 2014, Blackfriars Crown Court issued an order under s.32 of the Crime (International Co-Operation) Act 2003 requiring SCL to produce certain documents, as specified in the order, relating to the Zeta Plans. SCL responded by letter dated 12 December 2014 signed by Sanjay Shah as director. Mr Shah was cross-examined about the response, and suggestions were made that it was variously untruthful, or misleading, or unhelpful. However, no case was put that SCL had acted in breach of the order, indeed Mr Rabinowitz KC disavowed any such case and the cross-examination was conducted without even a copy of the order. In the absence of any case that SCL acted in breach of the order, I consider this episode not to be relevant even as to credibility.

In 2011/2012, Solo tried without success to find another bank willing to provide it with leverage for cum-ex trades. For example, mainly for reputational reasons, none of Brown Brothers Harriman, Bank of Ireland, State Street or BNP Paribas was willing to provide leverage or act as custodian for cum-ex trades that might be arranged by Solo. There was some London market sentiment that, due in some way to the Broadgate transaction, Solo was to blame for a widespread withdrawal of appetite on the part of major financial institutions to fund cum-ex trading.

On 9 April 2012, Sanjay Shah emailed Mr Markowitz of Argre to ask if he had any USPFs that could be used for trading equities and derivatives, and in a subsequent call, it was explained that SCP itself would be the custodian. The idea, appreciated within Solo by at least Sanjay Shah, the DWF Ds and Mr Klar (who was then on the point of departing Solo to go into business for himself), was to structure out any need for funding of any kind external to the structured trades themselves, by the use of internalised settlement at SCP, acting as clearer to settle the trades.

That was the magic ingredient identified within Solo that, as those working on this perceived it, could allow trades to be settled without access to external funding, or shares, i.e. with synthetic leverage, as Mr Dhorajiwala called it, and (equally, I would say) synthetic shares (see paragraph 71 above). There was no consistent body of evidence as to how widely it was made known within Solo even, let alone within the wider community of participants in the Solo Model, or later the Maple Point Model, that this was how trades would be and were being settled. In my judgment, the more powerful evidence was to the effect that it was considered the trade secret of those who designed and ran the trading that will not have been shared with anyone it was not thought needed to know.

I did not consider trustworthy, for example, claims made by Mr Horn that of course the Argre Principals knew all about it. In that evidence, which notably fell short of Mr Horn claiming that he had himself explained Solo’s settlement method to the Argre Principals, in my view Mr Horn was putting forward a line he liked for his defence that he saw no difficulty with the share-less settlement method, and believed that the resulting tax refund claims were or might be valid claims. If the very impressive likes of the Argre Principals knew what was going on, and did not object to taking part, so that line of argument would have it, Mr Horn could be reassured that it was an honest way of doing things. In my judgment, in line with evidence given by a number of the trial defendants, Solo had and deployed a plausible pitch that proved successful for recruiting participants, namely that they had a method of using stock lending at settlement to finance a trading structure so that the equity trade would settle and the tax-advantaged purchaser could be in a position to make a tax refund claim, the details of which were for Solo to take care of. That was not crazy or suspicious in concept; and if it was shown to work in practice, it was reason in principle for the willingness of participants to see the lion’s share of the profit generated through the tax reclaim going to those who were organising and implementing that trade. Moreover, as regards the Argre Principals in particular, in my view they would have been the last participants to whom Sanjay Shah would want the Solo method to be disclosed, as they had the experience, sophistication and contacts to replicate the method for themselves, depriving Sanjay Shah of his lion’s share of the profits (realised through Ganymede).

None of the Argre Principals was a witness in these proceedings, and though SKAT relied on some isolated passages in the deposition testimony of Mr Markowitz in its US proceedings, they did not include evidence that it was ever explained to him, or that he realised on some other basis, that a share-less model for settlement was being operated. Certain legal advice taken by or shared with the Argre Principals was in evidence, but it did not demonstrate awareness on their part that there would never be any shares or share transfers. On the evidence I had at trial, in my judgment SKAT’s submission that the Argre Principals should be found to have known of that was, in truth, an invitation to speculate that they sought to work out, and guessed correctly, how exactly Solo was settling Solo Model trades (and, therefore, how the Maple Point custodians, under direction from the DWF Ds, later settled Maple Point Model trades). I prefer and accept the Shah Ds’ submission that, as Sanjay Shah said in his evidence, the explanation of the trading given to the Argre Principals was limited in scope, such that:

The idea of doing things as they were done arose at Solo from discussions after Sanjay Shah had been told by someone at CACEIS Bank about its use of internalised settlement to settle transactions without the need for access to the asset, in the context of which Mr Klar mentioned a trade when he was at ABN Amro in which the Philippines branch bought government bonds from the London branch, selling short, and the purchase settled by the Philippines branch lending identical bonds to London on the settlement date, with the transactions settling against each other but being treated as having both been performed (equal and opposite gross ‘deliveries’), not cancelled. In evidence at trial that I accept, Mr Klar said that he cannot now recall the exact mechanism used, but (a) as I have just said, it meant the two trades were treated by ABN Amro as having been performed, not washed out, but (b) “nobody had to go out into the market to borrow Philippine government bonds”.

It takes only a moment to see that the trade Mr Klar described is functionally equivalent to a forward sale of government bonds by the London branch to the Philippines branch, for delivery at the end of what was in fact structured as a securities lending period (London borrowing bonds from the Philippines). The reason why it was structured as a mutually fulfilling sale and borrow-back (be it accounting, tax, regulatory, or as the case may have been) is a level of detail that Mr Klar does not recall (if he knew it at the time) about which, therefore, I am not in a position to make any finding.

As I have noted (paragraph 123 above; and see Appendix 3), one key element of the Solo Model was to have tax exempt clients that might be entitled in principle to the benefit of a DTT between their respective tax domiciles and Denmark. In the Solo Model 2012/2013 trades, the clients in question were all USPFs.

The first 30 USPFs to participate (the ‘Original Argre Plans’) were pension plans set up by the principals behind Argre, or contacts of theirs introduced to GSS following Sanjay Shah’s approach to Mr Markowitz early in April 2012 (paragraph 144 above), together with a few other USPFs introduced by other contacts of Sanjay Shah (the ‘Zeta Plans’). All of the Original Argre Plans were newly established entities with, so far as material, no assets. In late 2013, 10 further USPFs were taken on by GSS as custody clients and became involved in Solo Model trading, introduced by Mr Fletcher through Mr Murphy, Mr Murphy having been introduced to Mr Fletcher by Mr Devonshire.

All Solo Model trading in 2012/2013 (and again in 2014/2015) was planned and coordinated by Solo’s GSS team. The Solo Model could not have been implemented otherwise. The GSS team decided the terms and prices for the individual transactions in the initial and unwind phases that participants should trade. There was no option to trade on any different basis – the terms and prices chosen by the GSS team were not open to negotiation, the only trading decision to be made by the participating parties was whether to trade as proposed by the GSS team or not.

From the outset, that coordinated effort built in obfuscatory elements to hide the fact that many individual trades and their resulting tax refund claims were being generated centrally, in reality, by a single entity (SCP through the GSS team, for the benefit, primarily, of Ganymede). For example, the share volumes, and sometimes also the traded prices, varied a little across any given set of trades referencing the same dividend, to avoid them being all equal, which might look like coordinated trading. There was no independent decision-making there by any of the participants. They were given volumes to trade and prices to fix. The communication of those details was done largely off email to avoid creating evidence trails. That evidences, in my judgment, an instinctive expectation that what Solo was doing might be challenged if it was appreciated that indeed this was a coordinated trading scheme being run by SCP rather than SCP acting simply as custodian for trading conducted independently by clients and brokers.

Hence, for an illustration from the very first Solo Model Danish trade: 10 Argre USPFs traded between them 63.5m TDC shares and the relevant TDC dividend was DKK2.30 per share; there was no reason, except to serve that obfuscatory purpose, for the trading not to be 10 x 6.35m, leading to 10 tax refund claims each for DKK3,943,350; instead, 10 different volumes were traded, ranging from 5.5m to 7.75m, that did not in any meaningful way reflect requests from or choices by the USPFs (e.g. by the California Catalog Company Pension Plan that it wanted ‘only’ 5.5m shares, or by Michelle Investments Pension Plan that it wanted to go big and buy 7.75m shares).

Efforts by some of the defendants (including Sanjay Shah and, notably again, Mr Horn) to pretend in their witness evidence that there was independent trading beyond decisions to say yes to whatever Solo put in front of participants were rather lame, and did not withstand cross-examination. Likewise refusals to acknowledge the obfuscatory purpose of the otherwise unnecessary details, like variation in volumes or prices, or for that matter like having as many USPFs as possible, to spread the overall volume of activity across lots of different putative tax refund claimants.

The ingredients assembled, in the form of the traded transaction terms, did not blend completely harmoniously to a nil end result in Solo Model trading. This is the complex wrinkle on the cash side to which I referred in paragraph 74 above. The traded terms of the equity price hedge (cash-settled exchange traded futures in Solo Model 2012/2013 trading), as put on at the outset and closed out on the unwind, and the stock lending cost (cash collateral interest less stock lending fee), did not create such a perfectly balanced hedge that the amounts credited to and debited from the cum-ex buyer over the full life of the structured trade aggregated to nil. Rather, there was an overall net trading profit or loss to the buyer, on paper, and an equal and opposite net trading loss or profit to the short seller, on paper, leaving any tax refund claim amount out of account.

This wrinkle was dealt with in Solo Model 2012/2013 trading by retrospectively amending the stock lending terms either on the same trade, or on previous and otherwise unrelated trades, so as to create offsetting losses or gains for the long buyer (and their opposites for the short seller), so that, as intended by Solo, the only profit element anywhere in the structure would indeed be the tax refund amount to be claimed from SKAT. Successful such reclaims would be the only source of funds out of which any of the parties to any of the Solo Model trading transactions could or would ever be paid anything.

The Solo Model was not only deployed in respect of Danish dividend tax. A number of other jurisdictions were also considered, and Solo Model trading was in fact undertaken with reference to Austrian and Belgian shares. The Belgian trading gave rise to an episode on which SKAT cross-examined Sanjay Shah. It was not pleaded by SKAT as any part of its case against Mr Shah, so I treat the cross-examination as no more than a matter going to Mr Shah’s credibility:

I mentioned above that by late August 2013, Mr Horn had left Solo. His was not the only significant departure in 2013. Rajen Shah and Mr Dhorajiwala also left, following which they developed and implemented the Maple Point Model along similar lines to the Solo Model (see below). Solo’s Head of Compliance, Gary Pitts, also left, and he does not feature in the case after 2013.

In Rajen Shah’s case, his departure was triggered by Sanjay Shah’s failure to agree promptly to the withdrawal of Austrian tax refund claims derived from Solo Model trading following an email on 19 March 2013 from LeitnerLaw (‘Leitner’), tax law advisors in Vienna. Leitner had given advice as to Austrian tax law in 2012 that had been treated within Solo as supportive. They now wrote to Rajen Shah, unsolicited, to report “negative developments” for cum-ex trades, namely that “in some cases, no withholding tax refund has been granted and some issues have been raised with regard to the fiscal criminal rules”. By email a few days later, 22 March 2013, Rajen Shah asked for a ‘self-declaration’ form and any other documentation required to withdraw a tax refund claim. Leitner responded on 26 March 2013 with a written memo, indicating (in substance) that while in principle it should be permissible to withdraw a tax refund claim, the impact of doing so on any criminal charge relating to having made it in the first place was complex. A telcon followed on 28 March 2013, in which I infer Leitner was asked to draft a letter for the withdrawal of a tax reclaim, because on 1 April 2013, Rajen Shah by email asked Leitner to “forward the draft letter to withdrawn [sic.] the reclaim as soon as you can”. Leitner replied on 2 April 2013, saying that they were working on the matter, both senior partners for tax criminal matters were out of the office (it was Easter week), and they would revert the following week. Rajen Shah was not satisfied that Sanjay Shah would ensure, or allow, that any pending Austrian tax reclaims be withdrawn. He resigned from Solo with immediate effect on Friday 5 April 2013.

In the event, Leitner did not provide a draft tax reclaim withdrawal letter, instead concluding that they were conflicted and could not continue to assist SCP. Their email to Solo dated 9 April 2013 reporting that conclusion stated that without a detailed consideration of the actual facts and circumstances of whatever trades had been done, they could not exclude adverse tax consequences (refusal of a pending reclaim, claim to recover reclaims previously paid) or criminal issues (monetary penalties, theoretically imprisonment). What mattered, Leitner said, was exactly how Leitner’s original advice had been followed in the trading, if it had been; and a simple, informal withdrawal of the tax refund claim, without giving reasons, “may not be sufficient to avoid tax criminal elements” (if there were any, that is). Leitner recommended someone at Deloitte (Austria) who specialised in criminal tax matters, if Solo wanted further assistance. Now in Rajen Shah’s absence, Mr Horn forwarded that email to Mr Dhorajiwala, but no action was taken arising from it.

Turning then to Mr Pitts, on 16 July 2013, by email to Sanjay Shah, copied to Mr Dhorajiwala, he recommended Solo take advice “on the appropriateness of US based GSS clients entering into transactions that involve clearing on BClear and Eurex”. He did so because, as Chief Compliance Officer for SCP, he was uncomfortable signing off on the then recently recruited new USPF clients, since Solo Model trading at the time involved single stock futures being cleared via JP Morgan through Eurex, which might be unlawful under US law because Eurex was not listed as an approved means for interstate commerce by the US Commodity Futures Trading Commission (‘CFTC’). There was effectively an exemption if the USPFs were Qualified Institutional Buyers (‘QIB’); but Mr Pitts thought there was a risk that they were not, which in turn meant there was a risk that Solo was assisting US persons to circumvent US law and putting JP Morgan in a position where they were helping US persons breach US regulations and the rules of Eurex. Since no external legal advice had been taken before taking on USPF clients, Mr Pitts advised that SCP management was exposed to a risk of regulatory censure.

Mr Pitts’ email also noted, fairly, that no one law firm had been asked to opine on the Solo Model in its entirety. Some advice had been taken, but it was on particular points put to different lawyers from time to time. He recommended a full review of the model by one firm before any more extensive activity was undertaken. No such full review was ever commissioned. Cross-examination of Sanjay Shah as to that satisfied me that he had at the time, and can offer today, no good reason why not.

Focusing on the US law / QIB question, Sanjay Shah’s immediate response was to resist “burning money on legal fees at the drop of a hat”, suggesting that Solo had enough in house expertise for most issues and should recruit rather than outsource if that was not true. Mr Pitts responded inter alia that US securities law “is an ugly sprawling mess that is hard to navigate and has lots of pitfalls for the unwary”. He said he was likewise not fond of spending money on external lawyers, but advised that “if you or the firm are challenged about a course of action that has been taken, the only robust defence if we have got it wrong is that we took appropriate external counsel (as external counsel is not deemed to be under the delivery pressure that might be placed on a GC or CCO).”

Mr Pitts followed his own recommendation and took advice on behalf of SCP from Jacob Preiserowicz of Schulte Roth & Zabel LLP (‘SRZ’) in Washington, DC. That advice confirmed Mr Pitts’ understanding that US clients who did not meet the QIB definition would be acting in breach of US securities laws, facilitated by SCP; and it explained that a QIB had to own in aggregate at least US$100 million in securities. Mr Pitts recommended that onboarding of new clients should cease until the point was resolved and that any open trades should be unwound. Sanjay Shah did not accept that recommendation. He spoke to Mr Preiserowicz and declared himself unimpressed by him. He told Mr Pitts he regarded the QIB case as still open.

On 5 August 2013, Mr Pitts therefore sent Sanjay Shah draft instructions by which further advice might be sought from US counsel. They noted that when a USPF was taken on, it “might not” have US$100m in assets (in fact, more accurately, in the Solo Model it was certain not to have any substantial assets when approved for trading), and described SCP’s idea (this having been Sanjay Shah’s QIB theory) thus: “However, through the leverage facility of our custody platform, the [USPFs] will take on positions in excess of this size and thus become eligible to be considered QIBs, at which point we would like to categorise them as QIBs and retain this categorisation for them as their holdings would fluctuate above and below this level.” As Mr Pitts explained to Sanjay Shah when he queried it, the use of leverage obviously needed to be mentioned. Mr Shah’s theory depended on (a) the prima facie implausible, circular thought that a regulatory requirement to qualify for trading defined by a client’s assets might be satisfied by the value of the assets they would trade if they qualified, and then (b) an entitlement for that purpose to value assets gross of any leverage involved in ‘acquiring’ them.

The draft instructions also noted that the USPFs’ trades would clear at SCP “by netting off equal and opposite transactions in its omnibus custody account”. Sanjay Shah also baulked at the idea of mentioning that, asking why external counsel would need to know about it. Mr Pitts explained that “the advisor might assume that we put the assets in custody in the traditional model – this might have an impact on the advice given – if someone talked to me of custody and clearing I would certainly not have our model in mind. In the UK, the difference between our model and a traditional custody model would definitely have an impact on any advice I gave.”

In the event, Mr Pitts’ draft instructions were not used, and Solo Model trading continued on Sanjay Shah’s say so, contrary to Mr Pitts’ recommendation to pause activity until the US law / QIB issue was bottomed out. Sanjay Shah took some advice from Dechert LLP in New York in August 2013. Mr Shah first put the following to Dechert, as to part (a) of his theory (paragraph 166 above): “After some digging, we believe we can clear [European single stock futures] for QIBs. We found that a QIB can be an employee benefit plan defined by Erisa, as long as it manages $100mm of assets or more. The clients we deal with are 401k pension plans, and they would have over $100mm under management. I would like to know if these plans would be considered to be QIBs.” I think no lawyer reading that as a question for advice would guess that Mr Shah had in mind USPFs that would be signed up with SCP for trading when they had no or minimal assets, with the ‘qualifying’ US$100m in asset value then being the nominal value of assets traded through SCP, an integral part of that trading being the single stock futures that gave rise to the QIB issue. In response to the question put thus, Dechert advised that the USPF described would be a QIB.

At Mr Dhorajiwala’s prompting, Sanjay Shah replied, so as to test part (b) of the theory (paragraph 166 above), with “A quick question: is the $100mm test on gross or net assets? If the client owns $100mm of equities that were purchased through use of leverage, then do we deduct the leverage, or just look at the gross asset amount? In the event that the legislation isn’t specific on this point and open to our interpretation, can I assume we can use gross assets if we wish?”. Again, I think, any lawyer reading that would have assumed the assets in question existed independently of the trading SCP was considering that gave rise to the need to clear single stock futures and therefore the QIB issue, and not (synthetic) assets created only by that very trading. Dechert responded that the language of the relevant rules was not conclusive, but it was reasonable to conclude that a gross asset value test applied.

Given the way Sanjay Shah had framed his questions to them, Dechert’s positive advice could not reasonably have been understood to cover what SCP was doing through the Solo Model.

The QIB issue generated by clearing single stock futures on an exchange would not arise after 2013, as they were not used as the price hedge instrument in the Solo Model for 2014 or 2015. It is speculative to say whether Sanjay Shah’s mishandling of the issue was a resigning matter for Mr Pitts, who departed Solo thereafter. The concerns he highlighted in the work plan and handover notes he left as he went were bigger picture issues of corporate governance and compliance. For example, he thought there was uncertainty over transaction reporting, a potential for market abuse behaviours or fraud within the GSS business because Adam LaRosa traded for so many USPF clients under powers of attorney, and still no unified legal opinion “on the GSS platform as a whole”, that might be treated by the FCA as “fundamentally poor governance (perception of poor product development, no collegiate approach involving all stakeholders, apparent drive for income at all costs).” Or again, Mr Pitts was concerned that there was no documented business strategy at Solo and the Management Committee’s role was in reality limited to approving decisions made by Sanjay Shah.

As noted in Appendix 3, Solo Model trading in 2014/2015 evolved from the 2012/2013 pattern through four main developments, but the basic concept and method was the same in substance.

First, the price hedge became an OTC forward rather than an exchange traded futures contract. This followed the resignation of Solo’s futures clearers and its inability to find another clearer willing to act. In Solo Model 2012/2013 trading, JP Morgan cleared matched pairs of single stock futures through the London Stock Exchange (‘LSE’). On 16 May 2013, a Compliance Officer at JP Morgan emailed Mr Pitts at SCP to inform him that JP Morgan had been contacted by the LSE about the matched and crossed business submitted by JP Morgan in April 2013, that is to say in fact some of the pricing hedges from Solo Model 2012/2013 trading. The LSE’s enquiry was said to be a routine matter, but as a result the JP Morgan Compliance Officer wanted to understand the rationale behind the futures trades it had cleared for SCP.

On 17 May 2013, Mr Pitts forwarded the JP Morgan email to Mr Dhorajiwala and Mr Forsyth, copied to Mr Horn, asking for input and giving his view on what was probably happening: “It looks like some of the crossing business is triggering market abuse flags at the exchange.” In a prompt reply to all, adding Sanjay Shah as well, Mr Dhorajiwala provided the trade details Mr Pitts had requested, courtesy of Mr Forsyth, and said he had just had a call from JP Morgan, leading him to propose: “Can we get together to discuss? I’ll skype you all in 5.”

The upshot was a reply to JP Morgan on 22 May 2013, sent by Mr Pitts but drafted by Mr Dhorajiwala and approved by Sanjay Shah, stating that:

“As part of our [GSS] business, [SCP] provides clients with a futures clearing service through an omnibus client account held with JPM. Our client base is purely institutional and made up primarily of brokers and institutional investors. The commercial rationale behind the business is to charge clients a premium for a facility to clear through a Global Bank such as JPMorgan where they would not normally have direct access to such a provider. The business has been running for over a year now and continues to be profitable. Significant resources have been deployed to build adequate control systems and risk is closely managed by ensuring that we only clear equal and opposite trades crossed by a third party broker.”

That was a misleading description of the Danish cum-ex trading and the part played in it by the matched futures that SCP was getting JP Morgan to clear. Mr Dhorajiwala, Mr Horn and Sanjay Shah all knew that it did not give JP Morgan a fair or accurate description of the business rationale of the futures, which were in fact a price hedge on the equity leg of an aggressive cum-ex tax arbitrage strategy coordinated by SCP. I find that it was put forward to JP Morgan to avoid revealing that that was the truth of it, as they envisaged that JP Morgan would not wish to be connected to that type of activity.

The relevant enquiry will have been triggered by the fact that at the Exchange, matched ‘buys’ and ‘sells’ will have been observed, posted by JP Morgan for the same JP Morgan client account (i.e. SCP’s account). I accept evidence Mr Dhorajiwala gave that therefore, and as was explained to him in telephone discussions with JP Morgan, the key point for them was whether there were separate interests buying and selling, sitting behind SCP, for whom in turn JP Morgan was acting as General Clearing Member on the Exchange. The question in the JP Morgan Compliance Officer’s email, however, was not so narrowly framed; and the knowingly false answer is not excused by the fact that the question it addressed might have gone beyond the narrow point that triggered an alert.

I also accept evidence Mr Dhorajiwala gave in re-examination that through other lines of correspondence with JP Morgan, it was made aware that the SCP clients trading futures were not banks or substantial investment funds, as calling them “institutional investors” would be likely to convey. I cannot say on the evidence whether that discrepancy was identified by JP Morgan, or whether it otherwise identified or suspected the truth about the futures trades it had been clearing for SCP.

Whether it was satisfied by or had concerns over the business rationale provided by Mr Pitts, as to which I make no finding, in the event, on 1 August 2013, JP Morgan gave notice of the termination of its custody and clearing agreement with SCP.

SEB took over the role as futures clearer but again concern as to the underlying business surfaced. On 5 August 2013, a representative from SEB’s sales and marketing team emailed Ms Spoto at GSS to say that SEB’s risk team had questioned the fact that there seemed to be “no volume traded or open interest on the SSF [single stock futures]”, and so asked if it was possible “to give a bit of background to give risk some comfort”. That email was copied to, among others, Sanjay Shah and Mr Dhorajiwala. Mr Shah replied that “For single stock futures, there is typically no open volume at the exchange. Our clients will be buying and selling OTC with each other, and the exposure will be no more than 50k lots at any one time. The buying and selling interest won’t be shown to the open market on screen, but will be reported to the exchange as they are crossed.” (The immediate context of this exchange was a trade in relation to Belgian shares, but by its nature the concern over what underlay the matched futures trading was not specific to that jurisdiction.)

Two weeks later, there was an exchange of emails between SEB and SCP (for whom Mr Dhorajiwala took the lead) about whether, once crossed, the equal and opposite futures could be closed out, as Solo had requested, or had to be left open. SEB’s Head of Sales and Client Services took the view that, assuming there were two separate underlying clients on opposite sides of the futures trade, then the trade should not be closed out just because there was a complete match. Closing out, in their view, caused misreporting to the Exchange as SEB were then “not reflecting to the exchange the existence of one long and one short position in this contract”. Mr Dhorajiwala yielded to that view, saying on 2 September 2013 in relation to the contracts then most recently traded that “We will continue to leave them open so long as clients wish to keep their open interest position”. This was misleading spin, making it appear as if the two sides of the position might be acting, and might act in relation to their open positions, other than in tandem, as orchestrated by GSS.

On 28 November 2013, SEB’s compliance department forwarded to SCP, by email addressed to Sanjay Shah, among others, an incident report from the Exchange flagging the possibility that futures trades SEB had cleared for SCP represented a ‘wash trade’. The email said that SEB’s own monitoring system had also generated alerts for the relevant trades. SCP was asked to confirm that the buy and sell sides of the trades were executed on behalf of different clients and that those clients had different underlying interests. Sanjay Shah confirmed both elements by reply email the same day.

I infer that SEB’s risk management or compliance concerns were not alleviated entirely, because on 13 February 2014 SEB informed Sanjay Shah by telephone that it was not able to act as clearer for single stock futures crossing.

SCP also began a relationship with Citigroup Global Markets (‘Citi’) for futures clearing, developed from around September 2013, and Citi began clearing futures in relation to Danish shares in mid-November 2013. But on 26 November 2013, Citi asked a number of questions about SCP’s business model, including whether the trading was around or close to dividend record dates and whether the purpose was to make tax reclaims. In due course, this led Solo to share a draft legal opinion that might be provided by Hannes Snellman, the Danish firm to which I refer below, to try to get Citi comfortable with the trades. However, on 21 January 2014 Citi explained that the draft opinion “lack[ed] any description of the economic reasons for engaging in such purchases and sales of futures over the record date”, such that it was “of no help in evaluating the tax or franchise risk to Citi of these transactions”. Citi was not persuaded to change its view following a telephone discussion and it did not clear any further futures for SCP. Sanjay Shah was closely involved in this, and I do not accept evidence he gave in cross-examination claiming otherwise.

The factual basis given to Hannes Snellman and recited in the draft opinions included inaccurate descriptions of important elements of the Solo Model. In my judgment, that was deliberate on Mr Shah’s part because he did not want to disclose to Citi that this was coordinated cum-ex trading, organised centrally by Solo’s GSS team for the purpose of facilitating tax refund claims by the equity buyers and having no business rationale other than the creation of such claims. Thus:

The SSDs put a false spin on this interaction with Citi in their Defence. It was claimed that the Solo Model had been disclosed to Citi and that Citi had identified it as a competing business, causing it to decline to continue clearing single stock futures for SCP; and that was then alleged to have “reinforced Mr Sanjay Shah’s genuine and honest belief that the trading structures deployed in the GSS business were genuine, legitimate, would result in the [equity buyers] becoming shareholders in Equities for Danish tax purposes and permit [them] to make valid [tax refund claims]”. I consider that to have been all fiction.

The difficulties in finding an external clearer prompted the switch from futures to OTC forwards for the 2014 dividend season as OTC forwards did not need to be cleared by an external bank.

The second evolutionary change in the Solo Model for 2014/2015 was the use of additional parties, creating more links in the transaction chains and greater complexity, as described in Appendix 3. There was no legal or business reason for any of these changes. They represented an escalation of the obfuscatory effort to which I referred in paragraphs 153 to 155 above, to facilitate an escalation of scale in Solo Model activity. Those changes did not alter the reality, however, that the trading continued to be centrally coordinated at Solo, and constructed so that it would settle to zero in the absence of either cash or shares, all with the sole purpose of generating tax refund claims.

In cross-examination, after initial prevarication and unconvincing invention, Sanjay Shah conceded that having multiple Solo custodians with different addresses, and different-looking CANs, was an exercise in creating obscurity, but he insisted that he was not concerned to obscure anything from SKAT:

“MR RABINOWITZ: … So opacity was the order of the day Mr Shah, yes?

A: As regards competitors, yes.”

I do not accept that the concern here was competition. For 2014 and 2015, Mr Shah did have an eye on competition, from the DWF Ds and their Maple Point Model trading, but I do not consider that competitive activity to have formed part of the motivation towards increased complexity whereby to obscure what was happening. I accept submissions by SKAT, and find, (i) that the purpose behind all the complicating features in Solo Model 2014/2015, not just the multiplicity of custodians, was obfuscation, and (ii) that the only purpose for that deliberate obfuscation was to reduce the chance that SKAT might wake up to how much tax reclaim business was coming its way from a single effective source, i.e. Sanjay Shah’s GSS business, or that the FCA might take a dim view of such activity and intervene to stop it. However, I do not accept SKAT’s submission that that was in turn driven by a belief or understanding that a fraud was being practised on SKAT.

Nor though do I accept Mr Shah’s claim that he thought he was cleverly using a legal loophole in Danish tax law. In my judgment, he thought the tax refund claims might well not be valid claims, so that if SKAT ever challenged them or stopped paying that would be an end of the very lucrative business he had developed, and he assumed that SKAT would challenge claims or stop paying if the centrally coordinated nature of the Solo Model business became apparent to it (although, in the event, that was not said by SKAT in these proceedings to have invalidated the claims).

To facilitate the use of these more complex structures, additional counterparties were recruited, for example further short sellers, further stock lenders, and now forward counterparties. They were recruited in part from existing participants in the Solo Model and in part from new participants. For example, in 2014, Messrs Oakley and Mitchell, and Dilip Shah, each incorporated three new short sellers, Mr Murphy incorporated four new short sellers, Mr Smith incorporated six new entities to act as stock lenders, and Mr Körner brought new companies to participate. Also in 2014: Jason Browne (formerly of GSS) incorporated four new short sellers; Mr Klar’s company, Amalthea, which had been a Solo Model stock lender in 2013, acted as a forward counterparty; and the other forward counterparties were all incorporated by their principals to participate in Solo Model trading.

The third significant development for 2014/2015 was the scale of the Danish tax reclaim industry that the Solo Model became, as summarised in paragraphs 192 to 208 below, such that over 90% of the amount paid by SKAT in relation to the Solo Model came from 2014/2015 trading (DKK8.306bn, 92% of the Solo Model total).

In mid-2013, Mr Stein and Mr L’Hote left Argre to pursue cum-ex business separately through what became the Maple Point Model. Messrs Markowitz, Van Merkensteijn and Klugman incorporated 40 new USPFs for themselves, their friends and family, in July 2014, 39 of which traded under the Solo Model (the ‘New Argre Plans’) in place of the Original Argre Plans.

At about the same time, Sanjay Shah made efforts to find more US individuals who had or could establish USPFs that might participate in Solo Model trading. On 30 July 2013, in New York, as a result of Mr Devonshire’s introduction of Mr Fletcher to Mr Murphy, Sanjay Shah, with Mr Murphy, met Mr Fletcher and some of his work colleagues at a brokerage called Standard Credit: Mr Bradley, Mr Tucci, Mr Vergari, Mr Driscoll and Ms Anderson (the ‘Standard Credit Individuals’). Mr Shah described the opportunity as USPFs engaging in div-arb trading in Danish stocks in order to seek a tax refund. He did not explain that the opportunity was structured around cum-ex purchases by USPFs, or the detail of how the trades would be structured and settled. Mr Fletcher and the Standard Credit Individuals and their friends/family set up 31 USPFs, 10 of which traded under the Solo Model from late 2013 and others of which traded only in 2014/2015 (the ‘Standard Credit Plans’).

This New York recruitment trip is significant. Mr Shah had no reason to suppose that any of the Standard Credit Individuals would be willing to participate in a fraud. Neither did Mr Murphy or Mr Fletcher; and neither did Mr Devonshire have any reason to suppose that his long-standing good friend Mr Fletcher would have any interest in being involved in a fraud. I find that whatever was said to Mr Devonshire, causing him to put Mr Murphy in touch with Mr Fletcher, and whatever was then explained to Mr Fletcher and the Standard Credit Individuals, did not cause them to think that the Solo Model was or might be a means for practising fraud upon SKAT. Moreover, I consider it highly unlikely that Mr Shah would have gone about this client recruitment as he did if he had understood that what he was selling was in fact participation in such a fraud.

In relation to these introductions, initially Mr Murphy was paid 15% of the net profits of the USPFs’ tax reclaims, from which he made smaller payments to the Standard Credit Individuals and to Equilibrium Capital, an entity majority owned by Mr Devonshire of which Mr Fletcher was a director. I accept Mr Fletcher’s evidence that in the late summer of 2014, he agreed new terms with Sanjay Shah, having learned that Mr Murphy had dealt dishonestly with him in relation to the introduction of the Standard Credit Plans. He and Mr Devonshire would be paid US$1m per USPF for new introductions, to be paid in each case once the USPF had been set up, taken on by GSS as a client, and successfully traded. Fletcher-Devonshire companies received US$20m under that new arrangement (for 20 new USPFs at US$1m each), of which c.US$3.6m was paid on to Messrs Tucci and Bradley.

Messrs Tucci and Bradley later made their own direct arrangement with Sanjay Shah to be paid for introducing further USPFs, which were set up in December 2014 (the ‘Further Tucci/Bradley Plans’). Messrs Tucci and Bradley, through corporate vehicles, were paid c.US$10m for this by Ganymede or the mini-Ganymedes (as defined in paragraph 233 below).

In around October 2013, Sanjay Shah invited Mr Godson, at the time a partner in the broker FGC, to introduce potential investor clients for GSS, in return for a commission or fee. This led to the introduction of another 24 new USPFs (the ‘Godson Plans’). Mr Godson was the beneficiary or trustee of 13 of the Godson Plans, and he was paid €9,659,761 by Ganymede, some of which he paid on to the other individuals he had introduced.

At some point in 2014, Sanjay Shah agreed to pay introducer fees to one of the individuals introduced by Mr Godson, Gavin Crescenzo, as a result of which Mr Crescenzo’s companies were paid US$1.4m for the introduction to GSS of yet further new USPFs for Solo Model trading (the ‘Further Crescenzo Plans’).

Roger Lehman, mentioned in paragraph 128 above, was initially hired by Solo in 2013 to assist in the tax reclaim process for USPFs trading under the Solo Model, but also acted as trader for some of the USPFs, including those of Mr Fletcher and Mr Godson. Sanjay Shah subsequently agreed to pay Mr Lehman US$1m for introducing his own USPF (Valerius) and US$700-800k for each further new USPF introduced in 2014. As a result, Mr Lehman, friends of his, and his brother Kevin and friends of his, set up between them 20 USPFs that traded under the Solo Model in 2014/2015 (the ‘Lehman Plans’).

Also during 2014/2015, as noted in Appendix 3, the GSS cum-ex tax reclaim business expanded to Labuan. Some 24 LabCos were established, with the assistance of Labuan corporate service providers, to participate in Solo Model trading. They fall into two groups:

Again as noted in Appendix 3, Solo Model 2014/2015 trading involved four Solo custodians, rather than only SCP. The additional custodians were Telesto, incorporated by Solo in October 2013, West Point, 90% of which was acquired by Hooloomooloo, part of the Elysium Group, in September 2013, and Old Park Lane, acquired by Solo in mid-2014. Old Park Lane issued CANs from August 2014 in respect of Solo Model 2014/2015 trades on which it had acted as custodian; Telesto and West Point each did so from February 2015. As noted in paragraph 117(ii) above, SCP and the other Solo custodians shared middle/back office functions which were serviced by the Solo GSS team at SCP, later Genoa. Reflecting this, most of what Old Park Lane, West Point and Telesto charged in custody fees was passed on to SCP; and SCP also received the proceeds of successful refund claims supported by Old Park Lane, West Point and Telesto CANs, for onward distribution (mostly to Ganymede).

In conjunction with the expansion of the Solo Model business, Sanjay Shah sought control of or exclusivity with some of the Tax Agents and brokers:

To facilitate and accelerate the expansion of the Solo Model business, during 2014/2015 increasing automation was developed and implemented. That was a longer-standing goal for Sanjay Shah, but it received particular impetus after, in March 2013, some unwind phase transactions were not matched properly so that some first phase traded positions were not unwound.

On (Friday) 8 March 2013, Sanjay Shah emailed Mr Dhorajiwala and Rajen Shah that: “today’s trades/activity is in meltdown for about a dozen different reasons. Pogo [Mr Patterson], Nirav [Patel] and me are in the office trying to sort. I have said a million times before that we need a system. We need a developer hired asap”. Mr Shah asked Mr Patterson and Mr Patel for a timeline of what had gone wrong; and on the following Monday (11 March 2013), Mr Patel sent the requested timeline to Rajen Shah and Mr Dhorajiwala. He identified a number of issues that had arisen with the trading including issues with SCP’s systems, participants in the Solo Model failing to respond, and entry errors in the system. After this, developers were recruited by SCP to automate the GSS systems, namely Stuart Ervine, Albert Lacatz and Darren Hoggs. They were later reorganised into a different Sanjay Shah company, Theorem, working from the same office used by GSS, as reorganised into Genoa.

The first automated system was the Trade Approval System (‘TAS’), into which trades were inputted for approval, at first manually and later fully automated. This was in operation in 2013. It was updated and refined over time.

From late February 2015, two other systems came into use: (a) Brokermesh, used by the brokers, which matched orders and sent automated confirmations by email, all confirmed orders being executed at the published closing price for the Danish stock in question; and (b) Octave, an automated email system that included static data such as Danish companies’ dividend dates and amounts, and generated matched orders for all the trading counterparties (buyers, short sellers, stock lenders, forward counterparties). Octave could access and communicate with Brokermesh. As Ms Spoto of GSS explained on 23 February 2015 in an email to Mr O’Carroll, the trader for the IPG Individuals’ LabCos, those entities having “put a trade schedule in place”, “the new algo system will automate this. You will NOT need to contact anyone … unless the system goes down at any stage. …”.

The operation of the fully automated system can be illustrated by Sample Trade Solo 4 (see Appendix 3, below, at paragraph 18ff). All emails from Octave were sent from a single email address, octave@hydracapitallimited.com. Each was addressed to a designated email address for the addressee and set out a message to that addressee written as if it was a message from another Solo Model participant, to whose designated email address the email was copied. So, for example (and taking the initial trading phase as sufficient illustration), on 26 March 2015, the Solo 4 dividend declaration date:

These automated systems covered each of the Solo custodians, SCP, West Point, Old Park Lane and Telesto.

SKAT sought to make something of the fact that, in the manner just illustrated, Octave generated messages pretending that active processes of supply and demand, offer and acceptance, were in play, with parties looking to do trades and brokers looking to find liquidity, match principals, and confirm having done so. I agree that was all pretence; no doubt it would have been possible to design an automated trading system that documented things differently, for example it might just issue trade confirmations for trades that the embedded code ‘decided’ to allocate to participants within (it would have to be) parameters defined by the participation agreements, so as to commit the participants to such auto-allocated trades. But the pretence was transparent from the email correspondence that Octave generated. In other words, to put it less pejoratively, it was quite plain – and would have been so to any outside party, such as an auditor or regulator, if they looked at the trading records – that this was an automated system creating trades without input from the parties at the time of trading, programmed to document those trades in a way that mirrored non-automated trading processes.

I therefore do not find anything suspicious or untoward in the automation, or its detailed implementation, in itself. Its real significance, in the context of the Solo Model trading, is how it manifested the lack of real-world constraint upon the volume of trading, and consequent tax reclaims, that could be generated by using balanced, self-fulfilling, share-less settlement loops.

In April 2015, Anne Stratford-Martin (‘Ms Stratford’), a qualified lawyer by original background who had been CEO of SCP since January 2014, recruited by Sanjay Shah from Cantor Fitzgerald, although in truth Sanjay Shah continued to operate as de facto CEO of SCP for the GSS / Solo Model business, reviewed the draft opinion letter from Hannes Snellman to which I referred in paragraph 183 above (strictly, a slightly revised version of it provided by Hannes Snellman in February 2014 of which no use had been made). Ms Stratford first saw that draft opinion in late January 2015, when she was considering an enquiry from a Wall Street Journal journalist, Jenny Strasbourg, investigating Solo’s German and Danish cum-ex strategies.

SCP’s then General Counsel, Michael Herron, proposed that there should be a board paper, supported by legal advice, “in relation to the position under relevant tax laws as regards dividend arbitrage operations for which GSS was acting as the clearing agent”. That led Ms Stratford to look at the draft opinion letter from Hannes Snellman in detail in April 2015. Her review, as reflected by her marked-up comments sent to Priyan Shah and Gerard O’Callaghan on 20 April 2015, identified inter alia that:

No further legal advice was sought, either to address Ms Stratford’s concern about the adequacy of the draft opinion she had reviewed, or at all. Available records show that Ms Stratford and Sanjay Shah had a 44-minute telephone discussion on 21 April 2015, the day after she sent her comments on the draft opinion letter to Priyan Shah and Mr O’Callaghan. Mr Shah’s evidence was that he has no recollection of that call, and that it was unlikely that they discussed the draft opinion letter and Ms Stratford’s comments on it, as that would indeed have been a matter, at the time, being considered by the other Mr Shah and Mr O’Callaghan. I do not accept SKAT’s invitation to reject that as implausible. To the contrary, it is in my view both inherently plausible and supported by the fact that Ms Stratford’s email the previous day was not sent to Sanjay Shah, said that it followed a conversation she had had with Priyan Shah and Mr O’Callaghan on 20 April 2015, and asked for a further discussion with them the next day, i.e. 21 April 2015. It is of course plausible that in a long call with Sanjay Shah on that day, Ms Stratford might have mentioned that she was discussing what Hannes Snellman had said in the past with the other Mr Shah and Mr O’Callaghan; but it would be speculation, not a finding based on evidence, to say on that basis that the subject was mentioned on that call; and on any view the story around what must have been discussed that was put to Sanjay Shah by SKAT was speculative.

SKAT’s further, and imaginative, speculation was that Ms Stratford’s mark-up of the draft Hannes Snellman opinion letter, discussed (as SKAT would have it) in her call with Sanjay Shah the following day, was the catalyst for agreements Mr Shah made with Ms Stratford a month later to pay her £5 million, and Messrs Knott and Hoogewerf £2 million each. SKAT submitted that these were agreements for ‘hush money’ to buy the silence of Ms Stratford, Mr Knott and Mr Hoogewerf, about (as SKAT alleged) the fraud being perpetrated through Solo Model trading against SKAT (and other tax authorities). Mr Knott and Mr Hoogewerf each maintained that his agreement, agreed with Ms Stratford and negotiated by her with Sanjay Shah, was by way of retention bonus, documented as a loan but on the agreed understanding that it would be forgiven (i.e. repayment would be waived) if he had not left Solo by choice prior to the repayment date.

Ms Stratford pleaded an equivalent position in her Defence, as regards her £5 million. Ms Stratford was not a trial defendant, the proceedings against her having been stayed by consent between her and SKAT. Neither SKAT nor any defendant called her as a witness, so I had no evidence from her at trial. Sanjay Shah also maintained that the agreements were to incentivise Ms Stratford, Mr Knott and Mr Hoogewerf to stay at Solo, but he denied that there was any agreement about forgiving the loans if they did stay.

Save to confirm that I do reject as speculative SKAT’s suggestion that Ms Stratford’s consideration of the Hannes Snellman draft opinion letter had any connection to these agreements, I set out here only a few further basic facts relating to them, as follows:

As in relation to the Belgian tax authority’s enquiry in 2013 (see paragraph 158 above), without having pleaded any relevant case concerning them, SKAT cross-examined Sanjay Shah about enquiries of Solo received in June 2015 from the LSE, to whom all Solo Model equity trades (i.e. the purchase trades) were routinely reported, and the FCA, the regulator of SCP and the other Solo Model custodians. Those enquiries were mentioned in SKAT’s pleading, but only to say that SKAT would rely on the fact that Sanjay Shah dealt with them in support of its case that he remained the relevant directing mind and will of SCP (and related entities) notwithstanding Ms Stratford’s role as CEO in London. Therefore, as with the Belgian tax authority enquiry, I treat the cross-examination as going, if it went anywhere, only to Sanjay Shah’s credibility.

On 1 June 2015, the Market Supervision Department of the LSE asked SCP, Telesto and West Point to confirm that equity trades for TDC shares reported to the LSE on 27 May 2015 were “good trades”. These will have been trades for an unwind phase within the Solo Model. TDC had declared a dividend of DKK1 per share on 5 March 2015; and SCP, Telesto and West Point between them had been the custodian for 102 automated Solo Model trades around that dividend declaration date for an aggregate volume of 292,689,111 shares, some 36% of TDC’s total issued share capital at the time. The enquiry was passed up to Ms Stratford in London, who passed it on to Sanjay Shah for instruction on how to respond (hence SKAT’s reliance on this in support of its well-founded argument that Sanjay Shah was the directing will and mind of the custodians in relation to the GSS / Solo Model business). At Mr Shah’s direction, Ms Stratford caused confirmations to be given to the LSE that the trades were indeed “good trades”.

The LSE made a slightly fuller enquiry on 3 June 2015 in relation to 24 trades for Danske Bank shares reported that day, which again were unwind trades in the Solo Model, this time in respect of original trading around the Danske Bank dividend declaration on 18 March 2015. The 24 trades were each for 1,886,972 shares at DKK198 per share, after noting which the LSE’s email continued, “It’s quite unusual to [see] so many large trades at the same size and price; please could you confirm whether they are good trades, and if so, help us to understand the strategy behind these reports?”. An accurate response would have confirmed that the trades were “good trades” and explained that they were trades to unwind positions traded in March as part of a cum-ex trading strategy in relation to Danish shares. The response sent, as directed by Sanjay Shah, was a misleading half-truth: “We can confirm that the trades are good trades. Please be aware that the reported trades are the unwinding by clients of their positions [true]. They follow the closure of clients’ OTC derivative hedges which were closed with the clients’ brokers at the market closing price [misleading spin]. We as clearer had imposed a limit of €75m per ticket as so [sic.] you see the trades occurring in batches of the same size [misleading spin]. All trades were executed OTC by the clients’ brokers [half-truth at best, cloaking Solo’s role as director of all trading, and a gross spin on the fully automated Solo Model then in operation].”

In response, the LSE asked for clarification, and at Sanjay Shah’s direction was given the following response (questions posed by the LSE, answers added in reply):

“• Were these market facing trades? These trades were OTC and Broker facing

• Was there a change of legal ownership? Yes

• Was there a change of beneficial ownership? Yes”

The first answer was again a half-truth masking the reality that all trading was directed by Solo, now on a gargantuan scale via automation. The second and third answers were not true.

The LSE remained curious and set up a call with Alan Hadley (then Deputy Head of Compliance at Solo), on which, to reassure Sanjay Shah, Ms Stratford stood over Mr Hadley so she could, as she put it to Mr Shah in a Skype message, “stop the conversation if it strays beyond providing the answers we have already provided”. Although to his discredit Sanjay Shah refused to accept this in cross-examination, the obvious purpose was to ensure that as little information as possible about the Solo Model was disclosed.

On 11 June 2015, the LSE sent another email, attaching a simple table showing for each of 10 securities that the aggregate traded volumes reported by SCP between 1 May and 4 June 2015 represented significant percentages of the total issued stock, ranging between 5.01% to 25.42%. The email said that those traded volumes had resulted in several queries from the market, that the same applied for Telesto, Old Park Lane and West Point, and that the LSE wanted “to understand the origination of these trades; how the business was generated at the client level and the process in chronological order from the start to the point of trade reporting.” An accurate response could only have been to say that Solo was coordinating a cum-ex trading strategy to profit from tax refund claims, using automated processes to trade very large volumes, and the trades in question were part of the unwind phase of that strategy. Sanjay Shah instead approved, as a “good idea”, a proposal by Ms Stratford to say that “since we believe clients are taking advantage of tiny price differences by default the volumes are large to make the returns meaningful. Clients are not exposing themselves to market risk by undertaking such large trades as they are ensuring they are fully hedged.”

The full response, approved by Sanjay Shah and sent at his direction (via Ms Stratford) by Mr Hadley, was thoroughly dishonest:

“We have spoken with some of our clients and explained that we need to respond to your query. The clients are reluctant to share their intellectual property with us but we understand that the equity trades which we report are the hedges to OTC derivatives. The clients advise that their books are fully hedged for market risk.

(Nothing in the first two sentences was true. Solo’s clients had neither input nor even knowledge of these LSE enquiries. The party with ‘intellectual property’, who was indeed unwilling to share it, in my judgment vehemently so, was Sanjay Shah. On a point of detail for the second sentence, of course, in ordinary logic, if Product A hedges Product B, equally Product B hedges Product A. But it was misleading to suggest that there was a derivatives trading strategy, as part of which equity trades were used to hedge market risk.)

The motivation for trading is to take advantage of the small differences in the implied financing rates in the OTC derivatives market and the stock lending market. The clients buy or sell equities, which are then borrowed or lent on to their stock lending counterparties.

(The second sentence, taken on its own, could have been true. But it was in the context of Solo’s prior confirmation that legal and beneficial ownership of equities was transferred, making it misleading. The first sentence was false. Small pricing differences of that kind could in theory have been the target of an arbitrage strategy of some kind. But not only was that not the trading strategy in the Solo Model, the impact, if any, of those differences on how Solo Model trades would turn out was retrospectively adjusted out.)

Chronologically, it appears the equity hedges are only initiated when the clients have an OTC derivative and stock lending trade which they would like to execute.

(This was dishonest nonsense, given Sanjay Shah’s understanding of the Solo Model trading strategy, design and implementation.)

The clients noted that the trade volumes are not unusual compared to previous years.

(This was substantially false. First, again, Solo’s clients had not ‘noted’ any such thing, having not been involved at all in how the LSE’s enquiries should be answered. Second, while the total Solo Model traded volume being unwound in May-June 2015 (from trading around February-March 2015 dividend dates) was similar to the total traded volume for the whole of 2014 (c.2.3bn vs. c.2.1bn), it was larger by an order of magnitude against 2012 (c.79m, albeit only from the smaller second half of the year) and 2013 (c.381m for the full year).)

It is worth noting that the volume reported doesn’t reflect that we have many (over 100) clients trading so that each client’s volume is small compared to the overall volume that is reported.

(That was true but misleading, given that the large number of trading clients was a deliberate construct of the Solo Model precisely to mask the fact that a huge volume was being traded on a coordinated basis by Solo.)

We hope that this answers your queries in full.”

Sanjay Shah in cross-examination tried a number of lines of argument as to why that response was not thoroughly and deliberately misleading. They were all, I regret to say, feeble attempts to defend the indefensible. He resorted instead to an attempt to distance himself from it, effectively claiming to have been merely a post-box passing messages between Ms Stratford, on the one hand, and Priyan Shah and Mr O’Callaghan, on the other, whom he wished to say held the drafting pen. That flew in the face of the documentary record and was in my judgment an example of Sanjay Shah saying whatever occurred to him to say in the moment, question by question, in the hope that something he said might be thought plausible.

On 23 June 2015, Ms Stratford passed to Sanjay Shah an enquiry from the FCA as to why the bulk of SCP’s revenue in 2014 was generated between April and September. She suggested “replying (along the lines of the response to LSE) that “activity is client driven and clients do not share their trading strategies with us therefore I cannot provide a particular reason why there was more activity in that period”; and Sanjay Shah approved: “i think thats fine. this year the clearing fees are a flat monthly rate so there should be no spikes”. Mr Shah knew that proposed response to be false. The trading strategy was Solo’s strategy, given to the clients and directed by Solo in its execution; and the spike in revenues for SCP will have arisen because its major revenue stream was Solo Model trading, on which its fees were charged trade by trade prior to 2015, and that was tailored around dividend declaration dates and the generation of tax refund claims the chronological pattern of which across the calendar year was not smooth. I do not consider this to be evidence that Mr Shah knew or believed that the Solo Model amounted to or involved a fraud upon SKAT, rather than at most further corroboration that Mr Shah wanted to keep the reality of the Solo Model under the radar for fear that the FCA would seek to put a stop to it as an unacceptable form of trading.

The LSE and FCA enquiries are indications, however, of how the scale on which the Solo Model came to operate in 2014 and the first half of 2015 was going to make it impossible to stay under the radar much longer; and indeed, the lack of practical constraint in the Solo Model by 2015, to which I referred in paragraph 210 above, did lead to its undoing, and in consequence to the undoing of the Maple Point and Klar Models when SKAT’s investigations caught up with what had been happening.

There was an initial attempt in June 2015, not wholly effective, to tip SKAT off to what was happening, initiated (at the time anonymously) by Mr Bains. That was followed by a more concrete, detailed and informative tip-off from HMRC in July 2015. In different ways, the sheer scale of the Solo Model operation, as it had become in 2015, triggered both calls to SKAT to look into what was going on. Mr Bains had come to the view by the end of 2014 that Sanjay Shah was out to “bleed dry an innocent Nordic country [viz., Denmark] next year”, and the magnitude of the Solo Model by then was a spur to Mr Bains’ attempt to tip SKAT off. I cannot make any full finding as to what sparked HMRC’s initial interest to investigate, but the terms in which it tipped SKAT off in late July and early August 2015 evidence that the scale of Solo’s operation, as it had become, must have been part of it.

The almost total lack of control within SKAT over what it was paying, or why, by way of WHT refunds, described below as part of considering whether on the evidence SKAT can say it was misled into making payments, makes it possible, I think, that if Solo Model and Maple Point Model operations between them had been kept within (proportionately) more modest bounds, they might not have been stopped at all. From the main Danish dividend season in the first half of each year, the Solo Model generated c.DKK490m in WHT reclaims in 2013, but that leapt to DKK2.25bn in 2014 (with almost DKK1bn from Maple Point running independently alongside), then DKK5.71bn in 2015 (plus DKK1.75bn from Maple Point). Whether Sanjay Shah and others were greedy fraudsters, as SKAT claimed, or just greedy opportunists who had alighted upon a way of making a fortune without having to commit the fraud that SKAT alleged, as it seems to me greed was their downfall.

SKAT’s investigations following the HMRC tip-off led to a decision in August 2015 to suspend all WHT reclaim payments and eventually to these proceedings and all the other litigation around the world, civil and criminal, by which the Kingdom of Denmark has sought to make a recovery against what was paid out and to punish those whom it considers to have taken dishonest advantage of those lax controls.

For each USPF or LabCo that participated in Solo Model trading, there was a consultancy services agreement between the USPF/LabCo itself or the principal behind it (directly or through a corporate vehicle) and Ganymede, providing that the successful provision of consultancy services by Ganymede would entitle it to fees. Pursuant to those agreements, Ganymede invoiced for consultancy services supposedly provided by it, as the means by which it would take most of the proceeds of tax refund claims generated from Solo Model trading. Until mid-2013, the Tax Agents paid the USPFs (net of the Tax Agents’ fees) and the USPFs made onward payments to Ganymede; thereafter, the Tax Agents paid the amounts intended for Ganymede to SCP, and SCP paid Ganymede.

Ganymede thus received from the Original Argre Plans and the Zeta Plans 66.6% and 60% respectively of their net tax reclaim proceeds, 75% from the New Argre Plans, between 76.5% and 80% from the LabCos, and 95% from the Standard Credit Plans, Further Tucci/Bradley Plans, Godson Plans, Further Crescenzo Plans, and Lehman Plans. Ganymede was paid only if and after a tax refund payment had been received from SKAT.

There were in fact no consultancy services, or agreements for Ganymede to provide such services. The language of consultancy or advisory services was a euphemism for the provision by Sanjay Shah of access to the Solo Model, in return for which (through Ganymede) he would take most of any profit (see paragraph 118 above). The obvious dishonesty of documenting things in that way was relied on by SKAT in support of its allegations of fault and knowledge as they arise in the context of the causes of action it advances; but there was no freestanding point taken as to whether the true arrangement, i.e. the nature and extent of the Sanjay Shah/Ganymede profit share, itself affected the validity of the tax refund claims from which any profit to be shared would be and was generated. The point on the way the Ganymede profit share was documented was weakened substantially anyway by the fact that in the early period of the Solo Model, the agreements were explicit that what was to be shared between Ganymede and the client were tax refund claim proceeds, as remuneration for “tax reclaim advisory services provided by [Ganymede]”.

In turn, Ganymede made payments, directly or (from May 2015) via four BVI companies established for the purpose, Parla, Fire, Philo and Acai (the ‘mini-Ganymedes’). The mini-Ganymedes were owned and nominally controlled by Usha Shah, but she acted at all times solely on Sanjay Shah’s instructions in relation to them, and they were transferred to Elysium Global in October 2015. Payments were made to:

The arrangements with introducers and trading counterparties were documented as consultancy services agreements, treating them as consultants to Ganymede and providing that the successful provision of consultancy services would entitle them to fees payable by Ganymede. Here also, there were in reality no such agreements or services, and there was no honest reason for documenting the fiction. The true arrangement was simply that Sanjay Shah, acting through Ganymede, was happy to pay substantial sums (in absolute terms, even if at the same time modest percentages of the tax reclaim profits to be generated) for the availability of tax-exempt entities or trading counterparties for the Solo Model trading operation.

Between May 2014 and February 2016, funds totalling c.€46.8m were paid from accounts at SCP to fund the acquisition of c.80% of Varengold Bank AG, a German bank. Individual share acquisitions were coordinated by or on the instructions of Sanjay Shah, and on 26 August 2015 he was elected, together with Mr Barac and Mr Murphy, to the Supervisory Board of Varengold Bank. The various acquisitions were as follows:

Those Varengold Bank shareholders, except AESA and Elysium Dubai which were under Sanjay Shah’s control anyway, held their shares de facto as nominees for Sanjay Shah. Their acquisition of Varengold Bank shares was funded indirectly by him so he would obtain covertly, as thus he did, effective control of Varengold Bank, when an overt attempt would have failed because it would have required a regulatory approval that Sanjay Shah did not have and did not want to take the time to seek. This was disputed by the trial defendants in question, save for Mr Patterson and (effectively) Mr Preston (see below); but in my judgment it was comfortably demonstrated by the evidence, and the relevant defendants’ denials were not credible.

Several funding mechanisms were used for the Varengold Bank share acquisitions. The primary technique was the use of sham “High-Low Trades” in June 2014. Purported trades to buy and sell German shares, matching except as to price, were documented with Ganymede through Bastion, one of the brokers that participated in Solo Model trading, at prices chosen so that Ganymede would appear to have made an intra-day trading loss as the supposed commercial basis for transfers of sums totalling €23.7m from Ganymede’s account at SCP to fund the acquisition of Varengold Bank shares.

Some of the Varengold Bank share purchases were funded through loans and counter-loans between the purchasers and Elysium Dubai. For example, in relation to Mr Preston, on 18 January 2015 Elysium Dubai and Skyfall Holdings, the parent company of Skyfall, Mr Preston’s Varengold Bank buyer, entered into: (a) a loan facility for Elysium Dubai to lend Skyfall Holdings €5.31m; (b) a loan facility for Skyfall Holdings to lend Elysium Dubai €1.77m. The net effect was a loan of €3.54m from Elysium Dubai to Skyfall Holdings.

In other cases, funding was provided by Sanjay Shah’s companies ostensibly pursuant to invoices raised for purported consultancy services provided.

Whatever funding mechanism was used, the majority of the ultimate indirect owners of the Varengold Bank shares provided or were intended to provide security for the funding. Some arranged for a charge over the shares in the Varengold Bank share purchaser, or an entity with effective control of that purchaser, in favour of a Sanjay Shah entity. Others granted a Sanjay Shah entity an option to purchase the Varengold Bank shares.

Sanjay Shah accepted that he had the money to acquire all the shares in Varengold Bank and that his goal was to achieve sole ownership. He said that he believed an acquisition of Varengold Bank by him, or by a company of his, would “engage various regulatory rules which would be time consuming to resolve”. He claimed to have assembled the principals behind the other acquirers, therefore, to be “shareholders in their own right” as part of an “investment club”. This might have made sense if any of these ‘investment club’ members were co-investing with Sanjay Shah, but it was not a credible explanation for his funding of their investment on terms that entitled him to any upside. The transparent reality, confirmed by contemporaneous evidence, is that Sanjay Shah was using the other individuals, with their consent, to disguise his effective indirect acquisition of a controlling stake in the bank.

Mr Patterson admitted in his Defence that his companies’ Varengold Bank shares were held as nominee for Sanjay Shah. In October 2014, Mr Bains confirmed in a note to his new employer, Arunvill, that he and all the others held their Varengold Bank shares as nominees for Sanjay Shah. His denial of nomineeship in his Defence was fatally undermined by that. The cross-examination of Mr Bains on this aspect demonstrated, in my judgment, that the account in his note for Arunvill was substantially accurate. The further detail emerged that, at the time, Mr Bains intended not to honour the arrangement with Sanjay Shah, or at least to threaten not to do so, so that he would either sell his Varengold Bank shares, or force Mr Shah to buy him out, as a self-help means of enforcing what he (Mr Bains) believed to be bonus entitlements that Mr Shah had not honoured. That proved successful, as Mr Shah arranged for and funded Mr Barac to buy Mr Bains out.

Mr Preston’s Defence also denied that he was Sanjay Shah’s nominee but in his evidence he admitted, in substance, that he was, even if he did not use that label. His evidence was that it was agreed from the outset that he would “hand the shares back to Sanjay Shah in the future”, that they “were always going to be returned to Sanjay Shah”, and that at all times he took direction from Sanjay Shah in relation to the shares.

Other trial defendants who were Varengold Bank ‘investors’, for example Mr Jain and Mr Smith, clung in their evidence to a story that they were investing for their own account, not on behalf of Sanjay Shah. In my judgment, they did so as a dishonest response to SKAT’s allegation that participation in the Varengold Bank acquisition evidenced knowledge or understanding that the GSS cum-ex trading business involved, and was designed to generate, deceit practised upon SKAT. I do not accept that allegation. It did the defendants in question no credit that they did not tell the truth about the Varengold Bank arrangement, and trust the court with any decision over whether that took SKAT anywhere on the claims made in these proceedings. As juries are rightly and routinely directed, defendants may lie for any number of reasons other than that the material allegations against them in the case are true and they have no answer to them. In my view, that was the position here in relation to the Varengold Bank business.

To be clear, I am not considering whether the true arrangement, informal and undocumented as it was, that the Varengold Bank shareholders funded by Sanjay Shah would act as his nominees so that he had effective control over Varengold Bank, would have been legally enforceable, even apart from any impact of their purpose of circumventing regulatory takeover requirements, and I am not in a position to make, nor do I need to make, any definite finding as to whether those requirements were breached. My finding is only of the understanding, informal but accepted at the time, between Mr Shah and all of those shareholders (through their respective principals).

After 2015, Mr Patterson, Ms Bhudia, Mr Smith, and Mr Preston transferred the shares in their Varengold Bank vehicles, or parent companies, to entities controlled by Sanjay Shah. Mr Jain, and also former defendants Messrs Barac, Lui, Priyan Shah and O’Callaghan, resisted subsequent attempts by Sanjay Shah to obtain ‘their’ Varengold Bank shares. They were identified in an October 2016 note by Greg Nixon, a lawyer at Elysium Dubai, as “hostile” shareholders, apart from Mr Jain whom he identified as “non-hostile”, meaning that at that time Mr Jain was not yet pretending that he did not hold his Varengold Bank shares for the benefit of Sanjay Shah. The later refusals to transfer Varengold Bank shares to Sanjay Shah, or to his order, does not disprove that they were supposed to be being held as nominees for him. By that time, Solo had been the subject of publicly reported raids, and those “hostile” shareholders may well have considered that it was not in their interests to admit involvement in what had been a dishonest scheme by which Sanjay Shah took effective control of Varengold Bank.

The funding of the Varengold Bank share acquisitions by the companies of Mr Lui, Mr Jain, Mr Barac and Messrs Priyan Shah and O’Callaghan was never repaid, and their shares were later transferred to SKAT pursuant to proceedings brought in Germany by a prosecution authority in the context of an investigation into the Varengold Bank acquisition as possible money laundering.

Between November 2014 and June 2015, funds totalling approximately €59.48m were paid from accounts at SCP to an account in the name of Trillium Capital to fund its purchase, which completed in the second half of 2015, of 100% of the shares of Dero Bank, another German bank, for €28.573m.

Prior to 12 November 2014, Trillium Capital was wholly owned by Trillium Holdings, which was in turn owned and controlled by Sanjay Shah. Between 12 and 14 November 2014, Trillium Holdings sold shares in Trillium Capital, as follows:

Those new shareholders in Trillium Capital funded their acquisitions through loan and security agreements concluded with Elysium Global and/or Elysium Dubai. As with Varengold Bank, but in the different ownership structure used here, the ‘co-investors’ in Trillium Capital held their shares (and thus their indirect ownership interests in Dero) as nominees for Sanjay Shah. That was admitted by Messrs Patterson and Murphy. It was denied, but not credibly, by others. The acquisition was entirely directed and coordinated by Sanjay Shah for his benefit as a means by which to obtain sole control of Dero while making it appear that he had only a minority stake, to avoid the inconvenience (as Mr Shah saw it) of satisfying regulatory requirements that might have attached to the acquisition of a controlling stake.

The Maple Point Model trading was designed and implemented following the departure of Messrs Stein and L’Hote from Argre, and the DWF Ds from Solo, in 2013. As described in Appendix 3, it reproduced the essential features of the Solo Model 2012/2013 trading, using non-Solo custodians and OTC forwards rather than exchange-traded futures for the equity price hedge. It enabled those principally involved to take a greater share of the resulting profits than they took from the Solo Model trading.

Maple Point Model trading extended exclusively to USPFs associated with Maple Point, a US company formed by Messrs Stein, L’Hote, La Rosa and McGee, formerly of Argre, and Donald (‘Don’) Donaldson, a New York lawyer. Mr Donaldson was appointed to act, and acted, as the authorised representative of all of the Maple Point USPFs.

Other key individuals for Maple Point Model trading included:

As with Solo Model trading, the key to the whole structure for each transaction, so as to achieve internalised settlement to zero meaning that the trades would be treated as settled although no shares were acquired or transferred, was to have:

The complex wrinkle on the ‘cash’ side of unwind settlements at Solo, referred to in paragraphs 74 and 156-157 above, was ironed out for Maple Point trading by using a breakage fee. The unwind phase was traded for settlement prior to the scheduled expiry of the stock loans and forwards (as indeed it was in Solo Model trading). That involved an element of ‘early termination’. A breakage fee was calculated to match, and cancel out, what would otherwise have been a trading profit, on paper, for one side of the trade or the other (the short or the long), leaving aside the tax reclaim. To illustrate that, using Sample Trade NCB 2 relating to 920,000 Coloplast B shares:

A stock loan termination fee was charged to the USPF, and a smaller such fee was charged to the stock loan intermediary, as the mechanism by which to pay to the stock loan intermediary and the short seller their agreed percentages of the dividend amount referenced in relevant trade. If the stock loan intermediary was also the forward counterparty (NCB and Lindisfarne trades), this also dealt with that party’s profit share as forward counterparty. In Indigo trades, where the forward counterparty was different, a separate termination fee was charged to the USPF only, i.e. not also on the back-to-back forward with the short seller, so as to pay the forward counterparty its reward for participation.

As in Solo Model trading, no Maple Point custodian ever held any Danish shares, directly or via any sub-custodian, as custodian for any of the USPF buyers, short sellers or stock loan counterparties, or at all, or received any dividend payment as part of any distribution from a Danish company, via VPS and a custody chain. As Mr Sharma put it concisely in his expert report, “the trades were structured so that no shares or cash (other than the dividend reclaim [i.e. tax refund] amount) needed to be sourced externally”. In fact, “to be sourced externally” adds nothing (it is a clouding use of jargon: paragraph 70 above). Putting it plainly, by design there was no need for shares or cash to execute or settle the trades, given how they were in fact settled.

In around June 2013, Rajen Shah had discussions with Maple Point regarding the possibility of working together, with a view to profiting from the Solo Model strategy. Later, he invited Mr Horn and Mr Dhorajiwala to collaborate with him in replicating the Solo Model, for use to generate trades focused on USPFs that Maple Point might be able to offer. Mr Horn had left Solo in June 2013 when he had been effectively fired by Sanjay Shah with immediate effect. Mr Dhorajiwala did not leave Solo until 30 September 2013, but began discussions with Rajen Shah about the possibility of working with Maple Point in June 2013 and but for that opportunity may well have stayed at Solo.

The DWF Ds between them represented a complementary offering. Rajen Shah was an experienced structurer, who was eager for Mr Horn to be involved given his understanding of the ‘technology’, and Mr Dhorajiwala had expertise in the operational mechanics required for custodians to operate the trading model.

In mid to late 2013, agreement was reached with Maple Point whereby the DWF Ds, in exchange for a substantial share of anticipated profits, would identify and assist new custodians, implement trading, and facilitate the making of tax refund claims to SKAT. A significant amount of the DWF Ds’ time thereafter, in late 2013 and early 2014, was spent getting what I have been calling the Maple Point Model ready for action. I have used that label following the terminology used generally in the litigation, but it should not be thought that it was a trading model designed or developed by Maple Point (or its ex-Argre principals). On the evidence of who did what, and who had the relevant expertise and ‘IP’, it was the DWF Ds’ trading model, not Maple Point’s.

The DWF Ds worked through two entities established by Rajen Shah for the purpose:

The first order of business was obtaining the services of a custodian willing to approve and settle the trades, and issue CANs accordingly. From the beginning there was a desire for two such custodians. This was not, I find, with a view to obscuring things in the way that the creation of extra Solo Model custodians was. There were different profit splits with Maple Point as between NCB, brought to the activity by Messrs Stein and L’Hote, and Indigo (replaced for 2015 by Lindisfarne), brought by the DWF Ds (or in the case of Lindisfarne, by Mr Horn alone); and the involvement of a custodian not linked to Messrs Stein and L’Hote would mean that, if they wished to do so, the DWF Ds might be able to continue the business even if they (Stein/L’Hote) at some stage decided to stop and the connection to NCB was therefore lost.

The custodianship arrangements at NCB were set up first. It was a small German bank ultimately majority owned and controlled by Messrs Stein and L’Hote through their vehicle Oban. During late 2013 and early 2014, the DWF Ds worked closely with the management of NCB to set up its new custody business and develop the capabilities needed to execute the Maple Point Model. The DWF Ds, among other things, provided sample trading documents to NCB, explained and answered queries about how the Maple Point Model worked, and introduced brokers and trading counterparties. A separate email domain was set up for correspondence relating to these special trades, to facilitate the deletion of records given that anything passing through NCB email servers would be retained.

The second custodian was Indigo, a company of Mr Dhorajiwala’s brother-in-law, Nailesh Teraiya. Mr Teraiya had participated as stock lender in Solo Model trading through his entity, Aquila, but that does not mean he will have been aware of how the Model worked. He now needed, and received, assistance from the DWF Ds to understand the custody aspect, in which he had not previously engaged. After discussions with Rajen Shah and Mr Dhorajiwala in late 2013, Mr Teraiya agreed to seek the necessary approvals from the FCA so that Indigo could act as custodian for Maple Point Model trades. As with NCB, the DWF Ds then worked closely with Indigo to establish its new custody business. Mr Horn also joined Indigo as a director (in May 2014). The necessary approvals were obtained from the FCA by use of a business plan that misdescribed the business Indigo intended to conduct, so as to avoid giving the FCA the chance to refuse to approve the actual trading model.

When Maple Point Model trading then commenced, the DWF Ds were involved in coordinating it, planning details of the intended trading and communicating to the trading counterparties what they each respectively needed to know to execute the trades the DWF Ds needed them to execute, being on call to manage any issues on trading days, and managing the unwinds.

NCB’s CANs were generated by its systems and sent by it to Mr Horn at Indigo, at Rajen Shah’s request, so that he could provide them to the Tax Agents, which he did. Mr Horn had responsibility for Indigo CANs, but Rajen Shah also reviewed most of them, and Mr Dhorajiwala was aware at the time of their form.

The DWF Ds then managed the process by which tax refund claims supported by Indigo or NCB CANs were made to SKAT. Oryx dealt with the Tax Agents including by agreeing their fees, referring the Maple Point USPFs to them via Don Donaldson, and providing other assistance ad hoc, including with the on-boarding process. The DWF Ds also liaised with the Tax Agents as to the status of the tax refund claims and kept Maple Point informed.

In total, Maple Point Model 2014 trading resulted in CANs used to support successful tax refund claims to SKAT for: (a) DKK306,062,482.50 for CANs issued by NCB; and (b) DKK688,383,354.02 for CANs issued by Indigo. A significant proportion of those proceeds was ultimately received, indirectly, by the DWF Ds:

The Maple Point USPFs themselves received a “long fee”, typically (although not in every case) of 4%. In an email at the time, Rajen Shah wrote: “Pension funds end up with 4% of the P&L after all fees”. However, the spreadsheet to which I was referred at trial shows that it was more complex than that. A Maple Point USPF’s fee was not quite an agreed percentage of “the gross … reclaim less all fees due to the Tax Agents, Brokers and other trading counterparties”, as SKAT submitted. It was less than that, because:

In late 2014, Rajen Shah, Mr Dhorajiwala and Mr Teraiya ceased to be involved with Maple Point. Mr Horn took steps to ensure that Maple Point Model trading could continue in 2015.

First, Mr Horn found a custodian to replace Indigo. He incorporated Gold Eagle Securities Ltd (‘GESL’) for that purpose on 8 January 2015 and Natassia Kourousi was employed as compliance officer, a role she had held at Indigo. However, GESL did not obtain authorisation from the FCA to act as custodian until August 2015. As with Indigo for 2014 trading, the FCA authorisation application prepared by or under the direction of Mr Horn misdescribed the proposed business so as to avoid disclosing the true nature of the trading model, which (I find) Mr Horn expected that the FCA would not approve.

In the meantime, in early 2015 Mr Horn was introduced to Arthur Hogarth and Paul Baker at Lindisfarne. The introduction came from Mr Price, who had been a founding partner of Lindisfarne. Lindisfarne suggested Fox Davis Capital or Shard Capital as custodians, but discussions with them did not bear fruit. Lindisfarne agreed to act as custodian itself, on condition that all trading netted to zero so that their accounts would only ever show nil share and cash balances, meaning there would be no settled balances of shares and no connection to shares registered at VPS.

Mr Horn then worked with two ex-Syntax employees under his supervision, Gelara Avak and Harry Rowe, to assist Lindisfarne in preparing for Maple Point Model trading. For example, they introduced Maple Point USPFs, brokers, short sellers, stock lenders and forward counterparties to Lindisfarne, provided template or draft documentation, including custody agreements, fee schedules, email trading confirmations and CANs, commented on drafts of Lindisfarne CANs, custody agreements, client statements and internal record-keeping documents, explained to Lindisfarne how the Maple Point Model worked, and provided a trading calendar for Denmark. They assisted Lindisfarne with the onboarding of Maple Point USPFs and other compliance matters.

As a second main step, Mr Horn considered that a new Tax Agent should be found, given Sanjay Shah’s efforts to tie Tax Agents to Solo, as described above. In late 2014, Mr Horn agreed with Alba Brown, formerly of Acupay, that she would establish her own company, in the event Koi, to act as Tax Agent. Mr Horn provided a start-up loan to Koi of £36,000 in December 2014 to cover its establishment costs and initial operating expenses. He also concluded an agreement with Ms Brown, later extended to her brother, Alasdair, who became a co-owner of Koi, to refer clients to Koi. In the event, Koi’s only clients were Maple Point USPFs.

As a third main step, in early 2015 Mr Price introduced Mr Horn to Adnan Haider, a former UAE banker with div-arb experience who was operating through his Seychelles company, WWAM, which he later settled onto a discretionary trust organised under the law of Gibraltar, the WWAM Trust. Mr Haider also owned White Rock and was the authorised trader of Lannister, two companies which then participated as short sellers in Maple Point Model trading via the NCB platform in 2015. WWAM agreed to act as consultant to the Maple Point USPFs and together with Mr Horn essentially took over the role that had been played by the DWF Ds for Maple Point Model 2014 trading. Mr Horn later entered into a consultancy agreement with WWAM on 27 September 2015, which he assigned to his entity GMH Advisory LLP on 13 October 2015.

Again, the trading was pre-planned, as realistically it had to be if it was always to result in the desired and intended net zero settlement loops. As Mr Hogarth of Lindisfarne said in his evidence, “the trading strategy was known and when the trades were presented to Lindisfarne, we could work out the cash flows at maturity”, and the trading “appeared to be pre-planned in terms of which accounts would participate, the structure of the trade and the times of the trades”.

As in 2014, NCB CANs from Maple Point Model 2015 trading were sent to Mr Horn, and he provided them to the Tax Agents. Assisted by Mr Rowe and Ms Avak, he also reviewed Lindisfarne CANs before they were sent. He liaised with Maple Point and the Tax Agents relating to progress of the tax refund claims, as he had in 2014.

In total, Maple Point Model 2015 trades resulted in successful tax reclaims on behalf of Maple Point USPFs to an aggregate value of (a) DKK829,712,963.69 for claims supported by NCB CANs, and (b) DKK920,721,444.02 for claims supported by CANs from Lindisfarne.

Between July 2015 and September 2015, in satisfaction of the WWAM invoices for services provided, the Maple Point USPFs between them paid €83,286,000 to WWAM from their accounts at either NCB or Lindisfarne. Disclosure in the proceedings did not locate much evidence of the fees payable to short sellers, stock lenders and forward counterparties in Maple Point Model 2015 trading, but it seems probable that they were similar to the previous year. In the case of Mr Klar’s entities, Sherwood and Potala, indeed, there is good evidence that they were paid on the same basis in 2015 as in 2014.

Koi was also paid a bonus of €30,000 per Maple Point USPF, in total €900,000; and Lindisfarne received total fees of c.€4.5m.

As I indicated in paragraph 100 above, one main theme at trial was whether, and if so on what basis, trial defendants who were aware of it believed that the use of balanced share-less settlement loops of the type used in the Solo Model or Maple Point Model, settling to zero, could leave the USPF (or LabCo) with an entitlement to a dividend tax refund from SKAT. In that connection, it is natural to consider, among other things, whether advice was taken on Danish tax law, and if so by whom, on what basis, with what outcome. It is convenient to answer that now on the facts, before turning to deal with the Klar Model to complete this main narrative.

The short answer as regards legal advice, I find, is that none was ever sought on that question, and I conclude that Sanjay Shah, Mr Horn and Rajen Shah, at the time, knew that advice had not been sought on that question and expected that if it had been sought, it would probably have been negative.

The outcome of the Revenue Rule Trial, after SKAT’s successful appeal, is that none of the pleaded claims pursued by SKAT at the Main Trial is regarded as inadmissible in an English court as a claim to enforce, directly or indirectly, Danish tax law or some other sovereign interest of the Kingdom of Denmark. The tax law context is still relevant, as it may bear upon aspects of the case pursued by SKAT, for example: the purport or effect, considered objectively, of the submission of a tax refund claim to SKAT, or of the documents that went to SKAT; the assessment of SKAT’s actions or thinking said to justify a finding of inducement by representations; the investigation of the understanding or intention of those who played some part in the trading that resulted in tax refund claims being made to SKAT, or in the making of those claims.

Further, it is right to keep in mind throughout that the process scrutinised in this litigation was not a business negotiation or commercial transaction, between SKAT and anyone. It was a process created and operated by SKAT as a national tax authority for receiving claims for relief from Danish dividend tax and for a payment accordingly from Danish public funds. Persons considering or interacting with that process might or might not draw any or any clear distinction between questions of having, or not, an entitlement to a tax refund, on a proper understanding of Danish tax law, on the one hand, and questions of understanding and satisfying (or not) what SKAT had decided, or appeared to have decided, to treat as sufficient for it to accept a claim and make a payment.

With that in mind, I summarise not only the private legal advice, to the extent I had evidence of it, sought and received by parties involved in Solo Model or Maple Point Model cum-ex trading, but also, and first, the advice and guidance published at the time by SKAT and by Clearstream.

SKAT’s Legal Guide was published on its website only in Danish. Where I quote from it in English, therefore, I am quoting translations from the Main Trial, not anything that appeared in English at the time. Section C.B.3 of the Legal Guide concerned “Taxation of dividends and distributions of shares, etc.”, and included the following guidance:

Section C.B.2 of SKAT’s Legal Guide concerned “Taxation of gains and losses on disposal of shares”. Within that Section:

Section C.F.3.1.7 noted that in Denmark there was a customary withholding rule, so that the company paying the dividend would withhold (it said) 28%, “even if the shareholder is domiciled in a State with which Denmark has [a DTT]. The shareholder will then be able to apply for a refund of the dividend tax via SKAT.” Then section C.F.8.2.2.10.1.1, concerning “dividends covered by Article 10 of the Model Agreement [for DTTs]”, said that in that context:

“… dividends are the distribution of profits from the legal entities that can distribute profits to the persons who own shares in the legal entity. In Denmark, these include share dividends and dividends paid by a limited liability company to its shareholders. Abroad, the legal entities, the ownership interests in them and the distributions to the owners may have different designations. … The law of the country where the distributor of the dividends is resident determines which types of income are covered by Article 10 … .”

The Legal Guide also contained the following guidance about DTTs (and Article 10 in particular), and their operation:

SKAT’s website carried inter alia the following English language pages during the relevant period:

SKAT’s published stance and guidance at the material time, therefore, was that:

I found in SKAT (Validity Issues), supra, that there were in fact cumulative requirements. I held, on the language of s.69B(1) of the Danish Withholding Tax Act and the expert evidence of Danish tax law at the Validity Trial, that to be entitled to a refund a party had to have been the beneficial owner of a dividend for the purpose of a DTT and also the shareholder under Danish tax law when the dividend in question was declared. That was not SKAT’s published understanding at the time; and s.69B(1) of the Danish Withholding Tax Act was not mentioned in the Legal Guide, SKAT’s website, or Form 06.003. An applicant using Form 06.003 would claim either “In my capacity as beneficial owner” or “On behalf of the beneficial owner”, and the Form required the full name and address of the “Beneficial Owner” to be filled in. Those would all naturally be taken, in context, to refer to the ‘beneficial owner’ concept in DTTs, which concerned dividends and did not require share ownership. SKAT relied in argument on a different standard form published by SKAT at the time by which a declaration of beneficial ownership of securities might be made. No such declaration was required for a dividend tax refund claim, however, and in my view its existence does not affect the purport or implications of SKAT’s Legal Guide and Form 06.003, read sensibly in the context of Denmark’s DTTs.

As the DWF Ds submitted, SKAT’s published position and guidance therefore gave no hint that short selling, deferred settlement, or stock lending, might be relevant to the incidence of dividend tax liability or any entitlement to a tax refund. It was submitted for the DWF Ds that it likewise gave no hint (i) “that any dividend which was the subject of a refund application should be traceable to a payment by a company paying the dividend” or (ii) “that any shares which were the subject of an acquisition contract should be traceable to shares held by VP Securities”. As to those submissions:

One upshot of the above is that if someone with knowledge of cum-ex trading strategies consulted SKAT’s Legal Guide as part of considering whether cum-ex ‘worked’ for Denmark, they would reasonably have concluded that it did or at least might. If shares were acquired after a dividend record date under a purchase (contract to buy) concluded with a short seller before the ex-date, and lent out by the buyer under a stock loan concluded only after the ex-date, the buyer would have acquired, it might be thought, ownership rights to shares that according to SKAT would be treated for Danish tax law purposes as having been acquired prior to the ex-date, and the disposal under the stock loan would seemingly take effect for tax purposes, according to SKAT, only from after the ex-date.

On what was said in SKAT’s Legal Guide, that would have appeared to mean the buyer was the shareholder subject to liability for Danish dividend tax. The question would remain whether the buyer was the beneficial owner of the dividend for the purpose of any DTT, on which the Legal Guide provided no real assistance.

The magic ingredient of the Solo, Maple Point and Klar Models, of course, was that the buyer never acquired shares at all. SKAT’s Legal Guide said nothing to suggest that such a buyer might be entitled to a tax refund, except in the limited sense that it was not definitively ruled out as impossible given that the Legal Guide indicated that a beneficial owner of a dividend, for the purpose of a DTT, was entitled (if they fell within a tax-favoured category under the DTT) and need not necessarily have been a shareholder liable for dividend tax.

Clearstream material publicly available at the time supported the view that a cum-ex buyer of Danish shares should be credited, and their seller debited, with what I am calling a dividend compensation payment, if the sale settled. On one reading of the Clearstream material, cum-ex trading where settlement is always intended to be with ex-div shares was within the cases where what it called a ‘market claim’ should be processed on a settlement after the record date. In my view, on balance, the better reading of the relevant Clearstream material is that it dealt only with the processing of market claims as defined by the CAJWG Standards, but as I have just indicated I can see how it might have been read differently. Nothing in the Clearstream material advised that a cum-ex dividend compensation payment would be treated as a dividend for tax purposes in Denmark, although it did not expressly rule that out, as it did for certain other jurisdictions where Clearstream felt able to make a positive statement to the contrary, for example, for the Netherlands, “Note: Market claims are considered indemnities. Recipients of indemnities are not able to reclaim Dutch withholding tax”, and for Italy, “Compensations on the Italian market are not considered as income but as a price adjustment”.

A reasonable view that may have been gleaned from the Clearstream material at the time was that of Rajen Shah, expressed in an email to Sanjay Shah and Mr Horn on 30 August 2012, that for Danish shares: “Dividend entitlement is based on traded position at ED [i.e. ex-date]. The CSD [i.e. VPS] will credit accounts only when position settles (this should not impact entitlement)”. Again, however, the secret to the Solo Model trading that was by then underway (the first trading was around TDC’s dividend date of 8 August 2012) was the absence of shares and the related share-less settlement method. The Clearstream material did not help to answer the question whether trades settled in that way might generate a Danish tax refund entitlement.

The only Danish tax advice obtained during the design or implementation of Solo Model cum-ex trading was advice given by the Danish law firm Hannes Snellman, in: (a) an opinion letter dated 27 June 2012 (the ‘First HS Advice’); (b) an email dated 1 July 2013 (the ‘First HS Email’); (c) an updated opinion letter dated 20 December 2013 (the ‘Second HS Advice’); (d) an opinion letter dated 29 January 2014 (the ‘Third HS Advice’); and (d) an email dated 12 February 2014 (the ‘Second HS Email’). There were also drafts for a further Hannes Snellman opinion letter in January and February 2014, but they add nothing to the body of Danish legal advice obtained, given that they were only drafts, although they played a separate part in the factual narrative, as mentioned in paragraphs 183 and 211 above.

The First to Third HS Advices were addressed to Elysium Dubai, but stated that they had been prepared at the request of SCP. The First and Second HS Advices concerned possible trading by USPFs and the Danish tax consequences thereof. The Third HS Advice concerned LabCos.

Hannes Snellman were not asked to advise, by reference to any actual trading, on the validity or otherwise of the tax refund claim, actual or prospective, of any Solo Model USPF or LabCo. Nor were they given a detailed or complete transaction structure for Solo Model trading, as they could have been since every element of it was designed at, and would be directed by, Solo.

Hannes Snellman were asked instead to consider the “Danish tax implications” for a USPF (or LabCo), and for SCP, if the USPF (or LabCo) entered into trades described in general terms in the introductory parts of the advice, which did not disclose that those trades would be part of a structure designed and coordinated by Solo for the sole purpose of generating a possible tax refund claim by the USPF (or LabCo). Among the trial defendants, Sanjay Shah, Rajen Shah and Mr Horn had responsibility for what Hannes Snellman were asked and how they were asked it. In cross-examination, Rajen Shah and Mr Horn both argued that they sought advice candidly and obtained advice that Solo Model trading did generate valid tax refund entitlements, when perfectly obviously neither was the case.

Sanjay Shah in cross-examination was somewhat more frank about the limited nature of what Hannes Snellman were asked to consider, but not very coherent in his argument overall. In my judgment, he found himself unsure whether to push the line that although there were limitations to it, Hannes Snellman’s advice did confirm that Solo Model tax reclaims would be valid under Danish tax law, or to stick to the different line, closer to the truth, that Solo was not seeking that type of advice at all, but rather something more akin to a ‘marketing tool’ for recruiting potential clients, i.e. tax-advantaged equity buyers for Solo Model trading. That tension in Sanjay Shah’s evidence was created by the fact that he has claimed to have thought that Solo was cleverly exploiting a Danish tax law loophole, but he knows that Solo never sought advice on whether any loophole that might exist extended to cover transactions where no shares were ever acquired.

I was unimpressed by those efforts to explain away the fact that Solo did not do what one would expect if it had been looking for advice on whether the Solo Model generated valid tax reclaims. I draw the firm conclusion that that is not what Solo was doing. For a more positive finding, the best explanation for what they did, in my judgment – in fact, I think, the only plausible explanation – is that their focus was on whether Solo Model equity buyers, who would know and see only something like the transactional facts that Hannes Snellman were asked to assume, would or might reasonably conclude that they had a right to a tax refund, or at least an arguable position on the basis of which they might be comfortable filing a claim with SKAT. That is not dissimilar to Sanjay Shah’s sometime notion that the Hannes Snellman advice was a marketing tool. It is not quite the same in that, I find, it was never intended to be, and generally was not, shared outside Solo.

Rajen Shah sent a copy of the First HS Advice to the Argre Principals; but they were rather different to all other principals behind Solo Model (and Maple Point Model) equity buyers, given their backgrounds and history of involvement with Solo. In sending it to them, Rajen Shah invited them to find it a “riveting read”, obviously a bit of irony (Mr Shah said in cross-examination that it was “tongue in cheek”). In my judgment, there was only sensibly room for such flippancy if Mr Shah did not think the Argre Principals would be interested in it, because they would know it was not for them to rely on confidential legal advice where they were not the advisor’s client and would be deciding for themselves whether to participate, and if so by reference to what, if any, legal advice that they might choose to obtain. I considered whether, perhaps, it might corroborate Mr Horn’s evidence that the Argre Principals knew before they decided to participate, and when participating, how Solo was going to enable their USPFs, or the USPFs of others whom they might introduce, to trade without funds or funding external to the traded structure itself. In my view, it does not, although I think it plausible that the Argre Principals may have assumed that Solo had a full transaction structure in mind supporting the trading activity posited for the USPF and if so may have noted that that wider supporting structure was not set out or explained. That would naturally be considered to be Solo’s ‘IP’, and I do not think there is a sufficient basis in evidence to find that the Argre Principals were troubled by not having all the detail. To proceed on that basis would be speculative.

SKAT relied heavily on the inadequacy of Hannes Snellman’s advice, if it was supposed to be advice on whether the Solo Model generated valid tax reclaims. I agree with SKAT that it did not provide any reasonable basis for considering that Solo’s magic ingredient of share-less equity trade settlement had no impact on possible tax refund entitlement. However, I do not consider that the way Sanjay Shah and his colleagues went about obtaining Hannes Snellman’s advice evidences preparation to commit the fraud that SKAT has alleged in these proceedings. They did not consider (and SKAT did not allege) that tax reclaims would state that there was a tax refund entitlement. They did not tell the truth at trial about the Hannes Snellman advice, in the face of a certain relentlessness of SKAT’s insistence that it was central to the case. They had, in consequence, a general unwillingness to admit what might be thought the unattractive truth that they constructed what became a tax reclaim industry on a massive scale without ever having, or having any solid basis for, an opinion that the reclaims were valid under the applicable rules of Danish tax law. But in my judgment they never saw what they were doing as setting things up so that SKAT would be or were being told untruths.

Denmark was not the only jurisdiction that Solo investigated. The Solo Model was also deployed in Austria and Belgium, and a decision was made not to make the attempt in some other jurisdictions. In closing argument, Mr Graham KC for SKAT initially developed a submission that in that planning phase, Solo was actively seeking to identify soft targets to exploit. That submission was not pressed, however. Mr Graham accepted, on reflection, that it was a new point, materially different to anything put in cross-examination. However, I agree with the submission that remained open to SKAT, namely that the evidence on which it had been based negatived any suggestion that declining to extend the Solo Model to other jurisdictions evidenced a focus on the validity under local law of tax refund claims. Without contemplating the more cynical notion of picking soft targets, in my judgment the evidence did show that Solo’s focus was the tax authorities’ processes and the documentary requirements involved in them. If Solo Model trading would generate the documents enabling claims to be filed, the view at Solo, I find, was that it was worth putting Solo Model trades on and seeing whether they resulted in successful outcomes, i.e. tax reclaims that the tax authority decided to accept and pay.

The First, Second and Third HS Advices were all structured in the same way, so far as material:

The USPF’s transactions posited in the First HS Advice, then, were as follows:

“3 CONTEMPLATED TRANSACTIONS

USPF contemplates to make the following transactions:

1)

USPF purchases Danish exchange traded equities (the “Equities”) (either via a regulated inter-dealer broker or via purchasing Eurex single-stock, physically delivered listed flex future contracts regarding the relevant Equities);

2)

The purchase of the Equities will take place prior to the ex-dividend date but no later than the dividend approval date meaning the date where the dividend is finally approved for distribution (generally meaning the date of the annual general shareholder’s meeting) (“Dividend Approval Date”). The settlement date for the purchase of the Equities will be on or after the dividend record date. In the case of physically delivered listed futures contracts, the expiry date of the futures contract will be no later than the Dividend Approval Date;

3)

USPF will receive the dividend which has been declared on the Equities on Dividend Approval Date (the “Danish Dividend”);

4)

On purchase of the Equities, USPF will hedge its long exposure on the Equities by selling exchange-traded, cash-settled single stock futures over the Equities (the “Short Derivative”). The price of the Short Derivative will take into account a proportion of the expected Danish Dividend in calculating the strike price of the Short Derivative but no adjustment will be made if the actual dividend amount is more or less than the expected dividend amount;

5)

On or after the ex-dividend date USPF will sell the Equities via an inter-dealer broker. The settlement date for the sale of the Equities will be the same date as the settlement date of the purchase of the Equities in 2) above. At the same time, USPF will enter into a long futures position over the Equities in order to close out the short futures position created in step 4 above The long futures position will expire on the same date as the short futures position created in step 4 above;

6)

It is possible that in the period until sale, the Equities would be lent out under a stock loan (using a standard Global Master Securities Lending Agreement) to an unrelated party (which could be located in any jurisdiction) in order to minimize financing costs. The period of the stock loan is not expected to exceed 3 months and the stock loan would be cash collateralised. The stock loan transaction may be entered into on or after dividend record date (with settlement of the stock loan occurring on the same day) but always after Dividend Approval Date and the borrower of the Equities would not receive any dividend on the Equities.”

The language of ‘purchase’ and ‘sale’ seems to connote contracting (to buy and sell), as with my use of that language in this judgment, because the ‘purchase’ is said to occur prior to a dividend ex-date, for settlement on or after the associated record date. The transaction pattern posited, read literally, makes no sense. It proposes that:

That impossibility, then, is created by the fact that the posited sale, terminating the USPF’s investment, is for settlement on the same date as its (possibly cum-ex) purchase. That element of the fact pattern was introduced by Rajen Shah at the last minute, relatively speaking, in an iterative process of working up the First HS Advice with Hannes Snellman. Mr Shah introduced it as part of tracked changes he sent to Hannes Snellman on 25 June 2012, and the First HS Advice was finalised and issued on 27 June 2012.

Hannes Snellman were first given a fact pattern to consider in an email from Mr Horn on 23 May 2012. In that fact pattern, and thereafter until Rajen Shah’s late amendment, the clear impression given was that there would be an interval of unspecified length (but not expected to exceed three months) between the USPF acquiring the shares and the USPF later selling them, during which interval the USPF might lend the shares out to minimise financing costs. Although not much of his evidence about this, or anything else, was satisfactory, on balance I accept Rajen Shah’s evidence that, at all events in the final version, the stock loan possibility was intended to be a strict alternative to immediate re-sale. He accepted that it would have been better to spell that out, for example by amending paragraph 6) of the fact pattern so that it began, “It is possible that, instead, the Equities will be sold later, and until that sale, …”.

Rajen Shah claimed that he made that clear to Hannes Snellman on the telephone. I do not accept that evidence. It is clear from the correspondence that there were telephone discussions with Hannes Snellman (also involving Mr Horn), but there is no hint that Hannes Snellman had taken this ‘alternative structures’ point on board, and, as will become clear below, I find that Hannes Snellman did not do a good or careful job with this advice, so I do not infer that they must have understood the point because they did not protest that paragraph 6) of the fact pattern had become nonsensical. I think it more likely that Hannes Snellman did not spot the tension created by Mr Shah’s amendment to paragraph 5) of the fact pattern. In itself, that amendment was apt only to cause Hannes Snellman to apply their mind to whether settlement of an ex-div sale on the same date as settlement of (what might be) a cum-ex purchase would affect the Danish tax law analysis; and if the clash it created with paragraph 6), as it stood, had been noted, I think it unlikely that Hannes Snellman would have left it as it was.

On any view, the fact pattern gave as fixed points that:

The fact pattern did not clarify or elaborate what was meant by the proposition that because of the equity purchase the USPF would “receive the dividend”, either generally or in particular if the purchase was cum-ex, which was within the fact pattern.

The brief to Hannes Snellman therefore was, or included, a brief to advise whether a USPF which bought Danish shares cum-ex but nonetheless in some unspecified sense “receive[d] the dividend”, and sold ex-div for settlement at the same time as its purchase, would be entitled to a refund of Danish dividend tax, and whether a stock loan by the USPF entered into on or after the ex-date would affect the answer.

Against the background of the hypothetical transactions described in Section 3, in Section 4 the assumptions that are material for present purposes were stated as follows:

“4 ASSUMPTIONS

For the purpose of this Opinion, we have assumed that:

…

5)

On purchase of the Equities, USPF will obtain unconditional ownership to the Equities and will have full ownership rights over the Equities on Dividend Approval Date including the right to receive the dividend declared on that date;

6)

The Equities will be held by USPF through a custodian. USPF’s ownership to the Equities will be recorded with the custodian and, depending upon the custodian’s and sub-custodian’s other long and short positions, the Danish Securities Centre;

7)

The custodian’s records will show that USPF is the legal and beneficial owner of the Equities on Dividend Approval Date and that the Danish Dividend represents a real dividend on the Equities which is passed on to USPF by the custodian.”

Those assumptions are not ordinary matters of fact. For the most part, they state matters or consequences of law. Assumption 5), for example, is all law. It said that on purchase, i.e. by concluding the purchase trade, the USPF would obtain (a) unconditional ownership of Danish shares, with (b) full ownership rights over those shares on the dividend declaration date, and (c) the right to receive the dividend declared on that date. In an advice as to the Danish tax law consequences for the USPF, given those assumptions, the assumptions could not sensibly have been stating conclusions of Danish tax law. They could only sensibly have been understood as defining the position under the general law that Hannes Snellman said they were assuming would exist, upon the basis of which they were offering an opinion on the Danish tax law treatment.

Assumption 6) stated as assumption that the USPF’s “ownership to the Equities”, which harked back to Assumption 5), would be recorded with its custodian, and may or may not be recorded at VPS, depending on the detail of the custody chain. Assumption 7) reinforced the position, making clear that it was assumed that, in particular even though VPS may not have any record of the USPF’s share title, the USPF’s custodian’s records would show it to have had title on the dividend declaration date, and would show that the payment it received was a real dividend payment being passed on by the custodian. Sanjay Shah, Rajen Shah and Mr Horn all knew at the time that none of that would be true for Solo Model trades:

By the supporting explanation and analysis, Hannes Snellman advised that as a general rule, the liability to Danish dividend tax was that of “the person or entity … registered as the holder of the dividend paying shares on Dividend Approval Date”. They said they believed “that USPF must be considered the recipient of the Danish Dividend from a Danish domestic law perspective”, so that “the Danish Dividend should be taxed in the hands of USPF”; but that was “[b]ased on our understanding that USPF will have full ownership rights over the Equities on Dividend Approval Date and … would only lend out the Equities after Dividend Approval Date with no dividend rights attached”.

The First HS Advice did not refer to any special rule of Danish tax law as to when ownership of equities (and therefore entitlement to dividends) was taken to have accrued. It did not consider how, if at all, including in what settlement circumstances, it might be possible, given the transaction pattern described, for the USPF buyer to “have [under the general law] full ownership rights over the Equities on Dividend Approval Date” (Assumption 5), or to receive something that “represents a real dividend … passed on … by the custodian” (Assumption 7).

As I read the First HS Advice, “the Equities” meant throughout real Danish shares, not something synthetic or derivative, referable to but not constituting shares. That means that Assumption 5) was an impossibility, creating severe difficulties for Assumption 7), even without knowing in detail the Solo Model settlement mechanics. Yet Hannes Snellman did not qualify their opinion by reference to any of that, or refuse to offer an opinion because of it. In expressing their opinion, Hannes Snellman said in terms that it was on the basis of the stated assumptions, but it is not competent or careful advice for an expert adviser, giving advice on stated assumptions, to fail to note and explain that parts of them seem to be impossible, if it is within the adviser’s professional competence to identify and understand that. The obvious danger of stating assumptions without cautionary advice about them is that the adviser may be taken by the client to be advising, implicitly, that (other things being equal) the assumptions should be satisfied, or at all events that there is no reason that it would be within the adviser’s competence to identify why they could not be satisfied or it is otherwise unreasonable to make them as assumptions.

Assumption 7) was added by Hannes Snellman after exchanges and discussions between their Anne Becker-Christensen, Rajen Shah and Mr Horn. Ms Becker-Christensen explained in a covering email on 26 June 2012, attaching in final draft form what became the First HS Advice, that Hannes Snellman had added Assumption 7) because, “We need the opinion to reflect somehow that the payment for which USPF will claim treaty benefits is de facto a dividend payment which USPF has received, as a shareholder, from the dividend distributing Danish company (via the custodian)”. SKAT relied on the history to that observation, which was as follows:

The upshot, as SKAT submitted, is that (a) the First HS Advice can only sensibly be read as assuming that the USPF would acquire full title to real shares in respect of which (therefore, by definition) there would be a complete custody chain, and that, although VPS might or might not have a record that the USPF sat at the top of the custody chain, what was paid to the USPF would be a real dividend payment, passed on to it by its custodian, and (b) lest there might have been any doubt about that, Ms Becker-Christensen had made it clear in her emails that that was Hannes Snellman’s intent.

Hannes Snellman thus did not advise, because they were not asked to advise, that the Solo Model would or did result in the acquisition of any ownership rights to shares. They advised only that, if under the transactions described such rights were acquired, then for tax purposes they would be treated as having been acquired in time for the USPF to be considered the party liable to Danish dividend tax and entitled in principle to claim a refund.

The email dated 1 July 2013 referred to in paragraph 299 above was from Ms Becker-Christensen to Jessica Hammers in SCP’s legal department, following a telephone call between them. By the email, Ms Becker-Christensen confirmed her advice in these terms:

“As a general rule, dividends distributed by a Danish company are subject to Danish tax in the hands of the owner of the shares, i.e. the person who has legal ownership to the shares on the date where the dividend is declared … . The taxable event is the declaration of dividend and not the receipt of the dividend. Therefore, the Danish dividend tax is not cancelled if the dividend for some reason is never received by the owner of the shares.

With respect to dividends on Danish shares which are subject to a stock loan arrangement, … the dividends according to Danish practice are taxed in the hand of the lender and not the borrower if the borrower has an obligation to pay a manufactured dividend to the lender.”

That was inaccurate advice in that, without further explanation, one would not know that Danish tax law took a bespoke view, different from that of the general law, on the identification of the owner of shares. The email adopts the simple understanding of an accrual rule of taxation that I remain residually concerned might in fact be the true rule of Danish tax law, even though I was persuaded, on balance, to find otherwise, so that I am now bound to say that the email was inaccurate: see as to that, SKAT (Validity Issues), supra, at [184], [193]-[195], [308]. Be that as it may, there is no hint in the email that Hannes Snellman understood they might be advising in relation to trading in which no party ever owned shares at all.

The Second HS Advice was, for my purposes, materially similar to the First HS Advice, as regards the possible entitlement of the posited USPF to a refund of Danish dividend tax. It now made clear that SCP might be the USPF’s custodian and, again expressly on the basis of the hypothetical facts in Section 3 and the assumptions in Section 4, advised additionally in Section 5 that:

“1)

The custodian services which may be carried out by [SCP] in relation to the Equities (a) does [sic.] not entail that [SCP] has established a permanent establishment in Denmark for Danish tax purposes and (b) should not trigger any adverse Danish tax consequences for [SCP]; and

2)

In the event that USPF’s reclaim of Danish withholding tax withheld by the Danish tax authority is rejected, no economic penalties should be imposed on USPF or [SCP].”

The Third HS Advice was materially similar to the Second HS Advice, but it considered hypothetical share transactions by LabCos rather than by USPFs. At step 3 in the transaction description, it was made explicit that the dividend it was said would be received by the LabCo would be “of an amount equal to the gross dividend net of the relevant rate of Danish withholding tax”; and at step 6, a repo was stated to be an alternative to a stock loan: “Alternatively, the Equities may be refinanced under a sale and repurchase agreement (i.e. a repo) documented under a Global Master Repurchase Agreement … entered into on or after dividend record date (with settlement of the stock loan [sic.] occurring on the same day) but always after Dividend Approval Date and the buyer [viz., the repo buyer, not the USPF] would not receive any dividend …”. That, I think, reinforces the conclusion I reached, on balance, about the First HS Advice, namely that steps 5 and 6 of the hypothetical fact pattern in Section 3 should be read as alternatives to each other.

Finally, the email dated 12 February 2014 referred to in paragraph 299 above was from Ms Becker-Christensen to Priyan Shah at Agrius Capital (the services company through which he and Mr O’Callaghan worked for Sanjay Shah/Solo), but copied to Sanjay Shah, among others. By that email, Ms Becker-Christensen confirmed that the analysis detailed in the Second and Third HS Advices would not change if the price hedging was via cash settled forwards rather than exchange-traded futures, provided that “the counterparties to the cash settled forwards are: 1) unrelated, independent parties and 2) are not counterparties to the equity trades in question”.

I was also shown written opinions from Hannes Snellman in June 2013 and February 2014 addressed to Belador Advisers UK Ltd (the ‘Belador Advice’). The Belador Advice was not seen at the time by any of the trial defendants. It was provided by Hannes Snellman in relation to the structured trading organised by MCML (when it was still ED&F Man), on which see paragraph 26 above. The Belador Advice referred to a ‘USPP’ rather than a ‘USPF’ (i.e. ‘Plan’ rather than ‘Fund’). It had the same general structure and format as the First to Third HS Advices, and much of the content is similar, but there were differences.

One of the differences was an addition in Section 3 that, “USPP may receive the Danish Dividends as a compensation payment from its custodian based on the fact that (a) the Share Purchase may settle on a T+4 settlement basis or (b) the settlement of the transfer of the Danish Equity under the … Share Purchase … may be delayed because of settlement failure”. There was then an additional opinion given, relating to that possible fact, that “If the … Share Purchase due to an agreed settlement in accordance with a T+4 settlement cycle or a settlement failure settles after the dividend record date, USPP should be able to claim a full refund of the Danish dividend withholding tax on received dividend compensation payments pursuant to the Denmark – US Tax Treaty provided that the conditions discussed in Section 7.1 below are met”.

Hannes Snellman should not have used language suggesting that the dividend compensation payment would have been the subject of Danish dividend tax or that its receipt would have generated an entitlement to a tax refund. The point of substance in the additional opinion is advice that it would not defeat the USPF’s entitlement to a tax refund if the only payment it received, because of settlement being after the record date, was a dividend compensation payment. That entitlement, according to the advice, still was to a full refund of the Danish dividend tax that had been imposed on the real dividend, arising because of the assumption (as in the First to Third HS Advices) that the USPF became the full legal and beneficial owner of real shares prior to the dividend declaration.

In ‘Section 7.1’, setting out the conditions the USPF would need to satisfy to be entitled, Hannes Snellman advised inter alia that it would be “crucial” that it could provide, inter alia:

“Documentation that the Danish Equity was purchased pursuant to an unconditional and legally binding agreement entered into prior to the ex-dividend date and that the Danish Equity according to such agreement was purchased on a cum dividend basis. Further, USPP must be able to document that the Danish Equity has subsequently been delivered, ideally only a few days dividend record date [sic., after that date].”

That confirms that in Hannes Snellman’s view a cum-ex buyer whose purchase was settled by a transfer of shares on the settlement date was a shareholder for tax purposes in respect of a dividend declared between the trade date and the settlement date, under a ‘contract accruals rule’ of Danish tax law, if it had not committed prior to the dividend declaration to a disposal of those shares to another. That oddly contrasts with the view seemingly given to SCP in the First HS Email (see paragraphs 326 to 327 above). I am also bound to find that the view given in the Belador Advice was incorrect under Danish tax law as determined for these proceedings. That is because I held on the evidence at the Validity Trial that the contract accruals rule does not mean that a trade prior to a dividend declaration that settles by a transfer of shares in due course is sufficient to create tax ownership of shares, because the seller may have been short when the trade was entered into and when the dividend was declared, yet still have achieved that physical settlement when the time came: see SKAT (Validity Issues) at, e.g., [247], [306]-[307], [309] and [310]-[313].

I have to find that the Belador Advice was inaccurate in that way, but it evidences that it would have made no difference to any of the First to Third HS Advices if Section 3, the fact pattern put to Hannes Snellman, had made it clear that the USPF’s (or LabCo’s) equity purchase would always be a cum-ex trade. That does not mean it was Hannes Snellman’s view that tax ownership of shares could be created by a trade that did not settle by a share transfer, DVP.

Whether Hannes Snellman would have articulated it precisely as I did on the expert evidence at the Validity Trial, I consider it clear on the whole of the evidence available as to Hannes Snellman’s opinions that in their view, as I found, “if a buyer never receives, by physical performance (share transfer), the shareholding seemingly contracted to be sold to them, they will not incur dividend tax liability on dividends declared after the contract was concluded (or at all)” (SKAT (Validity Issues), supra, at [228]). I am confident overall, and find, that Hannes Snellman would not have issued a positive opinion letter on the eligibility of a USPF (or LabCo) for a dividend tax refund if they understood that on the facts (actual or assumed) by reference to which they were advising, no shares would ever be transferred to the USPF (or LabCo). As Ms Becker-Christensen made clear to Rajen Shah and Mr Horn, in Hannes Snellman’s analysis the USPF (or LabCo) had to have acquired a real shareholding. Therefore, not only can it be said that given the hypothetical factual basis and explicit assumptions upon which they advised, Hannes Snellman did not advise that the Solo Model share-less settlement method resulted in tax refund entitlement, in my view it is clear, and was clear at the time to Rajen Shah and Mr Horn, that, if asked, Hannes Snellman would have advised that it did not.

That factual basis, and those assumptions, mean that Hannes Snellman had been asked to assume that the USPF (or LabCo) was a dividend date shareholder (as I defined that term in SKAT (Validity Issues), supra, at [166]). In effect, all they were being asked to consider was whether, if a dividend date shareholder was hedged against share price movement and traded, after the ex-date, to sell their shares or lend them out ex-div, the price hedge or the ex-div disposal affected things. Their advice was nonetheless poor, irrespective of points of detail that might be made on the basis of SKAT (Validity Issues), because it should have been obvious to Hannes Snellman that as a matter of law the posited dividend date shareholding was an impossibility.

Not only did Hannes Snellman not qualify their opinion because of that, or explain it, they introduced the relevant assumption. It did not come from Solo, but arrived as Assumption 5) with Hannes Snellman’s first draft for their advice, and remained throughout the process of finalising it. It is a little difficult to know what to make of that, without evidence from Hannes Snellman to explain their thinking. It is plausible, I think, that Hannes Snellman may have introduced the assumption to ‘join the dots’ in the fact pattern between paragraphs 1) and 2) (describing the equity purchase trade) and paragraph 3) (asserting that as a consequence the buyer would “receive the dividend”); but there is in truth no evidence to go on and I do not think I can make any finding as to what was their thinking.

Whatever the thinking, I do consider it surprising that Hannes Snellman did not spell out that not only was their opinion an ‘IF’ opinion (if the facts are as stated, and if the assumptions stated hold true, then …), but also that, for reasons of law upon which they were qualified and might be expected to advise, on the face of it the key assumption they had themselves introduced could not hold true. It is like asking a firm of shipping market tax lawyers to advise on the tax implications of some cash flow under a time charter, and being given by them an opinion that states as an assumption that the time charterer enjoys a right to possession of the ship and opines that on that assumption the cash flow will have a certain tax treatment, without a note of caution that the assumption, on the face of things, could not hold true because a time charter does not grant a right to possession, it creates only a contractual right to direct employment.

Standing back, as I said at the start of this section of the judgment, it would have been simple, if Solo had been interested to know the tax treatment of the actual trading, to set it out, accurately and in full, and ask that question. That was not done, and in my view Sanjay Shah, Rajen Shah and Mr Horn knew it had not been done. In truth, they were only taking advice from Hannes Snellman to get an idea as to whether a Solo Model equity buyer who saw and knew about only their own trades might believe that they were entitled, or at least take the view that they had a ‘filing position’, i.e. a plausible argument that they might have an entitlement.

Some of the trial defendants relied on an absence of complaint or protest from SCP’s compliance or legal functions as fortifying the belief they say they had that there was nothing improper about the trading. How far that can take any given defendant would depend on their circumstances and knowledge, as well as those of the compliance or legal personnel said to have appeared to be content. I therefore do not deal with those defence arguments in this general narrative except to dispose of one point taken by SKAT, which goes back to Mr Pitts’ departure from Solo mentioned in paragraph 171 above.

Mr Pitts’ handover file note, prepared on 29 August 2013, said this as regards the Solo Model trading to that point:

“GSS Platform and legal advice – there is no evidence that the GSS platform as a whole has been reviewed by a single law firm to provide senior management with external comfort that it is fit for purpose. CMS Cameron McKenna opined on the client money/TTCA side (and concluded that we could just about justify it) and they opined on the Custody agreement, but this appears to have been in isolation and they did not see all the other agreements needed to make this work. JB [Mr Bains] and GH [Mr Horn] did agree to having a lawyer look at this, then backtracked. I raised the issue again only for SS [Sanjay Shah] to say no in writing. At present senior management is exposed if the model is wrong – and GP [Mr Pitts] has advised this. In my opinion the firm urgently needs to have a lawyer look at all the docs and have a full description of how the process works so they can confirm whether JB’s work and the assorted advice from different advisers works together.”

I accept Mr Dhorajiwala’s evidence that the holistic review by an external lawyer that Mr Pitts had advocated concerned SCP’s regulatory compliance, that the single law firm he (Mr Pitts) had in mind would have been a London firm advising under English law, and that his concerns were not directed at the foreign tax law validity or invalidity of the tax refund claims the SCP clients were intended to make on the back of Solo Model trading. I also accept Mr Bains’ evidence that he did not backtrack, as Mr Pitts put it, on support for the commissioning of such a review. Any reluctance will have been that of Sanjay Shah and Mr Horn.

Two legal opinions were referred to in evidence in connection with the setting up or implementation of Maple Point Model cum-ex trading:

The BB Advice was obtained for NCB by the Frankfurt office of Jones Day, which was advising NCB in connection with the possibility of acting as custodian in Maple Point Model trading. Dr Martin Bünning, the responsible partner at Jones Day, made initial contact with Anders Orbey Hansen of Bech-Brunn, explaining that advice was sought:

In the email exchange that followed, Bech-Bruun explained that “[t]he type of transaction you are describing, which may entail a possibility for two shareholders to reclaim the same withholding tax has been presented to us before (“cum-/ex trades”)” and that such trades were “generally problematic from a legal perspective”, “typically problematic under Danish law”, and “problematic from a Danish legal point of view”, so that providing a ‘should’ opinion, viz. that NCB “should not be subject to criminal prosecution” or similar, would require Bech-Bruun to have “a full overview of the transaction and the role of the client, including what takes place in Denmark”. When in mid-April 2014, a transaction description was given to Bech-Bruun by Jones Day, Bech-Bruun took it to indicate that “USPF acquires cum dividend shares from a Seller which does not own (or delivers) cum dividend shares” and made clear that if that was the position, then “by definition USPF cannot be a holder of cum dividend shares on the dividend approval date, which in our view is a condition for being considered owner of the shares and thus the dividend itself for Danish tax purposes.”

In the SSDs’ Defence, it was said that tax advice relating to other jurisdictions fortified a belief in Sanjay Shah that a share buyer becomes the owner of shares the moment a trade is entered into, whether or not the seller was selling short and whether or not shares were ever delivered. I do not think this was intended to suggest that legal advice (if there was any) that a particular cum-ex trading strategy would work for (say) German shares and German withholding tax refunds could inform a view that the same structure would work for Danish shares and Danish dividend tax refunds. Sanjay Shah’s evidence indeed was that “at no point did I (nor [did] anyone else at Solo to my knowledge) believe that a positive legal opinion about a cum-ex trading structure in one country could be relied upon to say that it would be valid in another country”. To similar effect, Rajen Shah acknowledged in evidence that “it was necessary to obtain legal advice in respect of each jurisdiction in which equities trading was to occur and … you could not assume that the tax position in one jurisdiction would be the same as the tax position in another”.

Klar Model trading, between March 2012 and May 2015, appears to have been Mr Klar’s principal occupation in 2012, while in 2013 to 2015 it sat alongside lucrative activity as a trading counterparty in both Solo Model and Maple Point Model trading. The Klar Model used Salgado as custodian. Salgado was a Comoros Islands company Mr Klar incorporated for the purpose. He used a simplified cum-ex trading model to generate successful tax refund claims dated between December 2012 and May 2015 for an aggregate amount paid out by SKAT of DKK321m.

Mr Klar qualified as a chartered accountant in 1997 and worked at Shell, as a tax structurer, then at Enron, as director of a tax department, between 1997 and 2001. After the collapse of Enron in 2001, he joined ABN Amro as director of Structured Debt & Equity working on tax-related projects, but not cum-ex. He was introduced to Sanjay Shah by Rajen Shah, probably in 2008, and became one of the first recruits to the Solo business.

In September 2010, Mr Klar described his role at SCL as “Principal of Solo Capital’s start-up fund management business” in charge of “structuring … vehicles to trade market neutral equities strategies over listed Western European shares”. He was a Co-Chief Investment Officer, together with Rajen Shah, and he was SCL’s Compliance Officer, as a result of which he was among other things the designated person with responsibility for the FCA-regulated controlled functions at SCL of money laundering reporting (from 20 September 2010 to 17 January 2011), significant management (from 20 September 2010 to 3 January 2012) and customer-dealing (from 7 January 2010 to 3 January 2012).

The basis on which Mr Klar was engaged was changed from employment by SCL to self-employed consultancy. This was documented only from late 2011 but the changed basis was backdated to December 2010, so that his employment by SCL was treated as having ended on 20 December 2010 and his engagement (by Sanjay Shah) as a consultant was treated as having started on 21 December 2010. The change in the basis of his engagement, whether before or after it was documented, did not make any difference to the work he was doing and did for Solo.

At Solo, the idea to use internalised settlement with settlement loops settling to zero, to avoid any need for funding or the acquisition of shares, as a foundation for dividend tax refund claims, came from Mr Klar and Sanjay Shah, as I have described above. Mr Klar was no longer at Solo, in any capacity, by the time Solo Model trading commenced, as he had left, with effect from 3 January 2012, to pursue his own interests. They came to include participation as a stock lender and (later) a forward counterparty, at Sanjay Shah’s invitation, in whatever trading strategy he (Sanjay Shah) had finally settled upon.

Mr Klar insisted, in evidence that I accept, that he was not told in any detail what that strategy was. However, as he accepted, he did not need to be told. When he did the trading himself (in 2013, before he had Mr Sethuraman trading for him), he was a stock lender borrowing and lending on back-to-back terms from and to SCP clients, where always the lender to him was a USPF and the borrower from him was not, trading on a dividend record date for next day settlement with cash collateral calculated at a cum-div price from before the ex-date, with no margin or haircut, and his reward for doing the trade was 0.5% of the dividend amount, payable not by his lender or his borrower but by Ganymede. To Mr Klar, it was obvious, and he in fact assumed, that he was participating in a coordinated cum-ex trading strategy designed at Solo around the idea of settling internally so that neither cash nor shares would be required, in order to facilitate the making of a tax refund claim to SKAT by the end buyer within the structure.

Mr Klar’s Solo Model activity was carried out through his company, Amalthea. Amalthea was incorporated on 14 February 2013 and over the rest of that year acted as stock lender on dozens of Solo Model 2012/2013 trades. In 2014, Amalthea was a forward counterparty on hundreds of Solo Model 2014/2015 trades, and another company of his, Cork Oak, after incorporation on 7 March 2014, also participated, as a forward counterparty, in 250 settlement loops in Solo Model 2014/2015 trades between March and December 2014. Mr Klar insisted, and I accept, that all of the individual decisions to conclude those 2014 and 2015 trades will have been made by Mr Sethuraman, as Mr Klar left him to get on with the trading once he was in post in time for the first 2014 trades.

Through Sherwood and Potala (incorporated, respectively, on 6 and 11 February 2014), Mr Klar also participated in Maple Point Model trading. Sherwood was onboarded by NCB, Potala by Indigo and (later) Lindisfarne. This participation came in response to an initial invitation from Rajen Shah, the later invitation to trade via Lindisfarne then coming from Mr Horn. Sherwood acted as both stock lender and forward counterparty via NCB; Potala acted as stock lender via Indigo, and as both stock lender and forward counterparty via Lindisfarne. Mr Klar appreciated that the transactions of Sherwood and Potala were, like those of Amalthea and Cork Oak, elements of coordinated structures designed to create circular settlement loops settling to zero that would generate CANs from the custodians.

Returning to the Klar Model, Mr Klar devised, orchestrated and participated in the entirety of the trading, involving himself in all aspects of it and preparing all the trading documents including the CANs issued by Salgado. The other key participants were his companies, Salgado, Europa and Khajuraho, plus Blue Ocean and Cole, which were corporate vehicles (limited liability companies incorporated in Illinois) of Kevin Kenning and Todd Bergeron.

Europa was Mr Klar’s English pension scheme, founded by him on 3 December 2010, and Khajuraho was a company he incorporated in Luxembourg on 23 May 2012. Under Denmark’s DTTs with, respectively, the UK and Luxembourg, each of Europa and Khajuraho was entitled not to pay more than 15% in Danish dividend tax. The tax refund claims made on their behalf, and paid by SKAT, are the only instances with which I am dealing where the reclaim was for a partial refund (12%, given the WHT rate of 27%) rather than a full refund. Blue Ocean and Cole, for their part, each established a USPF that became a client of Salgado, and in due course full refund claims were presented to SKAT on behalf of those USPFs, and paid by SKAT.

Salgado was incorporated on 28 February 2012 in Anjouan, a Comoros Island, with a share capital of US$1. None of its customers ever deposited any funds with it, its only receipts being proceeds received from successful dividend tax refund claims made to SKAT and its counterparts in other jurisdictions, in particular Belgium and Austria.

Mr Klar had control of Salgado, and was its directing mind and will, at all times. He was its only authorised trader, made all of its trading decisions and prepared all trading documents, was in charge of all brokerage and custody services, and prepared all broker and stock loan confirmations, and all Salgado CANs. In practical terms, Salgado was Mr Klar, documenting trades and their settlement using standard office software products, without any connection of any kind to any equity trading or financial market.

Nominal ownership of Salgado was not with Mr Klar, however. His evidence was that he wanted someone else to own Salgado because he intended to trade cum-ex and “did not want to give any tax authority an excuse, no matter how invalid it might be, to be difficult about approving WHT reclaims”. I consider it more likely that Mr Klar simply wanted Salgado to appear to be independent of Europa and Khajuraho, to disguise the fact that in Klar Model trades involving them, Mr Klar was effectively on both sides of the trade.

The result was that Mr Klar arranged for a corporate service business in Anjouan to hold the shares in Salgado, via Hillingdon Turner Nominees Ltd from 28 February 2012 to 11 June 2012 and by the Salgado Charitable Trust, a Mauritian trust, thereafter. Mr Klar was in my judgment at all times the shadow director and true, but hidden, ultimate owner of Salgado.

As I noted above, Europa was an English pension scheme established by Mr Klar. It was his personal pension fund. He was its sole trustee and beneficiary and had absolute control and effective indirect ownership of all the trust assets. Europa was eventually wound up only in April 2021, although Mr Klar decided to wind it up in August 2016. Khajuraho was incorporated in Luxembourg, with Mr Klar as sole shareholder and ultimate beneficial owner. It had a gérante and office manager (Janice Allgrove), but she had a purely administrative role and played no part in Khajuraho’s trading which was directed entirely by Mr Klar. In November 2020, Khajuraho was placed in compulsory liquidation following a claim by the Luxembourg tax authorities, and it was dissolved by a judgment of the Luxembourg District Court on 15 May 2023.

Between December 2012 and May 2015, Goal made tax refund claims to SKAT on behalf of Europa and Khajuraho, supported by Salgado CANs, on which SKAT paid out a total of DKK114,049,140. Tax Agents made tax refund claims on behalf of the Blue Ocean USPF dated between 13 March 2013 and 7 May 2015, and on behalf of the Cole USPF dated between 17 April 2013 and 7 May 2015, all supported by Salgado CANs, on which SKAT paid out totals of DKK105,633,585 and DKK101,076,120 respectively.

The involvement of Blue Ocean and Cole arose out of a meeting in Dublin between Mr Klar, Messrs Kenning and Bergeron, and Barry O’Sullivan in the second half of 2012. Mr O’Sullivan was the principal of QED Equity (‘QED’), an Irish investment firm, who had discussions with Solo (Mr Klar, Sanjay Shah and Rajen Shah) in early 2011 about QED’s involvement in one of the German transactions at Solo. Mr Klar explained the trading strategy, in broad terms at least. That included, and Messrs Kenning, Bergeron and O’Sullivan therefore cannot have failed to appreciate, that the submission of tax refund claims based on the trading carried out was fundamental to the strategy, as the only source of realisable profit for any of the participants. Mr Klar explained to them (and used a whiteboard to illustrate in some way) that the strategy involved the short sale to the equity buyer being settled by netting it against a stock loan by the buyer, through Salgado, to the short seller.

Mr Klar’s case was that from 2013, the short seller in Klar Model trading was Heber Securities Trading Ltd (‘Heber’), a Maltese company of Mr O’Sullivan’s, rather than Salgado, which therefore acted as a matched-principal broker and custodian only. His case was that the Dublin discussion resulted in agreement for Messrs Kenning and Bergeron to participate in the Klar Model strategy through USPFs they were in a position to establish, and for Mr O’Sullivan to participate as the short seller for all Klar Model trades.

Mr Klar disclosed no documents evidencing any involvement of Heber in Klar Model trading. His disclosed documents evidence only trading by Salgado with Europa, Khajuraho, and the Blue Ocean and Cole USPFs, with Salgado acting as custodian, broker, short seller and stock borrower / stock re-purchaser, and no external brokers, intermediaries or additional counterparties. However, the buyer-facing transaction records generated at Salgado would be the same whether it was backing the trades with Heber, as matched-principal broker, or trading for its own account. Mr Klar explained the absence of documents showing Heber’s back-to-back involvement thus:

In Mr Klar’s criminal trial in Denmark, he and Kevin Robinson, a solicitor, gave evidence that indeed Heber acted as short seller, while Messrs O’Sullivan and Kenning gave evidence that Mr O’Sullivan merely acted as introducer, to Salgado, of Blue Ocean and Cole (meaning, in practical terms, Messrs Kenning and Bergeron). Neither version of events is without difficulty. Heber was only incorporated in May 2014, so it could not have been trading as a short seller from 2013. On the other hand, there was no reason for Mr O’Sullivan to be given (through Heber or otherwise) a 30% profit share (or, it may have been, a 33.3% share) of tax refund claim proceeds achieved by the Blue Ocean and Cole USPFs, let alone such a share also of those proceeds achieved by Europa and Khajuraho, as an introducer’s fee. Mr Klar knew Mr Kenning already, they having worked together at ABN Amro, along with Mr O’Sullivan.

Mr Klar was in my judgment wrong to suggest there was no such profit share, and likewise he was wrong to suggest that no profit sharing was agreed with Messrs Kenning and Bergeron to split the balance of tax refund proceeds net of fees and Mr O’Sullivan’s share. His evidence in that regard was part of his misguided attempt, effectively abandoned during cross-examination, to claim that the purpose of the Klar Model trading was not to generate successful tax refund claims as the only source of (realisable) profit for any participant. The absence of documentary records, and the lack of complete consistency of the testimony variously given by those involved, in these proceedings and elsewhere, means I cannot make a clear finding as to the agreed profit share percentages. I consider it possible that there was, as SKAT asked me to find, a simple, even, three-way split, between Salgado, Blue Ocean/Cole, and Heber; but Heber’s share may have been only 30%, not 33.3%, leaving 70% rather than 66.7% split between Salgado and Blue Ocean/Cole.

On balance, and although it involves rejecting Mr Klar’s evidence on the profit sharing arrangements, I believed and am prepared to accept Mr Klar’s evidence that Mr O’Sullivan joined in the Klar Model activity so as to act, and did act, as the short seller on the opposite side of Klar Model trades from 2013, and that he did so acting through a corporate identity of ‘Heber’. The unavailability to Mr Klar now of a full documentary record makes it a mystery on which I am not in a position to make any finding that Heber, the Maltese entity, came into existence only in May 2014. Logical possibilities include the existence prior to that date of a different ‘Heber’ entity (and after Mr Klar’s cross-examination, SKAT did locate evidence that a BVI company called Heber Trading Ltd had existed since 1 September 2009), or the use of the ‘Heber’ name by Mr O’Sullivan for Klar Model trading prior to having incorporated (the Maltese) Heber as a separate legal entity (and in response to the additional evidence located by SKAT, Mr Klar has insisted that Heber, i.e. a company by the name of Heber Securities Trading Ltd, was onboarded by him at Salgado before 2013 trading in reliance on what appeared to be corporate constitution documents for it as an extant entity).

As appears from Appendix 3, Mr Klar’s transaction model, in both of its iterations, had an elegant simplicity that might be said to have taken the ideas behind all of the Solo Model trading variants, or for that matter the Maple Point Model trading, to a logical conclusion. If they worked, in the sense of generating tax refund entitlements, why not also Mr Klar’s simple ‘buy and lend’ or ‘buy and re-sell’ ideas for trading between parties that had neither assets nor funding?

For its part, SKAT adopted that logic for an argument that made its way back up the chain of reasoning. Mr Klar’s trading, SKAT said, involved the idea that two entities without assets or funding could conjure out of thin air an entitlement to payments by SKAT, by trading with each other on Mr Klar’s laptop, entering into mutually offsetting contracts with each other and having them settle against each other by book entries at a custodian that held neither shares nor cash for either of them. SKAT argued that any claim to an honest belief that that worked could not be credible, let alone a claim of honest belief on the part of an experienced financial professional. But then, SKAT contended, likewise any claim to an honest belief that Solo Model trading or Maple Point Model trading worked, because by design those Models achieved nothing of substance different to or more than Mr Klar’s simplified version.

If that reasoning were sound, it would not entitle SKAT to judgment on any claim that was before this court, because that required SKAT to prove, for any given claim, not some generalised allegation of dishonesty, but the necessary ingredients of the claim, as pleaded. On that aspect, SKAT did not suggest that the facilitation or arrangement of invalid Danish tax refund claims gave it any cause of action, even if those involved did what they did knowing or believing that they were causing SKAT to pay invalid claims.

In any event, one must not rush to accept that sort of reasoning, powerful though it can be. Staying in Denmark, Hans Christian Andersen was familiar, as is any judge, with the capacity of the mind for delusion. Some of the courtiers and townsfolk in The Emperor’s New Clothes, perhaps many or even most of them, persuaded themselves, so as to have an honest belief, that they were seeing the finest of new imperial robes, until the young child, free from the persuasive influences in play, pointed out the objectively evident truth.

At the centre of Hans Christian Andersen’s tale lay con men posing as fine clothmakers who busied themselves pretending to conjure robes out of thin air. Whether any analogy with The Emperor’s New Clothes extends that far depends on whether SKAT established on the evidence not just dishonest behaviour of one or more kinds in and about the Solo Model, Maple Point Model and/or Klar Model cum-ex trading businesses, but the particular deceit it alleged by its pleadings in these proceedings.

Trial Witnesses

The principal events out of which SKAT’s claims arose occurred between 9 and 12 years before the Main Trial. They have been litigated heavily and more or less continuously for much of the period since SKAT temporarily suspended all WHT reclaim payments in late 2015, bringing an end to all Solo Model, Maple Point Model and Klar Model trading activity. That litigation has included these proceedings and civil proceedings of various kinds in other jurisdictions, and criminal proceedings in a number of jurisdictions, including criminal cases in Denmark against some of the trial defendants. By the time of the Main Trial, Sanjay Shah, Mr Patterson and Mr Klar were all in custody in Denmark, although Mr Patterson was transferred during the course of the trial to a prison in this jurisdiction to complete his sentence here.

Mr Klar was convicted after a contested Danish criminal trial. Mr Patterson was convicted on his guilty plea entered at the start of what was to be a joint trial of the charges against him and the charges against Sanjay Shah. Mr Shah spent about 18 months in custody in Dubai fighting extradition to Denmark before being taken to Denmark in early December 2023. He contested the charges against him there, and his criminal trial proceeded in parallel, for him, to the Main Trial here. The conditions in which Mr Shah was held in Dubai were so restrictive as to make effective involvement by him in his defence of SKAT’s claims here all but impossible. The conditions of his detention in custody in Denmark were incomparably different, such that, thankfully, he was able to participate effectively in the final months prior to, and during, the Main Trial here. Given that, and his freedom to engage in the defence of the claims against him prior to May 2022, I was satisfied that the process overall was fair to him, the difficulties of his being in custody notwithstanding, and no application was ever made asserting otherwise so as to seek a dismissal, stay or adjournment of the proceedings against him or any part of them. Mr Shah was found guilty of the criminal charges pursued against him in Denmark, and was sentenced by the Danish criminal court, in early December 2024. He has appealed against that conviction and at the date of this judgment there has been no decision on that appeal.

The Danish criminal proceedings, then, explain why Sanjay Shah and Mr Klar gave evidence at trial remotely, from custody in Denmark. Had Mr Patterson given evidence at trial, that would have applied to him too. The threat of Danish criminal proceedings against the DWF Ds underlay their application, which I granted, to give their evidence at trial remotely from Dubai. At the first pre-trial review for the Main Trial, I refused their primary application which was that for the period scheduled for their factual evidence I should adjourn the trial and take evidence from them in Dubai, as a special examiner having appointed myself to that role for that purpose: SKAT (Sitting in Dubai) [2024] EWHC 19 (Comm).

Rajen Shah spent some time in custody in Dubai, partly on the initiative of the Danish authorities, very shortly before he was due to give evidence in the Main Trial. He was taken into custody in Dubai again just before oral closing argument, with a view to possible extradition to Denmark. Mr Head KC voiced understandable concerns over the fact and timing of both periods of remand in Dubai (the second of which continues at the date of this judgment), and the first resulted in an adjustment to the trial timetable in the interests of fairness to Rajen Shah. As in the case of Sanjay Shah, however, I was satisfied that the process overall remained fair, and no application was made asserting otherwise so as to seek a dismissal, stay or adjournment of the proceedings or any part of them.

In assessing the factual witness evidence, and what conclusions to draw from it, I had well in mind the strong likelihood that recollections may be faulty, both generally but particularly as to matters of detail: see Gestmin SGPS SA v Credit Suisse (UK) Ltd et al. [2013] EWHC 3560 (Comm) and many cases since. That said, in this case much turned more on why things were done, or not done, and in particular with what understanding (if any) of the legal effects of transactions or the workings of Danish tax law. Factual witness testimony as an element of assessing attitudes, working practices, or comprehension, had the capacity to be valuable, therefore, notwithstanding the fallibility of human memory.

I also found helpful, and adopted, the approach to judging a circumstantial case explained by Bryan J in JSC BM Bank v Kekhman et al. [2018] EWHC 791 (Comm) at [78]-[79], citing and in part developing what Teare J said in JSC BTA Bank v Ablyazov [2013] EWHC 510 (Comm) at [197]-[198]. It was also necessary in this case, when applying that approach, and generally, to bear in mind, and I did so, that disreputable though any dishonesty is, business people may engage in dishonest conduct for any number of reasons, and witnesses may fail to tell the whole truth, or may tell lies in their evidence, for different reasons. Contemporaneous dishonesty, or subsequent forensic dishonesty, does not necessarily come from, so as indirectly to evidence, awareness of the truth of some allegation made in legal proceedings; and in any event a finding of dishonesty, of contemporaneous conduct or in the witness box, is likely to be one of many elements in any given case, and does not reverse the burden of proof.

Having made those introductory comments in the main body of this judgment, I have indicated what I made of the factual witnesses who gave oral evidence at trial in Appendix 6, below, taking them in the order in which they were called. In Appendix 6, where I refer to trustworthiness, I refer to the assessment I made of whether, or how far, what was given to the court as their factual witness evidence at trial was an honest account of what the witnesses perceived, when giving evidence, to be their recollection (or lack of recollection) of matters of fact, including, where relevant, matters of their knowledge, understanding, belief, intention or thinking in the past. When judging the facts on the evidence as a whole, in the usual way I have borne well in mind that the testimony of a witness who was trustworthy, in that sense, still may be unreliable in whole or in part; the testimony of a witness who was generally trustworthy, in that sense, may nonetheless have been untrustworthy in certain respects, or in relation to particular topics; and the testimony of an entirely or generally untrustworthy witness may nonetheless be accurate (in whole or in part), respectively trustworthy, on some points.

In the case of some witnesses, what I say in Appendix 6 led me naturally into recording there, rather than only when I consider the factual case alleged against them in Appendix 7, my findings as to their contemporaneous knowledge or understanding of certain matters.

I should also say, for completeness, that when I refer in Appendix 6 to witnesses having adopted their witness statements as their evidence in chief, I mean to refer to their witness statements as may have been clarified or corrected orally in chief.

There was also written factual witness evidence from:

There was expert evidence at trial on Danish public law, concerning SKAT’s identity and capacity to bring the proceedings, from a forensic accountant, analysing payments made by SKAT and the flow of funds resulting from them, and on matters of market practice. There was scheduled to be expert evidence on Danish private law, but the need for that evidence fell away (see paragraph 42 above).

On Danish public law, the only evidence was that of Prof. Waage by his expert report, which was unchallenged and so was adduced without the need for SKAT to call him at trial (see paragraphs 14 to 16 above). Similarly, only SKAT put in expert evidence from a forensic accountant, Mr Jens Ringbӕk of Deloitte Denmark, whose evidence was also unchallenged. Pending my having read his report, time was reserved in the trial timetable for him to be called, in case I needed to ask him questions to ensure I understood what the report was saying. In the event, I did not identify any such questions, and so for Mr Ringbӕk also, his report was adduced at trial and SKAT did not call him.

On market practice, SKAT, the DWF Ds and the Shah Ds instructed experts who prepared reports, a joint memorandum, and supplementary reports. All were called at trial and cross-examined. SKAT and the DWF Ds called Mr Graham Wade and Mr Paul Sharma, respectively, as they did at the Validity Trial. The Shah Ds called Mr Simon Bird, who was not an expert witness at the Validity Trial, but was not new to the wider litigation, having an involvement as an expert witness instructed by various defendants in related proceedings in other jurisdictions.

In brief summary, no doubt not doing justice to their CVs, their primary careers, entitling them to be considered expert witnesses as to relevant market practice, were as follows:

The case management directions for the Main Trial permitted expert evidence as to market practice on the following issues:

“18A. To the extent (if at all) that it involves expertise to identify, what trading structures and/or series of transactions were (purportedly) constituted or provided for by the Sample Trades?

18B. To what extent were the Sample Trades consistent with any standard market practice?

19.

With respect to each of the Sample Trades, did the Custodians hold any Danish shares in their custody accounts, in the custody accounts of any sub-custodians, or with VP Securities (for themselves or on behalf of any of the participants in the transactions purportedly carried out under the relevant models)?

20.

With respect to each of the Sample Trades, did the Custodians receive any payments in the amount of dividends declared by Danish companies?”

With hindsight, expert issues 19 and 20 perhaps should have been qualified in the same way as issue 18A, i.e. “To the extent (if at all) that it involves expertise to identify …”. Having said that, such a qualification is implicit in any question upon which expert evidence is permitted; and in the present case I am confident that failing to spell it out in expert issues 19 and 20 did not make any difference to how the experts went about their reports. They included in their reports full, lengthy and detailed discussions of expert issues 18A, 19 and 20, although (as I now see it) none of them required consideration of matters of market practice on which expert evidence might have been of assistance.

As regards expert issue 18A, the work the experts did was useful nonetheless. Their reports served as a convenient vehicle for setting out detailed descriptions of the transactions involved in the Sample Trades and how they were operated. By the edited versions of the expert reports that were ultimately adopted as evidence in chief, that material was not adduced as expert evidence at trial. However, much of it was effectively adopted as trial material, as a convenient way of presenting the primary facts to the court. For example, it informed, and was cross-referenced in, the Sample Trades Summary agreed between SKAT, the Shah Ds and the DWF Ds that was helpful for Appendix 3 to this judgment.

As regards expert issues 19 and 20, the straightforward answer in each case is ‘No’. By design, the Sample Trades involved no Danish shares being held by any custodian or participant at any time. It follows inevitably, and was the case as a matter of fact, that no custodian ever received a payment in any dividend amount. If there is a qualification to that, it is only that a debit entry in a short seller’s account at a custodian, for simultaneous crediting to a buyer’s account at that custodian as a dividend compensation payment to the buyer, might be said to involve the custodian receiving and making matching ‘payments in the amount of dividends declared’. But that adds nothing for present purposes, that is to say for concluding that market practice expertise is not involved in or required to answer expert issue 20.

The focus of the edited versions of the expert reports adopted as evidence in chief was therefore expert issue 18B.

At the Validity Trial, on the matters dealt with and the evidence I heard then, I concluded that Mr Wade and Mr Sharma (and Dr Collier, the expert called by the Shah Ds at that trial) gave, in general, “carefully considered, honest opinions as to matters they had been asked to address, to the extent they could do so within their … expertise”, and that their evidence had been “helpful, interesting, and obviously expert” (SKAT (Validity Issues), supra, at [43]). I found that, on the issues addressed by them for the Validity Trial, there was “very little in the way of relevant but different opinion between them” (ibid at [51]), and that “their differences of background and perspective mean … that Mr Sharma was better placed to deal with matters of market structure and operation, and regulation, whereas Mr Wade was better placed to deal with matters of market practices and understanding” (ibid at [52]).

On the matters covered by the experts for the Main Trial, and the greater depth with which some of those matters were explored, I would revise that last assessment slightly to say that Mr Sharma was at least as well placed as Mr Wade to deal with matters of market practice and understanding, although their backgrounds still mean their perspective on points may be different without that necessarily meaning, on its own, that the view of either deserves to be accorded greater weight than that of the other. On that aspect – the ability in general to give well-informed and obviously expert opinions to assist the court – I consider that Mr Bird was also well placed.

All three experts, in differing ways, did not provide properly balanced written reports, uninfluenced by the fact that they were instructed, respectively, by SKAT (Mr Wade), the DWF Ds (Mr Sharma), and the Shah Ds (Mr Bird). This was most evident in the case of Mr Wade, much of whose written work was argument rather than expert evidence. Regrettably, in my view he carried into his oral evidence the same tendency and approach, to think first of how he should be putting the case for SKAT. I concluded, with regret, that he has become compromised by the nature and extent of his involvement for SKAT in its global litigation effort, such that he finds it difficult not to think and express himself as an advocate for SKAT’s position. He has, I think, lost detachment from the partisan interests of SKAT as his instructing client, and that left me unhappy to accept views of his that, on analysis, were genuinely matters of expert opinion where his views were not shared by at least one of Mr Sharma or Mr Bird.

In Mr Sharma’s case, my concern was less pronounced. There was in his case an unwelcome tendency, in writing, to fail to express qualifications or set out a complete expression of his view, such that the absence, or narrowness, of any real difference of expert opinion between him and Mr Wade was obscured, and the impression might have been gained from reading his reports that there was nothing at all unusual or contrary to typical market practice about the Maple Point Model trading (which was Mr Sharma’s focus). That exposed Mr Sharma to a cross-examination that should have been largely unnecessary to confirm the extent to which, in substance, he agrees with a range of matters on which SKAT relied. However, he dealt with the resulting questions impressively, fairly (with balance), and with an obvious depth of thought and expertise. I was not left with any general concern about according weight to Mr Sharma’s expert views.

My difficulty with Mr Bird’s written work was the extent to which it consisted of or included his views on contentious matters of primary fact; but that was largely, if not entirely, overcome by the edited versions of his reports that were adduced as his evidence in chief at trial. Like Mr Sharma, in my view Mr Bird dealt thoughtfully and fairly with questions put to him in cross-examination and questions from the court to clarify his evidence. In general, and so long as I took care to filter out remaining instances where Mr Bird continued to express views on matters of primary fact that were not for him, I was not troubled about placing reliance on Mr Bird’s expert views.

However, on the question of terminology to which I refer immediately below, Mr Bird put forward a view on the normal market use of the terms ‘market claim’ and ‘manufactured dividend’ that I concluded had not been his view at any material time. It was inconsistent with the views he had expressed in writing, including in his primary expert report in these proceedings, to which he had to proffer alterations when called to give his oral evidence to advance the different view. In my judgment, the evidence as given at trial by Mr Bird on this aspect did him no credit and was advanced by him for no good reason, but rather only because he had appreciated that his actual views coincided with Mr Wade’s on the point and so would be said by SKAT to assist its case. As it happens, the point is not central, and my finding on it is not an unqualified acceptance of SKAT’s position anyway. That does not make it acceptable for Mr Bird to have approached his oral evidence on it as he did.

That point concerns what I said in SKAT (Validity Issues) at [53], namely (now with some added numbering) that:

“On the most important points considered by the market experts [for the Validity Trial], they were in any event agreed. For example, they agreed that [(i)] market participants would understand a dividend (or ‘real dividend’) to be a distribution from a share issuer to its shareholders, and [(ii)] to be different from a ‘manufactured dividend’, [(iii)] viz. a contractual payment representing a dividend but arising under a contract for the sale or other transfer of securities as compensation for a dividend forgone; they agreed that [(iv)] depending on the context the word ‘dividend’ on its own might be used in the market so as to encompass manufactured dividends as well as real dividends; and they agreed that [(v)] in any event what ‘dividend’ might mean for Danish tax law, and what were the requirements for there to be dividend tax liability or dividend tax refund entitlement under Danish law, [(v)(a)] would be understood to be a matter of Danish tax law on which [(v)(b)] a market participant, if interested, would take specialist advice.”

A point arose at the Main Trial on the definition of a ‘manufactured dividend’ that I gave at [53(iii)], viz. a “contractual payment representing a dividend but arising under a contract for the sale or other transfer of securities as compensation for a dividend forgone”. I touched on this at paragraphs 87 to 91 above.

On the factual and expert evidence at the Main Trial, it was tolerably clear, and I find, that there was no fixed or uniform market usage concerning the meaning of ‘manufactured dividend’. I should be clear that the market in question here is a global equity trading and equity finance market relating to European equities, most pertinently Danish equities. Within that global market, there might be specific practices peculiar to a particular jurisdiction, be that the jurisdiction of the equities in question or the jurisdiction of the participant in question (for example, that of a custodian or that of the equity buyer). They might include practices to meet specific local legal or regulatory rules. Something of that sort could affect the use of terminology in particular contexts, but there was no suggestion by any party, or evidence, that any specific market practice existed at the material time concerning the nature or content of documentation generated in the context of Danish dividend tax and any possible tax refund claim made to SKAT.

Leaving aside any usage derived from or tailored to some jurisdiction-specific circumstance, in the market generally, as in part I noted at paragraph 87 above, some might not refer to what I am calling a dividend compensation payment as a ‘manufactured dividend’, because some would give the latter term a narrower scope than I gave it in SKAT (Validity Issues). In particular, they might refer to a dividend compensation payment as a (type of) ‘market claim’, even if more commonly that term would be used in a meaning that would match the CAJWG definition, or be intended to refer to it. On all of that, I prefer and accept Mr Sharma’s expert evidence to the extent that it differed from Mr Wade’s or Mr Bird’s (if it truly did, once their evidence had been explored at trial).

That variability of terminology was illustrated by a handout put together by Freshfields in March 2012 summarising and explaining legal issues concerning div-arb opportunities across Europe. Freshfields, it seems, put on seminars from time to time for commercial parties interested in the field. Rajen Shah and Mr Horn went to one, probably in early 2011 (since it led to some advice being taken from Freshfields in May 2011, in connection with Belgium), and Mr Shah remembers taking away from it what may have been a similar handout or at any rate a document containing similar commentary.

The 2012 Freshfields handout discussed inter alia the tax treatment of dividends and related payments in a range of different jurisdictions; it seems evident that the section of the document for any given jurisdiction will have been prepared by or under the supervision of the named Freshfields partner practising there, as identified at the end of the document as the point of contact for that jurisdiction. In relation to some jurisdictions, the document covered cum-ex trades. Where that was so, it was not consistent in its use of terminology for what I am calling dividend compensation payments under such trades. For example, in the French and Belgian sections, they were referred to as market claims, but in the section for the UK it was said to be a “difficult question … whether [the] compensation payment is just an accounting for the real dividend that is treated as belonging to the buyer, or … is a manufactured dividend, or … should be viewed for UK tax purposes as simply an adjustment to the contracted sale price”. The DWF Ds also drew attention, in closing, to an email from Freshfields in November 2011, seen at the time by Sanjay Shah, Rajen Shah, Mr Horn and Mr Klar, referring to a cum-ex buyer’s dividend compensation entitlement as a market claim. (SKAT relied on indications in the Freshfields handout that they may not have been contemplating cum-ex trades where the seller was and always remained short. That does not affect the current point, which concerns the use of the term ‘market claim’. It would be relevant to whether the Freshfields handout could be taken to express a view that a dividend compensation payment in the case of a short cum-ex sale of Belgian shares that was settled without any share transfer would give the buyer a valid Belgian tax refund claim.)

Those who would give manufactured dividend the fuller meaning I gave it in SKAT (Validity Issues) would be likely, as Mr Wade does and Mr Bird did prior to his unsatisfactory oral evidence on this point, to use the term ‘market claim’, as reflected and influenced by the CAJWG definition, to refer to the process by which a real dividend payment received through a custody chain is redistributed to a party contractually entitled to that payment.

Except for the specific point mentioned next, none of that would cause me to revisit anything I said in SKAT (Validity Issues), so long as it is borne in mind when reading it that my use of the term ‘manufactured dividend’, as I defined it in that judgment, is not a usage that was fixed in the market at the time. For completeness only, therefore, I would now re-state SKAT (Validity Issues) at [53] as follows:

“[(i)] market participants would understand a dividend (or ‘real dividend’) to be a distribution from a share issuer to its shareholders, and [(ii)] to be different from a contractual payment representing a dividend but arising under a contract for the sale or other transfer of securities as compensation for a dividend forgone, [(iii)] to which in this judgment I refer as a ‘manufactured dividend’, although that was not a fixed market usage; … [(iv)] depending on the context the word ‘dividend’ on its own might be used in the market so as to encompass manufactured dividends as well as real dividends; and … [(v)] in any event what ‘dividend’ might mean for Danish tax law, and what were the requirements for there to be dividend tax liability or dividend tax refund entitlement under Danish law, [(v)(a)] would be understood to be a matter of Danish tax law on which [(v)(b)] a market participant, if interested, would take specialist advice.”

That brings me to the further, specific point noted in passing at paragraph 91 above, which is more important to the case than the point on terminology that I have just revisited above. In SKAT (Validity Issues) at [269], I said that the experts at the Validity Trial had agreed in the joint memorandum, as they had, that it was the general practice of custodians to draw a distinction, in any CANs or tax vouchers they generated, between real and manufactured dividends. The context for that paragraph was a hypothetical example where a custodian holding 100 shares for Client A lent 40 to Client B so it could settle a short sale to a third party. I made the point that, other things being equal, in those circumstances any ‘dividend’ payment received by Client A during the stock loan term would be payment of a ‘real dividend’ on 60 shares and of a ‘manufactured dividend’ on 40 shares. I noted that in cross-examination Mr Sharma had said the practice of distinguishing between real and manufactured dividends in documents issued was limited to UK custodians in tax vouchers for possible submission to HMRC in relation to UK tax, but I rejected that evidence in the light of the unqualified joint memorandum opinion to which Mr Sharma had subscribed that, in the example given, “The market practice was that if [Client] A was provided with a confirmation of the amount withheld or a tax voucher, it would distinguish that A had received 60 dividends (net of withholding tax where relevant) and 40 manufactured dividends.”

At one stage it appeared that SKAT might contend that an issue estoppel arose from that, to the effect that the general practice of custodians was to draw a distinction in CANs between real and manufactured dividends (for that purpose, including dividend compensation payments, since that is how I was then using that term). It was not obvious how that might be correct, since nothing decided at the Validity Trial depended on that finding. I could set out additional analysis, but it suffices to say that the material decision, as reflected in the formal answers given to Validity Issues 10(a) to 10(d) in the Appendix to SKAT (Validity Issues), was that matters of market practice or understanding did not affect the relevant content of Danish tax law. In the event, an opportunity arose during the cross-examination of Mr Wade to check the position, and Mr Goldsmith KC for SKAT confirmed that no argument of issue estoppel was pursued.

That means I am free to consider the point afresh. Doing so, on the factual and expert evidence adduced at the Main Trial, I can see that Mr Sharma was in fact right to qualify what he had agreed for the Validity Trial. The way his attempt to do so came out at the Validity Trial was influential at the time; but that is now history. Naturally, his evidence on this point was tested at the Main Trial by reference inter alia to that history. I was persuaded by the cross-examination only to the view that Mr Sharma had done himself a disservice by aligning himself in the Validity Trial joint memorandum to an unqualified view that did not represent his full opinion. I am satisfied that in fact Mr Sharma was only ever aware of CANs or tax vouchers explicitly drawing the distinction, by describing income items as manufactured dividends, in the context of specific HMRC requirements to do so for UK tax purposes, that he was not aware of any practice of doing so outside that context, and that in fact his view always was that there was no general practice of doing so. The tenor of Mr Bird’s evidence was that there was no general market practice on the point, and now that Mr Wade had to defend his restated contrary opinion under cross-examination, I was not satisfied that he had any real basis for it. Tellingly, the CANs issued by Merrill Lynch under the Broadgate transaction, used by Solo as the template for CANs issued by SCP when Solo Model trading commenced, did not explicitly draw the distinction Mr Wade had in mind, yet they also concerned dividend compensation payments under a short selling cum-ex trading structure focused on facilitating a tax refund claim in Germany.

On the whole of the evidence I have now received, the true position, I find, is that there was no market standard for or general practice as to the form, format or content of a CAN. In particular, there was no established or general practice for CANs of expressing any distinction between real dividends and dividend compensation payments. The only rule of practice, in truth, was the obvious one that in issuing a CAN, if it did, a custodian should report accurately to its client the dividend-related credit it had booked to the client’s account. The fact that a CAN used the word ‘dividend’, and not the words ‘manufactured dividend’, would not convey that what was being reported was a real dividend payment rather then, for example, a dividend compensation payment.

Whether the CANs in this case merely reported accurately to their addressee clients dividend-related credits booked to their accounts, and if not why not, is an aspect of whether SKAT was misled, as it alleges, into making the payments it made in response to the tax refund claims that are the subject of these proceedings, to which I turn from paragraph 424 below. If a CAN, properly read, made a statement that was falsified by the fact that the payment credit reported was only a dividend compensation payment, then it will have been a misleading document in a way that could have been avoided by drawing the distinction explicitly. In terms of market practice, though, that will mean simply that the CAN failed to conform to the golden rule of factual accuracy, not that it failed to conform to a specific market practice about what words should be used to achieve it.

Sham Trading?

It was not always clear in the litigation what SKAT was alleging, in saying as it did that the trading activity in the case was ‘sham’, or why it was saying it. SKAT ultimately accepted that though it would still contend, if required, that what it said about that activity, if accurate, would make it proper to say that the trading transactions were shams, whether that was correct or not, a matter of the legal characterisation of the transactions, was not itself relevant to the claims it pursued in the litigation. What mattered was whether what SKAT said about the trading activity was true, and if so (where relevant) what state of knowledge or understanding any given individual had of that at the time, as part (and only part) of determining whether, if SKAT was misled, relevant individuals knew, believed or intended that to be the case. I shall therefore deal with this more shortly than I might have done if the question of characterisation (sham or not sham) was itself critical.

Taking what is for my purposes the most pertinent aspect, to illustrate the point about relevance, consider the equity purchase trade in each instance. Then:

In closing, SKAT referred to Chitty on Contracts, 35^th Edition, for a submission that English law would use the language of sham in two situations: firstly, where parties “create a document that purports to show a contract between them when in fact there was none” (ibid at 4-213, giving as examples the sham charterparty in The Good Helmsman, Harmony Shipping Co SA v Saudi-Europe Line Ltd [1981] 1 Lloyd’s Rep 377, and the sham ownership transfer in The Ocean Enterprise, Glatzer v Bradston Ltd [1997] 1 Lloyd’s Rep 449); secondly, where a document “is a sham in the sense of being designed by parties to give “[to] third parties or [to] the court the appearance of creating between the parties legal rights and obligations different from the actual legal rights and obligations … which the parties intend to create” [where] the parties do not intend the sham document to bind them, but they do intend to be bound by the “actual” agreement.” (ibid at 4-214, quoting Snook v West Riding Investments Ltd [1967] 2 QB 786, per Diplock LJ at 802D). Chitty acknowledges that on the facts of a given case, a correct characterisation might be that of collateral contract, at all events where there is no joint intention to give false appearances to third parties (ibid, at n.964, citing Coleman v Mundell [2020] EWHC 2852 (QB)).

SKAT submitted that distinct concepts are not involved, but rather in both of those situations the same general principle is applied, namely that there is a sham where there is a “common intention that the acts or documents are not to create the legal rights and obligations which they give the appearance of creating”: Snook, supra, per Diplock LJ at 802E; and see Antoniades v Villiers et al. [1990] 1 AC 417 (HL), in which one of the agreements purported to be a licence, with a view to evading the Rent Acts given that in truth a lease was agreed between the parties, and purported to reserve to the freeholder a right to place others into the property when no such agreement had been reached.

As Diplock LJ made clear in the same passage in Snook, supra, all parties to the relevant allegedly sham transaction must have the common intention to create a false appearance: “No unexpressed intentions of a “shammer” affect the rights of a party whom he deceived.” (ibid, at 802F). There was an express finding in Snook that the defendants knew nothing of the matters said to render sham the transactions they had concluded for the hire-purchase refinancing of a car, so the contention that they were a sham failed.

SKAT then submitted, in writing, that it alleged sham trading “in support of its case that various Defendants knew that they were participating in a scheme to defraud SKAT and/or in response to the Defendants’ allegations that they were acting honestly at all times”. However, on analysis – hence, ultimately, SKAT’s concession referred to in paragraph 414 above – SKAT was not relying on its allegation of sham trading for that purpose, rather it was relying on matters that might (also) support such an allegation (for example their case that parties other than custodians appreciated that no Danish shares were being or had ever been transferred) that either had significance for the claims being pursued or not, as the case may be, irrespective of whether they (also) led to a conclusion that one or more of the transactions was or were sham.

It may be unnecessary in the circumstances to say anything more about SKAT’s allegation that this was sham trading. However, I consider I should make clear that I do not accept it in its most extreme form. That is to say, I reject the claim, to the extent made, that any of the parties involved was engaged in a total charade, pretending to enter into contracts but not intending what they were doing to have legal effect, doing so simply as pretext for generating CANs that could be presented to SKAT to support tax refund claims.

I am satisfied that that is not what the architects of any of the trading models involved were doing. They were, all of them (Sanjay Shah, Mr Horn, Rajen Shah, Mr Klar), engaged in what was for them an evolutionary process in which they developed trading models previously used to the point reached where, they thought, real share purchases (i.e. contracts) could be settled in the absence of real shares. Whether they took on board the full implications of the synthetic trading systems they created (The Emperor’s New Clothes point perhaps) may be relevant to whether they believed that they generated or might generate valid tax refund claims or to their understanding of whether false statements were or would be made to SKAT when tax refund claims were submitted. However, I was left in no doubt but that they all considered and intended the transactions involved in the respective trading models to be real trades (binding contracts) between the parties to them.

All the more so, I am clear, those not in command of the closed trading ecosystems that the Solo Model and Maple Point Model trading created – the non-Solo custodians, the equity buyers, the short sellers, the stock loan and forward counterparties – intended to enter into, and will have thought they were entering into, legally binding trading transactions. That is so even if (which for the most part I do not accept anyway) those outside participants appreciated that their trades were being settled by the self-fulfilling, share-less loop method that was in fact being used. The Klar Model involved far fewer outside participants (just Blue Ocean, Cole and (possibly) Heber), in respect of whom there was very little evidence before me beyond Mr Klar’s own evidence. I do not accept it was shown by SKAT that any of them, to the extent they were involved at all, were merely pretending to trade in the sense I am considering now (paragraph 420 above).

The fact that the architects of the scheme, and others involved if they knew of that method, either realised or should have realised that no shares were being transferred, so that it might be at least questionable whether valid tax refund claims were generated, does not mean their trades were not real trades at all.

Was SKAT Misled?

As I noted in paragraph 105 above, SKAT alleged that it was induced to pay the 4,170 tax reclaims giving rise to these proceedings by misrepresentations made to it, in each instance, by the content of the documents it received, given that they were submitted so as to request from it, as the Danish national tax authority, a refund of Danish dividend tax. SKAT put the case on the basis that:

The law on what a claimant must show, concerning the effect upon it of a false statement, for a deceit claim to be capable of succeeding (subject to considering the defendant’s state of mind), is currently summarised in Clerk & Lindsell on Torts as follows (24^th Edition at 17-36 to 17-37, omitting immaterial footnotes):

“17-36 … a claimant … must show that he¹⁷⁶ acted (or in a suitable case refrained from acting) in reliance on the defendant’s misrepresentation. If he would have done the same thing even in the absence of it, he will fail. What is relevant here is what the claimant would have done had no representation at all been made. In particular, if the making of the representation in fact influenced the claimant, it is not open to the defendant to argue that the claimant might have acted in the same way had the representation been true.¹⁸¹ …

¹⁷⁶

Or a machine, such as a computer, under his control: see Renault UK Ltd v Fleetpro Technical Services [2007] EWHC 2541 (QB) … at [122] (defendants causing to be inserted into computer orders for cars with a fleet discount to which they knew they were not entitled).

¹⁸¹

See Downs v Chappell [1997] 1 W.L.R. 426 at 433; Bank of Tokyo-Mitsubishi UFJ Ltd v Baskan Gida Sanayi VE Pazarlama AS [2009] EWHC 1276 (Ch) … at [1005]; OMV Petrom SA v Glencore International AG [2016] EWCA Civ 778 … (Romanian state oil buyer deceived on an industrial scale with regard to what oil it was buying: nothing to the point, even if true, that it was so desperate for oil it would have bought in any case) (discussed in A. Summers, “Deceit, difference in value and date of assessment” (2017) 133 L.Q.R. 41). Note, however, that the point was left studiedly open by Males J in Leni Gas & Oil Investments Ltd v Malta Oil Pty Ltd [2014] EWHC 893 (Comm) at [20].

17-37 There is some issue whether a person can be said to have acted in reliance on an implicit misrepresentation about an issue that was never present to his mind at the time. In Leeds City Council v Barclays Bank Plc,¹⁸⁵ Cockerill J gave a negative answer to the question, thus discounting a plea by a local authority that it had entered into a swaps contract with a bank on the basis of an implied representation that the bank had not engaged in dishonest interest rate manipulation. At the time the prospect of such manipulation had not been in the authority’s contemplation at all; it therefore could not say that it had known about, let alone relied on, any representations about it. With respect, however, this must be open to some doubt. Such a holding seems inconsistent with the jurisprudence on half-truths and misrepresentation by deliberate concealment;¹⁸⁶ furthermore, there seems nothing incoherent in the idea of a party holding, and acting upon, an implicit if subconscious belief that there is nothing unusual or untoward about a given transaction.¹⁸⁷

¹⁸⁵

[2021] EWHC 363 (Comm); [2021] Q.B. 1027. See too Marme Inversiones 2007 SL v Natwest Markets Plc [2019] EWHC 366 (Comm) at [286], where Picken J had said, obiter, that a showing of reliance required proof that the representee had “given some contemporaneous conscious thought to the fact that some representations were being impliedly made”; Groen v Heath [2024] EWHC 1654 (Ch) at [37]. Note also ACL Netherlands BV v Hewlett-Packard The Hague BV [2022] EWHC 1178 (Ch) at [505]-[506] and Allianz Funds Multi-Strategy Trust v Barclays Plc [2024] EWHC 2710 (Ch) (cases under the Financial Services and Markets Act 2000, s.90A and Sch.10A Pt 2, respectively; same rule applied). On the other hand, for the rule in the Barclays case to apply the statement must be implied; an express statement that nothing is wrong, if brought to the attention of the representee or his agent, will potentially engage liability. See Patarkatsishvili v Woodward-Fisher [2025] EWHC 265 (Ch) at [86]-[103].

¹⁸⁶

See cases such as Schneider v Heath (1813) 3 Camp. 506 and Gordon v Selico Ltd (1986) 18 H.L.R. 219, where liability in deceit arose from acts intended precisely to hide the fact that any issue arose on which a representation could be made. This point was adverted to by Waksman J in the later Crossley v Volkswagen AG [2021] EWHC 3444 (QB) at [76] … .

¹⁸⁷

See Crossley … at [46]-[97]. There, in a claim arising out of the alleged concealment of untoward emission test results on cars, Waksman J declined despite the Leeds case to strike out a claim in deceit by purchasers claiming to have been duped; and this even though the purchasers presumably had never dreamt that anything might be wrong on the emissions front.”

Some care is needed to understand the reference in 17-036, at f.n.181, to a counterfactual of whether “the representation had been true”. That might be thought to posit a case where that which was represented had been true rather than false (as it must have been for there to be a question of liability for deceit). Reference to the authorities cited in f.n.181 makes clear, however, that the case posited is, rather, one in which the representee was told that which was in fact true, concerning the subject matter of the representation. For example, the statement of principle by Hobhouse LJ (as he was then) in Downs v Chappell [1997] 1 WLR 426 at 433D, which Clerk & Lindsell cite as the primary source for their proposition, is that the trial judge in that case “was wrong to ask how [the plaintiffs] would have acted if they had been told the truth” (my emphasis, i.e. not “… if that which they were told had been true”).

The currently fashionable controversy referred to by Clerk & Lindsell at 17-37 was the subject of argument in this case. SKAT contended that there was no general requirement of contemporaneous conscious awareness of the kind mentioned by Picken J in Marme Inversiones (see Clerk & Lindsell at f.n.185). There were submissions about that case, about Leeds City Council (Cockerill J) and Crossley (Waksman J), and about more recent cases, especially Loreley Financing (Jersey) No 30 Ltd v Credit Suisse Securities (Europe) Ltd [2023] EWHC 2759 (Comm) (Cockerill J, adhering obiter to the view she expressed in Leeds notwithstanding Crossley) and Farol Holdings Ltd v Clydesdale Bank Ltd [2024] EWHC 593 (Ch) (Zacaroli J, as he was then, also obiter on this point).

In the last of those, Zacaroli J noted at [221] that the issue with which the earlier judgments had been grappling is that “sometimes the court has found that a misrepresentation was relied on, apparently without a finding that the representee gave conscious or active thought to the representation (see for example the cases … summarised at §380 of Loreley). An extreme example (discussed from §105 of Leeds) is the representation by a diner at a restaurant, made by the conduct of ordering a meal, that they have an intention to pay for it (see DPP v Ray [1974] AC 370 …). It is highly unlikely that the waiter who took the order gave any thought to whether such an implied representation was being made.”

Zacaroli J doubted “the utility (as did Cockerill J) of breaking down this causation question into distinct elements and seeking to find a single universally applicable test for those elements. It is essential to keep in mind that in every case it is necessary to show, as a matter of fact, that the claimant’s decision to take the action (or refrain from taking action) which caused it loss must have been caused by the representation made by the defendant. The evidence required to satisfy the requirement will differ greatly depending on where on the spectrum the case lies (from “it goes without saying”, at one end, to a complex representation said to be implied from conduct and statements, at the other)” (ibid at [223]).

It is not necessary in this case to resolve this contemporaneous consciousness controversy (which may be more apparent than real anyway). There is a danger, I think, that the recent judgments considering it veer towards over-complication. I see force in the observation quoted immediately above (although I would say “was caused” rather than “must have been caused”), and in the fact that Zacaroli J set the controversy in the causation context of “[the] identification of the appropriate counterfactual if the statement had not been made”, a question of fact (ibid, at [218]) that falls to be considered because “[the] relevant question … is whether the claimant would have [acted as it did] if the representation had not been made …” (ibid, at [217], which I have generalised from “entered into the contract”, per Zacaroli J, which was the action alleged to have been induced in Farol Holdings).

The intent to pay that in DPP v Ray the House of Lords judged to be represented implicitly, if diners order a meal at a restaurant without confessing that they do not intend to pay, realistically can only not be conveyed (assessing the matter objectively) if the confession is made. To ask whether the apparent intent to pay influenced the waiter in putting the order through to the kitchen, for the purpose of a misrepresentation claim, is therefore, ultimately, to ask what would have happened if when ordering the meal the diners had said the like of, “oh, by the way, we don’t intend to pay”. Echoing the Editors of Clerk & Lindsell, supra, there would be nothing incoherent about concluding that there was reliance by the waiter on the representation regarded by the court as implicit if the answer was that the waiter would have refused to serve the table and asked the party to leave, or would have served the table further only if reassured that they were joking about not intending to pay, or would have called the restaurant manager over to deal with the situation. If the evidence revealed instead a disloyal waiter who loathed the restaurant and would have served the table anyway, taking pleasure in the prospect of the restaurant losing out, then other causes of action might exist, but a claim requiring proof that the impliedly represented intent to pay caused loss should fail.

As to whether any representations were made, and if so what representations, it was uncontentious at trial that in law a representation is a statement made to the representee on which it is intended and entitled to rely as a positive assertion by the representor that the matter stated is true. That means a question can arise as to the characterisation of a statement, not just as to its meaning: “Until one has decided that, in the circumstances of the particular case, [a] statement should be treated as a representation rather than, say, a contention or argument, any question of liability for misrepresentation cannot arise” (per Neuberger LJ, as he was then, in Kyle Bay Ltd t/a Astons Nightclub v Underwriters [2007] EWCA Civ 57, at [32]).

SKAT put its case on the basis that the representations it alleged were made by the documents submitted to it, construed objectively in their context, for which purpose the question to be asked was what impact the content of the documents, given their context, might be expected to have upon a reasonable representee in the position and with the known characteristics of the actual representee (applying the test stated in Casa di Risparmio della Republica di San Marino SpA v Barclays Bank Ltd [2011] EWHC 484 (Comm), per Hamblen J, as he was then, at [215]). In the present context of claims, using the Form Scheme, to be paid a refund of Danish dividend tax, the only possible putative representee was SKAT itself. It is thus convenient (and shorter) to speak simply of SKAT, acting reasonably, rather than of the reasonable representee in the position and with the known characteristics of the actual representee, namely SKAT.

A representation may be implied where although no statement to the effect of the representation is made expressly nonetheless clear words and/or conduct on the part of the representor convey such a statement in the equivalent objective sense, namely (here) that SKAT, acting reasonably, would be expected to take from what was said and/or done that something was being stated to it on the truth of which it was intended to rely as a positive assertion of the truth of the matter stated. I gave a very brief summary of the law as to that in SKAT v Goal Taxback [2020] EWHC 1624 (Comm), [2020] 4 WLR 98, at [114]-[115]. There were submissions in closing argument by reference to, among others, DPP v Ray, supra, Gordon v Selico [1985] 2 EGLR 79, Spice Girls Ltd v Aprilia World Service BV [2002] EWCA Civ 15, Property Alliance Group Ltd v Royal Bank of Scotland plc [2018] EWCA Civ 355, [2018] 1 WLR 3529, The C Challenger, SK Shipping Europe Ltd v Capital VLCC 3 Corp et al [2022] EWCA Civ 231, and various first instance cases considering implied representations alleged in the context of banking transactions. In my judgment, those submissions did not affect the accuracy or sufficiency for present purposes of my short summary.

SKAT pleaded that the representations it alleged were “partly express and partly implied”, but the only conduct on which SKAT relied was the submitting of the documents so as to make, as SKAT put it, “an application for the refund of tax supposedly withheld from dividends supposedly received on the Danish shares identified in the application”. It was plainly the case in each instance that a claim was being made, by the submitted Form and supporting materials, for a refund of Danish dividend tax; and that is the pleaded context in which those documents fall to be construed. As a result, there is no meaningful distinction here between asking what the documents would have conveyed to SKAT, acting reasonably, given their terms and the context, and what was impliedly stated to SKAT by submitting a dividend tax refund claim with those documents in support, by way (in either case) of a positive statement as to the truth of some matter upon which it was intended that SKAT should rely.

There can be liability in deceit where a statement is made, intending thereby to convey a falsity that, on that objective approach, would not be said to have been conveyed. No such case was pleaded here, however, and it was accepted by SKAT that if, objectively, the tax reclaim documents submitted to it did not convey the falsities it alleged, then all of its pleaded claims failed. Specifically, SKAT accepted in closing argument that on the case pleaded and pursued at trial, each claim it pressed against each trial defendant (a) required it to establish one or more of the pleaded representations, and therefore (b) failed in limine if, objectively, none of the pleaded representations was made. The first part of that is true because the many causes of action other than for damages for deceit that were pursued by SKAT all required deceit to have been committed, as alleged by SKAT, save only that there was a claim against Lindisfarne for damages for negligent misstatement, but that was still founded upon the same set of alleged representations said to have been made to SKAT (or a sub-set of them). For example (as to the general point, not the specific point about Lindisfarne), unjust enrichment claims were pressed, but only on the basis that SKAT had paid claims under a mistake induced by the fraud it alleged, if it proved that fraud; or again, knowing receipt, dishonest assistance and proprietary claims were made based on alleged constructive trusts, but the trusts alleged were trusts said to arise in respect of the proceeds of the fraud SKAT alleged, if it proved that fraud.

I noted at paragraph 104(ii) above that almost all of the 4,170 tax reclaims in the case were made on Form 06.003, the then current form published by SKAT for the purpose. To the extent SKAT pleaded reliance on the contents of the Form, it did so by reference only to Form 06.003. It would not be appropriate to entertain a case that any of the pleaded representations was made when Form 06.008 was used, if it was not made when Form 06.003 was used. Furthermore, if a pleaded representation was made when Form 06.003 was used, but only because of language in that Form that was not in Form 06.008, in my view any claim based on that representation in the Form 06.008 cases should be dismissed even though the equivalent claim for Form 06.003 cases would not fail in limine in that way.

Form 06.003 was entitled “Claim to Relief from Danish Dividend Tax”. That reflects, indeed it states, the nature of the process. Each instance involved a claim to receive a payment from public funds, made to an organ of the state with responsibility for those funds whose decision whether to meet a claim was final, unless a challenge to that decision was put before a Danish Tax Tribunal by a party with standing to challenge it under Danish law. It was not a negotiation or discussion, and it was for SKAT to choose what information it required to be provided and the basis upon which any decision whether to pay would be made.

When it was signed and submitted by a Tax Agent, as was always the position here, the simple statement made by Form 06.003 was that on behalf of a “beneficial owner” identified on the Form, the Tax Agent as applicant made a claim “for refund of Danish dividend tax” in an amount stated in DKK. That did not represent to SKAT that the Tax Agent’s named client was entitled to such a refund; nor did SKAT allege such a representation. It was SKAT’s function to decide for itself upon what basis it would accept and pay Danish dividend tax refund claims. The Form did not require the named client or the Tax Agent on its behalf to certify entitlement, or belief in entitlement. In the absence of any such requirement in its specified Form, SKAT could not reasonably have understood that the Tax Agent was making a statement to it, intending it to rely on the statement, that the named client was (or believed itself to be) entitled.

Likewise, SKAT did not allege a representation that the Tax Agent’s named client had incurred, or believed it had incurred, liability to Danish dividend tax. There may be room for the view that an assertion of tax liability was implicit in the making of a refund claim or claim for relief. However, if that view were taken, there would then be an important issue whether that was again in the nature of a contention by the Tax Agent on behalf of its client, rather than a statement that SKAT could reasonably have thought was being made to it as a positive assertion, on which it was intended to rely, of the true position under Danish tax law. These matters were not explored at trial since an implied representation of tax liability (or belief as to tax liability) was not how SKAT put the case.

None of the representations alleged by SKAT was stated in terms in the Form or in any CAN. The submission for SKAT was, in substance, that it would reasonably be expected to be looking only to pay where there was entitlement to be paid, and therefore things said in the Form or a CAN should reasonably be taken to have conveyed that which would have to be the case for there to be an entitlement. I do not accept the premise, and in any event the conclusion does not follow: “I claim a Danish dividend tax refund” does not translate as or imply “I hereby state to you that the requirements of Danish tax law for entitlement to the tax refund sought are satisfied”, or “I hereby state to you that the facts are [whatever the facts have to be for there to be entitlement]”.

SKAT’s contrary argument was set out for closing as follows:

“76.

A reasonable tax authority in SKAT’s position would naturally have only wished to pay ‘refunds’ of withheld dividend tax where the requirements under its national tax law for a valid WHT refund application were in fact satisfied.

77.

… the Court has held in the Validity Issues Judgment that, in order to be entitled to a refund of withheld dividend tax from SKAT under Danish tax law:

77.1

The WHT Client must have been entitled to a dividend when declared by a Danish company;

77.2

By virtue of being a shareholder (for tax purposes) of the relevant Danish company when the dividend was declared;

77.3

Such that the WHT Client suffered a withholding of tax on such dividends by the Danish company.

78.

The background to each WHT Application was that:

78.1

a reasonable representee in SKAT’s position would be expecting that that Application and the documents supporting it would be informing SKAT that the WHT Client was entitled to a dividend on shares held in a Danish company and had suffered a withholding of tax on such dividend, of which a ‘refund’ was being sought; and

78.2

any person applying for a refund would wish to represent and demonstrate those matters to SKAT.

79.

This is the context in which the statements relied on by SKAT in the documents comprising the WHT Applications … fall to be interpreted, including:

79.1

The Tax Refund Form which identified the amount of tax withheld of which a “refund” was sought on behalf of a named WHT Client;

79.2

The DCAs which specified the number of shares on which the gross dividend had been declared, the net dividend paid to that WHT Client, and a figure for tax or tax withheld.

80.

Moreover, the meaning of shares, dividends and tax in the context of a WHT Application is necessarily driven by the meaning of those concepts in Danish tax law. That is how a reasonable representee in SKAT’s position would have understood them as a tax authority in the context of an application for a refund of withheld Danish dividend tax.”

SKAT’s paragraph 76 was a baldly asserted conclusion for which no evidence was cited that might justify it as a finding of background fact. It is the premise to which I referred in paragraph 441 above. I found in SKAT (Validity Issues) that even if SKAT had established what might otherwise be an administrative practice, as known to Danish public law, it could not have bound itself to pay tax refund claims in circumstances where the requirements of s.69B(1) of the Danish Withholding Tax Act, as I found them to be, were not satisfied (ibid, at [97] to [101]; and the Appendix at [9]). That means that if SKAT had adopted a continuous and consistent course of deliberate conduct by which it accepted and paid tax refund claims where one or more of those requirements was not met, it always remained open to SKAT to refuse any such claim. It does not mean that SKAT must reasonably be taken not to have made policy choices about when it would accept and pay that might involve it paying claims where it knew that the requirements of s.69B(1) were not all met, or the information available to it did not establish that those requirements were met.

There was evidence that a policy approach of that kind was known to be taken by the Belgian tax authority at the time. SKAT noted that the known approach in Belgium did not extend to treating dividend compensation payments as taxable dividend income if they were payments made between parties neither of whom ever acquired any shares, but that does not prevent the Belgian example from undermining SKAT’s bald general assertion as to the approach it should be assumed a national tax authority would be taking when construing tax refund claim documents to identify what, if any, representations were made by them to that authority.

Similarly, SKAT’s Legal Guide, taken at face value since SKAT proffered no other evidence on the point, evidences a choice by SKAT to treat as irrelevant whether trading giving rise to a tax refund claim involved short selling or stock lending (see paragraph 293 above). That also contradicts SKAT’s bald assertion about context, and again that is so even though SKAT was correct to submit that the Legal Guide did not provide reason to think, more specifically, that the share-less trading involved in this case might generate a valid tax refund claim (see paragraph 296 above).

SKAT’s paragraph 78 likewise was a bare argument unsupported by evidence. It also proves too much. If true, the conclusion would be that a representation of entitlement was made by the submission of a tax refund claim. The reality is that SKAT asked for its Form 06.003 to be used, and therefore would reasonably expect to receive, neither more nor less, whatever use of that Form would mean it received, and a person making a tax refund claim using the Form would be expecting to provide to SKAT whatever the Form required it to provide, neither more nor less.

I also disagree with SKAT’s paragraph 80, to the extent it was meant as a submission that words used in Form 06.003 or in CANs submitted to SKAT should be read as having specialist Danish tax law meanings. To assess whether a tax refund claim met the strict requirements of s.69B(1), if that had been its approach, SKAT would have needed to consider whether the claim had been made by or on behalf of a shareholder for Danish dividend tax purposes when the dividend in question was declared. It does not follow that if the tax refund Form asked for information about the named client’s position, SKAT could reasonably think that the user of the Form would intend their language to have specialist Danish tax law meanings if that was not explained by SKAT in the Form, or in notes or guidance for completing it. For example:

I was therefore not persuaded that SKAT’s specific submissions about context had any force. It can be said, as a more general statement as to context, that when considering communications to SKAT from a Tax Agent on behalf of a client, relating to a particular type of income tax collected by it, SKAT might reasonably be expected to be interested in information about income of that type earned and/or received. That could influence how what was submitted to SKAT might reasonably be expected to be understood by it. However:

That answers SKAT’s reliance on the well-known “helpful test” articulated by Colman J in Geest plc v Fyffes plc [1999] 1 All ER (Comm) 672 at 683B of considering “whether a reasonable representee would naturally assume that the true state of facts did not exist and that, if it did, he would necessarily have been informed of it”. In oral closing argument in reply on this part of the case, Mr Graham KC put that reliance as follows on Day 82:

“We say that a reasonable Danish Tax Authority would expect that if a named client did not own any Danish shares and had not received any dividends from a Danish company and had not suffered any withholding of any dividend tax, then they would not be submitting a DCA in the form they did as part of an application for a refund of withheld dividend tax. …

On any view, … against the background of that conduct [the submission of a claim] and those words [the content of a DCA] a reasonable tax authority would expect to be told if there were no shares, no dividend and no withholding of tax.”

That argument, if well founded, might perhaps lead to a conclusion that there was an implied representation that the Tax Agent’s client had owned shares on which it had earned a dividend on which it had been taxed. However, SKAT alleged different representations, and as Mr Head KC put it in a submission for the DWF Ds, the trial would have been a different trial if it had been about materially different alleged representations.

Submissions by defendants as to context focused on the fact that a national tax authority does not operate in an ivory tower or a vacuum, and I agree. It would reasonably be expected to have a general knowledge of the international legal and market environment in which any WHT refund scheme operated by it sat. Thus, SKAT would reasonably be expected to be aware that DTTs on the OECD model used a Treaty notion of the ‘beneficial owner’ of dividends, not tied to or based on the legal concepts of any national system of law, to locate possible entitlement to tax relief. Such DTTs did not stipulate, in terms, for such an entitlement. They imposed Treaty obligations not to levy tax on dividends, either at all or above a specified maximum tax rate, where the ‘beneficial owner’ of the dividends was within a relevant Treaty category. The entitlement to relief therefore needed to be provided if Denmark was to comply in substance with those Treaty obligations, since (so far as material) it taxed dividends at source by a general withholding tax obligation imposed on the company and attaching to the dividend declaration.

It was also to be expected that SKAT, acting reasonably, would be aware of the information and guidance it had published concerning Danish dividend tax and the availability of tax relief on the basis of DTTs. I dealt with that as part of setting out the basic facts, at paragraphs 286 to 296 above, and referred to it also at paragraph 445 above. SKAT would also be expected to be aware that it could require further information from those seeking tax refunds from it, including if it considered that to be necessary to ensure (if this was its aim) that it only paid where there was a refund entitlement under Danish tax law.

As a national tax authority collecting dividend tax on a withholding basis and operating a process for dividend tax refund claims to be made to and paid by it, SKAT reasonably would have been expected to be aware and to have a general understanding of common equity trading practices. In receiving and processing dividend tax refund claims during the relevant period, SKAT would reasonably have been expected to be aware, therefore, of stock lending, short selling, deferred settlement, and (in outline at least) div-arb, including tax arbitrage and cum-ex, as practices engaged in by parties trading in equities markets.

SKAT reasonably would have been expected to be aware as a result that, as was in fact the position:

Finally, SKAT would reasonably have been expected to be aware that a shareholder in principle entitled to dividend income from a Danish company (it may be via a chain of custody) might in various ways transfer that entitlement or its benefit to another, so that even the receipt of real dividend income was not always tied to share ownership. Indeed, that was implicitly recognised in the beneficial ownership element of DTTs, the Treaty obligation being not to tax (or not to tax above a certain rate) where the beneficial owner of the dividend qualified under the DTT. It was also recognised explicitly by SKAT itself in Form 06.008, use of which it still accepted, in that it allowed for claims by owners or usufructuaries. (I have said it would be inappropriate to allow SKAT to rely on wording in Form 06.008 to found any of its alleged representations, having pleaded no such reliance. That does not mean I should ignore this evidential value of the Form on a matter going to context.)

The pleading of the representations that SKAT says were made to it by each of the tax refund claims submitted to it was complex. Care was needed to bear in mind not only the primary pleading, alleging the representations, but also both a preparatory plea in the Particulars of Claim and important Further Information provided by SKAT pursuant to a Request for Further Information served by the DWF Ds after the Validity Trial.

The preparatory plea, paragraph 17(c) of the Particulars of Claim, was that each CAN, so SKAT alleged, “purported to record that:”

Reading together paragraph 19 of the Particulars of Claim and the Further Information clarifying and confining SKAT’s case, the plea by SKAT was that, given certain of the content of the tax reclaim documents submitted to it and the fact that they were being submitted so as to make a tax refund claim in respect of Danish dividend tax, the Tax Agent, by submitting the claim, represented to SKAT that:

The dividend entitlement representation and dividend payment representation are set out above as separate representations. There is some convenience in that when considering whether the pleaded representations were made. However, they were pleaded as a single, composite representation, viz. (in the basic plea, before building in the Further Information) that “[the named client] had received the dividend, and in due course payments representing the dividends, net of tax, described in the Credit Advice Note (on the “payment date” recorded therein)”. The Further Information clarified that the pleaded case was of a representation of entitlement as a matter of Danish tax law and the receipt of a payment in respect of that entitlement. Therefore, the allegation of the dividend payment representation was not freestanding and it falls away if the dividend entitlement representation, as alleged, was not made.

Each of those core representations is detailed and specific. Nothing like any of them was stated expressly in any of the documents submitted to SKAT, as SKAT recognised by its plea that the representations were not made expressly, but “partly expressly and partly impliedly”, the sense of which I considered in paragraph 435 above and have sought to capture in the opening words of paragraph 457 above.

For closing argument, SKAT sought to recast the pleaded representations by a “Restatement of Alleged Representations” which drew a distinction, for each, between the representation pleaded and what was said to be the representation “in substance” alleged. I did not find that helpful. If the suggested ‘substance’ differed materially from the pleading, the ‘Restatement’, if pressed, would have been an objectionable attempt to amend beyond the eleventh hour; if it did not differ materially from the pleading, it would have been unobjectionable, in the final analysis, but an unwelcome, unnecessary distraction nonetheless. In fact, the Restatement did articulate a ‘substance’, in my view, that differed materially from the pleading.

After closing argument, there was no real dispute but that an understanding of the material essence of a representation suffices where the tort of deceit requires awareness or understanding that the representation was being or had been made. It therefore does not matter if an individual whose awareness or understanding falls to be considered might not have articulated their thinking in terms that match precisely those in which the court chooses to articulate the representation, or those in which it may have been pleaded. Moreover, the variety and flexibility of language mean that in any given case, there may be room for articulations that look quite different on the page but convey, in truth, essentially the same message. When applying that principle, however, what matters is the essence of the pleaded representation, not therefore, in this case, the materially different ‘substance’ in SKAT’s Restatement document. This judgment determines the claims pleaded by SKAT, and I consider that paragraph 457 above sets out accurately the core representations pleaded.

In oral closing argument, Mr Graham KC accepted the force of concerns expressed about the Restatement document, and rested SKAT’s case on the core representations as pleaded, so long as the ‘material essence’ principle applied, as in my judgment it did. He also provided an articulation of the material essence of those pleaded representations, as SKAT would have it for the purpose of applying that principle. I consider those articulations when dealing with each of the pleaded representations in turn, below.

Returning to the formulation of the core representations in paragraph 457 above, there was an issue as to whether the tax ownership representation, as pleaded, was, even more specifically, an alleged representation that in respect of the shares described in the CAN, the Tax Agent’s named client was either (a) a dividend date shareholder who was not to be treated by Danish tax law as having disposed of those shares prior to the dividend declaration, or (b) a party, when the dividend was declared, to whom a dividend date shareholding was to be treated by Danish tax law as having been disposed and by whom that shareholding was not to be treated by Danish tax law as having been disposed, those being the categories of legal person that the Validity Trial determined to be shareholders when a dividend was declared, in the eyes of Danish tax law, subject therefore to Danish dividend tax liability on the declared dividend.

That issue arose because in the Further Information that I have said needs to be taken into account, as regards the tax ownership representation SKAT:-

That continuation also said (in line with paragraph 17(c)(i) of the Particulars of Claim (see paragraph 456 above)) that, so SKAT alleged, each CAN recorded that on the date of the CAN there was a settled positive balance of shares in the referenced Danish company held by the custodian for the Tax Agent’s named client, from which the Shah Ds derived a submission that the tax ownership representation, as pleaded, was also defined, and confined, by that.

In my judgment, the sense of the relevant RFI Response was not that SKAT was alleging a representation quite as specific and detailed as those submissions would have it. SKAT’s reference to the requirements for Danish tax law share ownership as determined by SKAT (Validity Issues) was not part of the pleaded representation. It was a plea making clear what SKAT said it would take for the pleaded representation to have been true, viz. that those requirements would need to have been satisfied. It is like an allegation that a job applicant represented by their c.v. that they had achieved a particular award (say, a Gold Duke of Edinburgh’s Award). If that were followed by a plea stating, as alleged by the claimant, the requirements for achieving that award, one would not naturally interpret that as alleging, as such, a representation or set of representations that the applicant had satisfied each of those requirements. The supplemental plea would be likely to be taken to identify the criteria by reference to which, according to the claimant, the truth or falsity of the pleaded representation that the applicant had a Gold DofE Award fell to be judged.

The plea in the Further Information harking back to the preparatory plea about the contents of CANs was a reiteration of one of the alleged planks upon which SKAT claimed that the tax ownership representation fell to be constructed. That involved an unhelpful jumbling of pleaded points; but it is not a sensible reading of the pleaded case as a whole, I think, to say that SKAT was confining itself to a representation of tax ownership on the dividend declaration date, in respect of a settled balance of shares on the date of the CAN. A CAN reported a payment, and was therefore issued on or after a payment date. On any view, it did not say or imply anything about the existence of a settled balance of shares on the date of the CAN.

There was also a question, raised by Mr Jones KC for the Shah Ds, over the formulation of the dividend entitlement representation, as pleaded, in paragraph 457(ii) above. Was the representation, as alleged by SKAT, that the client had been (as I have it there) the shareholder for tax purposes on the dividend declaration date, or that it had been the dividend date shareholder (as I defined that term in SKAT (Validity Issues) at [166])?

That arose because in the same Further Information, SKAT clarified that the pleaded representation that the Tax Agent’s client “had received the dividend” was a representation, as alleged by SKAT, of “receipt [by that client] of a dividend for tax law purposes by the owner of the shares on the dividend declaration date (as explained at paragraph 179 of the Validity Issues Judgment)”. SKAT (Validity Issues) at [179] completed and summarised my “conclusions as to the accrual and transferability of the right to dividends under Danish company law” (ibid at [183]) and did not concern Danish tax law. The cross-reference to [179] in SKAT’s Further Information was potentially confusing, therefore, although if there was any confusion the root cause may have been that in framing the relevant RFI Request, the DWF Ds said that [179] set out a conclusion of Danish tax law. Be that as it may, the only way to make sense of SKAT’s Response is if the cross-reference to [179] relates to and qualifies only the words “dividend declaration date”. That is to say, the Further Information clarified that the dividend entitlement representation, as pleaded, was defined as and confined to a representation that the Tax Agent’s client had received a dividend entitlement by being the relevant owner, i.e. the shareholder in the eyes of Danish tax law, at the relevant time, i.e. at market close on the dividend declaration date.

Finally as to the articulation of the alleged representations in paragraph 457 above, Mr Head KC queried in oral closing argument whether, as pleaded, the tax representation referred to (payment in respect of) the dividend declared by the referenced company, rather than the referenced dividend (as I have it in paragraph 457(iv) above). I was during argument and remain unable to discern any meaningful difference between the two.

SKAT also alleged that each CAN itself made to SKAT the core representations set out in paragraph 457 above, plus an implied representation that the CAN was an honest and accurate statement by the custodian issuing it of the facts set out in it (the ‘honest custodian representation’). The Further Information clarified that the honest custodian representation alleged by SKAT was in fact only, and more narrowly, a representation that the custodian had issued the CAN with an honest belief as to the truth of the core representations as (allegedly) made by the CAN. In final argument, having reviewed the trial transcripts, SKAT withdrew reliance on the honest custodian representation against Mr Horn, because in Mr Horn’s case Mr Graham KC said that he had not dealt with it sufficiently in cross-examination.

In my judgment, it was a misguided and unnecessary complication for SKAT to plead representations made to SKAT by CANs as some separate or additional case. It was always said by SKAT, rightly so, that any representations made to it as a result of what was stated in a CAN were made because of the content of the CAN, interpreted as a document received by SKAT in support of a claim for a Danish dividend tax refund. A question might arise whether a custodian issuing a CAN could have a primary liability arising from what it stated to its equity buyer client by issuing a CAN to it, if the CAN was submitted to SKAT. But that does not mean that focusing on a CAN, to the exclusion of the tax reclaim form with which it was submitted, is a worthwhile exercise when considering what, if any, representations were made to and relied on by SKAT. In my judgment, it is not, indeed it is not in truth a meaningful exercise. As the DWF Ds put it in their written closing submissions, “It makes no sense to speak of any representation to SKAT that could arise from a [CAN] other than as part of a WHT Application, given that each [CAN] was only ever received and read by SKAT together with the other WHT Application documents that accompanied it”.

SKAT also pleaded implied representations that the Tax Agent’s named client had an honest belief as to the truth of the core representations, as (allegedly) made by the tax refund claim documentation, and that the signatory of a CAN (where the CAN was signed) believed the representations (allegedly) made by the CAN to be true (including the alleged honest custodian representation). In its written closing submissions, SKAT withdrew reliance on either of those as a representation said to found any cause of action. It was said that SKAT did not concede that those further alleged representations were not made, or could not have founded a claim, but considered that they could not add anything. I make no comment on that. All that matters for my purposes is that the further alleged representations were not pressed in closing, so I do not need to deal with them.

Nothing turns on the documents provided to SKAT with the subject tax refund claims that confirmed the tax status in their respective home jurisdictions of the Tax Agent’s named clients, for example U.S. Form 6166s by which the IRS certified that, to the best of its knowledge, USPFs qualified under section 401(a) of the U.S. Internal Revenue Code to be exempt from taxation in the U.S. under section 501(a). Likewise, the Powers of Attorney by which the Tax Agents evidenced their authority to submit claims on behalf of their clients.

Therefore, the tax refund claim documents for each claim were, so far as might be material, the Tax Agent’s cover letter, a completed Form 06.003 (or, in the few cases where it was used, Form 06.008), and a CAN.

In its pleadings and argument, SKAT relied on particular turns of phrase used by the Tax Agents in their cover letters, in support of the submission that the core representations were made to it. For example, Goal’s cover letters described CANs as “evidence of payment and tax deduction paid on the client’s securities”, or again Syntax’s and Koi’s cover letters described the claim being made as a claim for a “refund of Danish Dividend Tax that was previously withheld in relation to their investments”.

However, Mr Nielsen’s evidence left me in no doubt but that he paid no regard at all to what was said in the cover letters when processing any of the tax refund claims. There was no evidence, or reason to suppose, that anyone at SKAT other than Mr Nielsen paid any attention to the cover letters, save that in the few cases where Form 06.008 was used, someone may have had to take the Tax Agent’s bank details for payment from the cover letter as those details were not asked for in the Form itself, as they were in Form 06.003, and the Tax Agent may or may not have thought to enter them onto the Form. In reality, SKAT, through Mr Nielsen, received and processed tax refund claims as claims consisting of the completed Form, supported by a CAN or CANs and the documents referred to in paragraph 474 above. The cover letters may as well have said just, “Please find herewith completed Form 06.003 [or 06.008] for consideration” (and, where Form 06.008 was used, “Bank details for any payment as follows: …”, if the Tax Agent had not added those details to the Form). They would have been processed by Mr Nielsen, and paid by SKAT, exactly as they were in fact.

It follows that if, considering the matter objectively, one of the representations alleged by SKAT was conveyed by the tax refund claim documents only due to something said in the cover letter, it did not induce Mr Nielsen’s approval of the claim, and any claim founded on that representation fails for that reason.

The documents submitted to SKAT did not refer to or provide any information about any particular trade or trading structure to which the Tax Agent’s named client had been or was privy. SKAT did not require any such information to be provided. That needs to be taken into account when considering the objective purport of those documents, such as might give rise to a representation made to SKAT. The question is what would the documents convey to SKAT, acting reasonably, bearing in mind inter alia that they did not say anything about any trades or trading structures involved.

Finally before considering each representation in turn, I deal with SKAT’s preparatory plea about CANs (paragraph 456 above). That was a plea as to what, SKAT says, the CANs were, and what they conveyed, as documents issued by a custodian to a client, which may inform any consideration of whether the tax refund claim documents conveyed the pleaded representations to SKAT, given that each claim was supported by a CAN.

Examples of the different forms of CAN, it will be recalled, appear in Appendix 5, below. It was variously contended both by SKAT and also by defendants that the content and connotations of the different forms of CAN were materially similar. However, each of those contentions was made as part of an argument that the representations alleged by SKAT were made (according to SKAT) or were not made (according to defendants). The rival submissions about similarity cannot be put together to say that it was common ground that the CANs were materially similar in purport. Neither submission was a concession that the case on which, if any, of the pleaded representations was made stands or falls uniformly across all custodians. I therefore reject attempts that were made, on occasion, on either side, to pair a submission about one form of CAN with the opposing side’s submission as to similarity, for an argument that if the former was persuasive, its conclusion could be applied across the board. SKAT’s preparatory plea, as to the purport of CANs in themselves, and then its claim that the representations it alleged were made to it, must be judged separately, in my view, for each form of CAN.

SKAT alleged, firstly, that each CAN purported to record that a stated number of shares in a named Danish company was held for the client to whom the CAN was issued (paragraph 456(i) above). In my judgment, that was not true for any of the CANs, although for the NCB CANs it may be that is so only because SKAT made clear, by the Further Information to which I have been making repeated reference, that the allegation was, more specifically, that CANs reported share ownership as a matter of Danish tax law on and before the ex-date.

A CAN issued by SCP advised the addressee that SCP had credited its account and that “This payment represents the dividend as shown below:”, under which details were set out of dividend income that it might meaningfully be said that a payment credit represented, namely dividend income, net of withholding tax, on a stated number of shares in an identified Danish company (also identifying, in early versions of the SCP CAN, the “Pay Date” for the referenced dividend, and in later versions also the “Ex Date” and the “Record Date”). None of that purports to record any holding of shares by the custodian for the addressee. It is a cash account credit advice, not a custody statement setting out securities holdings. The receipt of a payment referable to a dividend declared by a listed company did not imply that the recipient had a shareholding in that company (see paragraph 453(i) above). An SCP CAN was entitled “DIVIDEND CREDIT ADVICE”, but that does not change the limited substance of what it stated, and the use of the word ‘dividend’ in a CAN did not mean or imply that there had been an entitlement to a real dividend (see paragraph 453(iii) above).

I noted in paragraph 411 above that the SCP CAN was modelled on the CANs issued by Merrill Lynch in connection with the Broadgate transaction. I am not trying any claim or issue concerning that transaction, which related to German shares and German tax and was not investigated in depth at the Main Trial. I cannot make any firm finding as to this, therefore, but it seems plausible that the language may have been chosen by Merrill Lynch to ensure that it should not be taken to suggest dividend income earned on a shareholding rather than a more derivative type of income indirectly reflecting or relating to dividends that had been declared.

An Old Park Lane CAN contained no narrative statement as to what it was reporting. It was entitled “INCOME ADVICE”, above the name and address of a client as addressee, and did not refer to or give information about any account of the client’s at Old Park Lane. It thus took the form of a report of income, rather than that of a credit advice reporting the crediting of a payment to an account. By way of content, then, it tabulated details of a Danish share security, the “Ex-dividend Date”, “Record Date” and “Payment Date” for a “Dividend” declared on that security, and “Payment details” identifying a “Dividend Per Share”, a “No. of shares”, and “Gross”, “Tax” and “Net” amounts.

Read sensibly, as a report of income by a financial services firm to its client, in my judgment an Old Park Lane CAN thus conveyed that the client had received payment on the stated “Payment Date” of the “Net” amount shown, and how that was calculated from the referenced dividend. The CAN did not say anything about the nature of or basis for any entitlement the client may have had to such an income payment; and the receipt of such a payment does not imply that the recipient had a shareholding (see paragraphs 453 and 454 above). In particular, in the context of the referenced Danish share security, it was a given that the only party that would ever deduct anything from a ‘dividend’ payment made by it would be the Danish company itself, when paying VPS. In a payment credit advice remote from the company, and given the matters referred to in paragraph 453(i) to 453(iv) above, those “Payment details” in an Old Park Lane CAN did not, in my judgment, purport to confirm anything more than how the reported income had been calculated.

A West Point CAN was entitled “DIVIDEND CREDIT ADVICE”, above a date and a “Tran Ref” (presumably ‘transaction reference’, not explained further), all above the name and address of the client as addressee. Like the Old Park Lane form of CAN, a West Point CAN contained no narrative sentence identifying what it was reporting. It likewise did not refer to or give any detail of any account of the client’s at West Point; however, the document title identified that it was advising the client that West Point had credited a ‘dividend’ payment to the client, implying (I think) the existence of such an account. The West Point CAN content was materially the same as, albeit formatted marginally differently from, that of an Old Park CAN. As before, the receipt of a ‘dividend’ credit does not mean or imply the existence of any shareholding, and in my view a West Point CAN therefore conveyed materially the same information and message as was conveyed by an Old Park CAN, plus (only) the (implicit) statement that the relevant payment had been credited to an account of the client’s at West Point.

A Telesto CAN was entitled simply “CREDIT ADVICE”, above a date and “ID” number (not explained further), all above the name and address of the client as addressee. By way of content, it identified a “Security” (always a Danish share issue, identified by name and ISIN), an “Ex Date”, “Record Date” and “Pay Date” relating to that security, and then data for “Dividend Per Share”, “No of Shares”, “Gross Dividend”, “Withholding tax deducted” and “Net Dividend”, without any sub-heading. In my judgment, the purport of a Telesto CAN was the same as that of an Old Park CAN, for the same reasons, plus (as with the West Point form of CAN) an implicit statement that the payment being reported had been credited to an account of the client’s at Telesto.

Indigo and Lindisfarne CANs were materially identical to Old Park Lane CANs, except that an Indigo CAN was entitled “CREDIT ADVICE – DIVIDEND” and a Lindisfarne CAN was entitled “DIVIDEND CREDIT ADVICE”, rather than the Old Park Lane CAN’s “INCOME ADVICE”. Again, therefore, no statement was made or implied that shares were held; and in my judgment Indigo and Lindisfarne CANs conveyed the same information and message as did Old Park Lane CANs, plus (implicitly) the information that the reported payment credit was to an account of the client’s at Indigo or Lindisfarne, respectively.

A Salgado CAN was entitled “CREDIT ADVICE”, and set out information relating to a security and dividend from which, it is evident, “Due dividend payment details” of a “Due payment amount” and “Due payment date” had been derived. That purported to do no more than explain the calculation of the amount of the credit being reported, and did not make or imply any statement about shares being held for the client. The client’s name was entered against “Name of beneficial owner”, which (if anything) reinforces the view that the CAN could not be read as purporting to record any shareholding (cf paragraph 454 above).

The CANs issued by NCB in the Maple Point Model were materially different in form and content. They purported to record “Dividend income” in respect of “Holdings as at” a stated date (always in fact the dividend declaration date) of a stated number of Danish shares. In my view, that could only reasonably be read as confirmation that the client held the shares referred to, on the dividend declaration date, and was being credited with a ‘dividend’ income payment, calculated net of tax deducted at source, on that holding. The fact that similar content, as to a ‘dividend’ payment, might not by itself imply anything about shareholding does not detract from the clear sense of the express “Holdings as at” statement that an NCB CAN made.

An NCB CAN stated on its face that it was “Not a tax certificate”. That meant NCB made no statement, and the CAN was not to be relied on, as to whether tax had been deducted at source, or otherwise as to the tax consequences of the “Actual payment” to the client of the amount stated. In my view, that did not negative or qualify the “Holdings as at” statement, which was not a statement concerning the tax status or consequences of the reported holding. Nor did anything in the small print on the second page of an NCB CAN serve to negative or qualify the “Holdings as at” statement (so long as it is not taken to say anything about tax status or treatment).

It is not an answer, as was submitted on behalf of the DWF Ds in respect of the objective purport of an NCB CAN, that NCB used transaction recording and reporting software used by a number of German banks so that its CANs were what that software package generated from the inputs it was given on Maple Point Model trades. The evidence at trial disclosed no reason to suppose that whatever inputs the software required were not drawn accurately from the Maple Point trades as executed. Equally, however, on the evidence, I cannot say whether the required inputs would have ‘told’ the software that there were no shares, yet it generated the “Holdings as at” statement nonetheless, or whether the software in effect wrongly assumed that any share trade settlement will have been DVP, i.e. will have involved the transfer of a shareholding. Either way, the Maple Point Model, it seems, exposed a weakness in the software, as the plain fact is that it generated CANs purporting to record shareholdings where there had been none.

That brings me to the point arising from SKAT’s Further Information. Asked to clarify what SKAT meant by alleging that CANs purported to record “shares … held”, the Response was “that the shares specified in the [CAN] (i) were owned (as a matter of Danish tax law) on and before the Ex-Date …; and (ii) by the time the [CAN] was produced constituted a settled positive balance of shares …”. In my view, there is no warrant for reading anything like that into the NCB CANs, or any of the CANs for that matter, none of them having stated anything like it expressly. Furthermore, in the case of an NCB CAN, the claim that it made a statement about Danish tax law ownership is directly negatived by the disclaimer that the CAN was “Not a tax certificate”.

Secondly, SKAT alleged that each CAN purported to record that a specific payment representing a dividend had been received for the account of the client to whom the CAN was issued (paragraph 456(ii) above). I do not agree. Each CAN reported to the client that the custodian had credited it with an amount reflecting or based on the dividend identified, or (in the case of an Old Park CAN) that the client had received an income payment in such an amount. None stated that the custodian had “received for your account …”, or similar, or had to be based on a receipt by the custodian so that some such statement might be implied.

Thirdly, then, by the final part of its preparatory plea concerning the purport of the CANs themselves, SKAT alleged that they purported to record that “such payment of dividend” had been received by the client “net of a specific amount of tax that had been withheld by the … Danish company” (paragraph 456(iii) above). The subject matter (“such payment …”) is the payment received by the custodian for the client’s account that CANs reported, according to SKAT. But as I have just stated, in my judgment CANs did not make such a report.

I therefore do not accept any part of SKAT’s preparatory plea at paragraph 17(c) of the Particulars of Claim, quoted in paragraph 456 above, as clarified by its Further Information, concerning the purport of CANs, as issued to custodians’ clients.

Mr Graham KC articulated the essence (see paragraph 462 above) as being an allegation that SKAT was told that “the named client was the Danish tax law owner of the shares when the dividend was declared”. That is to the same effect but uses fewer words, so there could be no objection to considering the case by reference to it.

There was no express statement to that or any similar specific effect in any of the tax reclaim documents. The argument for SKAT was that it was nonetheless conveyed by them, so that on an objective assessment the alleged representation was made to it, because:

For those reasons, I am not persuaded that the tax ownership representation was made to SKAT as it alleged.

As will be clear from paragraph 499 above, SKAT placed no reliance in closing argument on the language of any of the Tax Agent’s cover letters. There was pleaded reliance, in support of the tax ownership representation, on the fact that Goal’s letters stated that they enclosed “a tax reclaim form together with evidence of … tax deduction paid on the above client’s securities”, and that Acupay’s, Syntax’s and Koi’s letters referred to a “reclaim application” from a “qualifying” entity for a “refund of Danish Dividend Tax that was previously withheld in relation to their investments”. Those introductory words do not in my judgment amount to or convey a statement to the detailed and specific effect of the alleged tax ownership representation, or anything like it.

The essence of that was said by Mr Graham KC to be that SKAT was told that “the named client (a) was entitled as a matter of Danish tax law to the dividend referred to in the CAN and (b) had received directly or indirectly from the Danish company a payment net of tax in respect of that entitlement”. Part (a) of that is to the same essential effect as part (a) of my longer formulation, above, distilled from the pleadings. I agree with the submission that was made, in explaining the formulation I have just quoted, that part (b) is also essentially the same: the primary element is still the receipt of a payment, net of tax, in respect of the entitlement referred to in part (a); the additional detail is essentially the same, although quite different words are used, because the fuller formulation reproduced in the heading above merely spells out how it is that dividend payments from the dividend declaring company are distributed.

I agree with a submission by Mr Head KC that there is a mis-match between part (a) and part (b). Depending on the detail of their transactions: on the one hand, a party in receipt, directly or indirectly, of a dividend payment from the company, might or might not have been the shareholder entitled to the dividend in the eyes of Danish tax law; and on the other hand, that shareholder might or might not receive such a payment. That incongruence is relevant to whether the alleged representations were made, but it is not a reason to reject Mr Graham KC’s simplified formulation as in any way unfair or materially inaccurate as an encapsulation of the essence of the pleaded case.

I did not find SKAT’s argument that the dividend entitlement representation was conveyed by the tax reclaim documents persuasive. The main flaw, once again, was its reliance on the non sequitur I identified in paragraph 441 above. SKAT submitted that:

The key point, once again, is that the making of a claim, or even the assertion by one party to another of a right against them, is not by nature a representation. Even if there might be situations where that does not hold, it is in my view true of the presentations of claims here, to SKAT as a national tax authority with responsibility for deciding for itself on what basis or bases it was willing to recognise a claim, and what information or evidence it required to receive in order to make a decision. If SKAT meant, by its reference to “asserting a right”, a statement by the Tax Agent to the effect that “our named client is entitled to a tax refund”, the simple fact is that SKAT did not ask for or receive any statement of that kind, and did not allege that a representation to that effect was made to it.

SKAT developed, or unpacked, that basic submission, as set out below, but did not improve it by doing so. In what follows, I have removed reliance SKAT sought to make in closing upon specific aspects of the language of Form 06.008 not found in Form 06.003 (see again paragraph 437 above):

For those reasons, in my judgment the dividend entitlement representation was not made to SKAT as it alleged. For completeness, it should be clear from my reasoning, above, that SKAT’s reliance on the contents of the Tax Agent’s cover letters in the argument for the dividend entitlement representation did not affect the conclusion.

The dividend payment representation, as alleged, supplemented the dividend entitlement representation, if made, and could not stand alone. It therefore falls away, as I explained in paragraph 458 above. For completeness only, I add that even if the dividend entitlement representation had been made, I would have rejected the argument that the dividend payment representation followed. It did not, because of the mis-match correctly identified by Mr Head KC (paragraph 503 above).

Mr Graham KC shortened that, in his oral argument as to the essence of the representations alleged by SKAT, to a case that SKAT was told that “the Danish company had withheld the tax identified in the CAN from the dividend paid to the named client”. There can be no objection to that paraphrase of the pleaded case, as long as care is taken to keep in mind what must be meant by the notion of the Danish company withholding tax from a payment made to the Tax Agent’s client.

Mr Graham KC’s paraphrase was close to the primary pleading, at paragraph 19(c) of the Particulars of Claim: “The Danish company had withheld the tax described in the [CAN] from that payment”. SKAT was asked to make clear what it meant by that, and in particular whether the representation alleged was to the effect that (i) the Danish company had withheld dividend tax when paying VPS, (ii) the Danish company had withheld dividend tax from a payment made directly to the Tax Agent’s client, or (iii) something else. In response, SKAT said that it alleged a representation “to the effect that the Danish company which had declared the dividend had withheld the amount of tax identified in the [CAN] so that the payment received by the [Tax Agent’s client] was paid net of the tax withheld” (my emphasis).

That was an important clarification. The basic pleading might have been taken to allege a statement that the Danish company had deducted tax from a dividend payment made by it to the Tax Agent’s client. The words I have emphasised made clear that SKAT was not alleging a statement to that effect. Rather, the statement alleged was to the effect that the Danish company had withheld tax with the result that the payment received by the Tax Agent’s named client was net of that withheld tax. In my view, it is only meaningful to speak of a payment received remotely from the Danish company (such as by one of the Tax Agents’ clients here) as having been received net of tax that had been withheld by the Danish company, if the payment was at the end of some kind of payment chain starting with the Danish company.

I am therefore content to adopt Mr Graham KC’s formulation of the essence of the tax representation for the purpose of considering the claim, but always bearing in mind that clarification as to what it means. The tax representation, as alleged, thus depended on the notion, pleaded by SKAT in its preparatory plea about CANs, that a CAN purported to record the receipt by a custodian of a payment for the account of the client. The tax representation plea was that, more particularly, a CAN reported the receipt by the custodian, for the client’s account, of a payment net of tax at the end of a payment chain originating with the Danish company.

I did not accept the claim that a CAN stated or implied the receipt of a payment for the client’s account, or therefore the more particular claim, similar perhaps to the allegation that the tax representation was made, that a CAN reported a receipt by the custodian net of tax withheld by the company (see paragraphs 495 and 496 above). In my view, that is fatal to the claim that the tax representation was made.

The argument for the tax representation was put by SKAT as follows (again removing reference to language in Form 06.008 that does not appear in Form 06.003 (see paragraph 437 above)):

My findings, therefore, considering the matter objectively as called for by the way SKAT put the case, are that:

As SKAT rightly submitted, a representation of fact may carry with it an implied representation that the representor honestly believes the fact to be true. In other cases, it may be that what is claimed to have been a representation of fact is held not to have been, because the representee would not reasonably be expected to take the representor to be speaking to the truth of the matter (see paragraph 432 above for the basic concept of a representation, which was not contentious in this case). In such cases, there might nonetheless be (instead) an implied representation that the representor honestly believed the fact stated to be true, or (perhaps) had no reason to think it not true.

SKAT submitted that the honest custodian representation was implied here on that basis, and that the ‘helpful test’ in Geest v Fyffes was met (paragraph 449 above). According to that argument, “a reasonable representee [in SKAT’s position] would have naturally assumed that the true state of facts did not exist (i.e. that the Custodians did not honestly believe the statements made in the relevant documents) and that if that were the true position they would have been informed of it or the DCAs would not have been issued in the first place.”

That argument, however, misunderstands and misapplies Colman J’s idea. The key point here, in each instance, is that the tax refund claim was being made by the Tax Agent, expressly acting on behalf of its named client, and not by or on behalf of the custodian that had issued the CAN. Any representation to SKAT was a representation by the Tax Agent, acting in that capacity. Any implied representation as to the belief of the representor would be a representation about the Tax Agent’s belief, or (perhaps) about its named client’s belief, not a representation about the custodian’s belief. Applied to the present case, Colman J’s dictum would invite consideration of whether SKAT naturally would assume that if the custodian did not believe the core representations (if they were made by the CAN), SKAT would have been told as much by the Tax Agent.

The short answer, in my judgment, is no. There might be room for saying that it would naturally be taken as implicit that, so far as the Tax Agent (perhaps also its client) was aware, the custodian was acting honestly in issuing the CAN, or that the Tax Agent (again, perhaps also its client) had no reason to suppose the custodian was not acting honestly. There might also, or instead, be room for the view that a representation of some kind was implicit that the CAN was genuine, i.e. had been issued by the custodian to the client in respect of an account held by the client with the custodian. However, SKAT would have no reason to suppose and could not reasonably expect or naturally assume that the Tax Agent (or its client) would know, if this were the case, that the custodian was not acting honestly. The Tax Agent’s conduct in submitting the tax refund claim could not reasonably be thought to convey, as a statement of fact made to SKAT, that the custodian was acting honestly.

I find therefore that the honest custodian representation alleged by SKAT was not made.

The result, and my decision, is that of the various representations alleged by SKAT and pursued by it in closing argument, considering in the first instance only whether they were made at all in the objective sense that SKAT accepted was a threshold requirement for all of its claims:

SKAT’s claims concerning tax refund claims submitted by Acupay therefore fail in limine. There were 1,232 such claims (out of the 4,170 with which I am dealing), seeking payment in aggregate of DKK3,543,799,028.53, which was paid by SKAT.

SKAT’s claims concerning the other 2,938 tax refund claims, on which it paid in aggregate DKK8,547,045,919.80, also fail in limine to the extent they were founded on alleged representations other than the tax representation. If the tax representation was made in those claims, because of the Tax Agents’ cover letters, it was always a misrepresentation. The dividend compensation payment that had been credited to the Tax Agent’s named client, and which therefore was the subject of the CAN submitted to SKAT, was not a payment from which tax had been withheld by the Danish company in question in the sense conveyed by the representation.

Having mentioned this already in passing at paragraph 478 above, I can say immediately that the tax representation, if thus made (considering the matter objectively), did not induce SKAT in any way. In those instances, it could only be said that the representation was made at all, taking an objective approach, because that approach effectively assumes that attention would be paid to the Tax Agent’s cover letter, potentially influencing what might be gleaned from the documents sent to SKAT, taken as a whole. But SKAT’s process, as implemented by Mr Nielsen, treated the cover letter as irrelevant; and Mr Nielsen in fact paid what was or was not said in the cover letter no attention at all. As I said in paragraph 477 above, my finding is that if the cover letter had just said, “Herewith completed Form 06.003, for your attention”, the tax refund claim would have been treated in exactly the same way. What each Tax Agent chose to say in its cover letter had no influence at all on what occurred.

The result is that SKAT’s claims in these proceedings all fail, without the need to explore in any greater detail how it dealt with tax refund claims during the relevant period and whether, therefore, its case on inducement was made out at trial. I deal with that fully nonetheless, below, to explain why I concluded that indeed SKAT’s case on inducement was not made out.

I made some initial observations about the requirement of inducement when introducing this main section of the judgment (paragraphs 424 to 431 above). I consider that Zacaroli J stated the legal test accurately, and sufficiently for the present case, when he said as follows in Farol Holdings at [216] (but, again, I generalise it from the particular case of inducing entry into a contract):

“… the representee must in fact have been induced to take action … in reliance on the representation. The misrepresentation need not be the only reason for the representee’s decision to [act as it did], but the representee will have no cause of action if it would have [acted as it did] even if the representation had not been made. If it is proved that a false statement is made which was material – in the sense that it was likely to induce [the action said to have been induced] – then there is an evidential presumption (of fact, not law) that the representee was so induced. The presumption is stronger if the representation was made fraudulently.”

Of the strength of the presumption of inducement where a representation has been made fraudulently, i.e. where the representor (or party responsible for a representation made through another) intended the representation to induce the claimant to act and knew the representation to be false (or was reckless as to its falsity, not caring whether it was true or false), it has been said that:

The existence and strength of the presumption explains why it has long been recognised that there is no rule of law that the relevant decision-maker(s) must be called to give evidence and swear that they acted upon the inducement: see Smith v Chadwick (1889) 9 App Cas 187 (HL), in which, at 196, Lord Blackburn explained (again generalising away from the particular case of inducing entry into a contract) that “… if it is proved that the defendants with a view to induce the plaintiff to [act] made a statement to the plaintiff of such a nature as would be likely to induce a person to [act in that way], and it is proved that the plaintiff did [so act], it is a fair inference of fact that he was induced to do so by the statement.” The present case does not give rise to any issue about how close a match there must be between the (type of) action likely to be induced and the action in fact taken by the claimant. The case concerns the approval by SKAT, and therefore payment, of the submitted tax refund claim in each instance; and if representations were made with intent to induce SKAT to act, obviously that was the intended outcome.

Powerful though the presumption is, it remains a factual presumption only, not reversing the burden of proof, and the evidence in any given case may paint a sufficiently clear picture of how a decision was made, or why some action was taken (or not taken), that the finding is that a misrepresentation, though material, and even if fraudulently made, in fact had no influence, and the same decision would have been made, the same action would have been taken (or not taken), without it.

That is the position, and my finding, in the present case, for the reasons set out in the remainder of this part of the judgment.

Before turning to the facts, I should mention one other aspect of the law, namely that it recognises the possibility of mechanistic or automatic reliance that may be sufficient, so that there is inducement, although the making of the individual decision said to have been induced does not involve thought being applied by any human mind to what is or is not being conveyed by the words and/or conduct constituting or giving rise to the representation. In that regard, SKAT referred me to: Renault v Fleetpro (Renault UK Ltd v Fleetpro Technical Services [2007] EWHC 2541 (QB), noted by Clerk & Lindsell at f.n.176 in the passage quoted at paragraph 425 above; Grant & Mumford, “Civil Fraud: Law, Practice & Procedure”, 1^st Supplement (2022) to the 1^st Edition (2018), at 1-133A; de Verneuil Smith & Day, “Reliance: a comparison between the common law and s90A FSMA” (2021) 6 JIBFL 389 at 291; and Day, “Recent travails of fraudulent misrepresentation” [2021] LMCLQ 636 at 644.

Those references all consider the possible conundrum of machine (computer) processing, and conclude (rightly, in my view) that there is no real conundrum, so that, for example and as SKAT submitted, if within an online application a box is ticked such that (objectively speaking) a representation is made by the applicant, and the system is coded to reject any application in which that box is not ticked, it would only be “an orthodox view of the law” (per Day, supra) to conclude that the representation induced the (computerised) acceptance of the application. The point of principle (if rightly described as such) is that an output generated by the routine operation of a system that produces such outputs in response to inputs obtained from another may properly be held to have been induced by a representation by that other, made by the provision of those inputs with a view to obtaining the output, if the system was designed to deliver that output from those inputs because the view was taken that the provider of the inputs would make that representation by providing them. The system design, on the basis of that view, can be seen as anticipatory reliance on representations that will be made, which is completed to create an individual instance of actual reliance when a particular application is routinely processed.

Furthermore, if documents required to be submitted for a payment claim to be processed by the party to whom it is made will necessarily make a representation material to a decision whether to pay the claim (interpreting the purport of the documents objectively, in context), and evidence shows that individual claims are processed more or less unthinkingly through some automated system, it may be no more than an orthodox application of the presumption of inducement in the particular case to say that the representation is likely to have induced the claim payment. The inevitability that a material representation will be made may render it inherently likely that the system was designed to respond as it does on the view that the representation would be so made.

The logic of paragraphs 531 to 533 above is not confined to machine processing, in my view. Suppose in the example posed by SKAT, the application checking was not automated but was nonetheless a mechanistic clerical exercise that did not require the checker to apply their mind to (the meaning of) the contents of the completed form, but rather required them just to verify that a certain set of boxes was ticked. If the application would have been rejected if the box had been left unticked that, by being ticked, created the representation (speaking objectively), again it would be orthodox to say that the representation induced the (clerical, mechanistic) acceptance of the application. Moreover, that would be so, other things being equal, even if the checker was called to give evidence and said, as might be expected, that they gave no thought to what the completed form said. Indeed, that would still be so, other things being equal, if the checker gave evidence that they did give thought to that and, having done so, understood the completed form to be saying something that bore no relation to the representation objectively made by it, perhaps even that it was saying something quite inconsistent with the representation objectively made.

That is all so because, in the adapted example: (i) the checker’s understanding, if they had any, of what was conveyed by the completed form is irrelevant to the approval of the application; and (ii) other things in the case being equal would result in the drawing of a fair inference that those who designed the process made the ticking of the box in question a pre-requisite because by ticking it the applicant would be stating that which, objectively speaking, ticking it would be taken to represent. In the nature of such a presumptive inference as to the facts, other evidence in the case might rebut it. For example, suppose, in the most extreme version of the example, the checker’s evidence was that their (objectively wild) understanding of what the completed form conveyed came from what they had been told by a more senior person, who set the system up, that the online form was designed to elicit from the applicant.

I therefore do not accept submissions that were made by defendants that there can only be systemic reliance (to give it a convenient label) of the type found in Renault v Fleetpro, supra, in the context of machine processing or decision-making. For example, the DWF Ds submitted that the representations alleged here “are not wholly express, simple or clear and are not the result of specific questions asked. They require interpretation by a human mind to discern their meaning alleged to be implicit in the words used and which on any realistic view are not the only potential interpretation of the (limited) words used.” I agree with all of that. It followed, the DWF Ds submitted, that Renault v Fleetpro and the commentators’ views on this point were irrelevant, incapable of application. I do not agree with that. All must depend on the facts; but I see no difficulty in principle in a finding of inducement based on, say, evidence that the view was taken by those with authority over the matter that documents such as those routinely submitted did involve pertinent statements being made, which they regarded as required to justify a payment being made, and that is why a clerical process involving a document checker who did not have to think about such things was adopted and regarded as acceptable, or (in a suitable case) based ultimately on a conclusion that a presumption of inducement was not rebutted.

That is also why, in particular, it does not without more defeat a case of systemic reliance, as the DWF Ds argued, if the document checker does in fact read the documents and make something, in their own mind, of what they read. If that had a bearing on whether they processed the documents as approved, that would drive the court away from the possibility of systemic reliance and the focus would be on whether the way in which their interpretation of the material affected their actions established that they had been induced by an alleged misrepresentation; but if it had no bearing on their relevant conduct, then causally speaking that internal consideration of what the documents were saying may as well not have happened and, as it seems to me in principle, systemic reliance would be capable of being found if justified by the evidence taken in the round.

The sufficiency in law of systemic reliance mattered in the present case because, on the evidence of Mr Nielsen and Ms Rømer, I find that Mr Nielsen was tasked with and performed a mechanistic clerical task of the kind I have just discussed. He was billed by SKAT’s pleadings, and by the witness statement he adopted as his evidence in chief, as a man whose job it was to consider what, if anything, was conveyed by the tax reclaim documents he processed, as regards the factual basis upon which tax refund claims were being made, who performed that job, and who concluded that the documents conveyed that the Tax Agent’s clients had been shareholders for Danish tax purposes on the dividend declaration date who by virtue thereof had received dividends for the purposes of Danish tax law, and a payment net of tax in respect of that entitlement, and that the payments by reference to which claims were made had been received by the Tax Agent’s clients net of tax withheld by the respective Danish companies, in the sense contended for by SKAT in its pleading of the tax representation. In my judgment, that bears no relation to the reality of Mr Nielsen’s role or approach.

Before dealing further with the facts, I now summarise how SKAT’s case as to inducement was pleaded. References in what follows are to the Particulars of Claim unless otherwise indicated.

At paragraph 20, it was alleged that “… SKAT in fact understood … the [core representations] to have been made, given: (i) the terms of the … Form completed by the Agents …; (ii) the terms of the covering letter provided by the Agents …; and (iii) the terms of the [CAN(s)] provided with the … Form …; and (iv) the fact that such documents were being submitted to SKAT as part of [an application] seeking to reclaim WHT deducted by a Danish company in respect of a dividend purportedly received by the [named client].” Paragraph 22 added (so far as material) a plea that SKAT in fact understood the honest custodian representation to have been made in each instance, given the terms of the CAN(s) and the fact that the CAN(s) “were being produced for the purpose of being submitted to SKAT as part of [an application] seeking to reclaim WHT deducted by a Danish company from a dividend purportedly received by the [named client] on shares owned by it.”

In respect of both of those pleas, in the Further Information to which I referred when considering whether, objectively, the representations were made, SKAT pleaded thus, as to whether its case was that any natural person at SKAT “was actively aware” of the representations alleged: “It is SKAT’s case that Mr Sven Nielsen was actively aware of the representations at the material times.” In a different set of Further Information Responses, in answer to a Request served by the SSDs, SKAT said its case was that Mr Nielsen at the time had a conscious understanding that the CANs he dealt with conveyed what SKAT alleged that they conveyed (by the preparatory plea quoted in paragraph 456 above). The RFIs anticipated argument at trial over whether liability for deceit requires contemporary conscious awareness on the part of the claimant of the allegedly deceitful misrepresentation; so they did not ask in simpler, more traditional terms, by whom (which individual(s)) SKAT alleged that it “in fact understood” at the time that each representation was being made. Nonetheless, the Responses left no doubt that SKAT’s case that it in fact understood the representations to be being made was confined to a case that Mr Nielsen had that understanding, and the trial was conducted on that basis.

At paragraph 25, it was alleged that, “In reliance on the [representations alleged] (and acting under a mistake of fact induced thereby), SKAT paid DKK12.091 billion, which it would not otherwise have done”, with brief particulars breaking the figure down into the aggregate amounts respectively paid to each of the four Tax Agents (DKK4.282bn to Goal, DKK3.043bn to Syntax, DKK1.22bn to Koi, and DKK3.544bn to Acupay). At paragraph 63, it was said that the fraudulent misrepresentations made to it (as SKAT alleged) “were material and were likely to induce, and in fact induced, SKAT to pay DKK9.025 billion … in respect of the Solo WHT Scheme, DKK2.745 billion … in respect of the Maple Point WHT Scheme, and DKK321 million … in respect of the Klar WHT Scheme, which it would not otherwise have done. Paragraph 25 above is repeated.”

SKAT was asked by the DWF Ds to clarify the plea at paragraph 63 by giving “full particulars of the allegation that [the alleged] misrepresentations [and each of them] [(i)] were “material”; and [(ii)] “in fact induced” SKAT to make the payments alleged.” In response, SKAT pleaded, respectively, that:

The case opened by SKAT at the Main Trial was that:

SKAT thus opened four alternative cases:

I reject the first and second alternative cases because in my judgment Mr Nielsen’s job was indeed, in relevant respect, mechanistic and unthinking.

So far as Ms Rømer was concerned, if there was a correctly completed Form and the documentation it asked for was provided, Accounting II’s function, and therefore Mr Nielsen’s job, was to process the claim as approved, meaning it would be paid by SKAT. It follows, and this was in any event the clear tenor of her and Mr Nielsen’s oral evidence overall, that it was no part of Accounting II’s function, or of Mr Nielsen’s job, to have an opinion about, or think about, what (if anything) a properly completed Form accompanied by a CAN said about the factual or legal circumstances of the claim being processed. SKAT submitted that Mr Nielsen was “a thoughtful man who did his job with care [and who] understood the essential requirements for a valid refund and … relied on the Core Representations in processing the WHT Applications.” In my view, it was not an exaggeration for Mr Jones KC to submit, as he did during closing argument on Day 80, that this made it appear that SKAT’s representatives had attended a different trial. I respectfully agree with Mr Jones, continuing, that:

“… Mr Nielsen was, with respect to him, a man who operated at the most basic level of document processing, considering up to 80 applications a day or thereabouts, based on out-of-date training which had occurred 30 years earlier, whose dedication to his employer was such as to lead him to steal from that employer over a lengthy period of time, and fail to cooperate with that employer when they wanted to investigate how he had done his job.”

(I would add, though, that Mr Nielsen’s involvement in ‘stealing’ from SKAT was his participation in a fraud against SKAT unrelated to cum-ex trading or any of the parties involved in it, and was only peripherally relevant as possibly going to credit.)

I summarise the management structure of SKAT at paragraphs 568 to 571 below. As a result of it, Ms Rømer had no effective supervision from any of her superiors in relation to dividend tax. For his part, in practice Mr Nielsen had no real supervision even from Ms Rømer. The retirements of Ms Rømer and Mr Cramer made no difference to Mr Nielsen or his work, as he had been doing it alone for a long time. He never met Ms Held and had no need to contact Ms Madsen in the course of his work. (Mr Cramer, Ms Held and Ms Madsen are all mentioned below.)

Mr Nielsen joined SKAT in about October 1970, after four years in shipping, without academic, professional or vocational training or background in finance, law or tax. He began dealing with dividend tax reclaims in around 1985-1986 with only very basic training, given informally by his then manager, Mr Boding, from whom he learned on the job by being told what he had to do to process a claim. Mr Boding gave Mr Nielsen a rudimentary notion that an applicant was supposed to have owned shares in the relevant company, received a dividend and paid too much tax. However, he was given no instruction as to what share ownership meant, either at all or as a matter of Danish tax law, on what was required to establish share ownership, or on what it meant to say that a dividend had been received. If it had been any part of the job to consider and assess, as a matter of fact, general law and/or tax law, the conformity of dividend tax refund claims against criteria for eligibility, Mr Nielsen would have needed proper training in those matters. There is no reason to suppose he would not then have been given it. He was not given it, I infer, because it was not part of the job to consider such matters.

As a result, Mr Nielsen had no knowledge of equity lending, short selling, omnibus accounting, the existence or possible impact of trading around a dividend date or of different settlement periods, or the possibility that a payment credited by a custodian in respect of a dividend might not involve the custodian passing on cash received (indirectly) from the company declaring the dividend. Whether such things might occur was, for Mr Nielsen, a technical matter that it was not necessary for him to know about or understand to carry out his job. He had no notion even that what mattered for tax purposes was a right to a dividend and not a payment.

Mr Nielsen’s witness statement asserted that, although he was not a lawyer, he was aware of the requirements which governed whether or not a claimant was entitled to a refund of dividend tax. His oral evidence confirmed that he had in truth no real notion at all about those requirements, nor needed one to do his job. He was not even aware that SKAT published a legal guide on its website, and had never read it.

By the relevant period, and for many years before, Mr Nielsen was processing dividend tax reclaims alone and with minimal (and no effective) supervision. Two new employees or trainees, Ms Bidstrup and Ms Fridberg, processed a limited number of tax reclaims under Mr Nielsen’s supervision, to gain a general awareness of the work; so it is impossible to say whether Mr Nielsen alone processed every one of the 4,170 reclaims with which I am concerned. But that might have been the case in fact, and anyway it might as well have been the case since he certainly processed personally well over 90% of the dividend tax refund claims received by SKAT under the Form Scheme during the relevant period, and Ms Bidstrup and Ms Fridberg will only have been replicating Mr Nielsen’s process if they did deal with any of the reclaims that are now before me.

The volume and value of tax reclaims submitted to SKAT under the Form Scheme, and therefore processed by Mr Nielsen (or possibly in a few cases under his supervision), grew hugely during the relevant period:

Assuming (say) 46 working weeks (230 days), processing 6,000-7,000 claims would mean averaging c.25-30 a day; 10,000 would be 44 per day on average (and indeed Mr Nielsen agreed in evidence that by 2012 he was processing 40-50 reclaims a day); 16,000 would be an average of 70 per day (and again, to the best of his recollection, Mr Nielsen said he thought that 70-80 reclaims a day by 2015 was about right).

That rate of processing was possible because the task was purely clerical. Accounting II was operating a simple bookkeeping, data recording function, not any kind of claims assessment or claims control function. If a function of that kind was carried out, it had to have been elsewhere within SKAT, and in truth SKAT adduced no evidence that it was being carried out at all during the relevant period. My conclusion, below, when considering SKAT’s alternative case of systemic reliance is that, at the time, SKAT was simply not enquiring whether what was reported by claimants as net dividend amounts received by them was connected in any way to any tax withheld by the Danish company in question.

Mr Nielsen’s introduction to the job by Mr Boding in the mid-1980s left him with a simple notion that dividends are paid to shareholders and therefore only a shareholder would receive a document like a CAN. But that was not relevant to the job he performed at SKAT. It had no influence on whether, why or how he processed claims, so that they came to be paid by SKAT. Furthermore, I am satisfied that he did not think at the time, as the witness statement he adopted as his evidence in chief claimed, that a CAN said anything about share ownership, or about the basis upon which the dividend-related payment reported by it had been made to the client to whom it was addressed.

Mr Nielsen gave evidence in cross-examination about what he would have understood example CANs to convey. Some of it was not even of superficial assistance to SKAT. For example, in one answer he said of an ex-date stated in a CAN, “We didn’t use that information. What was important to us was the year because that was the relevant tax year” (by which he meant the financial year of the Danish company in respect of which the referenced dividend was declared). Or again, he said of the example NCB CAN reproduced in Appendix 5, below, that he would not have understood it to be conveying any information about ownership of shares (even though in fact the NCB form of CAN did say something about that, albeit it did not make the tax ownership representation alleged by SKAT).

Other answers Mr Nielsen gave were more superficially helpful to SKAT. For example, he said of that NCB CAN that he would have understood it to be conveying information about the withholding of tax from the dividend the subject of the CAN: “What I see here is that the amount has been distributed as dividends to Grace Bay and that 27% has been withheld in tax.” On that point, Mr Nielsen’s evidence was less clear and possibly less seemingly helpful to SKAT in relation to other forms of CAN, although there was also more than a touch of general confusion on his part about it. For example, he agreed of a Lindisfarne CAN that he did not understand it to be talking about any payment so far as Danish tax law was concerned, and said of an Indigo CAN that he saw it no differently, but then also gave these answers about that Indigo CAN:

“Q: You didn’t understand this one to be making any statement to you about the deduction of tax from this particular credit?

A: That appears from the numbers.

Q: But again you don’t understand this to be making any statement about the Danish company withholding the tax described in this particular advice?

A: It doesn’t appear that tax has been withheld, but at some point in time the custodian has received dividends and has -- has received the net dividends.

Q: You are not suggesting – sorry, you did not understand at the time that this credit advice showed the receipt of a dividend from the company itself?

A: We know which company distributed dividends but of course they did not distribute dividends directly to DWM Pension Plan [the client named in the CAN he was being shown].

Q: … to clarify … an answer you gave earlier … when we [were] looking at [an SCP CAN] …:

“Question: So you did not understand this dividend credit advice to be making statements to SKAT about the withholding of tax or the payment of dividend, did you?”

And your answer was:

“Answer: He received it from his custodian and tax had been withheld.”

… It is right, isn’t it, that you didn’t understand the [SCP] credit advice to be making statements about the withholding of tax from the credit identified in the statement?

A: If it appears from the CAN the tax has been withheld, we had to trust that.

Q: You didn’t understand the [SCP] credit advice to be talking about the payment or receipt of the dividend, did you?

A: As I remember it, it shows the distributing company and what has been withheld in taxes.”

My assessment was that, for the most part when giving evidence about all of this, Mr Nielsen was engaging in a mental exercise he did not undertake at the time, and that it is not realistic to try to disentangle from within his various answers how much, if any, of them might be recollection of contemporaneous thought. Most pertinently, I am clear that that exercise, if or to whatever extent undertaken by him at the time, was irrelevant to Mr Nielsen’s task during the relevant period. If the correct Form had been used (and even that has to be qualified by the continued acceptance of a misuse of the outdated Form), and there was a document calling itself a ‘dividend credit advice’ or similar recording that the applicant (or its client if the applicant was a Tax Agent) had been credited with a payment referable to a dividend, being a dividend from which tax would have been withheld by the company, Mr Nielsen’s job was simply to process the claim by checking the consistency of the figures and that there was evidence of the tax favoured status of the payee in a jurisdiction covered by a Danish DTT (and a power of attorney, if the Form had been submitted by a Tax Agent).

That is why the Tax Agent’s cover letters are irrelevant. It was not part of Mr Nielsen’s task to consider their content, so he did not do so; and it was not part of his task to consider their content because it was not part of his task to give thought to what the tax reclaim documents submitted to SKAT did or did not communicate. His was a clerical task in a bookkeeping department whose functions and responsibilities did not extend to deciding, or even knowing, the criteria upon which SKAT considered that tax refund claims should be paid, or assessing whether those criteria were met, claim by claim, or at all. If asked at the time, he would have said he had a general understanding that shareholders received dividend payments and it they were tax favoured foreign shareholders they might be entitled to a partial or total refund based on a DTT; but that was irrelevant to, and played no part in, the job he was charged with doing and did. If he had had a more sophisticated, or just different, or no, understanding about those matters, his job and the outcome on the facts of the case would have been the same.

I therefore reject the claim that Mr Nielsen relied on any of the representations alleged by SKAT to have been made by the documents submitted to it.

Turning then to SKAT’s final alternative case of systemic reliance, the first thing to say is that the cover letters remain irrelevant. SKAT’s system did not require a cover letter at all, let alone a cover letter with some specified content about which a view might have been taken that if that content was present, some material representation would result. SKAT recognised this in closing, as the argument on this last alternative was put exclusively on the basis of a systemic requirement of Form plus CAN.

SKAT opened no factual case and led no evidence on the design of the Form Scheme generally, or as to the information it was intended or thought that a completed Form and CAN would convey in particular. SKAT’s positive case was essentially an optimistic submission that the presumption of inducement would suffice, on an argument of the type identified in paragraph 533 above: Form plus CAN was required; Form plus CAN could not fail to make the representations alleged (considering the matter objectively); therefore the court should conclude that the system was “designed to ensure that a successful application made the Core Representations, which corresponded to the essential elements of a valid WHT refund application under Danish tax law. Indeed, it is hard to see that less could have been represented to SKAT by the required documents filed in support of a claim for the refund of withheld Danish dividend tax. Further, the requirement of a DCA from the relevant custodian was designed to ensure that applications were made honestly and therefore necessarily made the [honest custodian representation] to SKAT.” It is no coincidence that SKAT cited no evidence in support of those contentions about system design. It was not asking for a finding of fact derived from witness testimony or documentary evidence that SKAT had in fact designed the system with those intentions or understandings as to what it was doing. SKAT was in substance contending that the presumption arose, and resting on that.

As a result, defendants were left to make the running as best they could from the documentary evidence in the case, cross-examination of SKAT’s witnesses (none of whom could speak to system design first hand), and expert evidence. In the event, that running was done mostly by the Shah Ds and the DWF Ds, especially the latter.

The DWF Ds submitted that in a governmental authority of SKAT’s size and status, there must have been senior individuals who could have provided the court with full and informed explanations of SKAT’s policy decisions in the design and operation of the Danish dividend WHT and tax refund claim system during the relevant period. I agree. The DWF Ds said that SKAT’s failure to provide such evidence in a case of such large dimensions was disrespectful to the court. That is not a characterisation I would adopt. I am not exercising any public law jurisdiction (not that I would have any in relation to SKAT). SKAT owed the court no duty of candour in respect of the evidence it chose to lead. However, SKAT’s failure to call evidence it must have been in a position to obtain, if so advised, properly weighs against it in judging whether on the evidence as a whole inducement is not established, even having had due regard to the existence and strength of the presumption of inducement weighing in SKAT’s favour.

Ms Rømer said in her evidence that Form 06.003 was created by SKAT’s Legal Department and approved by Head Office, but I make no finding to that effect. I accept it as a statement of what Ms Rømer thinks likely to be true, but I do not think she knows now, or knew at the time, anything more than that the design and approval of the Form, likewise any policy decisions involved (explicit or implicit) were not matters for Accounting II, let alone for her personally.

Ms Rømer also said (both at this trial and in a deposition for the proceedings before Judge Kaplan in New York) that it was a Danish government priority to encourage foreign investors to invest in Denmark, and the tax system had been made as simple and user-friendly as possible to serve that purpose. Ms Rømer extended that thought, when asked about SKAT’s failure to supplement its process by requiring information about underlying trading, by saying that “it might be that the government chooses not to introduce matters or to change matters that might prove complex.” That was speculation, an example of Ms Rømer’s willingness to say things she imagined might be true, thinking that to be helpful. It seems to me a plausible possibility, but it was not evidence of fact (except in so far as it evidenced indirectly Ms Rømer’s perspective at the time, corroborating her and Mr Nielsen’s evidence generally that the processing and approval of tax refund claims by Accounting II did not involve consideration of whether they were valid claims as a matter of fact and Danish tax law).

Before 2002, dividend tax had been dealt with in a regional office of SKAT. In 2002, various departments merged to form a new Dividend Tax Department, led by Ms Rømer, which was the first time she had any involvement with dividend tax. The scope of work and the reporting lines were continuous thereafter, although in 2009 the Dividend Tax Department became part of the Accounting Department under the name Accounting II. Ms Rømer continued as head until she retired in December 2013. She was succeeded by Dorthe Pannerup Madsen (or possibly first by Hanne Held and soon thereafter by Ms Madsen, which was Mr Nielsen’s recollection). I did not have any evidence from Ms Madsen or Ms Held, or from Laurits Cramer, who was immediately senior to Mr Nielsen, reporting to Ms Rømer. Mr Cramer worked with Mr Nielsen in the development of SKAT’s 3S data system, into which Mr Nielsen entered details of the tax refund claims he approved, but he was not involved in the processing of individual claims.

The head of the Accounting Department was Lars Nørding, then René Frahm Jørgensen (from the end of 2012), from neither of whom I had any evidence. Ms Rømer told the court she had little contact with either of them.

The Accounting Department was one of four departments that in turn reported to SKAT’s director for Central Jutland, Mr Sørensen. He was part of a senior level of management, referred to informally within SKAT as “Head Office” (or strictly, I assume, some Danish equivalent), which sat above what Ms Rømer described as the day-to-day “production” level of “departments dealing with the work”, like Accounting II, but still below the top tier of management within SKAT, the “Board”, above which again sat the “Department”, i.e. the Danish Ministry of Taxation. Until 2012 or 2013, Mr Sørensen reported to Board directors Ole Kjær and Steffen Normann Hansen (Mr Kjær being the “top director” responsible for SKAT as a whole), neither of whom gave evidence. From 2012 or 2013 onwards he reported to a new “top director” Jesper Rønnow Simonsen, who did not give evidence.

Mr Sørensen had remarkably little knowledge of the work of Accounting II or the administration of dividend tax. He had no contact with Mr Nielsen and no knowledge of what he did, or of the tax reclaim Form or how it was prepared or processed. He never discussed dividend tax even with Ms Rømer, although she had meetings with him. He had no responsibility for dividend tax matters prior to 2009, and thereafter nothing was reported to him about it (aside from periodic internal accounting reports), and he had no involvement with it, prior to the summer of 2015. From 2013 he was almost wholly focussed on developing an IT system within SKAT that had nothing to do with dividend tax; and dividend tax was simply not a priority of his.

When considering the context in which SKAT’s dividend tax reclaim system was set, I identified in paragraphs 450 to 454 above various matters that SKAT would reasonably have been expected to appreciate. Even though SKAT did not provide evidence from the senior individuals who must exist who could have spoken directly to this, I was satisfied from the documents and Ms Rømer’s evidence (where she was able to assist) that SKAT was aware of all of those matters during the relevant period.

SKAT knew share trading could be complex and might affect tax refund entitlements. It knew about stock lending, short selling and deferred settlement as widespread practices. SKAT knew that it could require those seeking refunds to provide information about their trading, or documentary evidence of it, if it wanted to assess their entitlement, and that it did not do so under the Form Scheme. That was not just an audit weakness of failing to interrogate or check, or seek better evidence of, information provided by the documents required by and routinely submitted under the Form Scheme. It extended to an appreciation by SKAT that what it required, and received, under that Scheme, did not give it information that, if accurate, established entitlement. SKAT was aware, in short, that a completed Form 06.003 accompanied by a CAN did not involve the Tax Agent stating to SKAT that the client had been a shareholder as a matter of Danish tax law (or at all) on the dividend declaration date (or on any other date), or that the client had, for Danish tax law purposes, received a dividend by being the tax law shareholder on the dividend declaration date (or a payment, net of tax, in respect of such an entitlement), or that the Danish company had withheld tax from a payment received by the client (in the sense relevant to SKAT’s claims (paragraphs 509 to 512 above)).

In closing argument, it was submitted that SKAT adopted a general trust-based approach, believing that those who claimed to be taxpayers, or to be acting on behalf of taxpayers, would “do the right thing”. There was some support for that in SKAT’s witness evidence. Mr Sørensen said in his witness statement that SKAT’s approach, in general, was “to inform the taxpayer about how to do things right, and then trust the taxpayers to do things right.” Ms Rømer said in hers that SKAT “trusted the applicants, as verified by the banks” (by which she meant the “bank or institution that had provided the dividend advice”). In a deceit claim, generalised ideas like those are no substitute for reliance on the particular representations alleged in the claim. (It was not alleged that there was any implied, general representation of honesty.)

In any event, I do not consider that witness evidence to be a reliable guide to SKAT’s approach. In the relevant passage of her witness statement, Ms Rømer said that at the time her approach was that: “If the bank confirmed that the amount claimed was all true and accurate, we were instructed by Head Office to process the payment to the applicant. If we thought the bank might have made a mistake, we could ask the bank that had provided the dividend advice about it; if the bank said that the information was correct, we accepted that the application was valid and had to pay. … If the applicant did not own shares or did not receive a dividend, they were not entitled to a refund of withheld dividend tax and should not have been applying to SKAT for a refund in the first place. If the form and the dividend advice said that the applicant owned the shares, had received the dividend and was certified as resident in a country that had a DTA giving the applicant the right to a reduced rate of tax, then we trusted those documents and took the view that the applicant was entitled to a refund of withheld dividend tax.” (my emphasis). In fact, as I find, Accounting II (and Ms Rømer personally) did not have that approach. The Form Scheme did not call for, and SKAT routinely did not receive, any confirmation from a bank or other financial institution of information that, if accurate, would establish any tax refund entitlement. There was no instruction from Head Office to the effect suggested by Ms Rømer. It was no part of Accounting II’s function to consider whether a Form and CAN, as received by SKAT, said that the applicant (strictly, the applicant’s client, where the applicant was a Tax Agent) owned shares and had received the dividend, and Mr Nielsen did not do so, not that a statement that the client “owned the shares [and] received the dividend” would translate to the representations alleged by SKAT anyway.

Mr Sørensen had no knowledge of SKAT’s approach to dividend tax refund claims (and should not have suggested otherwise, as he did in his witness statement). He may have been senior enough to be able to speak to a general approach to things at SKAT, expressed at the high level of abstraction of his comment quoted in paragraph 574 above. If so, particularly when read with the sentence from Ms Rømer emphasised in paragraph 575 above, that only revealed that SKAT’s claims pursued at trial lacked substance. If SKAT’s approach was to assume that a Tax Agent’s client was entitled to a refund, because otherwise the Tax Agent would not be putting in a claim, that might have assisted if the claim was of reliance on a representation, express or implied, of entitlement; but no representation to that effect was alleged by SKAT.

Furthermore, on the evidence as a whole, in my judgment it was clear that SKAT was not operating a dividend tax reclaim process on a trust-based approach like that suggested by Ms Rømer and Mr Sørensen. Rather, it was operating a system well knowing that it did not involve the provision to SKAT, in any form, of evidence that, taken at face value, established entitlement; and appreciating as a result that it was very likely accepting and paying claims where there was no entitlement, to an extent and value it was not in a position to assess. That is why I said in paragraph 573 above that this was not just an audit weakness. It was a more fundamental tax control weakness that was apt to be exploited, and came to be exploited at great cost to SKAT during the relevant period.

Those conclusions are, in particular, the only sensible way to explain the striking matters established by the evidence summarised in the remainder of this section of the judgment.

The sheer volume, and value, of claims approved and paid by SKAT during the relevant period is remarkable. I summarised some of the available data on that in paragraphs 553 to 555 above. That summary is already striking, and telling as to the nature of Mr Nielsen’s task. But in addition:

There is no evidence that SKAT monitored what was happening at Accounting II such that features and trends of that kind were identified. The point is not that SKAT in fact identified them prior to September 2015 (I could not find that it did), nor is it a question of criticising SKAT for failing to do so, since it is no defence to a deceit claim to show that the victim should have spotted the fraud, if there was one. Rather the point is a specific factual one for this case. Those extraordinary outcomes are explicable only on the basis that SKAT’s approval and payment of claims simply did not have reference to, and so was not influenced at all by, the matters about which it claimed in this litigation that the tax reclaims made representations. In my judgment it is impossible that SKAT’s processes could have resulted in those outcomes if those processes were looking for tax ownership of shares, tax entitlement to dividends, or a transmission of the burden of the tax deduction at source, before paying out.

SKAT was well aware during the relevant period that ‘dividend’ advices were issued, in form and content indistinguishable inter se, to parties who might or might not have been shareholders (either at all, or from the perspective of Danish tax law in particular).

Telling evidence on that came from SKAT’s disclosure concerning an episode in 2006-2007 in which Ms Rømer had an involvement, so that she was able to assist the court in relation to it in cross-examination although she had not been asked by SKAT to say anything about it in chief. Bankers Trust Opera Trading SA (‘BT Opera’), a French entity in some way associated with Deutsche Bank (London), sought and ultimately obtained from SKAT a dividend tax refund relating to dividends declared by TDC (at the time still called TeleDenmark). In the course of dealing with the BT Opera claim, SKAT recognised and discussed at senior levels:

SKAT did not redesign or tighten up its system, for example by amending the Form to require statements to be made, or evidence to be provided, about share ownership, dividend entitlement, or related trading. In fact, it did the opposite, simplifying from Form 06.008 to Form 06.003, dropping the requirement for the Danish company to certify tax deduction, and stipulating nothing about the material to be provided (so far as might be pertinent) except that there be a ‘dividend advice’. The BT Opera episode, and other evidence, showed that SKAT appreciated when doing so that the ‘Form + CAN’ document set it would thus normally receive would not make statements to it of the kind alleged in the litigation to have been representations always made.

For present purposes, the BT Opera episode begins with an internal memo dated 29 September 2006 prepared by Ms Rømer. On BT Opera’s behalf, Deutsche Bank had claimed just over DKK405m from SKAT in connection with TDC’s dividend of c.DKK2.576bn on which TDC had withheld c.DKK721m by way of dividend tax (the WHT rate was 28% at the time). Ms Rømer had noted that this meant BT Opera was seeking payment equal to more than 50% of the total tax due on the dividend. She noted that given the lack of visibility over foreign shareholders, and the use of cross-border share lending and nominee accounts, SKAT could not say from the information provided whether the refund, if paid, would be going to the dividend recipient, and it could be that it would “thereby secure a refund that is not really in accordance with the [DTT] and the legislation”. Her initial proposal was to pay the claim anyway, in view of the then requirement on SKAT to pay within 30 days (rather than 6 months as during the relevant period), “but that a similar situation be prevented by making changes to the relevant regulatory framework”.

Ms Rømer’s concerns were escalated, and on 17 October 2006, Leif Jeppesen, a Legal Services Director at Head Office, passed her memo to colleagues in an email expressing concern about speculating on Danish shares by moving them between jurisdictions for short periods, for example by a purchase and next-day sale, around the dividend date, so as to use the France-Denmark DTT to obtain a refund. He considered it possible that there was trading that would be considered pro forma under Danish law: see SKAT (Validity Issues), supra, at [83], [225]. I could not say that Mr Jeppesen had in mind a doctrine articulated exactly similarly; his later correspondence suggests that he viewed as pro forma a purchase just before the dividend date, with sale back immediately after, and that he was troubled, given the language of Form 06.008 (then current) that perhaps BT Opera had “simply ‘bought’ the dividend right without any connection to the right of ownership”. Ms Rømer rightly took him, on any view, to be concerned that there may have been a structured trade or series of trades intended to bring about a particular consequence, trading around the dividend date in an artificial way simply to generate a tax refund claim.

Mr Jeppesen suggested that underlying documents should be checked; and Ms Rømer agreed that the only way SKAT could learn what the trading had been was to ask for the detail, and so further documents were sought from BT Opera and Danske Bank. Mr Jeppesen noted that according to TDC’s website, Nordic Telephone Co ApS at the time owned 88.2% of TDC’s shares (which obviously rendered it prima facie implausible that BT Opera had a 56% shareholding).

Ms Rømer’s first request for additional documents was in fact very limited. On 30 November 2006, by email she asked Deutsche Bank for “a signed letter from the beneficial owner” (stating what, she did not say), and “the dividend note issued by the paying bank along with the payment of the dividend”. She told Deutsche Bank that would “satisfy the Danish Tax Authorities and refund will be prosecuted”. In cross-examination, Ms Rømer said that was not her decision, it was what she had been told to do by Head Office.

On 20 December 2006, SKAT received a “Month-to-Date Cash Statement” as of 28 April 2006 for an account of BT Opera’s at Deutsche Bank (London), showing credit entries apparently referable to dividends declared by TDC; and on 12 January 2007 SKAT received a declaration of beneficial ownership “of the underlying Danish equities at the time of the entitlement to the relevant dividend” from BT Opera, Deutsche Bank having given a similar confirmation on behalf of BT Opera in its letter enclosing the cash account statement.

On 16 January 2007, Ms Rømer wrote to Mr Jeppesen (among others) proposing that SKAT now could not refuse to pay the claim; but he disagreed, protesting that to follow the transactions SKAT should have asked for “both purchase and sale documents, copies of settlements for purchases and sales and detailed explanations of the terms and conditions”. He noted that the declaration of beneficial ownership said nothing about the basis of the claimed interest (including whether it was under Danish, German or French law), or for how long any interest was held, and considered that varying treatments of stock lending in different jurisdictions might be relevant.

On 18 January 2007 Mr Jeppesen followed up internally by email, attaching a draft early warning notice in relation to stock lending and suggesting disclosure of documents going to beneficial ownership and stock loans. It discussed the tax treatment of stock loans, the possibilities of stock lending without the knowledge of the owner in the custody chain and of multiple holdings of the same shares, and the need to review the form and content of claim forms “so that it is clear what requirements we have for ownership, etc. and documentation for this in order to get a refund of dividend tax”. In reply, on 19 January 2007, Ms Rømer agreed that stock loans were “a serious problem” and said that the BT Opera claim met “the requirements we set out in our forms” and that those forms “have been sought to be changed for several years as they are insufficient”. (On the basis of Ms Rømer’s oral evidence, I find that even prior to the BT Opera episode, SKAT had identified that the declarations made to SKAT by tax refund claimants were inadequate, that there were discussions and working groups, but that no change of process so as to require informative declarations by claimants was ever made. She said her efforts in that regard were not always well received by Head Office or SKAT’s Legal Department, and that even Mr Jeppesen’s suggestions for systematic change were resisted.)

The documentary evidence now available about the BT Opera claim is more limited thereafter. By early July 2007, Mr Jeppesen appears to have relinquished responsibility for the unit dealing with the case; and SKAT eventually received what was described internally within SKAT as “comprehensive documentation” that included “further documentation regarding BT Opera Trading’s purchase of shares”, following which, on 7 August 2007, SKAT decided to pay out. Ms Rømer was unable to help as to what the further documentation showed, or was thought by SKAT at the time to show, that unlocked the payment.

Ms Rømer’s oral evidence also showed that SKAT was aware, during the relevant period, of key matters of market practice in the context of which tax refund claims might be made to it, and that they meant that statements of the type SKAT has wished to say in the litigation were made to it simply were not made by the basic Form 06.003 plus CAN document sets that SKAT received. I accept Ms Rømer’s evidence on those aspects, despite my concerns about her as a witness, because it was all plausible, in the light of the expert evidence, it was adverse to SKAT in the case, and SKAT adduced no contrary evidence from any potentially more reliable source, e.g. contemporaneous documents or witness evidence from better informed, more senior personnel. Thus:

In mid-February 2012, just as Mr Klar was preparing to engage in the earliest Klar Model trading in respect of Danish shares, which came first in time out of all the trading with which I am directly concerned, and as Sanjay Shah, Mr Horn and Rajen Shah were working up the first iteration of Solo Model trading, Ms Zester emailed Ms Rømer, in the context of possible reform of financial reporting, to get her best estimate of the financial impact in relation to dividend tax administration. In Ms Rømer’s absence, Ms Zester asked Mr Cramer to respond, and he replied that ongoing reporting would provide “better opportunities to ensure that refund applications are from dividend recipients who have had dividend tax withheld at a percentage that entitles them to reimbursement. I dare not quantify our losses from reimbursement to unjustified refund applications” (my emphasis).

Mr Nielsen confirmed that it had been Mr Cramer’s view, voiced at SKAT at the time, that indeed SKAT must be paying out unjustified claims to an extent he could not quantify. Ms Rømer said she did not recall the particular exchange of emails or, therefore, that particular example of Mr Cramer expressing concern, but volunteered that “we had been concerned for a long time”, it had been discussed many times whether SKAT could or should seek further documentation or information about underlying trading, and that the decision at a top level (she could not say by whom exactly) had been to wait for what the OECD came up with (which, in the event, was nothing, during the relevant period). In that passage of evidence, Ms Rømer also referred to SKAT standing on the “three legs” of “the declaration, the signature of the tax authorities, and the justification from the bank”. However, that was argumentative. Ms Rømer knew that no relevant ‘declaration’ was made by or on behalf of the claimant, the only tax authority signature received concerned the tax status of the claimant in its home jurisdiction, and a CAN in typical form was not a ‘justification’ of anything. Pressed to agree that she understood at the time that the terms of the underlying trading might mean the claimant was not entitled, Ms Rømer said, “I prefer to think that people are doing what is right, but of course there is a risk that it is not the case”, which was still a touch argumentative but in my judgment contained within it a possible grain of truth, namely that she may perhaps have imagined that refund claims would only be made by those who thought themselves entitled. If that was her view at the time, it had no influence on any of the decisions Mr Nielsen made to approve the tax reclaims with which I am concerned; and it would not have been evidence of reliance on any of the representations alleged by SKAT, even if it had influenced anything.

In my judgment, in the light of the matters of context of which SKAT was well aware, and the BT Opera episode a few years previously, the matters summarised above did not evidence trusting reliance on statements being made to SKAT of the kind from which SKAT sought to construct its claims in this litigation (or anything similar). Rather, it demonstrated or confirmed awareness within SKAT that statements of that kind were not made, that it therefore had no idea whether they would or might be true or false for any given tax reclaim, and that its system, by design and in practice, paid out anyway.

That is illustrated (and further confirmed) for the tax ownership representation in particular by a draft report circulated within SKAT (but not to anyone who gave evidence) at the end of September 2015, on the handling of dividend tax refund claims. It concluded generally that control (as regards whether only valid claims were paid) was flimsy. It suggested that SKAT had no real possibility to check whether the refund claimant was the rightful owner other than by asking that claimant. I do not agree with that, but what matters for present purposes is that the report recommended that refund claimants be required to declare that they were shareholders when the dividend was distributed, in substance on the basis that no such statement was being made to SKAT under the system as it stood.

In closing, SKAT argued that the core representations were so simple as to go without saying, such that it was impossible to make a tax refund claim without making them. It submitted the Form Scheme was “designed to ensure that a successful application made the Core Representations”, so they could not have been missed by SKAT. Defendants submitted to the contrary; by way of example, the DWF Ds suggested that “This ambitious argument is to equate the Core Representations … with examples of very straightforward implied representations mentioned in the authorities, e.g. … bidders at auction or diners in a restaurant who impliedly represent that they have funds to pay. It rests on an unrealistic reading of the authorities and in any event has no sensible application to the present facts.” Save that it is unnecessary to pass comment on whether SKAT was misreading the case-law, I agree with the DWF Ds.

SKAT’s final alternative case on reliance, a case of systemic reliance, therefore also fails on the facts.

That conclusion on systemic reliance is further reinforced, although it would have been my conclusion without this, by the indifference of SKAT’s process, as designed and as then duly operated by Mr Nielsen, to any narrative language in the reclaim documents. It would be very surprising if a system designed to be influenced by statements made by Tax Agents in support of tax refund claims did not require those operating the system to pay close attention to what was said in those Agents’ covering letters, or to notice narrative wording such as in the SCP CANs. I found, above, that the Tax Agents’ cover letters may as well have said nothing. Further than that, however, I was satisfied on the evidence as a whole that it would have made no difference if those letters had said “Please find attached a claim relating to a dividend compensation payment [or manufactured dividend or payment in lieu of dividend] received by our client”, or if a CAN had stated, in terms, that it concerned such a receipt.

Having rejected the suggested case of systemic reliance on the facts, although I took the view that such a case could be sound in principle (paragraphs 531 to 535 above), it is not necessary to deal with an argument developed by Mr Jones KC on behalf of the Shah Ds that it was not fairly open to SKAT to ask for liability to be found on the basis of systemic reliance, because of the way it had pleaded its case on inducement.

The evidence allowed me to find that, and in truth left me in no real doubt but that, “(i) Mr Nielsen processed the WHT Applications as a straightforward clerical matter without regard to the matters alleged to form the content of the Core Representations; and (ii) no-one within SKAT instructed or understood him (or … the documents submitted to SKAT) to be collecting information confirming the content of the alleged Core Representations”, which is how the DWF Ds put it in their written closing submissions.

The examination of SKAT’s claim process afforded by the Main Trial in my judgment demonstrated that SKAT’s approval and payment of the tax refund claims at issue in these proceedings was not influenced in any way by any of the representations that SKAT alleged. That is so even if the presumption of inducement here is the strong presumption, not easily rebutted, that fraudulent misrepresentations are likely to hit their mark, i.e. have an influential impact as intended. I had that presumption well in mind when considering the matters summarised above. I recall it now, separately, as the foundation of SKAT’s primary case, as opened at trial (see paragraph 545 above). The strength of the evidence overcomes by a clear margin the very strong inherent likelihood, other things being equal, that an intended fraud will have had some operative impact.

Mr Jones KC submitted that it was not open to SKAT to fall back on the presumption, having pleaded and particularised as its positive case that the alleged representations were relied on by Mr Nielsen, in the sense that he was consciously aware of them as communicated by the tax reclaim documents and they influenced his decision in each case to process the reclaim so that it would be paid by SKAT. Whether or not an argument of that kind might work in a given case, I do not accept it in this case. SKAT did not admit in its pleadings any requirement in law for there to have been contemporaneous conscious awareness; and it could be coherent for a claimant to fail to establish that some identified individual(s) relied, as alleged by way of primary case or at any rate particularised positive factual case, without the plausible inherent likelihood of reliance being decisively negatived.

On the evidence concerning SKAT’s approach to and understanding of the tax refund claim system it was operating, in practice through the clerical processing efforts of Mr Nielsen, however, the notion that there was reliance on (anything like) the representations alleged by SKAT was, in my judgment, decisively negatived.

I was referred to judicial dicta critical of lawyer-led constructs concerning what it might be said had been represented, for example in Farol Holdings, supra, per Zacaroli J at [409]-[410]:

“Although, chronologically, the pleading comes before witness statements, it is an essential prerequisite of a claim in implied representation … that it is based on the evidence of the relevant representees that they understood the implied representations to have been made to them. … Legal ingenuity might establish that all sorts of other statements are to be implied from the words and conduct of the representor, but unless the representee was led to the same conclusions at the time, a claim in misrepresentation cannot be made out” (Farol Holdings, supra, per Zacaroli J at [409]-[410]).

I do not mean now to open the issue I have not needed to resolve of whether or when there must be contemporaneous conscious awareness that a representation has been or is being made. The presently pertinent sting of observations such as Zacaroli J’s is that a litigation claim that harmful action (or a harmful failure to act) was induced by misrepresentation should be driven by an understanding, and realistic assessment, of why in fact the harmful action was taken (or why action was not taken, such that harm resulted from the failure to act). SKAT’s case indeed had the opposite appearance, as defendants submitted, of a lawyer-led case as to what representations, objectively, it might be said were made, in the absence of anything simple like a representation of entitlement to a refund or a representation that tax liability had been incurred, onto which the fitting of a case as to inducement then had to be attempted.

Mr Nielsen’s witness statement, if accepted uncritically, might have seemed up to that task; but in substance he failed to come up to proof, in the modern fashion that cross-examination showed he should not have given that seemingly supportive evidence by adopting his witness statement as his evidence in chief, rather than in the stricter or old-fashioned sense of a witness failing to say in chief what they had been billed as able to say. I am satisfied that if simply asked to give an account of how he did his job, and why (to the extent he can remember and articulate) he processed all these tax refund claims for payment by SKAT, and if there were no implicit pressure to be helpful to SKAT by his answers, Mr Nielsen could not honestly have given an answer that might have led to a plea that he relied on the representations alleged by SKAT, or anything like them. An absence of reliance by Mr Nielsen would not in itself make the pleaded claims impossible, because in principle there might have been a provable case of systemic reliance, or the presumption of inducement might have done enough for SKAT. In the event, however, as I have said, the facts revealed by the trial process demonstrated the contrary.

That conclusion may have the consequence that SKAT’s dividend tax refund control was, at the time, so flimsy as to be wide open to what might be seen as exploitation, and that, at all events when claim volumes and values were ramped up in 2014, the Solo and Maple Point Models in particular exploited it ruthlessly (especially the fully automated Solo Model in 2015). In fact that is what, in my judgment, Sanjay Shah, for one, thought he was doing at the time. I think his 2021 television soundbite was candid and revealing (paragraph 2 above): “If there’s a big sign on the street saying, “please help yourself”, then me or somebody else would go and help themselves.” I consider that Mr Bains saw it that way too. He accurately assessed it to be Sanjay Shah’s thinking, when he said in December 2014, in his email to a personal counsellor to which I referred in paragraph 227 above, that in his view the reason Sanjay Shah had made him a substantial offer to re-join Solo, an offer that Mr Bains had refused, was that “he [Sanjay] only ever wanted to keep me out of the market so he could bleed dry an innocent Nordic country next year [2015]”.

In my judgment, however, even bearing that also well in mind, it remains the case that on the evidence, there was no need for any of the representations alleged by SKAT to be made for SKAT to pay out on tax refund claims submitted under the Form Scheme, through the flimsy system it had put in place (and as it happens, my conclusion, in substance, is that they were not made, given the irrelevance in practice of the Tax Agents’ cover letters). The evidence made it very clear to me that SKAT knew that representations of that sort were not being made to it, and that it was paying claims anyway. SKAT was not induced to act, at all, by the representations it alleged in these proceedings.

For completeness only in the present context, I would add in relation to the remarks quoted in paragraph 607 above that I do not judge them to evidence a view at the time that SKAT was being deceived into paying Solo Model tax refund claims, i.e. tricked into paying by falsehoods told to it. Neither Mr Shah nor Mr Bains thought that at the time, I conclude, although they both had come to realise that Solo had hit upon a way to take advantage of what had to be a very slack process at SKAT that would be enormously lucrative so long as it continued. Mr Shah was comfortable to play that advantage as long as it lasted; Mr Bains was more conflicted, initially making a play to profit substantially from it at Arunvill, but rendered uncomfortable when Arunvill’s reaction was to think it must amount in some way to fraud, and then feeling that because of the scale to which Mr Shah was taking things at Solo, SKAT should be tipped off to look into what was happening.

Result (except SKAT vs. Syntax)

Syntax is in a different position to all other trial defendants, because in its case liability to SKAT stands established by a default judgment (see paragraph 615ff below).

In relation to all trial defendants other than Syntax, the result of the conclusions set out above is that, in summary, and subject to paragraph 612 below:

That all has to be qualified, however, because of the present non-existence of some of the SSDs. As noted in Appendix 1 below and paragraph 38 above, at the date of this judgment Ampersand, Elysium Properties, Skyfall, Skyfall Holdings, Woodfields, and Woodfields Holdings all stand struck off in their respective jurisdictions of incorporation. That was the position throughout the Main Trial, and indeed since before the Validity Trial in January/February 2023. That means in turn that although I have referred to them as trial defendants, strictly none of them is before the court and I cannot make, and am not making, any decision affecting them. Moreover, the position may be complicated further by the fact that they were struck off after my decision, but before the Court of Appeal’s decision, on the Revenue Rule Trial.

The upshot is that there will now be judgment dismissing all of SKAT’s claims against all trial defendants listed in Appendix 1 other than (i) Syntax, dealt with immediately below, and (ii) the six corporate SSDs listed immediately above. I shall ask for the assistance of counsel for SKAT, and counsel for the SSDs (if they consider it possible and proper to make any submissions), as to what, if anything, the court can or should now do in relation to those six SSDs. I shall also require assistance in relation to defendants who were not trial defendants but who are still parties to the proceedings against whom they have not yet been finally determined (for example defendants, if there are any, against whom stays are in place that will now come to an end or be affected in some other way by the handing down of this judgment).

In that circumstance, I have not lengthened the main body of this judgment by setting out conditional conclusions of fact and law that might have determined, trial defendant by trial defendant, whether any (and if so which) of SKAT’s claims would have succeeded, had SKAT established the major premise for all of them, viz. that in each case it was induced to pay the invalid tax refund claim(s) the subject matter of the claim by one or more of the misrepresentations it alleged. I have done that to a certain extent in Appendix 7, below, principally to identify findings of fact as to trial defendants’ knowledge or understanding at the time, but more briefly than I would have done had the points considered been decisive, in case what I would have found may be said by any party to matter in relation to costs or any other consequential matter that may now arise.

SKAT vs. Syntax

Judgment in default was entered by SKAT against a number of corporate defendants, so that what was left for each of those defendants was only a need to measure damages or determine to what other remedies, if any, SKAT might be entitled under its relevant default judgment. The proceedings against all but one of the default judgment defendants were later stayed, leaving the question of remedies for determination at the Main Trial, and now by this judgment, only against Syntax. For these purposes (see paragraph 30 above and Appendix 1, below), Syntax was the 22^nd defendant to the Second Claim, in which the default judgment was entered against it on 10 May 2019.

I do not know whether Syntax has any assets such that any judgment as to remedies and consequent relief to which SKAT may be entitled will be of any practical value. It is also an unusual case in that the conclusions now reached after trial, on liability issues that would have been common, mean that Syntax would have had a good defence to the claims pleaded against it. However, SKAT stood on its entitlement under a properly entered default judgment, and there was no application on behalf of Syntax seeking the exercise of a discretion to set it aside, nor any appearance for Syntax for any other purpose.

The effect of the default judgment is that liability on the claims pleaded by SKAT against Syntax stands established and my conclusions on liability issues as between SKAT and other trial defendants are irrelevant to Syntax’s position (subject, in both respects, to the complication noted at the end of paragraph 626 below).

SKAT has recovered substantial sums from various parties under settlements or through legal process of one kind or another. It prepared a detailed spreadsheet together with an explanatory ‘Note on Recoveries’ setting out how it offered to give credit for its recoveries to date. SKAT sought thereby to allocate amounts recovered in a logical manner, taking into account the source of the recovery and any particular terms on which the recovery was made that might affect how it should be allocated across the tax refund claims giving rise to the litigation claims.

The vast majority by value of what SKAT has recovered to date has been under settlements with confidentiality terms as to the paying parties, and as to how much each of them would pay. SKAT’s spreadsheet and the Note on Recoveries explained SKAT’s approach to allocation, and presented the detailed figures, in such a way as not to infringe that confidentiality. It was not necessary, for the purpose of judging the fitness for purpose of SKAT’s approach, to require SKAT to reveal more, and no trial defendant sought any such order. In summary, SKAT has recovered to date:

There was no obvious error in SKAT’s approach that might have caused me to reject it in the absence of any active objection to it. I would therefore have relied on SKAT’s analysis and the detailed accounting in the spreadsheet in respect of any trial defendant, had the question of giving credit for recoveries arisen as a result of some claim against that defendant having succeeded. In the event, the point now arises only as part of quantifying relief under the default judgment against Syntax.

In quantifying SKAT’s entitlement to damages, or assessing its entitlement to any other form of relief, under the default judgment, SKAT is entitled to rely on its primary pleaded case as to the causes of action available to it. Its primary pleaded case, on an allegation that Mr Shah took over effective control of Syntax on 19 September 2014, was that Syntax was liable for damages for deceit and/or unlawful means conspiracy in respect of tax refund claims submitted in respect of Solo Model trading on or after that date. In oral closing argument, however, that claim was abandoned, and Ms Nanchahal rested SKAT’s request for quantified relief on the alternative pleaded case of liability for deceit and/or conspiracy in respect of Solo Model tax refund claims submitted by Syntax on or after that date.

Under the default judgment, therefore, the only proof required of SKAT for its common law damages claim was proof of the amount it paid on Solo Model tax refund claims submitted by Syntax on or after 19 September 2014. At trial, SKAT proved that amount to have been DKK2,763,859,045.79.

Credit for recoveries by SKAT to date has to be given against that total. The recoveries spreadsheet gave credit of DKK1,264,918,064.76 for recoveries referable to losses allegedly caused by Syntax. However, that was on the basis of the primary case not pursued in closing of liability for the total paid by SKAT on all Solo Model tax refund claims submitted on or after 19 September 2014, which at trial SKAT proved to have been DKK5,960,889,149.91. A revised version of the spreadsheet now identifies that recoveries allocated against Syntax reclaims, rather than all Solo Model reclaims, submitted on or after 19 September 2014 amount to DKK316,998,798.95. As a result, the loss suffered by SKAT and recoverable as damages from Syntax under the default judgment is DKK2,446,860,246.84, subject to the point in paragraph 625 below.

On the basis of the liability case pleaded, Syntax was unjustly enriched through the payment to it, by SKAT, of Solo Model tax refund claims submitted to it on or after 19 September 2014. At trial, as I have just stated, SKAT proved that amount to have been DKK2,763,859.045.79. The question of apportionment of recoveries already made by SKAT again arises, reducing the claim to DKK2,446,860,246.84.

The question will need to be addressed whether SKAT can ask for judgment to be entered for substantial damages if it asks for judgment for payment by way of restitution in respect of unjust enrichment. In Pisante v Logothetis (no.2) [2022] EWHC 2575 (Comm), for example, I concluded that entering judgment, by way of restitution consequent upon rescission of an investment transaction, that required Libra to pay Swindon the full value of the consideration it had paid for the rescinded transaction, some US$8,991,250, left Libra with a nominal damages liability only for fraud, since Swindon’s only pleaded damages claim was for the loss it suffered by way of paying that consideration and I refused an application to amend. I reasoned that there should be no substantial damages judgment, because “with the dismissal of the amendment application there is neither claim nor proof of loss beyond the US$8,991,250 that … Libra will have been ordered to restore to Swindon consequent upon rescission” (ibid, at [78]). The position was different as against Mr Logothetis personally, as there was no restitutionary claim against him, so that in his case an undischarged judgment against Libra could not be said to have reduced Swindon’s loss or to give rise to any logical difficulty over an award of damages in respect of that loss.

There was a proprietary claim pleaded against Syntax, on which at trial SKAT asked for hypothetical declaratory relief, i.e. a declaration that if Syntax had retained sums paid to it by SKAT, or their traceable proceeds, then it held the same on trust for SKAT. At trial, SKAT made no attempt that I could identify to prove that Syntax has retained anything traceable to sums paid to it by SKAT now some 10 years or so ago. I am not satisfied that it is appropriate to grant declaratory relief in those circumstances. That makes it unnecessary to consider a possible complication I identified in preparing this judgment, but which was not addressed at trial, namely that the default judgment in any event was only for an amount to be determined by the court in respect of SKAT’s claims for monetary relief. It was obtained upon an application by SKAT supported by a solicitors’ witness statement that made no mention of the claims for declaratory relief. If I might otherwise have contemplated granting any declaration, I would have given SKAT an opportunity to address that complication before making a final decision.

Finally, as well as being the subject of that default judgment in the Second Claim, Syntax was also a defendant to the First Claim (see paragraph 30 above, and Appendix 1 below). Judgment in default was entered against Syntax in the First Claim on 31 August 2018, but by a Consent Order dated 6 February 2024 all further proceedings against Syntax in that Claim were stayed, with no order as to costs. I shall ask counsel for SKAT to assist me as to whether any further order needs to be or should be made now in respect of the First Claim as against Syntax.