Skatteforvaltningen v Solo Capital Partners LLP & Ors

This litigation concerns the Danish tax regime in respect of dividends declared by Danish companies, an aspect of the taxation of income by the Kingdom of Denmark. The claimant is the Danish national tax authority. Without deciding the point, I understand that to mean the claimant is not a separate legal person from the Kingdom of Denmark. I refer to the claimant as ‘SKAT’ without deciding any question of its true legal nature or identity that goes beyond what I have just said.

Many national tax regimes use a notion of tax domicile (whether or not that label is used), such that:

The case concerns exclusively shares in Danish listed companies that existed at all material times only as a legal construct, in dematerialised and fungible form. Everything I say therefore refers only to such shares and such companies, during the period of interest in the case, which is August 2012 to July 2015.

Under Danish tax law, Danish company dividends were taxed (a species of income tax to which I shall refer as ‘Danish dividend tax’) as 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; and in respect of Danish dividend tax, Danish tax law operated on a withholding tax basis. Danish companies, upon declaring a dividend, were obliged to pay 27% of the dividend to SKAT and only the balance (73%) to VP Securities, the Danish Central Securities Depository (‘CSD’), for distribution to shareholders. There were exceptions to the general withholding obligation imposed on Danish companies declaring dividends, but they are not relevant for my purposes.

The payment by a Danish company to SKAT of the 27% it withheld from what it paid to VP Securities for distribution discharged the Danish dividend tax liability of all those liable to Danish dividend tax on the dividend in question. 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 specified rate below 27%. Of particular relevance in this case, for example, tax-exempt US pension plans were entitled under the Denmark-US DTT not to be taxed on Danish dividends; likewise tax-exempt Labuan corporations under the Denmark-Malaysia DTT.

That entitlement, where it existed, was or gave rise to a right under Danish law, enforceable against SKAT, because the DTTs were given effect, by statute, under Danish domestic law. The statutory technique by which that was achieved varied over time. The upshot was that if a tax-exempt US pension plan or tax-exempt Labuan corporation incurred a Danish dividend tax liability that was discharged by a withholding and payment to SKAT by the Danish company in question, it had a right under Danish law to be refunded by SKAT.

There was a difference in the expert evidence over the legal characterisation of that right. Prof Laursen, called by SKAT, thinks it is a restitutionary entitlement based upon a principle of unjust enrichment. On that specific point, I prefer the view of Mr Bachmann, called by the Sanjay Shah Defendants, that there is no unjust enrichment of SKAT by its receipt of a payment to which it is entitled as of right under primary tax legislation. I accept Mr Bachmann’s evidence that the claim for a tax refund is not a restitutionary claim, but a statutory claim under Danish tax law.

Prior to 1 July 2012, there was no specific statutory provision in Danish tax law referring or giving effect to that refund entitlement. It therefore arose simply as a necessary incident of the enactment into Danish domestic law of a DTT. Since that date, s.69B(1) of the Danish Withholding Tax Act has provided as follows (in translation):

This litigation (and related litigation in various other jurisdictions, in particular the US, Malaysia and Dubai) concerns claims for repayments under s.69B(1) presented to SKAT and paid between August 2012 and July 2015 that SKAT says were not claims it was obliged to honour. SKAT says it was wrongfully induced to pay those claims by misrepresentations made in or implied by the reclaim forms and/or supporting documents presented to it. It says it thus paid out (in aggregate) over DKK12.5 billion (c.£1.5 billion) as (purported) dividend tax refunds it was not liable to pay, 90% or more of which was paid out in the second half of the relevant period, from March 2014. For this judgment, it is not necessary to introduce the defendants or summarise their different degrees of involvement (as alleged by SKAT) or the bases upon which SKAT says they each have a liability in connection with the payments it says it was wrongfully induced to make.

Five Claims have been consolidated: CL-2018-000297 (70 defendants); CL-2018-000404 (25 defendants); CL-2018-000590 (8 defendants); CL-2019-000487 (9 defendants); and CL-2020-000369 (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 (1) below). At the time of this second preliminary issues trial, there were 17 separate legal teams from 16 firms of solicitors responding to SKAT’s claims, representing between them 65 of the defendants.

The remaining 41 defendants, at the time of this trial hearing, were:

I was appointed as designated judge for the litigation, pursuant to what is now section D1.4 of the Commercial Court Guide (11^th Edition – at the time, section D4 of the 10^th Edition), prior to the first main CMC in January 2020, at and after which I have sought actively to manage the case, with Bryan J initially, subsequently Foxton J, as alternate. I described the structure of SKAT’s claims, and what it seemed trying them would involve, in a ruling at the second main CMC in July 2020: [2020] EWHC 2022 (Comm). The main case management decision taken then, for the reasons given in that judgment, was that there should be three trial hearings:

The specific questions directed to be determined at the Validity Trial, with the answers I now give them, are set out in an Appendix to this judgment. The main point of contention at trial was whether the existence of a sale contract by which a seller contracted to sell shares in a Danish company to a buyer was sufficient to render the buyer liable for Danish dividend tax on a dividend declared by the company after the trade was concluded but before it was completed (if it was) by a share transfer to the buyer. The answer, I find, is no.

This judgment is considerably longer than it would need to be to explain that answer. That is because the Validity Issues require me not to stop there, but to find and state what were the requirements of a valid claim under s.69B(1) of the Danish Withholding Tax Act, for a refund of Danish dividend tax. One of those requirements was that the applicant had had a Danish dividend tax liability. So I am charged with finding when an applicant did have such a liability, not only with determining that the existence of a share sale contract was not enough. That way, this judgment is intended to define ‘the law for the case’, as regards the requirements of a valid refund claim under s.69B(1), so that there should be no need at the Main Trial for expert evidence of Danish law relating to the original validity or invalidity of the tax refund claims I shall be considering.

In paragraphs 10-11 above, I mentioned the number of defendants to these proceedings and summarised the extent to which they are or are not currently represented. All defendants had the opportunity to plead a case for and participate in the Validity Trial. In the event, and using labels that have been adopted in the litigation, pleaded cases were served in response to SKAT’s relevant pleading by the Sanjay Shah Defendants, the DWF Defendants, Messrs Jain and Godson, the PS/GoC Defendants, and the HK Defendants. As the heading of this judgment records, of those, only the Sanjay Shah Defendants, the DWF Defendants, and Messrs Jain and Godson, chose to participate at the Validity Trial.

It is a necessary averment for all of the causes of action alleged by SKAT in the proceedings that the individual refund claims were not valid claims under s.69B(1) of the Danish Withholding Tax Act that SKAT was obliged to pay. Strictly, that is an averment made severally against each of the several thousand refund claims that SKAT’s claims impugn. It will be essential for the Main Trial to have sample claims that between them are sufficiently representative of the full set of impugned claims that investigating the sample claims in detail will do justice to both sides of any remaining disputes as to validity, applying the findings of fact as to Danish tax law I am making now.

It follows from what I have just said that it is SKAT’s burden to prove the invalidity of the dividend tax refund claims it has impugned. It is therefore SKAT’s burden at this stage to prove (a matter of fact in this court) the material requirements under Danish tax law of a valid Danish dividend tax refund claim. That is as true against defendants who pleaded no case for the Validity Trial, or who pleaded a case but played no part at trial, as it is for the defendants who played an active part. Having said that, the opportunity that all defendants had to participate fully, if they did not want their fortunes to follow those of the defendants who did so, makes it fair to determine the Validity Issues as between SKAT and all defendants on the basis of the cases presented by the defendants who played a full part at this stage. It is fair for the court to proceed on the basis that if there were reasons not to accept SKAT’s case, or to make some finding contrary to SKAT’s case, they will have been put before the court by the active defendants; and that is the approach I have adopted.

I do not deal in this judgment with every point raised or argument presented at trial, but attempt to resolve the differences between the parties it is necessary to resolve, and to make the findings it is necessary to make, in order to answer the Validity Issues.

SKAT’s claims in these proceedings concern ‘cum-ex’ transaction structures by reference to which dividend credit advice notes were issued. SKAT will claim at the Main Trial that on their own, or when read together with the relevant s.69B(1) Form submitted to SKAT and/or any covering letter, the credit advice notes falsely stated or implied that the tax-exempt non-Danish reclaim applicant (e.g. a US pension plan) had received a Danish dividend from which Danish dividend tax had been withheld. SKAT’s pleaded case is more complex, involving several seemingly separate alleged misrepresentations. There may be a question whether SKAT can and should simplify its case, in the light of the findings made by this judgment as to the content of applicable Danish tax law.

SKAT has pleaded allegations that transactions arranged by some of the defendants and purportedly entered into, by reference to which dividend credit advice notes were generated, were shams. Nothing I say in this judgment is intended to pre-judge such allegations; but I shall not litter the judgment with ‘purported’ or ‘purportedly’ because of them.

A feature of Danish law and practice relating to Danish shares was the ‘record date’ for a dividend. If a dividend was declared on date ‘D’, then from October 2014 the record date for that dividend was D+2, i.e. 2 business days after D. Prior to October 2014, it was D+3. VP Securities, as the Danish CSD, was obliged to consult its records for 17:59:59 hrs (Danish time) on the record date to identify those to whom to distribute the dividend, a time chosen to ensure that the processing of all share trading settlements due on that date would have been completed so that the records consulted would reflect the position when the Danish stock exchange closed for trading at 17:00 hrs that day.

The standard settlement cycle for an ‘immediate’ sale of Danish shares (S sells to B ‘now’) was ‘T+2’, i.e. 2 business days from the trade date ‘T’, from October 2014, and T+3 before that. Therefore, an immediate sale concluded on or before a dividend declaration date, if completed on time by a transfer of shares, would mean the buyer would receive that transfer on or before the record date for that dividend.

A very common feature of cum-ex transaction structures was the use of short selling. At a simple level, an idea of what it means to sell short might be that it means to sell that which the seller does not own when the transaction is entered into, so that the seller must acquire it by the end of the settlement cycle if they are to effect a normal physical settlement (cash against delivery, or as the jargon has it, ‘delivery vs payment’ or ‘DVP’).

The notion of short selling that is relevant is more sophisticated, however, because a sale commitment can leave the seller with a short exposure even if they have a long securities position, i.e. they have the shares to sell, when the transaction is entered into. The most important illustration of that is where the seller’s long securities position comes from shares ‘borrowed’ under a stock loan. Leaving insolvency and its consequences aside, a stock loan on typical terms has an economic effect equivalent to a loan, as between the parties to it, but involves a full transfer to the ‘borrower’ of title to the ‘borrowed’ shares, with full right of disposal, and an obligation at the end of the loan period only to deliver equivalent securities, not an obligation to redeliver ‘the same shares’ as were delivered, which is generally not a meaningful concept in relation to dematerialised, fungible shares.

As is a common usage, for the sake of simplicity of language I may variously refer to ‘shares’, ‘a share’, or ‘the shares’, in a way that if taken literally might convey something more tangible (e.g. I am about to mention the case of a stock loan borrower who sells ‘the shares’ he has borrowed). Wherever I do that, I do not intend to imply that any one share is distinguishable from any other share, let alone to imply that shares exist other than in the abstract as legal rights, albeit rights the owner of which from time to time may be identifiable from records that have a real (digital electronic) form.

It follows from the full title transfer effected under a typical stock loan that the borrower who has taken delivery under such a loan, and then sells the shares, goes short by doing so, although they have the shares to sell when they conclude the sale. Having sold, the borrower/seller has obligations to deliver to the buyer now and to deliver to the lender at the end of the loan period. Their exposure position is short, and the sale creates that short position, but they are not selling something they do not own when they enter into the sale transaction. In market parlance, and under the Short Selling Regulation (I envisage – see below), they are a short seller.

The Short Selling Regulation is Regulation (EU) No.236/2012 of the European Parliament and of the Council of 14 March 2012 on short selling and certain aspects of credit default swaps. Article 2.1(b) defines a ‘short sale’ in relation to a share or debt instrument to mean “any sale of the share or debt instrument which the seller does not own at the time of entering into the agreement to sell including such a sale where at the time of entering into the agreement to sell the seller has borrowed or agreed to borrow the share or debt instrument for delivery at settlement , not including (i) [a repo]; (ii) a transfer of securities under a securities lending agreement; or (iii) entry into a futures contract or other derivative contract where it is agreed to sell securities at a specified price at a future date” (my emphasis).

There is a subtlety on that language that I do not need finally to resolve, on which there were brief submissions at trial. At first sight, “for delivery at settlement” appears to qualify both “has borrowed” and “has … agreed to borrow”. However, that would leave a sale by a seller who has already taken delivery under a stock loan as not a short sale if the stock loan resulted, as normally it will, in a full title transfer. That in turn would mean that such a sale would not have to be brought into the calculation of short positions under Article 3.1 of the Short Selling Regulation, and I do not think that can be the right reading of the Regulation.

Provisionally, I consider that the true sense of Article 2.1(b) is likely to be “… has borrowed [the share or debt instrument] or agreed to borrow [it] for delivery [no later than] at settlement”; and that the sense of “ such a sale where …” (my emphasis) is likely to be a sale for which the seller does not own the thing sold independently of the stock loan to which the following language refers. I recognise that is somewhat clumsy, and thankfully, as I have said, I do not need to determine finally whether it is correct.

The importance of short selling and stock lending to the transaction structures it is said were used in this case had the consequence that in exploring the Validity Issues at trial, illustrative examples involving short sales and stock loans were debated. Mr Jones KC for the Sanjay Shah Defendants went so far in opening as to suggest that whilst answering all the individual issues would be useful, in large part the trial came down to a single (tongue-twisting) question, namely “Does a short seller sell shares?”.

I do not agree. The answer to that question is, “Yes, of course, if they perform by a physical settlement (share transfer)”, but that begs all the issues that matter. A short seller who duly delivers shares as promised at settlement (having ex hypothesi covered themselves successfully so as to be able to do so) will have sold shares, in the event; and the question whether he was short selling at the time the trade was entered into is not relevant to that conclusion. Important issues begged by Mr Jones KC’s over-simplification include:

Before I turn to analyse and reach conclusions about the requirements under Danish tax law of a valid refund claim under s.69B(1), I first introduce the legal and market background at greater length, after saying something about the witness evidence at this trial, all of which was expert evidence.

The expert evidence came from eight witnesses across three disciplines: Danish tax law; Danish securities law; and market practice. I am grateful to all of them for their hard work preparing written materials, meeting one another to discuss their views, and attending court to give evidence in person. I set out now an overview of their backgrounds and expertise, before giving my assessment of the quality of their evidence.

SKAT called Professor Anders Laursen, Professor of Law at the University of Aarhus. He has great expertise in Danish and international tax law, having specialised in those subjects since 2011, and also has some experience of tax policy through his membership of various international committees.

The Sanjay Shah Defendants called Mr Christian Bachmann, a lawyer specialising in Danish tax, who is the founder and managing partner of Bachmann/Partners Law Firm. He has long professional experience of Danish tax law and was a member of the Danish Minister of Taxation’s committee on tax legislation from 2016 to 2020.

The DWF Defendants called Mr Robert Mikelsons, a partner and head of the tax group at NJORD Law Firm, having practised in the field for over 20 years. Like Mr Bachmann, he therefore has a wealth of professional experience in addition to his engagement in issues of tax law more widely, such as through representing the General Counsel of the Danish Bar in liaison committees organised by SKAT.

Mr Henning Aasmul-Olsen, called by SKAT, is head of Capital Markets and Corporate Finance at Moalem Weitemeyer Advokatpartnerselskab. His expertise in Danish securities law derives from practice as a lawyer in Denmark since 1985, as well as from time spent as an investment banker and chairman of general meetings for publicly traded companies.

The Sanjay Shah Defendants did not serve expert evidence on Danish securities law.

The DWF Defendants called Ms Catherine Tholstrup, a partner at Permin & Tholstrup who has practised in capital markets law for over 25 years. From this she has garnered a wide knowledge of Danish securities law, with particular expertise in financial contracts and the trading and clearing of over-the-counter (‘OTC’) derivatives.

SKAT called Mr Graham Wade, who began his career as a manager at Deloitte & Touche before working in the Structured Capital Markets department at Barclays Investment Bank until 2014. After three years as CEO of an investment management firm, since 2019 he has been involved in various financial technology start-up businesses. He therefore has a wealth of expertise in tax structured products and the interplay between tax and financial markets transactions generally.

Dr Richard Collier was called by the Sanjay Shah Defendants. From 1983 to 2015, he specialised in financial sector taxation as an accountant at PricewaterhouseCoopers. Since then, he has had a number of tax and tax policy roles, most recently with the OECD and various universities. His specialist expertise lies mainly in the tax treatment and tax policy implications of financial markets trading. In September 2020, his book was published, Banking on Failure: Cum-Ex and Why and How the Banks Game the System, examining banks’ exploitation of tax structured products in the cum-ex scandal, primarily focusing on what happened in Germany and how that was dealt with.

Mr Paul Sharma was called by the DWF Defendants. Now a Managing Partner at Alvarez & Marsal, he was formerly an accountant, Deputy Head of the UK’s Prudential Regulation Authority, and an Executive Director of the Bank of England. He has over 20 years’ experience of financial services practice from a regulatory perspective.

With the exception of Mr Mikelsons, I am satisfied that the expert witnesses all gave, in general, carefully considered, honest opinions as to the matters they had been asked to address, to the extent they could do so within their respective fields of expertise. With that same exception, the experts’ evidence was helpful, interesting, and obviously expert.

I have already mentioned, and preferred Mr Bachmann’s view on, a difference between him and Prof Laursen as to the correct legal characterisation of a dividend tax refund claim under Danish law. That point of difference has no bearing, however, on the requirements of such a claim, for it to be valid. There was therefore no material difference of opinion between Prof Laursen and Mr Bachmann on matters of Danish tax law, at all events as regards Mr Bachmann’s main report and his views set out in the joint memorandum. To the extent that some things stated there by Mr Bachmann might have indicated a possible view that a seller with no shares or rights to shares can make his buyer a shareholder, for tax purposes, simply by concluding a sale contract with him, I do not think that is the correct reading of his written evidence. The statements in question, as I read them, always took it that the seller had shares to sell when contracting to sell. In any event, it was clear in cross-examination that it was not (and never has been) Mr Bachmann’s view that a short sale, without more, could make the buyer a shareholder for tax purposes.

Cross-examination of Mr Bachmann demonstrated that there were some errors and unsupportable conclusions in his supplementary report. I do not consider it a reliable piece of work, to the extent it did not align with Prof Laursen’s views. Most notably, Mr Bachmann posited, in very Danish fashion, a long-term milk supply contract between a dairy farmer and a butter manufacturer, but he had to accept that he had made a bad point out of it, by misstating the law fairly fundamentally.

Mr Mikelsons’ evidence was generally unsatisfactory. None of the material opinions he had expressed in writing that appeared different from the views expressed by Prof Laursen (and Mr Bachmann) withstood scrutiny. Very few of them were even adhered to by Mr Mikelsons himself when pressed to explain them; and those few to which he did stick he was unable coherently to justify. He sought refuge in obviously incorrect suggestions that things he had said that were unhelpful to the DWF Defendants’ cause concerned the position under the general law in Denmark, not under Danish tax law. He declined to answer at all, or admitted to having no answer for, a number of perfectly fair questions with which, if his relevant views had substance, he would have been able and should have been happy to deal.

Mr Graham KC properly focused in cross-examination on exploring, so as to expose the substantial inadequacy of, the content of the views Mr Mikelsons had expressed, where they differed materially from those of Prof Laursen. It was not the occasion for prolonging the challenge, for Mr Mikelsons, of his opinions being found so wanting, by interrogating him at any length over how that had come to pass. In the circumstances, I do not consider I could make a positive finding that Mr Mikelsons’ written evidence, or his oral evidence to the extent he stood by his written evidence where it was challenged, was not honestly given.

That said, it is not easy to see that his written evidence would have been any different than it was had Mr Mikelsons seen it as his task to argue a case for the DWF Defendants in which he did not believe rather than to provide the court with an impartial account of Danish tax law, to the best of his expert ability. The problems with his evidence revealed by cross-examination were sufficiently numerous and of sufficient substance as almost to make one wonder if his reports had been written by someone not properly qualified to express expert views as to Danish tax law – yet Mr Mikelsons plainly was properly qualified. Nor (and meaning no disrespect to the skill of the cross-examiner) was it a cross-examination in which a corrective light bulb seemed to have been switched on in Mr Mikelsons’ mind, for reasons of subtlety or complexity that he might have overlooked. On the other hand, the extent to which opinions Mr Mikelsons had expressed in writing were withdrawn by him when tested is perhaps, ultimately, more consistent with a very poor job having been done than with dishonesty.

Whatever the truth behind the difficulties, the upshot, I regret to say, is that I do not regard Mr Mikelsons’ evidence as any kind of reliable guide to Danish tax law, on any point that matters, and I do not consider it appropriate to treat his opinions as having any weight, except where they accorded with those of at least one of Prof Laursen and Mr Bachmann. Given the extent to which their views coincided, that means that overall Mr Mikelsons’ evidence was of no material assistance.

There was an apparently significant difference of opinion between Mr Aasmul-Olsen and Ms Tholstrup concerning when a trading counterparty could be considered under Danish law to be, or to have become, a shareholder in a Danish company. However, that difference was found to be one of terminology, not substance. Ms Tholstrup used the language of ‘ownership’, in the context of shares, in an extended sense to include contractual rights between a seller and buyer of shares that have no impact on the company. That apparent difference having proved unreal (so far as is material for this judgment), there was only one real point of distinction between their respective expert views. I deal with that in context, below, and I prefer Mr Aasmul-Olsen’s opinion on it.

The market practice experts provided different, and generally complementary, not conflicting, perspectives, in consequence of their different backgrounds and experience, as summarised above. In the event, there was very little in the way of relevant but different opinion between them. Although the matters they were asked to address were labelled ‘market practice’, for want of any better single term, their opinions in fact covered a range of rather different types of issue:

As a well-researched investigator of and commentator upon the phenomenon, especially as it affected Germany, Dr Collier is an interesting and well-informed rapporteur by whom to be introduced to cum-ex trading, its causes and possible effects, and its evolution over time. However, I do not consider that his expertise matched that of Mr Sharma or Mr Wade on the issues on which the court in truth required the assistance of market experts. As between Mr Sharma and Mr Wade, their differences of background and perspective mean, in my judgment, 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.

On the most important points considered by the market experts, they were in any event agreed. For example, they agreed that market participants would understand a dividend (or ‘real dividend’) to be a distribution from a share issuer to its shareholders, and to be different from a ‘manufactured dividend’, 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 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 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, would be understood to be a matter of Danish tax law on which a market participant, if interested, would take specialist advice.

I now summarise the basic positions of the parties who actively contested this trial on the central issues (which are all matters of Danish tax law).

SKAT submitted that there were four requirements in section 69B(1) of the Danish Withholding Tax Act, viz. that:

Taking those four requirements in turn:

In relation to the receipt (or obtaining) of dividends, SKAT said that had to be understood as referring to the acquisition of an entitlement to receive dividends, and not to cash received (or not received, as the case may be). Danish dividend tax on an accruals basis entails that rights to a dividend accrue to the current shareholders at the time when the dividend is declared. The key question is then whom Danish tax law treats as a current shareholder from time to time, applying its accruals (acquisition of rights) principle of taxation.

SKAT said that principle could mean that a buyer became a shareholder under Danish law, for tax purposes, when concluding a contract for the acquisition of shares, but only if the contract gave the buyer a final and binding right to a shareholding, which in turn required the seller to have had at that moment a right to transfer a shareholding to the buyer (albeit the seller would only have to effect or procure a share transfer to the buyer at settlement). If when a buyer contracts for a shareholding, their seller had nothing to sell, Danish tax law did not treat anything as having yet been sold; it did not treat the buyer as having become a shareholder by entering into the contract.

Thus, SKAT submitted, a short seller did not without more transfer to their buyer, for tax law purposes, ownership of any shares at the point of entering into a contract to do so; they only did so if and when they acquired shares that could be transferred to the buyer, or a right to such shares.

The Sanjay Shah Defendants noted that the register of shareholdings in Danish companies at VP Securities will be incomplete and inaccurate, for dividend tax purposes: incomplete, because it will include custodians holding as nominees; inaccurate, because (a) a borrower under a stock loan may be the party for whom the custody chain is holding shares, but is not treated as a shareholder by Danish tax law, and (b) a share buyer whose purchase had not yet settled will not appear in the custody chain as shareholders but might be treated as a shareholder by Danish tax law.

They submitted, further, that it is often, indeed typically, impossible to trace a share from its original issue to a given shareholder at some later point in time. Neither Danish tax law nor any DTT, they said, required an applicant for a refund of withholding tax to go through such a tracing exercise to prove their entitlement. Therefore, it was suggested, because Danish courts apply a dynamic, realistic and functional interpretation of tax legislation, it cannot be the law that the applicable tax legislation required an applicant to do such a thing to be granted a refund.

All the Danish law experts agreed about the general application of the accruals principle; and SKAT’s Legal Guide confirmed it. It dated back at least to the Danish legislation that established income tax in 1922. It also accorded with market practice in Denmark. The principle, it was submitted, was not altered just because a trade takes place on or shortly before a dividend declaration date. It was said to follow that:

The Sanjay Shah Defendants argued that SKAT was seeking, unjustifiably and without support in any Danish legal source, to create exceptions to the accruals principle, as applied to Danish dividend tax and the withholding thereof:

The DWF Defendants submitted that SKAT was reading s.16A(2)(1) of the Danish Tax Assessment Act and s.65(1) of the Danish Withholding Tax Act too literally and too narrowly. They said SKAT’s case took insufficient account of the realities of holding and trading intermediated shares internationally and of market understandings of dividends, while giving too much weight to Danish company law and insisting on an unstated requirement of traceability.

As the Danish courts adopt a dynamic and realistic approach to interpreting Danish taxing statutes, especially older legislation, the DWF Defendants said that market practice is key to understanding the requirements for a withholding tax refund. SKAT, they suggested, was ignoring this and adopting an unrealistic view of the law according to which refund applicants would have had to achieve the often impossible aim of demonstrating a right to dividends through many layers of custodians, even where the market would understand the applicant to have been a shareholder when a dividend was declared.

The near impossibility of doing this, it was said, was a product of numerous aspects of market practice, including dematerialisation, the fungibility of shares, intermediation, omnibus accounting, internalised net settlement, short selling and stock loans, and manufactured dividends. To import it as a requirement for a withholding tax refund would be contrary to market practice and would amount to reading into the tax legislation a requirement that was not there.

The DWF Defendants contended in writing that the meaning of ‘dividend’ in s.16A(2)(1) of the Tax Assessment Act and Article 10 of the relevant DTTs encompassed different types of payments equivalent or analogous to dividends, such as manufactured dividends, which the market treats in the same way as real dividends. That contention did not have substance and did not make it into closing submissions.

Like the Sanjay Shah Defendants, the DWF Defendants submitted that Danish tax law, pursuant to the accrual or acquisition of rights principle, applied in unqualified terms so that the determinative time for share ownership for tax purposes is when a contract for the acquisition of shares was concluded. In principle, any such contract would do, even though that meant that Danish tax law could recognise shareholders for tax purposes whose holdings (recognised as such for tax purposes) exceeded the entire share issue. It was said that if that was unintended or regarded as unacceptable by the Kingdom of Denmark, the solution was legislation not litigation.

The DWF Defendants, like the Sanjay Shah Defendants, complained that SKAT was seeking to create exceptions to the accruals principle that were unsupported by Danish case law or academic literature, and that were unjustified in principle. They would also create unworkability in practice, it was said, because if they existed, it would be impracticable if not impossible for a buyer to ascertain their tax liability (if any) and therefore their entitlement to a tax refund (if any).

Messrs Jain and Godson (jointly in writing, and Mr Jain orally in some cross-examination of Mr Wade and in closing argument) focused on whether the market practices alleged by the actively participating defendants were established. Substantial parts of what they said, especially in writing, were more of an introduction to what they may be saying at the Main Trial as to the nature and extent of their factual involvement and why (therefore) they will contend that they behaved honestly at the time. It is evident from the pleadings for the Main Trial that those aspects of what Messrs Jain and Godson said are contentious; in any event they were not for this trial.

The broad thrust of those submissions was that, so they say, Mr Jain and Mr Godson only ever engaged or were involved in trading on market-standard terms, generating documentation from reputable sources that they took at face value as evidencing that shares and dividends had been acquired, tax had been withheld, and so the tax-exempt US pension plans or Labuan corporations in question had a proper basis for claiming refunds. SKAT asserts a fundamentally different narrative as to Mr Jain’s and Mr Godson’s conduct. All of that is for the Main Trial, not for now.

Finally, Messrs Jain and Godson submitted, other jurisdictions had acted against this sort of trading in the past, and Denmark had the option to change its refund scheme (as in the event it did), or collect withholding tax on manufactured dividends as well as real dividends, to solve any problem it considered there was of more being paid out in dividend tax refunds than it regarded as desirable.

Danish law is a civil law system. The Danish Constitution is the supreme source of law, and ordinary statutes rank below it. Pursuant to s.3 of the Danish Constitution, the legislative authority to enact statutes in the form of Acts is vested jointly in the Danish Government and the Danish Parliament. This authority is in most cases non-delegable, but statutes on occasion delegate authority over minor matters to an appropriate Minister. Executive orders issued in the exercise of such a delegated power have the same legal effect as statutes and so bind the courts. However, executive orders can be issued only within the authority granted by the relevant statute and cannot conflict with it.

Case law is a secondary source of law, in that case precedents will often interpret statutes. But construing legislation is not the courts’ sole function. Certain aspects of Danish law, such as some basic general principles and remedies, are embodied only in case law. Lower courts will usually follow precedents set by superior courts under a tripartite court hierarchy, with the Supreme Court at the top, the High Court of Western Denmark and the High Court of Eastern Denmark below it, and the City Courts below them. There is no formal system of binding precedent as known to English law, however.

Danish courts interpret statutes by considering, first and foremost, the statutory language. If that yields a clear meaning, it will prevail over a meaning that might be suggested by secondary material adduced as an aid to construction. If the statutory language is ambiguous, the court looks to the object and purpose of the provisions in question, and if ambiguity remains to other interpretive aids, including the preparatory work that preceded the legislation (usually in the form of legislative notes, expert reports and Parliamentary remarks about the bill), general legal principles, case precedents, academic literature and, in some cases, market practice.

The Danish Constitution provides, by s.63(1), that “[t]he courts of justice shall be empowered to decide any question relating to the scope of the executive’s authority”. That applies to SKAT as a branch of the executive. However, the Danish system of tax law has its own decision-making bodies that stand outside the traditional court system. SKAT itself, as an executive body, must follow the law and cannot bind the courts with its decisions. The same goes for the Tax Council, a body within SKAT which decides cases involving fundamental questions of legal interpretation. Decisions of SKAT or the Tax Council can be appealed to the independent National Tax Tribunal (and during the period with which I am concerned, taxpayers could also appeal a decision of SKAT (but not of the Tax Council) to a Tax Appeals Board). The National Tax Tribunal also has no power to bind the courts, but its decisions can be appealed to them, and the courts often consider its decisions persuasive.

Danish securities law is a patchwork of general civil law, contract law, company law and personal property law, as those areas of law apply to securities, with an admixture of more specific regulatory provisions.

Under Danish securities law, the notion of a “share” is familiar enough. It is a security that represents a notional relative stake in the capital of a company and that provides the holder with various financial and governance rights over that company. Included among the financial rights is an entitlement to a percentage of the company’s distributed dividends in proportion to the shareholding (as a percentage of the entire share issue).

Shares listed for trading on the Danish stock exchange must be issued in dematerialised form (i.e. in the form of electronic book entries rather than physical certificates) through VP Securities, with a register of ownership maintained there in electronic form which may evidence share ownership. That has been the system in Denmark since 1988. Each share is allocated to an account at VP Securities maintained by an “Account Holding Institution”, which may be a Danish or foreign financial institution.

Shareholdings registered at VP Securities, however, might be held on a nominee basis. Furthermore, the party for whom an Account Holding Institution might be holding as nominee may itself be a custodian holding as nominee. Custodians did not need to be Danish; and their custody relationship with clients did not need to be governed by Danish law. Therefore, a shareholder might hold their shares through a chain of custody from a head custodian, holding on their behalf from VP Securities, possibly via one or more sub-custodians. In that case, the owner of the shares was the end investor, i.e. the custody client of the head custodian. A right under an account or custody agreement against a custodian did not per se constitute “ownership” of shares; but if there was accurate record-keeping all round, an investor should only have appeared to be a shareholder, by having shares credited to them in a securities custody account with their head custodian, where a nominee custodianship chain existed from that accounting record down to a holding at VP Securities and, therefore, down to the company. (It is possible to visualise custody chains as leading up from the company at the bottom to shareholders at the top, or vice versa. In this judgment, the language used assumes the former.)

A legal person could become an owner of shares in a listed Danish company by subscribing to newly issued shares, by a share purchase (whether on- or off-exchange), or by enforcing creditor rights. They might also become an owner by taking delivery of shares under a stock loan, as typically that would involve a full title transfer, but in practice that would typically be a transient ownership, as the usual purpose of stock lending is to complete a short sale, the seller/borrower settling both transactions at the same time, i.e. borrowing in and immediately transferring on.

Danish property law recognises a general principle that a buyer cannot acquire better rights than his seller. The Latin tag nemo dat quod non habet, familiar to an English lawyer, is not used in Danish law, but the principle is recognised. It is not without exceptions in Danish law. The law provides for certain cases in which, where a person has misappropriated the property from its true owner and sold it on, the third party buyer may obtain ownership and extinguish that of the true owner, if he acts in good faith and the misappropriating party has provided apparent (but incorrect) evidence of authority to sell the property.

Danish civil law generally also recognises a principle that where parties to a contract did not intend a transfer of ownership of an asset for which the contract appears to provide, the contract may be treated as a ‘pro forma’ and fall to be recharacterised in line with the parties’ true intentions, which may mean that ownership is not transferred at all. However, the buyer under a contract treated as pro forma may sometimes still pass ownership in the underlying assets to a third party buyer, if the third party buyer acts in good faith and the buyer under the pro forma contract has ostensible authority to transfer ownership.

In Danish securities law, dividends are a distribution of equity by a company to its shareholders. The Danish Companies Act, in s.179(1)(1)-(4), permits four types of dividend to be distributed. Only two of those matter to the present case, ordinary dividends under s.179(1)(1) and extraordinary dividends under s.179(1)(2). The former are declared by corporate resolution of the shareholders in general meeting (s.180(1)); the latter are declared by resolution of the board of directors with special authority granted by the general meeting (s.183(1)-(2)).

For listed shares, declared dividends were paid, net of withholding tax, on behalf of the company by reference to the registered shareholdings at VP Securities. That initial payment discharged the company’s financial obligations to its shareholders in respect of the declared dividend. As I have mentioned already, VP Securities distributed the net dividends based on registrations at 17:59:59 hrs on the record date. Those net dividend amounts were then paid on a set payment date, after the record date.

The Danish Constitution provides, by s.43(1), that only statutes may impose, alter or repeal tax liability.

Legal persons not resident in Denmark have limited Danish tax liability on income that has a source in Denmark falling within one of the categories listed in s.2 of the Danish Withholding Tax Act (natural persons) or s.2 of the Danish Corporation Tax Act (corporate entities). Dividends are listed at s.2(1)(6) and s.2(1)(c) respectively. An applicant for a refund of dividend tax under s.69B(1) of the Danish Withholding Tax Act must have been liable to tax under s.2(1)(6) or s.2(1)(c), as relevant, in respect of the dividend to which their refund claim related in order to be capable of being eligible for a refund. I consider the Danish statutory definition of dividends for these purposes later.

To be eligible for a refund under s.69B(1) of the Danish Withholding Tax Act, an applicant who had been taxed on Danish company dividends under s.2(1)(6) or s.2(1)(c), as the case may be, had to be able to say that the tax thus imposed on them exceeded the maximum tax liability permitted under a DTT applicable to them. It is of the essence of a valid tax refund claim under s.69B(1) that the applicant has been excessively taxed, by the application to them of Danish tax legislation such that tax has been levied from them beyond that permitted by a DTT of which they have the benefit.

Denmark, like the UK, adopts a dualistic approach to international law treaties. So treaties are generally not directly enforceable under Danish law, and DTTs are no exception. Before 1994, the Danish Authorisation Act authorised the Danish Government to conclude DTTs granting relief from double taxation in such terms as to give direct effect to any DTTs thus authorised without the need for any further legislation specific to the DTT in question. The Denmark-Malaysia and Denmark-Luxembourg DTTs with which I may be concerned at the Main Trial were entered into under this authority.

The Danish Authorisation Act was repealed in 1994, since when DTTs entered into by the Danish Government have needed to be ratified individually by the Danish Parliament to be enforceable under domestic law. The Denmark-US DTT (including its 2006 protocol) and the Denmark-UK DTT with which I may be concerned at the Main Trial were ratified in that way.

It is recognised as a ‘golden rule’ under Danish tax law, which was reflected in the language of the Authorisation Act and of the preparatory work for the Act that repealed it, that DTTs can only ever grant relief from taxation otherwise imposed by Danish law, i.e. by Danish tax legislation. DTTs cannot impose tax or widen what would otherwise be the scope of a tax imposed on non-Danes by Danish tax legislation. Thus, contrary to one of the themes in Mr Mikelsons’ expert report, if a DTT defined ‘dividends’ more widely than did Danish tax law imposing dividend tax on non-Danes, that could not mean that a ‘dividend’ falling within the DTT definition was taxable as a dividend under Danish law. It would mean only that if – whether the Danish law called it a ‘dividend’ or called it something else – Danish tax law taxed non-Danes on income that fell within the DTT definition, then those who had the benefit of the DTT in question might be entitled to relief, and therefore a refund, subject to satisfying any other applicable requirements.

That makes it unnecessary for this judgment to consider in detail the provisions of any particular DTT, or the OECD model DTT on which most DTTs are based. I do though mention one point on the language of Article 10 of the applicable DTTs. I do so for completeness only, since its significance fell away, save that Mr Mikelsons placed misconceived reliance on it that did him no credit. The point is that Article 10 grants relief, so far as material, only where dividends paid by a Danish company have been taxed in the hands of the “beneficial owner” of the taxed dividends. This ‘beneficial ownership’ requirement is an additional pre-requisite for entitlement to tax relief and, therefore, would be an extra condition to be satisfied for a tax refund claim under s.69B(1) of the Danish Withholding Tax Act to be valid. It does not refer to concepts of trusts and beneficial interests as known, for example, to English law, and has nothing to do with questions of ultimate beneficial ownership of a corporate entity (the contrary being the obviously bad point put forward by Mr Mikelsons). However, it is not now said by SKAT (it was at an earlier stage) that if the impugned refund claims in this case were otherwise valid, they were bad because of a failure to satisfy a DTT ‘beneficial ownership’ requirement.

Under Danish law, tax rules ordinarily follow any applicable (non-tax) civil law as regards the meaning and effect of common legal terms or of transactions. That is a general principle of Danish tax law agreed by the Danish tax law expert witnesses at trial. Unless the language of the relevant taxing statutes or some specific principle of Danish tax law requires otherwise, a ‘share’, or a ‘shareholder’, or a ‘dividend’, if such terms are used in defining a tax liability or entitlement, have the same meaning under Danish tax law as under applicable Danish civil law. For those terms in particular, that means Danish company law.

It was also agreed by the Danish tax law experts at trial, and so I find (on the balance of probabilities, albeit with the temerity to express a note of doubt, as I explain below) that under a specific principle of Danish tax law, it did depart from company law, to at least some extent, in identifying the shareholders of a company from time to time. It was therefore possible for a party to be regarded by Danish tax law as a shareholder to whom a dividend entitlement accrued, so as to be liable to Danish dividend tax, who was not the real shareholder to whom that dividend entitlement accrued as a matter of Danish company law.

SKAT admitted and averred that possibility, but said that whatever its precise ambit, a seller who neither owned a shareholding nor had contracted to acquire a shareholding could not make the buyer a shareholder for tax purposes merely by contracting to sell them a shareholding. The defendants contended that the Danish tax law principle did go that far and have that effect.

The principle under Danish law that tax law ordinarily follows civil law makes it unnecessary to say more at this stage about the meaning or effect of applicable concepts and legal terms, for tax purposes. I leave that for the decisive section of this judgment, below, which concerns the accrual of dividend rights, where I discuss and make findings as to the extent to which Danish tax law does take a different view from that of Danish company law.

Tax law, while generally operating by reference to legal constructs of the civil law, e.g. companies and contracts, is a species of public law. It was common ground that a Danish public authority can through continuous and consistent, deliberate conduct establish an “administrative practice” on which individuals and corporations might be entitled to rely (at all events unless and until it was publicly departed from), because of a principle of legitimate expectation or a principle of equal treatment. SKAT as a Danish public authority could therefore establish administrative practices in the field of Danish tax on which parties might be entitled to rely.

The party seeking to rely on an alleged administrative practice bears the burden of proving its existence. Statements in SKAT’s Legal Guide, as published by it from time to time, are presumed, but not irrefutable, proof that a practice in the terms of those statements exists.

However, (at least) two key criteria must be satisfied for a public authority’s actions to constitute an administrative practice:

On the basis of the expert evidence, as explored in cross-examination, I consider it the better view, and my finding is, that the second criterion is not confined to cases where the mandatory legal outcome or requirement was ‘clear’ or ‘obvious’. Articulations of the criterion can be found in Danish sources to the effect that an administrative practice must not be ‘clearly contrary to law’ (or the like, my emphasis). However, the better view of the expert evidence, in my judgment, is that the sense is not that of the obviousness or otherwise of the departure from the law, rather the concern is as to the nature of the discrepancy. An administrative practice cannot (purport to) depart from that which is definitively stipulated by mandatory law. It must be possible to say that the public authority, by adopting the practice, did no more than operate within the bounds of its discretion under a permissible interpretation of the law. That is so whether it was straightforward, or (to the contrary) it was complex or obscure, to identify what outcome or requirement was definitively stipulated by the law (for example on the proper construction of a tax statute).

That means, so I shall say no more of administrative practice in the remainder of this judgment, that there is no possibility here of any administrative practice established by SKAT altering or modifying the requirements that had to be satisfied for a valid tax refund claim under s.69B(1) of the Danish Withholding Tax Act. Those were requirements mandated by statute as to the circumstances in which SKAT was obliged to make a refund payment that no administrative practice could change. It was not a matter for this trial whether, if those requirements could be so affected, SKAT had established an administrative practice having that effect. I shall say nothing about that, therefore, although it was the subject of some submissions at this trial, in particular from the DWF Defendants and (in response) from SKAT.

There may have been an over-interpretation there, by the DWF Defendants, of how far the Validity Issues went. How SKAT dealt with dividend tax refund claims under s.69B(1) over the years has the capacity, in principle, to be relevant to contentious issues of primary fact for the Main Trial as to awareness of wrongdoing, as alleged by SKAT against many of the defendants. It may or may not prove to be of any real relevance, and if relevant it may or may not carry much weight; but that capacity in principle for overlap makes it inappropriate, I think, to say more about it now.

Whether or not I identified that at the time as a reason for distinguishing between administrative practice allegations and market practice allegations, as regards the scope of this trial, it is a sound and sufficient reason for that distinction being maintained. For the market practice allegations, but not for the administrative practice allegations, the Validity Issues as ordered include the question whether the matters of practice alleged on the pleadings existed in fact.

As I mentioned in paragraph 75 above, market practice is capable of being an interpretive aid, under Danish law, when a court is construing legislation, including tax legislation, that is unclear. This is, of course, subject to s.43(1) of the Danish Constitution, which reserves to primary legislation the power to impose, alter or repeal taxes.

In the event, none of the market practice experts gave evidence of the existence of a market understanding as to the requirements of Danish tax law, or of any fixed feature of the operation of the market requiring Danish tax law to have a certain content for the market to function. It would have been by no means an easy question whether such an understanding or practice, if there had been any, would in fact have affected the correct interpretation of the relevant Danish tax legislation. As it is, the question does not arise. The market practice expert evidence was interesting nonetheless, and not unhelpful for the forensic exercise of exploring the implications of the parties’ respective cases as to the content of applicable Danish tax law. That in turn was a legitimate aspect of testing the expert evidence of Danish tax law upon which any decisions I make must be founded.

I come back to that question of practical implications at the end of this judgment, but I set out now various details of how the Danish share market operated that were mostly if not entirely uncontentious.

Custodians operated a number of different types of custody arrangements. To be clear, whatever the arrangement, an end investor would expect to have a securities custody account with their head custodian accurately recording from time to time any shareholding they had in any Danish company. The different types of custody arrangements concern how the custody chain was structured. In particular, the distinction was drawn in the expert evidence between omnibus accounts and segregated accounts.

With omnibus accounts, multiple custody clients’ holdings in a given Danish company are commingled at one or more levels in the custody chain. At its most simple, if X and Y both hold 100 shares in DanCo, and use the same head custodian (H-C), who uses a sub-custodian (S-C), X and Y will each have a securities custody account with H-C showing them holding 100 shares, but if there is omnibus accounting between H-C and S-C, then H-C will have a single securities custody account with S-C showing them (H-C) holding at least 200 shares, without reference to or distinction between X and Y (or between either of them and any other custody client of H-C with shares in DanCo) in the record-keeping between H-C and S-C.

In turn, S-C’s account at VP Securities (or at a sub-sub-custodian) might be an omnibus account such that if S-C also has custody of the shareholdings of others (be they end investors or other custodians who in turn hold as nominees), S-C’s account will contain the aggregate holdings of all their custody clients, without reference to or distinction between them in the record-keeping between S-C and VP Securities (or the sub-sub-custodian).

One convenience of omnibus accounting is that some trades will be capable of ‘internalised’ settlement by book entries at one level of the custody chain. Thus, in my simplest of examples just discussed, if X contracts to sell 50 of their shares to Y, that sale can be completed by matching debit and credit entries in X’s and Y’s respective securities custody accounts with H-C. S-C, let alone VP Securities (or any sub-sub-custodian between the two), need not and ordinarily would not be involved, or have reason to have any record of the X-Y trade. It was common ground that no reporting down the custody chain was required by law or market practice.

With segregated accounting, a given custodian keeps separate accounts with VP Securities or their sub-custodian (as the case may be) in respect of their different clients’ separate holdings. Thus, in the above simplest of examples, if H-C maintained segregated accounts with S-C, they would have separate accounts there, each showing 100 shares, identifiable as separate accounts as nominee for X and as nominee for Y. In that case, if X sold 50 shares to Y, H-C would need to make S-C aware of that trade and ensure that their segregated accounts at S-C were appropriately debited and credited at settlement.

The market practice experts drew a distinction in their joint memorandum, in the context of segregated accounting, between ‘individual client segregation’, in which “[a] separate account is clearly and beneficially segregated for the individual client”, and ‘ultimate investor segregation’, in which “[the] client holds a separate account in its own name at the custodian or CSD”. I do not consider that a useful distinction to draw, or at all events those definitions were not very helpful. They derived from Mr Sharma’s first report, but they omitted his explanatory comment that in ‘ultimate investor segregation’, as he had called it, “[the] broker’s role is merely to send securities to, or receive them from, [the client’s account in its own name at a custodian or CSD] when instructed by the client”. That means Mr Sharma was not describing a special species or instance of segregated accounting, he was just saying that a broker may or may not act as a custodian.

A more helpful description, for my purposes, was given by Mr Wade in his first report, agreeing in substance with that element of Mr Sharma’s report, but articulating the effect as being “that different legal combinations of custody were possible ranging from full client segregation (where all custody accounts in the chain were clearly and beneficially segregated for the relevant client) to omnibus accounts where all client positions were co-mingled at one or more levels in the custody chain.” Thus, if X’s 100 shares in DanCo were held by way of account entry with H-C, H-C held segregated custody accounts with S-C, and S-C held segregated accounts with VP Securities (or with a sub-sub-custodian, who, and likewise all further sub-…-custodians, down to VP Securities, held segregated accounts), there would be a full custody chain of segregated accounts each holding 100 shares, representing X’s shareholding. If at any level of a given custody chain, omnibus accounting was in use, that would not be true.

The experts also recorded agreement in their joint memorandum that “Compared to segregated accounts, omnibus accounts were often a cheaper solution for the client because the custodian could offset long and short positions and reuse client securities as collateral or for rehypothecation.” Some care is needed to understand what that does and does not mean. There is no such thing as a ‘short’ securities custody position – securities custody accounts can never have a negative (‘overdrawn’) balance. A custody client either holds or does not hold securities at any given time. Their securities custody account will either have securities in it or not.

What the experts had in mind is different and more limited. Going back to my simplest of examples, X and Y each holding 100 shares in custody accounts at H-C, now suppose that Z, another custody client of H-C, contracts to sell 50 shares to Y, but has no shares. If H-C’s custody account terms and conditions allow this, H-C might borrow 50 shares from X to enable Z to deliver to Y. In concept, H-C takes a transfer from X of 50 shares (as a stock loan), H-C transfers those 50 shares to Z (as a stock loan), and Z then transfers them to Y (as a sale completion); or perhaps H-C causes X to transfer to Z (by way of stock loan), who transfers to Y (by way of sale), if the authorised rehypothecation entitles H-C to match X and Z as principals on the stock loan rather than have H-C itself sit as principal between them. I say that is the concept, rather than make a definite finding that it is precisely how settlement would be documented, in case it might be possible, on the terms of all relevant custody contracts and within the systems operated by H-C, to effect settlement simply by debiting X, and crediting Y, with 50 shares.

After that settlement, X’s exposure position is still long 100 shares; Y’s is long 150 shares; and Z’s is short 50 shares. The aggregate net exposure position across the three clients is long 200 shares. The aggregate shareholding across the three clients is also 200 shares at all times, not 250 shares (the aggregate of the two long exposure positions). That is because, though X’s exposure position is long 100 shares throughout, its securities position (shareholding) changes from a holding of 100 shares to a holding of 50 shares. The balance of X’s long exposure position (50 shares) has become constituted by a forward delivery obligation owed by H-C rather than by any current shareholding (or, it may be, a forward obligation owed to it by Z, if H-C is not interposed as principal between Z and X on the stock loan).

Settlement of securities transactions carries the risk that either side of the transaction will default on an obligation to transfer securities or make payment so as to cause the settlement to fail. Physical settlement, as opposed to netting off or cashing out, if authorised and possible on the facts, is normally on a DVP basis. For that, the CSD or custodian handling the settlement needs to see that the party obliged to deliver securities and the party obliged to make payment have the securities and cash, respectively, in their account to deliver; otherwise, settlement will fail.

Net settlement was a known practice. By definition, a net settlement, necessarily involving two or more otherwise separate transactions, involves treating the transactions as performed without delivery of all the securities otherwise required to be delivered and/or without making all cash payments otherwise required to be made. Going back to X and Y with their 100 shares each at H-C, if X sells 50 shares to Y today, for settlement tomorrow, and Y sells 50 shares to X tomorrow, for immediate settlement, the two trades would be capable of being ‘net settled’, meaning that no shares would be transferred, and any difference in price only would be paid. Assuming the legal basis for net settlement was in place, as a matter of contract, both trades would have been performed in a manner permitted by their terms or by the subsequent consent to net settlement – there would have been no default – but there would have been no movement of shares.

If the same contracts were entered into, but neither X nor Y held any shares, or ever acquired any shares from elsewhere with which to perform, and their terms of trade allowed those trades to ‘net settle’ rather than default (in the absence of any shares for delivery), so that a difference in price was still cashed out, again X and Y would be in a position to say that they had performed, not defaulted on, the two trades. But no shares would ever have been transferred; and neither X nor Y would ever have been a shareholder.

At the risk of introducing confusion, there are several recognised DVP models and one of them (Model 3), which was used by VP Securities, allowed multi-lateral net settlement, i.e. settlement across transactions that are not all between the same counterparties. That risks confusion because Model 3 is therefore a ‘DVP’ model that may and often will involve fewer securities being delivered, and less cash being paid, than the aggregate otherwise required by the transactions thus settled.

Mr Sharma was the best placed of the experts to assist the court with the practical realities of that. In short, and possibly over-simplifying, under DVP Model 3, all cash payment and securities delivery instructions accepted for settlement in a given settlement batch at (say) VP Securities are pooled. As long as all account-holding parties submitting those instructions and therefore due to participate in that settlement batch have in their accounts (a) at least the net volume of securities they are due to deliver (if their delivery obligations exceed their delivery entitlements) and (b) at least the net cash they are due to pay (if their payment obligations exceed their payment entitlements), in that settlement batch, then the batch will be processed as a success (no settlement failure).

The settlement processing will then replicate a collection from all net deliverers of their respective net delivery volumes, into a central basket, and the distribution of the contents of that basket (without distinction as to source) to the net recipients, each receiving their respective net entitlement; and the collection from all net payors of their respective net cash commitments, into a single bucket, and the distribution from there to net payees, without distinction as to source, of net cash entitlements.

If any net deliverer or net payor does not have the securities or cash to cover their net delivery or payment commitment, there will be at least some measure of settlement failure in that batch. In that case, complex rules will kick in to allocate the failure amongst the parties participating in the batch, the detail of which I am glad to say does not matter for my purposes, but the broad aim of which will be to maximise the degree to which the settlement batch still processes, i.e. to minimise the extent and therefore the impact of the settlement failure.

In case it matters for the Main Trial, I should be clear that in paragraph 121 above, the reference to cash payment and securities delivery instructions that had been ‘accepted for settlement in a given settlement batch’ was deliberate. I am not making now any finding as to what would be required for that under DVP Model 3, or at VP Securities in particular. If any of the parties considers that the evidence on this trial was sufficient for such a finding to be made, and that it would assist for it to be made, I would be content to entertain submissions upon this judgment that invited me to supplement it in that way.

Stock loans are commonly used for a number of purposes, including to help sellers of securities settle their trades. A seller who will otherwise lack the securities needed to settle a transaction may be able to borrow the necessary quantity of securities to deliver them to the buyer. This may be planned, because the seller was deliberately selling short, or accidental, as where the seller sells on the basis of a purchase but their seller defaults.

The market for stock loans was and is chiefly the preserve of large, well-capitalised financial institutions, pension plans and investment managers. Smaller market participants would typically access stock lending, if at all, indirectly through larger prime brokers, entering into a Global Master Securities Lending Agreement (“GMSLA”) with them on standardised terms settled by the International Securities Lending Association.

The terms of a typical stock loan provide for a transfer of title to securities by the lender in exchange for a fee and collateral, at the start of the loan period, and a transfer of title to equivalent securities by the borrower, at the end of that period. The borrower has free use of the securities transferred to them at the start and is not obliged to return them in specie at the end – nor would that be a meaningful notion for dematerialised and fungible securities unless (possibly, but which would rather defeat the purpose of the loan) the borrower in fact held them, under a fully segregated custody chain, doing nothing with them, for the duration of the loan.

As will be clear from what I have said already, the use of stock lending which will be of relevance in this case is to facilitate short selling. The classic short selling strategy, unrelated to dividend taxation or dividend arbitrage, is that of a seller looking to profit from what they predict is going to be a fall in a share price, by borrowing and selling today, with a view to buying back after the price has fallen to cover the stock loan ‘redelivery’ obligation, making a capital gain of the price difference less borrowing costs (and any other transaction costs).

Different jurisdictions regulate short selling in different ways, including outright prohibition, permission in certain instances, and universal permission. In the period with which this litigation is concerned, short selling in the EU was permitted so long as it complied with the Short Selling Regulation. In opening, Mr Jones KC suggested that in closing he would be inviting the court to say that a short sale that was permitted (so as not to be unlawful) by Article 12.1 of the Short Selling Regulation, was necessarily effective, from the moment it was concluded, to make the buyer a shareholder for the purpose of Danish tax law.

In the event, I did not detect that invitation in closing, for good reason. The Short Selling Regulation serves to regulate the lawfulness of short selling. It neither provides nor assumes anything about whether and if so when a buyer from a short seller trading lawfully, or for that matter unlawfully, from the perspective of the Regulation, is or should be treated as a shareholder for the purposes of national dividend tax rules.

It may be the Sanjay Shah Defendants will wish to say at the Main Trial that no transaction in which they had any involvement (if real and not sham) was or resulted in an unlawful short sale contrary to the Short Selling Regulation. If (as to which I have formed no view) that might be relevant to the issues at the Main Trial, so be it; and if it is both relevant and disputed by SKAT, it may need to be explored then, enabling findings to be made as part of a Main Trial judgment. However, it has no bearing on the Validity Issues, which concern the content of Danish tax law as regards Danish dividend tax and claims for refunds thereof.

Subject to the impact of any illegality if parties concluded a prohibited trade (for example, an uncovered short sale in breach of the Short Selling Regulation), trading parties have freedom of contract. On-exchange trading will be on standard terms, set by the exchange and consented to by trading there. However, parties trading with each other OTC might, in principle, contract on whatever terms they were content to agree. That might result in transaction structures utilising, in combination, a number of common types of transaction, potentially in novel or unusual ways. The effect of doing so, if the parties were not engaging in sham trading, and for that matter any question whether they were engaging in sham trading, would need to be assessed on the facts and circumstances of any given instance.

No party has suggested that the transaction structures upon the basis of which the dividend tax refund claims were put forward that SKAT now impugns were an accepted or standard form of trading, let alone that there was any market understanding as to their legal effects, either generally or as regards Danish tax law in particular. To the extent the DWF Defendants’ Validity Trial pleading might have appeared to make such a suggestion, it was withdrawn by Mr Onslow KC at the trial. For completeness, I acknowledge again the submission by Messrs Jain and Godson that they were only ever involved in the placing of standard share purchase orders on ordinary terms, for clearance and settlement by a UK regulated clearing and custody broker. That is disputed by SKAT, but that is a dispute for the Main Trial. If it is right, however, it would mean that Messrs Jain and Godson had involvement only in recognised or standard forms of trading, but that does not contradict what I said at the start of this paragraph, as it would mean that Messrs Jain and Godson, contrary to SKAT’s case on the facts, had not engaged in the sorts of complex transaction structuring that SKAT says is problematic.

To say that a transaction is on ‘cum-dividend’ terms (likewise ‘cum-div’, or simply ‘cum’), or that a price is a ‘cum-dividend’ (or similar) price, is to say, with reference to a given dividend, that under those terms the buyer is to receive the benefit of that dividend, respectively that the transaction has been priced on that basis. The contrast is with ‘ex-dividend’, ‘ex-div’, or ‘ex’, terms or pricing. The fixed terms for on-exchange spot purchases will include a cum/ex cut-off point. Any such purchase concluded prior to the cut-off will be a cum-div trade, any such purchase concluded after the cut-off will be an ex-div trade.

For Danish shares, the cut-off point was the market close on the dividend declaration date (17:00 hrs (Danish time), when the exchange closed). An ordinary on-exchange spot purchase prior to that point would be cum-div and would necessarily be for settlement (T+2) on or before the record date. That coincidence was by design. An ordinary on-exchange spot purchase after that point would be ex-div and would necessarily be for settlement after the record date. That was also by design.

Having all of that in mind, the ‘ex-date’ for a security, in relation to a given dividend, is the first exchange trading date on which a standard on-exchange spot purchase will be on ex-div terms. The ex-date for Danish shares was D+1, the first business day on the Danish exchange after the dividend declaration date for the dividend in question.

A ‘cum-ex’ trade, then, is a sale traded before the ex-date but for (non-standard) settlement after the record date. It will normally be on cum-div terms, as a matter of contract, meaning the seller will promise the buyer the economic benefit of the dividend in question; but if it is physically settled at all, on its settlement date, it will necessarily be with shares that the market would understand to be, by then, ex-div shares, i.e. shares that do not carry with them any right to that dividend. Any payment by the seller to the buyer referable to the dividend will necessarily be a ‘manufactured dividend’, and market participants would see it that way. As I explain below, the market understanding in that regard accords with the legal position under Danish securities law. Whether Danish dividend tax rules meant that nonetheless the buyer would or might be treated as having been the party entitled to the real dividend, so as to be liable to tax on it, is a matter of Danish tax law, not a matter of market practice or understanding.

An ordinary share investor, looking to invest for dividend yield and/or capital growth (share price increase), and in a position to fund a share acquisition, would have no economic interest in buying on cum-ex terms. If they were a tax-exempt US pension plan (say), one element of an investment decision to put their money into shares in a Danish company might be an assessment of likely dividend yield, bearing in mind the entitlement to reclaim dividend tax collected at source by SKAT. If anything, that would logically represent an anticipated slight reduction in likely yield, because of transaction costs or fees that might be incurred in making the tax refund claim. There would be no reason to agree to a possibly complicating or controversial deferred settlement date.

The cum-ex trade, therefore, with any accompanying suite of other transactions by which it was implemented, was likely to be of interest only to those who sought to generate a dividend tax refund claim in the cum-ex buyer although the seller had no shares to sell either when entering into the trade or on the dividend declaration date, and without needing either seller or buyer to be in a position to purchase on ordinary terms the volume of shares by reference to which the dividend tax refund claim was to be made. That is the sense in which Mr Onslow KC acknowledged that, if they could be effective to produce valid dividend tax refund claims, depending on their precise terms, short cum-ex trading strategies were ways of generating ‘free money’.

I noted in paragraph 51 above that several different types of matter were collected for this trial, perhaps imprecisely, under the label of ‘market practice’. Having just summarised various such matters that were not contentious, it is convenient to deal now with the limited number of primary ‘market practice’ disputes between the parties. The matters of market fact pleaded by defendants but disputed by SKAT were, and my findings in relation to them are, as follows.

It was pleaded that it was not possible to identify, follow or trace the ownership of a single share from a CSD to a particular shareholder. The debate at trial, and the cross-examination of the experts, about ‘traceability’ was confused by a lack of clarity at times over what was being debated. My conclusion, as will be seen below, is that as a matter of Danish company law, all shareholdings are by nature traceable to a holding at VP Securities, however easy or not it might be for any given shareholder to know or discover the full custody chain pursuant to which their holding was constituted. That is the starting point, of the general law, from which any suggestion that Danish tax law adopts a different approach or creates a different effect falls to be considered.

To the extent that this was intended by the allegation, I agree (and I am not sure that SKAT disputed) that shareholders other than direct account holders at VP Securities, holding for their own account, might well not know their full custody chain, or find it easy to discover. But I do not consider the evidence enabled me to find that it was, or would generally be, an impossibility, if for some reason evidencing that full custody chain was important.

Three further aspects of the slightly phoney war over traceability were:

There was then a more directly contentious series of pleas supposedly of ‘market practice’ that were either no such thing, or were not made good on the evidence.

Firstly, it was alleged that under a trade for the purchase of dematerialised shares executed prior to a dividend ex-date, the buyer “became entitled to the receipt of dividends as a matter of contract law, and as a matter of well-established market practice, as reflected in SKAT’s legal guide”. It is trite to say that the rights a buyer under any given share trade was promised, or purportedly granted, would depend on the terms of the contract. A standard spot sale prior to the ex-date, timely settled by a transfer of shares to the buyer, would necessarily complete on or before the record date, and so would entitle the buyer to the dividend receivable (subject to the timing and therefore the impact of any on-sale by the buyer). If the settlement date was also prior to the ex-date, then that completion would mean that the buyer would acquire the dividend right upon its accrual and not merely the accrued receivable, likewise subject to the impact of any on-sale.

Below, I accept Mr Wade’s evidence that the general market understanding was that by receiving a shareholding, completing a purchase, on or before the record date (and subject to the impact of any on-sale) a buyer would receive a real dividend, not a manufactured dividend, and I find that to match the legal analysis under Danish securities law. But the general market understanding would also be that whether that means the buyer was liable to Danish dividend tax or (therefore) could be entitled to a tax refund from SKAT would be a question of Danish tax law and not a matter of market practice.

If the trade were on non-standard settlement terms such that although traded before the ex-date, settlement was after the record date, the general market understanding would be that only a manufactured dividend would be generated for the buyer (if the contract provided for a dividend-related payment), which again matches the analysis of Danish securities law. As before, any question whether in those circumstances the buyer might nonetheless be liable for Danish dividend tax would be understood to be a question of Danish tax law, not a question of market practice.

Secondly, it was alleged that a dividend payment made by a custodian to an investor, whether foreign or domestic (relative to the company in question), would be in a net amount reflecting a deduction for the maximum rate of withholding tax. That is not quite right. More accurately, and focusing on Danish shares since they are my only interest, VP Securities would receive, and trigger a distribution up the custody chain of, a dividend payment made by the company. That distribution would be to the shareholders of record at the end of the record date. The amount that would therefore come to be distributed, as a dividend payment, in proportion to shareholdings as they then stood (totalling 100%) at the head of all custody chains, would be the amount paid by the company to VP Securities. It would therefore be an amount net of a dividend tax amount if and to the extent that the company withheld such an amount. There was no evidence that if the company did not withhold any such amount (rightly or wrongly, as regards its obligations under the Danish Withholding Tax Act), VP Securities would then do so. If the company paid VP Securities without any dividend tax withholding, I envisage the distribution would likewise be gross.

Thirdly, it was alleged that investors have no directly enforceable rights to dividends against the company. That is not a question of market practice.

Fourthly, and said to be related to the previous point, it was alleged that any dividend payments received by investors would be received by credit entry in an account with their custodian calculated on a net basis, that is to say net of the maximum rate of withholding tax. I agree that a dividend-related payment to an end investor would ordinarily be effected by and reflected in an account entry at their custodian, be it payment of a real dividend or payment of a manufactured dividend. Beyond that, I cannot agree with the generalisation pleaded. If what is being paid is a real dividend, the amount will be in proportion to the amount paid to VP Securities by the company, so it will be net of any dividend tax withholding to the extent, but not otherwise, that the company effected such a withholding. If what is being paid is a manufactured dividend, the amount will depend on the terms of the contract under which it is being paid.

Fifthly, it was alleged that an investor had no need to trace particular receipts to particular shares, or to any payment made by the company, that a custodian’s obligation to credit a dividend was the same whether the client was buying from a short seller or from a seller with shares to sell at the time when the trade was executed, and that all such credits to investors by custodians were treated by market participants as dividends. The problem with all of that is that it glosses over the fact that the market understood there to be a distinction between real and manufactured dividends, and appreciated that what counted as a dividend, for the purpose of Danish dividend tax and possible claims for refunds thereof, would be a matter of Danish tax law, not a matter of market practice. As anyone in the market would have appreciated, it was not and is not a matter of market practice to say whether a real dividend payment would necessarily be a payment to a party liable to Danish dividend tax (answer, not so), or whether a manufactured dividend payment would necessarily be a payment to a party not liable to Danish dividend tax (answer also, not so).

The assertion that an investor had no need to trace particular receipts to particular shares, or to a payment by the company, is too general to be capable of meaningful finding one way or the other. It is irrelevant to the Danish dividend tax position, since the tax liability attached to the accrual of dividend entitlements, not to the receipt of payments. None of that is to deny that in the ordinary case of an investor holding Danish shares, their receipt of a dividend payment through the custody chain will reflect their having been a shareholder on the record date, and on the dividend declaration date, and no complexity will arise. If a party, whether an ordinary investor or a party engaged in a cum-ex trading strategy, trades around the dividend declaration date so as to create potential for complexity or controversy, and if being certain of the Danish tax law position, in their situation, would or might require some species of ‘tracing’ exercise (whether of shareholdings, or of contracts, or of payments), that would not be reason to modify the rule of Danish tax law otherwise applicable.

The requirements of a valid claim for a repayment of Danish dividend tax were a matter of Danish tax law, under s.69B(1) of the Danish Withholding Tax Act, the key being (it will be seen, below) determining who were treated by Danish tax law as shareholders in the application of s.16A(2)(1) of the Danish Tax Assessment Act. As I have already found, under Danish law, tax law follows general civil law, for example contract law, company law or property law, where it operates by reference to legal concepts or terms created by or known to that general law; and as I have already noted, it was common ground that in relation to Danish dividend tax, Danish tax law departs to at least some extent from the general law. The essential issue at this trial was how far that departure extended at the material time.

Since the general law (in relevant respect, Danish company law) provides the starting point, I start with a consideration of its relevant meaning and effect. As I said when introducing the expert witnesses, there was hardly any real difference between the views of Mr Aasmul-Olsen and Ms Tholstrup. A substantial part of Ms Tholstrup’s main expert report, and correspondingly a significant proportion of her and Mr Aasmul-Olsen’s joint memorandum, to the extent it recorded and explained differences of view, was devoted to Ms Tholstrup’s notion of the ‘contractual ownership’ of a share, and its implications. It was apparent, however, that although she used the language of ‘ownership’, Ms Tholstrup was describing contractual rights between parties to a share sale that have no impact on the company. As she made clear in the joint memorandum, she accepted that:

Ms Tholstrup further accepted – again, as was clear from the joint memorandum – that under Danish law, a transfer of ownership of dematerialised and fungible shares requires the individualisation (identification and segregation) of ‘the shares’ being transferred that in general would occur only on delivery. She expressed the view, with which Mr Aasmul-Olsen did not agree, that: “No such requirement applies to the bilateral relation between seller and buyer.” This again was only a view concerning what could be the content of a contractual right of a share buyer against their seller, not having any impact on the company and thus not amounting to or having any bearing upon the location of share ownership from time to time.

Before Mr Aasmul-Olsen and Ms Tholstrup gave oral evidence, Mr Onslow KC for the DWF Defendants confirmed that his clients would not pursue “the “contractual ownership” versus “segregated ownership” argument”, on the basis that “determination of that issue cannot realistically affect the outcome of any of the issues for decision at this Validity Trial”. As Mr Onslow KC acknowledged, this amounted to or involved an acceptance that under Danish law, leaving any special tax rule aside, in general a buyer (or borrower) of fungible, dematerialised shares in a Danish company becomes a shareholder only if and when their purchase (or stock loan) successfully completes with a share transfer. (There is no suggestion that when the facts are considered at the Main Trial, there will be any case of an individualisation of shares potentially capable of having transferred ownership under the general law prior to a share sale (or stock loan) settlement date; so that general rule will be the applicable rule in the case, as I understand it.)

Mr Onslow KC further confirmed that although there is a special rule in Danish tax law that can mean that a buyer of shares is treated as having become a shareholder, for tax purposes, from the moment they conclude their purchase contract, his non-reliance on Ms Tholstrup’s ‘contractual ownership’ analysis amounted to or involved an acceptance that it was not relevant to any differences of view between the tax law experts about the scope, effect or implications of that tax law rule.

The material remaining difference of view between Mr Aasmul-Olsen and Ms Tholstrup concerned the accrual of a shareholder’s right to a dividend declared by the company.

So as to be clear about my use of terminology, it is convenient to introduce at this point a simplified shareholding diagram illustrating the holding of shares via custodians, leading ultimately to VP Securities:

Illustrative Shareholding Diagram

The diagram should be understood to depict the state of the share custody record at a given moment in time, at which moment:

Assuming accurate record-keeping all round:

In practice, the number of shareholders might be huge; likewise the number of custodianship chains from shareholders equivalent to S1 to S10, leading to VP Securities; and any given custodianship chain might have many more links between the shareholder in question and VP Securities than one (for S5 to S8) or two (for S9 and S10) as shown in my diagram. However, by definition any shareholder either will hold their shares directly with VP Securities (like S1 to S4) or will sit at the end of a matched custodianship chain, as illustrated by:

In that sense, any shareholder will have rights that are traceable to the company as the share issuer. In my diagram, S9 is a shareholder because, and only because, C2 holds 20% of the shares in DanCo (⅔ of its 30% shareholding) as nominee for C3 and C3 holds 10% (½ of its 20% holding) as nominee for S9. Likewise S10.

I should mention one potential wrinkle concerning stock loans. As I have noted several times already, where shares are delivered to the borrower under a stock loan on normal terms, the transfer to the borrower will be a full title transfer. If the borrower has not yet delivered those shares to another (often that will be to a buyer because the borrower is a short seller, borrowing so as to complete the short sale), the borrower and not the lender should be shown as holding the shares. There was some evidence of a practice in the Belgian custody market, to which I return much later in this judgment, of showing both lender and borrower as simultaneous shareholders of the quantity lent. If that practice were being followed, then the additional information that the holding was the subject of a stock loan would be needed in order to resolve which of the lender and the borrower was the shareholder. For completeness, therefore, as to my illustrative diagram, I should make clear that it assumes lenders under stock loans where delivery has occurred are either not shown, or have been eliminated from, the custody record.

When, therefore, I refer to being a shareholder ‘of record’ or ‘on the record’ at a given time, I mean being the equivalent of my S1 to S10. Although their rights may in practice have to be exercised through their custody chain, and thus through the series of bilateral contractual arrangements by which their custody chain has its existence, they will have a complete custody chain leading back to VP Securities, and thus the company in question, and their rights are, in concept, rights against the company, held for them by that custody chain, and not merely contractual rights against their head custodian.

I shall refer to a shareholder of record at the end (close of market) of a dividend record date as a ‘record date shareholder’, and a shareholder of record at the end (close of market) of a dividend declaration date as a ‘dividend date shareholder’. The record date shareholders are the shareholders to whom the declared dividend is to be distributed by VP Securities for the company.

That brings me back to the difference of view between Mr Aasmul-Olsen and Ms Tholstrup in relation to the accrual of dividend rights. They were agreed that the obligation on the company to pay a declared dividend accrues there and then, upon the dividend being declared. There is a separate point whether that means at the exact time when the dividend is declared, and I come back to that below.

Whatever one makes of that timing point, Mr Aasmul-Olsen and Ms Tholstrup differed (or may have – Ms Tholstrup’s explanations were not all easy to follow) on the nature, and therefore the time of accrual, of the corresponding right granted to shareholders to a share of the declared dividend.

Mr Aasmul-Olsen’s view was that rights to a declared dividend accrue to the company’s shareholders when the dividend is declared. He acknowledged that the accrued right (as he has it) is then a right to receive payment on the business day after the record date, which is to be made to the record date shareholders. That means a transfer of shares completed at any time up to the end of the record date is a transfer of cum-div shares, i.e. a transfer of a shareholding to the new shareholder that carries with it the accrued right to the dividend in question; whereas a transfer of shares completed later than that will be a transfer of ex-div shares. Mr Aasmul-Olsen’s analysis is that the transfer of cum-div shares, after the dividend has been declared, involves the transfer of a receivable (so far as the dividend is concerned) and does not mean that the accrual of the dividend right is somehow deferred.

I note that this is a transfer of a receivable, if that is the correct characterisation, that goes with the transfer of a share. It is not a freestanding transfer, divorced from the ownership of the share. Nor is it a right against the company the transfer of which, with the transfer of the share, can be avoided, if the share transfer is completed in time, or a right against the company that can be passed by a share transfer not completed in time. My focus is the creation of, and possible transfer between successive shareholders of, rights against the company. The terms of a share sale may purport or promise to grant such rights, or may include pricing or other financial terms that expressly or implicitly seek to put the parties in an economic position as if such rights had been or were granted to the buyer. That is a different matter. Such contractual rights do not themselves affect the company; they cannot create share ownership, or a right against the company to a declared dividend as an adjunct to share ownership.

For Ms Tholstrup, the right to a declared dividend being a right to share in a distribution by the company, and the distribution occurring by a payment made to the record date shareholders following the record date, when a dividend is declared shareholders obtain either no rights corresponding to the company’s obligation to pay the declared dividend, or only rights conditional on their remaining shareholders until at least the end of the record date. I found Ms Tholstrup’s explanations somewhat difficult to follow in places, and not always consistent. They included in particular the convolution that, strictly, the parties entitled to a dividend are the shareholders at the time the dividend is distributed (i.e., as she had it, when payment is made in the case of an ordinary cash dividend paid on time), yet the company obtains a good discharge against those shareholders as long as it has paid the record date shareholders. That is why I said the condition in her conditional right thesis was that the shareholders on the dividend declaration date must remain shareholders until ‘at least’ the end of the record date.

I do not accept Ms Tholstrup’s evidence, if and to the extent that she meant to suggest that between the dividend declaration and the end of the record date (or, possibly, until payment of the dividend), there is an inchoate obligation owed by the company declaring a dividend, but not owed to anyone. I prefer and accept Mr Aasmul-Olsen’s evidence that the right to the declared dividend, owed to the company’s then current shareholders, accrues along with the obligation on the company to pay it, when the dividend is declared.

In the context of this litigation, I regard it then as academic whether the fact that the accrued right to a declared dividend will transfer with the share to which it is attached, if the share transfer is completed prior to the close of the market on the record date, means that the right to a dividend should be characterised as conditional, as Ms Tholstrup would have it.

I turn to the question of when precisely a dividend is declared. As Mr Aasmul-Olsen explained in his written evidence, “One significant market convention relevant for the Dispute is the convention that shares in Danish companies are sold cum-dividends up to and including the close of market on the day that a dividend is declared.” There is a simple, deliberate harmony between the settlement cycle for on-exchange spot sales, the record date for declared dividends, and the set terms for on-exchange spot sales whereby any sale concluded before the market closes on the dividend declaration date is a cum-div trade in respect of that dividend, the next trading day being the ex-date.

Other things being equal (which includes an assumption that the buyer has no further transaction that might affect the analysis), a timely settlement of an on-exchange spot sale concluded on the dividend declaration date will make the buyer a record date shareholder, to whom therefore a dividend will be payable by the company, and to whom the appropriate share of the dividend payment made to VP Securities will find its way through the custody chain at the end of which, ex hypothesi, the buyer then sat. That is as it should be – since the sale called for settlement on the record date and was timely settled, it was a sale to be completed and in fact completed by the delivery of cum-div shares, i.e. by a share transfer carrying with it an accrued dividend right.

That will be so although, again ex hypothesi, the buyer will not have been a shareholder when the dividend was declared so as to have acquired a right to a dividend when those rights accrued; and their seller may or may not have been such a shareholder either, because the seller could have been a short seller who completed by delivering shares acquired after the dividend declaration date (for example, by taking delivery under a stock loan, with delivery on the ex-date, or for that matter on the record date).

To complete the deliberately harmonious structuring of ordinary on-exchange spot sales, and by contrast to a sale concluded on the dividend declaration date, an on-exchange spot sale concluded after that date will be a sale the completion of which entitles the seller to deliver and requires only the delivery of ex-div shares – a share transfer that cannot carry with it an accrued dividend right, since it will occur after the end of the record date.

For the legal analysis under the general Danish law, what matters is that in that deliberately harmonious set-up, the entirety of the final exchange trading day on which a dividend is declared is treated as a ‘cum date’. That date is not treated as lasting only until the moment in time when (if during exchange trading hours) the dividend was in fact declared. In Mr Aasmul-Olsen’s view, the impact of that market convention is that the dividend is treated as having been declared at the moment of the close of the market, normally 17:00 hrs (Danish time), on the dividend declaration date, whenever precisely, be it before or after the market closed, the dividend resolution was passed at the AGM (or as the case may be for dividends adopted in some other way). Ms Tholstrup agreed, describing the market convention as “almost a perfect example of something that could lead to the rule of customary law” (the sense of which was “… could give rise to a rule of customary law”).

My conclusion and finding, therefore, is that under the relevant general Danish law, that is to say Danish company law, the right to a declared dividend accrues to the dividend date shareholders. Between the close of the market on the dividend declaration date, when rights to a dividend accrue, and the close of the market on the record date, a transfer of shares in the company will carry with it that accrued dividend right. After the close of the market on the record date, a transfer of shares in the company will no longer carry with it any dividend right. The terms of any contract pursuant to which such a transfer occurs may have the effect, as between the contracting parties, that in one way or another the transferor must put the transferee in the same financial position as if cum-div shares had been transferred. But that has no impact on the rights and obligations of the company; it does not mean that a dividend right is transferred.

The market practice experts agreed as a matter of general market terminology that a cum-ex trade is a trade concluded prior to the ex-date for the security in question, for settlement after the record date. A cum-ex trade in Danish shares is therefore a trade that, if performed in accordance with its terms, necessarily will not transfer to the buyer any dividend right. That is so even if the terms of the trade require physical settlement and the trade is settled in accordance with those terms, in other words a share transfer is required and does occur, because ex hypothesi that will be a share transfer after the record date. Any financial adjustment between the parties, whether built into the pricing or by way of separate credit or payment by the seller to the buyer, reflecting an intention that the buyer have the economic benefit of the dividend in question, is necessarily a matter of contract only, having no impact on the rights and obligations of the company; it is not a dividend and it does not mean that any dividend right is transferred to the buyer.

The market practice experts were also agreed that the market recognised a distinction between a ‘real’ and a ‘manufactured’ dividend, with the latter also referred to as a payment in lieu or compensation payment, the former being a right against the company, the latter being only a contractual right between trading counterparties. A market participant using the word ‘dividend’, without qualifier, might intend thereby to include or to exclude manufactured dividends, depending on the context. For the market practice experts, a real dividend means “a distribution from a share issuer that is paid to the shareholder of record on the Record Date either directly or through a chain of custody”, while a manufactured dividend is a “Contractual payment representative of a dividend arising under a contract for the sale or transfer or securities, which is compensation for a dividend foregone.” In that definition, a payment by a cum-ex seller to their buyer referable to the dividend is necessarily a manufactured dividend – the buyer has foregone the real dividend by agreeing to settlement after the record date, so that if they receive a share transfer at all it is bound to be a transfer of ex-div shares, not a transfer of cum-div shares.

I accept Mr Wade’s further evidence as to market understanding (which sits with the agreed evidence I have just summarised, and in respect of which neither Mr Sharma nor Dr Collier gave contrary evidence), to the effect that:

My conclusions as to the accrual and transferability of the right to dividends under Danish company law (paragraph 179 above) mean that the general market understanding just described was congruent with the legal analysis under the general law in Denmark.

The Danish tax law experts were agreed that under Danish law, taxation operates on an accruals basis. There is nothing unusual or exotic about that, it might be thought. Applied to dividends, it need not mean anything more than that the accrual of a right to a dividend is the subject matter of dividend tax, not the receipt of cash or other value by virtue of that right. In view of my conclusions as to the accrual and transferability of rights to dividends, it is easy to envisage a choice for the legislature, as a matter of tax policy, whether to levy dividend tax from the dividend date shareholders or from the record date shareholders. Subject to that policy choice, there might seem to be no reason why tax law should have to depart from the general law in identifying the parties to whom rights to dividends are treated as accruing, viz. (as I have held) the dividend date shareholders.

As I stated above, the most basic principle of Danish tax law is that under s.43 of the Danish Constitution, taxes may only be imposed by primary legislation; and the Validity Issues that I am required to determine now are ultimately all questions of the proper construction of Danish primary tax legislation. The relevant statutory provisions at the material time were the following:

The definition of dividends for tax purposes as anything distributed by a company to its shareholders ensured that disguised dividends (distributions by a company to its members that are dividends in substance though not declared as such) are taxed as dividends for tax purposes. If that is to be viewed as defining dividends differently in tax law than in company law, it is no more than an application of the plain meaning of the language of the tax statute and is unremarkable as an expression of tax policy.

The Danish tax law experts were agreed that nothing turns on the difference of language as between corporations and natural persons (in translation, they who “receive” dividends and they who “obtain” dividends); the meaning and effect is the same. That language (either word), read with the definition of dividend as anything “distributed”, might perhaps be taken to mean that the dividend tax event is the dividend distribution (ordinarily, that will be a cash payment) made by the company following a dividend declaration.

However, that would fail to give effect to the clear provision in s.66 of the Danish Withholding Tax Act that dividend tax is due upon the adoption by the company of a resolution to declare the dividend. Furthermore, the preparatory work to the amendment to s.16A(2)(1) in 2012 by which the word “current” was added, explained as follows (and the experts were agreed that preparatory works are important, admissible aids to the interpretation of Danish statutes):

Bearing in mind those provisions and also the principle that taxation in Denmark is on an accruals basis, the Danish tax law experts were agreed that the taxable event for ordinary dividends declared by a company is the declaration of the dividend, not the cash distribution that follows. The right to share in that distribution accrues there and then, when the dividend is declared, in the eyes of Danish tax law, which accords with the general law in Denmark (i.e. Danish company law), on my prior findings in relation to that.

In his evidence, Prof Laursen focused on the exact moment when a dividend is declared. However, he was not asked to consider the impact of the convention (as I have found it to be) under Danish securities law that that moment is taken to be the close of the market on the dividend declaration date. Nothing in Prof Laursen’s evidence, in my judgment, provided reason why that should not be followed by Danish tax law. In those circumstances, in my judgment it was not established that Danish tax law provides for a different rule than that of the general law as to when, on the dividend declaration date, the right to a dividend accrues (that is to say, strictly, is by convention treated as accruing).

That means that a dividend date shareholder is liable to dividend tax on the dividend in question, subject to the tax twist that follows, below, even if they sell their shareholding the next day (D+1), and even if they do so OTC with immediate or next-day settlement (T+0 or T+1) such that their buyer will be the record date shareholder and they will not. In that case the buyer will acquire the accrued right to the dividend, as it will come with the share transfer that is completed in time for the buyer to be the record date shareholder; and any contractual payment by the buyer to the seller referable to the dividend will be a manufactured dividend. The seller will be liable to dividend tax nonetheless; but that is not a tax on the manufactured dividend, it is a tax on the accrual to the seller, not to the buyer, on the dividend declaration date, of the (right to the) real dividend.

So far, so good, and nothing I have said about the position under Danish tax law was contentious between the parties, except for the deemed timing of the accrual of rights on the dividend declaration date, on which I am with the defendants. The next element of the analysis was also, up to a point, non-contentious. Its true nature, and in consequence the nature and extent of its implications, however, was contentious, and was the essential point of dispute requiring to be resolved by the court on this trial.

The next element is that Danish tax law, the experts agreed, took the notion of taxation on an accruals basis – taxing the acquisition of rights, not the receipt of payments – a stage further than simply insisting that the dividend declaration, not the receipt of a dividend payment, is the taxable event. The experts all hold that where a dividend date shareholder had entered into a final and binding contract, prior to the accrual of a right to the dividend in question, to sell the shareholding to which that accrued right attached, then for tax purposes, other things being equal, the buyer fell to be treated as the rightful recipient of the dividend entitlement liable to dividend tax. The selling shareholder, the experts agreed, to whom the dividend entitlement in fact accrued, would fall to be treated for tax purposes as having disposed of their shareholding before the accrual so as not to be liable to dividend tax on that entitlement. I shall refer to this extended notion of taxing the acquisition of rights as the ‘contract accruals rule’.

I confess to a nagging doubt whether there must be such a rule. There is nothing in the language of the applicable Danish legislation (at any rate, reading them in English translation), or in the principle of taxing on an accruals basis, that requires a tax fiction about the parties to whom dividend entitlements accrue, which is what the contract accruals rule creates, whether it goes only as far as SKAT contended, or further than that, as the defendants contended. Indeed, at first sight, it might perhaps be thought that the legislative provisions simply give effect to an accruals basis for dividend tax by providing (when read in the light of the convention as to when a dividend is treated as being declared) that the liability to tax is that of the dividend date shareholders, to whom the dividend right accrues, rather than that of the record date shareholders (if different, due to share transfers after the dividend declaration date) to whom the resulting receivable is payable by the company. Nor did the experts cite any clear source in Danish case law or academic writing for this tax law fiction as to dividend accruals.

Prof Laursen explained that the accruals, or acquisition of rights, principle, is that “the relevant time by reference to which income is taxed (i.e. the income is considered realised or earned for tax purposes) is when the taxpayer acquires a final right to the income. In other words, pursuant to the acquisition of rights principle, the taxable event is not the payment or receipt of income. Rather, the taxable event is the acquisition of a final right to income. It follows that the actual cash settlement of the dividend is not decisive with respect to timing for taxation purposes.” Nothing in that articulation of the principle requires that a party to whom the relevant right to income did not accrue, i.e. (here) the right to a dividend, should be regarded for tax purposes as having acquired that right; or to require that a party to whom that right to income did accrue should be regarded for tax purposes as not having acquired it.

The experts cited and discussed Danish case law considering whether, for tax purposes, certain transactions fell to be characterised as derivatives (financial contracts) rather than share sales, or the capital gains tax treatment of certain particular situations, in which the date on which the final and binding contract was concluded, pursuant to which shares were acquired or disposed of, had a decisive significance as to when a gain or loss should be treated as having been realised or incurred. It did not seem to me that they necessarily had any impact on the question of the incidence of liability to dividend tax, with taxation on an accruals basis.

Further, the doctrine espoused by Prof Laursen, and to at least this extent not challenged by the defendants, has a striking consequence. For it holds that the contract does not have to be a contract that provides for the buyer ever to acquire the dividend entitlement upon which it is said they will be taxed. Thus, for example:

I should be clear that in the above examples, I am assuming contracts that on their terms may only be settled, and that do in fact settle, by a share transfer to the buyer on the settlement date. I accept Prof Laursen’s evidence that if those conditions were not both satisfied, then either the contract will have been, or will fall to be treated as having been, a financial contract, in which case no question of dividend tax liability would arise, or there will have been no share transfer at settlement, and then any tax consequence that might otherwise have attached to the contract on the basis that it provided for a share transfer will be reversed.

I turn next to the tax treatment of stock loans, since I find that, come what may, Danish tax law and Danish company law do there identify differently, in some circumstances, the party to whom a dividend entitlement accrues. Since a stock loan on normal terms provides for a share transfer, with full title transfer to the borrower, at the start of the ‘loan’ period, it requires and involves exactly the same performance, on the securities side of the transaction, by the lender at the start of the ‘loan’ period, as that of a share seller completing a sale; likewise, the share transfer required of the borrower at the end of the ‘loan’ period.

If and to the extent that the shareholding on which (under the general law) a dividend date shareholder accrues a right to a dividend is the subject of a current stock loan (where the dividend date shareholder is the borrower), then on the Danish tax law expert evidence I find that the dividend date shareholder’s lender, and not the dividend date shareholder, is the party liable to dividend tax.

The correctness of that proposition is not settled law in Denmark. Prof Laursen and Mr Mikelsons take the view that it is correct. Mr Bachmann declined to commit to a view, having recounted at some length how the approach of the tax authorities in Denmark has evolved (if not flip-flopped) over time. He would go no further, ultimately, than an opinion that the true legal position is uncertain. The weight of the expert evidence before the court therefore favours the tax treatment of a shareholding held by the borrower under a stock loan as being that the lender is still the shareholder for Danish dividend tax purposes, with the consequence that the lender, and not the borrower, is the party liable. That is so even though it is evident that the tax authorities’ attitude in favour of that treatment in modern times (over the last 30 years or so) has been driven by the consideration that treating the ‘delivery’ and ‘redelivery’ of shares under a stock loan as disposals for capital gains tax purposes might stymie stock lending in Danish shares, the widespread availability of stock lending being regarded, generally speaking, as desirable. On the basis of the expert evidence at this trial, that observation notwithstanding, I consider it more probable than not that ‘the lender is the shareholder’ is the rule of Danish tax law, for dividend tax purposes, where the shareholding of the shareholder of record is covered by a forward ‘redelivery’ obligation under a stock loan.

I have articulated that with some care because it must cater for the case where – assuming no other transactions that might complicate the analysis – L transfers shares to S under a stock loan and S transfers them to B under a share sale. For that case, all three Danish tax law experts agreed that B – and not either S or L – is the shareholder for dividend tax purposes at the latest from the completion by share transfer of the sale by S to B. When, if at all, that becomes true prior to that moment would take me into the contentious territory on this trial, hence ‘at the latest’ in that formulation.

Where the dividend date shareholder has forward ‘redelivery’ obligations as a stock borrower from more than one stock lender that exceed, in aggregate, their (the borrower’s) dividend date shareholding, no doubt there must be a rule, or rules, to ascertain which lender or lenders have the dividend tax liability. I am not in a position on the expert evidence to make any finding as to what that rule or those rules were, but I do not expect that will matter in the litigation.

The rule that the lender is the shareholder, as just explained, does not mean that Danish tax law imposed dividend tax liability on shareholders for tax purposes whose aggregate shareholdings for tax purposes exceeded the share issue. It only decides, as between lender, borrower and (if there is one) buyer (from borrower), of shares that the lender did have and did transfer (by way of stock loan) to the borrower, which of them is regarded as the current shareholder, for tax purposes.

The Danish dividend tax rules for the simple cases (where no other transactions are involved) of ‘L to S’ and ‘L to S to B’, where L is a shareholder, ‘L to S’ is a stock loan, and ‘S to B’ is a share sale, have relevant implications. Thus:

That brings me to Prof Laursen’s articulation of the extent of the contract accruals principle that in his view exists under Danish tax law, as it applies to a buyer of shares. It means that the principle is somewhat complex, particularly if an attempt is to be made to express the effect comprehensively in definitional form, viz. “you are a shareholder for tax purposes if and only if …”, something the Validity Issues require of me in a way, it would appear, that has not so far been demanded of the Danish courts. Thus, Prof Laursen said:

The relative complexity of the overall result notwithstanding, I am satisfied that each element of Prof Laursen’s formulation was well-considered, coherent and principled. Mr Graham KC demonstrated through his cross-examination of Mr Bachmann that there was no material difference between him and Prof Laursen on any of it. I prefer that – effectively joint – evidence to that of Mr Mikelsons, to the extent he expressed a different opinion.

Before I look in detail at the elements of Prof Laursen’s explanation, I should add that I accept, as did SKAT, that there is no Danish case law deciding, or published Danish legal source stating, that Prof Laursen’s analysis is correct. But nor is there any Danish case law deciding, or published Danish legal source stating, that Prof Laursen’s view is incorrect, or that the defendants’ contrary case is correct. It was common ground that this judgment will be the first occasion when the point has been raised and considered by a higher court (there have been Danish Tax Tribunal decisions that on one reading are supportive of Prof Laursen’s views, but they can also be read as deciding only that the tax refund applicants in question had failed to prove that they were not engaged in pro forma transactions (sham trading)).

The absence of specific, authoritative corroboration for Prof Laursen’s opinion is a less weighty consideration, in my view, than the absence of corroborative material suggesting that Danish tax law has ever recognised, or would ever recognise or by doctrine create, excess shareholdings for tax purposes, given the unequivocal agreement of the experts whose evidence carried weight on the impossibility of the latter.

It is evident, given the third qualification stated by Prof Laursen, that in his fourth qualification, when he contemplates the buyer becoming a shareholder for tax purposes at the time of delivery, that is on the assumption that there is at that time a physical delivery (share transfer) to the buyer.

The second qualification needs a little unpacking. A simple spot sale by a current shareholder does not terminate the seller’s (actual) share ownership when the contract is entered into, only when shares are transferred at settlement. But it was Prof Laursen’s evidence that all things being equal such a sale would be treated as transferring share ownership, for tax purposes, at that point (i.e. when the sale contract was concluded). In that, I note for clarity, ‘all things being equal’ involves an assumption that the sale to the buyer is the seller’s only current transaction.

Furthermore, it was not Prof Laursen’s opinion that the sale contract has to be on terms that purport to convey ownership to the buyer immediately, albeit the transfer of title under the general law will not occur until settlement. If that were his view, he could not hold the opinion he holds in relation to forward sales (see paragraph (1) above).

With those points in mind, the second qualification on any view involves this, namely that a contract the performance of which in accordance with its terms need not involve a transfer of share ownership to the buyer cannot make the buyer a shareholder for tax purposes until (if at all) it is in fact performed by a share transfer. To that extent it is just explanation or application of the notion that a final and binding right to a shareholding is required. If whatever contract the party in question may have does not entitle them to a shareholding, i.e. does not require (if performed in accordance with its terms) that they will be made a shareholder, then there is no such right.

Beyond that narrow and obvious aspect, I find it easier to consider the first and second qualifications together, which I do below, as they go hand in hand.

I take next Prof Laursen’s fourth qualification. It was not in dispute, and I find it proved. In case the detail matters on the facts at the Main Trial, I set out more precisely what the fourth qualification entails. That too, as explained by Prof Laursen, was not disputed, and it becomes apparent that this fourth point is also not so much qualification as explanation or application of the requirement of a final and binding right to a shareholding.

Under Danish tax law, a contract is prima facie a financial contract for tax purposes if it settles after the time of the final and binding agreement, at a price settled at the time of that agreement. Where such a contract is settled by delivery, then for tax purposes, by statute (s.33(1) of the Danish Capital Gains on Claims and Debt Act), “the transferred asset or liability is deemed to have been acquired or relinquished on the delivery date and at the market value on the settlement date.”

For that purpose, a sale contract providing for settlement within any customary time limit for immediate sales is treated as a contract that settles at the time of the final and binding agreement. In the case of Danish shares, that customary time is T+2 (T+3 prior to October 2014). Thus, a share sale providing for settlement at T+3 or later (T+4 or later prior to October 2014) falls within the basic definition of a financial contract to which s.33(1) of the Capital Gains on Claims and Debt Act applies.

However, s.30(1)(5) of that Act provides that a share sale is not to be considered a financial contract if the requirements of s.30(3) and (4) are met. Those sub-sections create three requirements, namely that:

The combined effect of those requirements, if met, will be that the buyer can say their contract gives them a final and binding right to be left (as a result of performance) holding shares. If any of them is not met, the buyer will be unable to say that. I note that Prof Laursen said that the third requirement was the absence of an ‘equal and opposite’ sale contract. He did not spell this out, but I understood him deliberately not to say the same about a stock loan. A ‘sale and lease-back’, were there to be such a thing in respect of Danish shares, in which S (a current shareholder) contracted to sell shares to F (a financier), for forward settlement, and to take equivalent shares from F on a stock loan, for settlement at the same time as the sale, would seem therefore to meet Prof Laursen’s third requirement (no opposing contracts). If there were no other transactions involving S or F that might need to be taken into account, such an arrangement, when it settled, would leave S as the shareholder still (as a matter of Danish company law), but F as the shareholder for Danish dividend tax purposes, by application of the Danish tax law rule that the lender is the shareholder.

In that ‘sale and lease-back’ example, therefore, so long as the first requirement for not being a financial contract for tax purposes would be met by the sale contract (S selling to F), the contract accruals principle as explained by Prof Laursen would mean that F was the shareholder for dividend tax purposes from the conclusion of the contract, and F (not S) would be the party liable to dividend tax on a dividend declared on a date between the trade date and settlement, if the trade duly settled.

Mr Sharma gave helpful evidence about net settlement procedures, including in particular the net settlement procedures operated by VP Securities in relation to Danish shares. One upshot is that no distinction will be drawn between share sale and stock loan delivery obligations. If matched share transfer orders have been accepted by the settlement system such that a party (X) is due to deliver 200 shares pursuant to a sale of shares (to Y) and is due to receive delivery of 100 shares (from Z, who is not Y) on a stock loan, for settlement at the same time, and none of X, Y or Z has any other involvement in that settlement batch, then as long as Z has 100 shares to deliver, X need only have its net delivery volume of 100 shares to avoid a settlement failure. Y will receive 200 shares at settlement, and both delivery obligations (100 shares Z to X (loan), 200 shares X to Y (sale)) can be regarded as having been performed.

If instead Z = Y (i.e. X has contracted to sell 200 shares to Y, and to take 100 shares from Y on a stock loan, for simultaneous settlement), it may be (I make no finding as to this here) that the system would not process a settlement failure if Y has no shares to deliver and X has only 100 shares to deliver, since X has a net delivery obligation of 100 shares and Y has a net delivery entitlement of 100 shares. It may be that whether that could occur turns on what is required for a delivery order to be accepted for settlement rather than what is required for an accepted order to settle rather than fail (see again, paragraph 124 above). If it could (and did) occur, though, Y would receive 100 shares at settlement, but obviously it could not be said that X had delivered 200 shares (to Y, or at all), or that Y had delivered 100 shares (to X, or at all). Rather, the effect of net settlement in that case would be that Y’s obligation to deliver 100 shares to X was offset against X’s obligation to deliver 200 shares to Y, so that there was physical settlement only of the net delivery obligation of 100 shares, X to Y. The forward delivery obligation of 100 shares, X to Y again, would still exist, of course, for settlement at the end of the stock loan period, and X would be short in respect of it; but that would be because, by agreement, Y’s delivery obligation at the start of the loan period had been satisfied by the offsetting so that, as a matter of contract between X and Y, it had been performed although there had been no delivery of shares to X. The aggregate holding of X and Y would be 100 shares immediately prior to, at, and immediately following, the settlement that involved that offsetting; it would not at any time have been 300 shares (the aggregate total of the delivery obligations that were treated as satisfied).

Returning, then, to the requirements under Danish tax law for a share sale contract not to be a financial contract to which the rule of s.33(1) of the Danish Capital Gains on Claims and Debt Act would then apply rather than any contract accruals rule, Prof Laursen’s evidence, which I accept, is that the first requirement (the contract must require physical settlement) “is not “formalistic”; if the contract calls for the delivery of shares, but the buyer nevertheless accepts a cash settlement, the requirement would not be met. Consequently, if a contract for the sale of shares meets the general requirements for being a financial contract [paragraph 216 above], and if it is settled by cash settlement, ownership to shares will have never passed … .” In the previous paragraph, therefore, share ownership for tax purposes would likewise pass, from X to Y, only in 100 shares; Y could not claim, for tax purposes, to have been a shareholder with any larger holding; and X and Y would not fall to be treated, in aggregate, as having had a larger holding than 100 shares, for tax purposes.

In that example, if Y’s holding (of 100 shares) is treated, for tax purposes, as having been acquired when the sale contract was concluded, that is matched by X’s holding (of 100 shares) being treated, for tax purposes, as having been transferred to Y at that same moment. The essential dispute before me is whether the former can be true if the latter will not be true, because the seller has no shareholding (even for tax purposes) when the sale contract is concluded. That is the subject matter of Prof Laursen’s first qualification to the contract accruals rule (paragraph 206 above), to which I come below, and indicates why it goes somewhat hand in hand with his second qualification.

Dealing next with Prof Laursen’s third qualification, I have already indicated that I accept his evidence about it (see paragraphs 198 and 207 above). The absence of any physical performance could be because the parties mutually never intended any such performance, even if they appear to have contracted for it. In that case, Danish tax law would treat the apparent trade as a ‘pro forma contract’ (i.e., in substance, a sham). Prof Laursen referred to that possibility, in order to provide a complete analysis, but it adds nothing for my purposes – an apparent, but unreal, trade documented between parties obviously would not create any final and binding rights between them.

Leaving that possibility aside, then, Prof Laursen identified two other instances where it would be relevant to the final tax law analysis whether a sale contract had been performed:

Prof Laursen continued as follows:

The upshot is that 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), because either:

It follows that agreeing to and accepting a net settlement of a share purchase could affect the incidence of liability to Danish dividend tax and, therefore, the potential ability of a buyer to make a valid tax refund claim under s.69B(1) of the Danish Withholding Tax Act. To illustrate that:

That brings me to two illustrative examples considered at trial, at my invitation.

X owns 100 shares in a Danish company that is due to declare a dividend on date D. On date D-1, X contracts separately with Y and Z to sell each of them 100 shares in the company, for settlement on D+1. X takes no step to acquire (rights to) any additional quantity. On the share settlement date, therefore, X transfers only 100 shares in total, defaulting either totally vis-à-vis one of Y or Z, or partially vis-à-vis both of them. Which of X, Y and Z is liable to dividend tax on the declared dividend, by reference to what share volume?

Prof Laursen and Mr Bachmann agreed, and I find, that the answer must be that there is a dividend tax liability in respect of a share volume of 100 shares only. If X performed with Y and defaulted totally with Z, the dividend tax liability would be that of Y. If X performed with Z and defaulted totally with Y, the dividend tax liability would be that of Z. If there were partial performance and partial default, there would be a dividend tax liability on the part of both Y and Z, in each case by reference to the share volume in fact transferred to them by X.

X owns 100 shares in a Danish company that is due to declare a dividend on date D, and X has borrowed a further 200 shares, on a stock loan, from L, under which delivery (by L to X) has occurred. On D-1, holding therefore 300 shares, X contracts separately with Y and Z to sell each of them 100 shares in the company, for settlement on D+1. X takes no step to acquire or dispose of (rights to) any additional quantity, so that on the share sale settlement date X transfers 100 shares to Y and 100 shares to Z, and is left holding a balance of 100 shares.

For that example, it was common ground that Y and Z have a dividend tax liability in respect of a shareholding of 100 shares, because they each acquired on D-1 a final and binding right to the shareholding transferred to them on D+1. It was also common ground that X and L between them have a dividend tax liability in respect of the 100 shares held by X throughout. It was suggested by the DWF Defendants that it was unclear whether that liability would be a liability on the part of X or on the part of L. I disagree. As SKAT submitted, on the Danish tax law analysis of stock lending transactions, the 100 shares held by X throughout are matched by (50% of) X’s forward delivery obligation as stock borrower. The dividend tax liability in respect of that 100-share balance of the 300 shares held by X at the start of D-1 is therefore that of L.

Crucially also, that is the limit of L’s dividend tax liability, even though it has a final and binding right to delivery at the end of the stock loan of 200 shares, not 100 shares. As regards the balance of that entitlement, X is short when the dividend is declared, and L therefore has no accrued right to a shareholding at that moment, in the sense required for it to be treated as a current shareholder then, for tax purposes, so as to incur dividend tax liability.

A key element in Prof Laursen and Mr Bachmann’s agreed analysis of Example A is their agreement that under Danish tax law, as under Danish company law, there cannot be more shares in circulation than the number of shares issued by the company, and consequently there cannot be more dividend entitlements than the declared dividend multiplied by that number of shares. As Mr Bachmann put it, for example, in his contribution to the joint memorandum: “Under Danish company and tax law, the size of the share capital and thus the number of shareholders is determined at the time of the incorporation of the company and any subsequent capital increases or reductions. Therefore, under the Danish tax law there can’t exist more shares or shareholders than the corresponding number under company law.”

Mr Mikelsons either did not in truth dispute that proposition, or did so on spurious grounds. As he summarised his view, in the joint memorandum: “There cannot be more shares, owners of shares or ownership interests in shares in circulation than the number of shares issued by the Danish company.” This was one of the opinions he suggested, when cross-examined, was a statement of the position under civil law, not under tax law. That was plainly neither the question being addressed by the Danish tax law experts nor the tenor of Mr Mikelsons’ answer to it in the joint memorandum. The experts were addressing, and giving evidence as to, what constituted ‘ownership’ of ‘shares’ in a Danish company for the purposes of the Danish Withholding Tax Act, the Danish Corporation Tax Act, and the making of a tax refund application (i.e. a refund claim under s.69B(1) of the Withholding Tax Act) during the relevant period.

In the joint memorandum, Mr Mikelsons qualified his answer as to the position under Danish tax law by asserting that there was a “generally accepted market phenomenon … that trading activity taking place along standard and accepted lines can result in claims to ownership of shares – or ownership interests – in excess of the number of shares issued by a company”. This was unsatisfactory evidence, in that Mr Mikelsons was not qualified to comment upon whether phenomena were or were not generally accepted by the market. In any event, however, a possibility that short selling (that being the relevant accepted market phenomenon) might mean that buyers who between them have contracted to purchase more than the entire issued share capital of a company might (in theory) all believe they had a final and binding contractual right to become shareholders does not mean there is excess share ownership for tax purposes.

As Mr Mikelsons said in the joint memorandum, all that could do is create a situation where there were claims to share ownership in excess of the number of shares issued by the company. Ex hypothesi, not all such claims could be valid. As Mr Bachmann said, it is then a tax control problem to ascertain in any given case where share ownership for tax purposes resided. I do not accept the inventive suggestion by Mr Jones KC in closing argument that Mr Bachmann accepted the possibility of excess share ownership for dividend tax purposes, and his references to a problem of tax control extended to the availability of anti-avoidance doctrines or amending legislation. To the contrary, in my judgment, Mr Bachmann was referring only to the practical problem, for the tax administration, of proof or disproof, created by a need to separate wheat from chaff; he was not admitting the possibility that there could be more wheat, for tax purposes, than a given dividend harvest had produced.

Necessity being the mother of invention, Mr Jones KC was forced into proposing an unrealistic interpretation of Mr Bachmann’s evidence because in truth Mr Bachmann agreed with Prof Laursen on all points that matter. That was tolerably clear from the written expert evidence, although Mr Bachmann’s supplementary report perhaps might have made one pause. It was completely clear from Mr Bachmann’s oral evidence. Mr Bachmann gave evidence, therefore, as did Prof Laursen, that entirely undermined the thesis that short selling could generate dividend entitlements, for tax purposes, in excess of the aggregate declared dividend. That left the Sanjay Shah Defendants’ attack on SKAT’s case without expert ammunition, as was most evident in Mr Jones KC’s bold invitation in closing that I should answer Example A, above, by saying that both Y and Z were shareholders for dividend tax purposes even though Mr Bachmann’s opinion, agreeing with Prof Laursen, was that that is impossible.

The axiomatic truth that Danish tax law does not recognise excess shareholding means that the “current shareholders”, a distribution to whom by the company is a dividend as defined by s.16A(2)(1) of the Danish Tax Assessment Act, must be a set of shareholders holding between them 100% of the share issue. That in turn brings me back to the first and second qualifications within Prof Laursen’s explanation of the contract accruals rule. Again it will be seen, on analysis, that Prof Laursen was not so much qualifying the rule as explaining its true nature.

That arises (explanation more than qualification) because Prof Laursen was not suggesting exceptions from, or qualifications to the applicability of, an accruals rule based on the acquisition of rights. His four qualifications were expressed to be necessary modifications to the proposition that in application of that rule, the decisive moment was the conclusion of a contract for the acquisition of a shareholding. Hence (see paragraph 206 above): the moment when a contract is concluded, he said, is only “generally” the decisive moment; and his four propositions serve to modify, because it “cannot stand alone”, the criterion of the conclusion of a contract being the moment when a relevant right is acquired.

With that in mind, Prof Laursen explained his first qualification inter alia in these terms, namely that:

Thus, Prof Laursen was not describing an exception to the acquisition of rights principle. He was explaining what must be acquired, prior to the declaration of a dividend, for there to have been a relevant accrual for dividend tax purposes, namely a final and binding right to a shareholding that the seller is then in a position to transfer.

I have found that under Danish company law the dividend entitlement accrues to the dividend date shareholders. The starting point of Danish tax law being to follow the general law, and the axiomatic truth under both the general law and tax law being that there can only ever be shareholdings equal in aggregate to 100% of the share issue, the dividend tax liability created by s.16A(2)(1) of the Danish Tax Assessment Act likewise accrues to the dividend date shareholders, except if and to the extent that a Danish tax law rule locates it elsewhere. That is why, in my view, Prof Laursen must be right to say, which became his second qualification, that a sale contract cannot, without more, make the buyer a shareholder for tax purposes unless it terminates what, absent the contract, would have been a shareholding for tax purposes of another. Anything else would be conjuring shareholdings up out of thin air.

The way the arguments were presented at trial meant that the implications were considered mostly from the perspective of an investor contracting to buy shares without having reason to concern themselves, so long as their share purchase settled properly, over whether their seller was long (sufficient at least to cover their purchase) when the trade was entered into. To see the coherence of Prof Laursen’s analysis, it may be easier to start with the dividend date shareholders illustrated by S1 to S10 in my diagram (paragraph 159 above), if that is taken for this purpose to represent the position when a dividend is deemed declared at the end of the dividend declaration date. Doing so fits exactly with Mr Bachmann’s helpful articulation of the question that, in this context, Danish tax law asks and answers by the contract accruals rule: “Since it is the current shareholders of a dividend-distribution company who are liable to tax on the distributed dividend, it is important to clarify when one shareholder’s ownership of a share ends, and another shareholder’s ownership begins ” (my emphasis).

Going back, then, to my diagram, taking it to represent the dividend declaration share register for a particular dividend, each of S1 to S10 has a dividend tax liability in respect of their 10% shareholding except to the extent that, though they were on the share register when the dividend was declared, they are treated for tax law purposes as having by then disposed of that shareholding to another. That would require that either:

As regards shareholdings that shareholders of record have committed to buyers, there will be contractual chains each starting with a shareholder of record selling or authorising sale from their long (real shareholding) position and reaching end buyers, who to that extent are the shareholders for tax purposes in place of the shareholder of record at the head of the chain, subject to the unravelling impact ex tunc of any lack of settlement by share transfer in due course. To the extent that dividend date shareholders’ shareholdings have not been committed to a sale for later settlement, the dividend date shareholders will be the shareholders for tax purposes subject to the dividend tax liability, unless, in turn, their shares are only borrowed. No sale contract divorced from any such commitment, i.e. not connected via a chain of contracts to such a commitment and therefore to a (real) current shareholding, could affect that conclusion.

Mr Bachmann confirmed that Danish tax law used a ‘FIFO’ (first in, first out) rule in the context of sales or purchases from multiple suppliers or to multiple customers. Applying that rule, to the extent required, and starting with the dividend date shareholders, there will always be a unique set of shareholders for tax purposes at the same dividend moment, accounting between them for 100% (and only 100%) of the share issue. Whether the company, or VP Securities, or SKAT, could readily identify all of those shareholders is a different point, although I envisage the answer will usually be that they could not, because there is no centralised, single depository of share trading information from which the company, VP Securities, or SKAT, could construct and verify all the contractual chains the end buyers of which would populate the set, together with dividend date shareholders (or their stock lenders) to the extent their shareholdings were not committed to sales.

I accept Prof Laursen’s evidence on all of this. Mr Bachmann agreed with it. Mr Mikelsons claimed not to agree with it, but had no coherent basis for disputing it, if he really did. Despite Mr Bachmann’s agreement with it, and Mr Mikelsons’ incoherence in refusing to agree with it, it was said I should not accept Prof Laursen’s evidence because of the practical difficulties, if it is right, in the way of establishing the parties liable to dividend tax on any given dividend.

Prof Laursen was unimpressed by the plea to impracticability, and so am I. But as a logically prior matter, I would say that:

Expanding on that last step, SKAT’s burden of proof does not mean that if it failed to persuade the court of the correctness of Prof Laursen’s analysis, the defendants’ case would be proved, to the effect that a seller with no shareholding to sell makes his buyer a shareholder for tax purposes merely by entering into a sale contract. The defendants’ case to that effect is disproved, without more, by the fact that it offends an axiomatic principle of Danish tax law (as I have found it to be, on the evidence) of 100% shareholding. It confounds the applicable tax legislation, since “current shareholders” in s.16A(2)(1) of the Danish Tax Assessment Act means shareholders for tax purposes who between them hold 100% of the share issue, and therefore the defendants’ case cannot be right. Were neither SKAT’s case nor the defendants’ case proved as to the content of a contract accruals rule under Danish tax law, the position would be that no such rule had been proved to displace the presumption under Danish law that has been proved, viz. that the tax law rule will follow the general law.

That is not my final analysis, and conclusion, however, because I am persuaded that the asserted practical difficulties do not justify the proposed conclusion that Prof Laursen’s evidence explaining the applicable tax law accruals rule is not correct and should be rejected.

The case as to impracticalities focused on the position of an investor wishing to know whether they had incurred a Danish dividend tax liability, so as to be in a position to seek a refund under s.69B(1) of the Danish Withholding Tax Act if under an applicable DTT they were either exempt from such tax or liable at most to tax at a rate below the rate at which the company had withheld tax under s.65 of the Act. It is therefore convenient to start by setting out the upshot of the main discussion, above, from such an investor’s perspective.

Firstly, an investor could be expected to know and be able to show whether they were a dividend date shareholder, or received a share transfer after the dividend declaration date to complete a share purchase they had concluded on or before that date. If neither of those was the case, they could not have been a shareholder for tax purposes when rights to the dividend in question accrued.

Secondly, an investor who was a dividend date shareholder could be expected to know and be able to show:

Thirdly, an investor who was not a dividend date shareholder, having delivered their shares by share transfer to a stock borrower on or prior to the dividend declaration date for ‘redelivery’ at a later date, would expect that they were not the party liable, since they would envisage that their borrower was borrowing so as to transfer immediately to a third party, not so as to hold the shares over the dividend declaration date. To be certain of the position, though, they would need information as to the borrower’s transactions (etc), as in paragraph (2) above.

Fourthly, an investor who became a shareholder by a share transfer only after the dividend declaration date, by the completion of a purchase concluded on or before that date, could not be certain, without more information, whether they were therefore the shareholder for tax purposes when the rights to the dividend accrued with the corresponding dividend tax liability. That would depend on how and when their seller became able to complete the sale, which could in turn require information to be obtained about the nature, timing and completion (or not) of the transactions of the seller’s counterparties, and so on.

Mr Wade’s evidence, which I accept, was that if issues around exactly when a share purchase needed to be conducted to ensure receipt of a real dividend entitlement were a concern, in particular for tax (or tax reclaim) purposes, the general market practice was to “arrange your trades to settle comfortably before whenever the relevant cut-off was”; settling early to ensure receipt of dividends for tax purposes was “the most common thing”. Mr Sharma agreed that it would be straightforward for a buyer, if it mattered to them for certainty as to their tax position, to ensure that their purchase settled before the dividend declaration date. He was right to do so. Public listed companies in Denmark are required to give at least three weeks’ notice of their general meetings, specifying the time, place and agenda for the meeting, including any proposed dividend declaration and its amount, all of which is published on the company’s website.

The ultimate litigation context for this case, at the Main Trial, will be cum-ex trading based on short selling, i.e. contracting to sell shares for settlement after the record date, when any share transfer will necessarily be a transfer of ex-div shares, where the shares transferred (if any were transferred) were borrowed only after the dividend declaration date (indeed, it may be, only after the record date). An ordinary investor seeking to buy Danish shares would not be expected to buy with a non-standard settlement cycle so as to be trading before the ex-date yet settling only after the record date. Such an investor who was offered a trade before the ex-date for settlement after the record date would see they were being offered a cum-ex trade, signalling the probable existence of a short selling transaction structure the tax effects of which might be contentious.

If the cum-ex buyer were aware of the short selling transaction structure or its probable existence, and were trading in order to claim that the transaction structure generated a tax refund windfall for them (the generation, again as Mr Onslow KC put it, of ‘free money’), then they could be expected to take legal advice, especially Danish tax law advice, before trading, or at any rate before pursuing any claim against SKAT for the tax refund windfall. There is no reason why the Danish tax law rule of which such a buyer would be informed, if advised accurately, should be one that would make the strategy work.

If a buyer were determined both to make a claim for a tax refund and also to trade on or just before the dividend declaration date, and traded on standard settlement terms, with physical settlement so that they became a record date shareholder, they would be paid the amount of the dividend, net of withholding tax, and would be likely to understand that they had received, as indeed they would have received, a real dividend, not a manufactured dividend. For that case:

The first of those suggested possible difficulties does not arise, since I have concluded that Danish tax law aligns with the general law in treating the dividend as declared at the close of the market on the dividend declaration date.

There is no perfect solution to the second potential difficulty. A buyer to whom it matters could seek information from their seller, or a representation or warranty. The identity of the seller, the size of the position, or the pricing of the trade might provide relevant information in any event. Having transacted and performed, the buyer might seek further detail before making a tax refund claim by speaking to its custodian, which would have records of which trades, with whom, were allocated to which client and when. If the parties used the same custodian, the custodian should know whether the seller was long or short at the relevant time.

Ultimately, however, it is true, as Mr Goldsmith KC for SKAT acknowledged, that there might be cases (‘edge cases’, he called them) in which a buyer might in good faith believe or assume they had been the party taxed on a dividend, so as to entitle them to make a tax refund claim if they were otherwise eligible (e.g. because they were a tax-exempt US pension plan), but be mistaken about that; or occasions where a party entitled to a dividend tax refund might find it difficult to obtain evidence sufficient to demonstrate it, because they had chosen to trade on or close to the dividend declaration date. The evidence did not demonstrate that this would be a widespread problem, let alone that it would throw the market into disarray, as hyperbolically suggested in some of the submissions for the defendants. I am not persuaded that it is reason not to accept Prof Laursen’s statement and explanation of the Danish tax law rule.

Reliance was also placed on the fact that the volume of securities held by a custodian with a sub-custodian, or VP Securities if the custodian was an Account Holding Institution, might be smaller than the aggregate of the long positions held by its clients. It was said that such a custody shortfall against client’s positions was possible and might be permissible.

The only instance that was identified was that of rehypothecation, that is to say where a client’s custody contract entitles the custodian to borrow the client’s securities (by way of stock loan). That would lead to a custody shortfall against ‘long positions’ where the rehypothecated shareholding was transferred to a third party, for example if the custodian has borrowed Client A’s shares to enable Client B, a short seller, to complete their sale to some third party. In that case, Client A’s ‘long position’, to the extent of the rehypothecated volume, does not amount to a shareholding. There is no true custody shortfall, i.e. the holding of fewer shares than a custodian should be holding; there is only a client with a long exposure position exceeding their current shareholding. In respect of the rehypothecated volume, Client A is just a stock lender like any other stock lender whose borrower has gone short against the ‘redelivery’ obligation by selling or authorising the sale of the borrowed stock.

However the custodian records Client A’s positions, it should not be recording that they currently hold the stock that has been borrowed from them and sold away. Client A therefore should not be misled into thinking that it is receiving real dividends on that portion of its exposure position. The defendants’ case was that custodians could ‘re-use’ clients’ stock in this way without updating the clients’ account statements, so that if a custodian lent out 40 of Client A’s 100 shares, Client A’s custody account statement would still show them holding 100 shares, rather than 60 shares.

If it were the established market practice to maintain inaccurate and misleading custody account records like that, that would still not assist the defendants’ attack on Prof Laursen’s opinion as to Danish tax law. Even if a custodian did not update the client’s custody account statement, they would know of the rehypothecation and therefore could distinguish in any credit advice note or tax voucher between, in the example just given, real dividends in respect of the 60 shares and manufactured dividends in respect of the 40 shares. Indeed, the market practice experts agreed in their joint memorandum that it was the general practice of custodians to draw that distinction in any credit advice notes or tax vouchers they generated. I do not accept the suggestion that Mr Sharma made in cross-examination that that was limited to UK custodians when issuing tax vouchers for possible presentation to HMRC for UK tax. No such qualification was included, implied or even hinted at in the joint memorandum he agreed with Dr Collier and Mr Wade, where their joint expert evidence was that:

In any event, I agree with SKAT that the defendants did not establish a market practice to the effect that custodians re-used clients’ shares without updating clients’ account statements. I have referred already to some evidence suggesting that Belgian custodians may have had a practice of crediting borrowed stock to both lender and borrower simultaneously. However, there was no evidence that this was accepted or established as a correct practice, even in Belgium, and to the contrary it was described in terms by the EU’s Directorate-General (Internal Market and Services) in 2011 as an incorrect practice. Moreover, the limited evidence (such as it was) of the possible practice in Belgium acknowledged that at least where the borrowed stock was reused in the books of the custodian, “there should be a debit of the account of the client having authorised the re-use and a corresponding credit”, so that the possible issue of misleading custody records after a reuse would only arise, it seems, where the reuse was not by the custodian itself but by the custodian’s sub-custodian.

There is no evidence that the limited practice described by the Belgian authorities was market practice universally, or indeed that it was anything other than unique to Belgium. When asked about UK custodians, for example, Mr Sharma confirmed that it was not standard market practice of UK custodians in 2012 to 2015 to adopt that approach.

Net settlement, as an established market practice, was also relied on by the defendants. However, it has no bearing on the correctness or otherwise of Prof Laursen’s analysis of the applicable Danish tax law. Circular trades between parties who do not own any shares cannot result in the creation or transfer of any shareholding for tax purposes, any more than they could create or transfer share ownership under the general law. The net settlement of equal but opposite delivery (share transfer) obligations cannot increase the total aggregate shareholding of the parties involved, for tax purposes, any more than it could increase the actual size of the share issue.

An example was considered with Mr Wade of a buyer (A) who buys 50 shares each from eight sellers (B1 to B8), and sells 200 shares (to C, who is not any of B1 to B8), with successful settlement (no settlement failure) at VP Securities or at a custodian common to all parties using multilateral netting (which would have to have the consent of all parties). Successful net settlement would see A credited with a transfer to them of 200 shares, their net entitlement, having bought 400 shares and sold 200, C also credited with a transfer to them of 200 shares, their net entitlement, having simply bought 200, and each of B1 to B8 being debited, on the settlement date, with a transfer from them of 50 shares, their net obligation. That could be so even if some of B1 to B8 were short sellers who covered their position only on the settlement date, e.g. by stock borrowing with immediate settlement; say (for the sake of the example) half of them were in that position.

I agree with SKAT’s analysis that, on the basis of Prof Laursen’s evidence, if correct, A is not the rightful recipient of any dividends in relation to a dividend declaration after the purchase and sale contracts have been concluded but before the short selling Bs have borrowed stock to be in a position to perform their sales. On that basis, applying Prof Laursen’s analysis, A acquires, in the eyes of Danish tax law, a final and binding right to 200 shares (only) from the purchase contracts with B1 to B8, but the sale contract to C transfers that right away. C is therefore the rightful recipient of dividends on 200 shares. As regards the balance of 200 that (in aggregate) A contracted to purchase, neither he nor whichever of B1 to B8 sold short will have acquired a final and binding right to any shares before the dividend declaration. Since those Bs in fact settle by delivering 50 shares each, they must each have sourced 50 shares, directly or indirectly, from a party who was a dividend date shareholder; but that does not mean any of those Bs was a rightful recipient of dividends on 50 shares, subject to the impact of their sale to A, such that that sale might have relocated that entitlement in A.

In that case (successful settlement), it is not relevant to the tax analysis, with its focus on the acquisition of rights in the extended sense explained by Prof Laursen, to ask whether the 200 shares received by A at settlement came from the Bs who were selling short or the Bs who were selling shares they already had to sell.

However, if there was a settlement failure, it would be relevant to know which seller had failed. I accept on the basis of Mr Wade’s evidence that that should be ascertainable from trading records that should have been created, including order allocation records (for example, if A’s purchase of 400 shares were placed simply as a purchase order that their broker then matched sale orders from B1 to B8 of 50 shares each). On Prof Laursen’s evidence, the settlement failure nullifies ex tunc what might otherwise have been the dividend tax consequences of the failed trade.

For example, if one of the Bs who was selling short and one of the Bs who was selling long failed, as I understood the evidence about net settlement procedures, A’s sale to C would still settle successfully, but there would be a short delivery to A. In that case, C would need to know that A had a final and binding right before the dividend was declared only to 150 shares, so that C likewise acquired only such a right; and if C otherwise qualified for a tax refund because of a double taxation treaty, they should be making a refund claim only in respect of 150 shares. It seems to me doubtful that that subtlety would be known to C; but that makes it only an ‘edge case’ where C might make in good faith an excessive tax refund claim. It does not mean that Prof Laursen’s opinion on the content of the applicable Danish tax law rule should not be accepted.

Before leaving such examples, I add that there was no evidence that multilateral net settlement, such as practised at VP Securities (by its use of DVP Model 3 settlement), was common at custody levels above VP Securities. The only entity mentioned as having been able to operate such multilateral netting was Clearstream (referred to in this context by Mr Wade, who noted that even the likes of Barclays and JP Morgan could not do it). So the discussion here is all rather theoretical. It is certainly no basis for doubting Prof Laursen’s opinion on the meaning and effect of the language of the Danish tax legislation I have to determine.

I should deal for completeness with the defendants’ case (and Mr Sharma’s view) that share sales and purchases, and loans, can be net settled internally, if all the contractual counterparties have accounts with the same custodian, even if the custodian holds no shares. The only situation in which it was suggested that this could occur was where the transactions the custodian was treating as settling netted to zero. It is obvious, and Mr Sharma was clear, that if a set of transactions does not net to zero, there cannot be a DVP settlement without shares. The following would be an example, therefore: A, B and C share a custodian; none of A, B and C holds any shares, and the custodian has no shares; A contracts to sell 100 shares to B, B contracts to sell 100 shares to C, and C contracts to sell 100 shares to A; the three trades net to zero and, the defendants say, can or might be treated as having settled, not failed. I agree with SKAT that even if that is true, it does not mean that any of A, B or C ever becomes a shareholder, whether for tax purposes or otherwise.

In that case, none of A, B or C holds shares at any time, and no shares have been transferred between them, or between any two of them. None of them is the rightful recipient of any dividend declared at any time. Any contrary contention would be fantasy. After all, netting to zero is the same as saying that, on the securities side, the trades have cancelled each other out. Mr Sharma said that A, B and C might be expected to “treat what has happened as, for all practical purposes, the equivalent of having delivered shares”. I can see a sense in which that might be so, especially if there are pricing differences and the net cash is paid as required thereby. Each of A, B and C can say that they have the same number of securities as before (none), which is where physical settlement would have left them, and the net cash they would have had if they had each both received a transfer of shares against payment and effected a transfer of shares against payment.

But the substance, then, is that the self-balancing loop of trades has been cash settled. The Danish tax law result of that, on Prof Laursen’s evidence, is that the trades would fall to be treated as financial contracts under which no shares were in fact delivered, so no shareholding was ever acquired for tax purposes.

I find it convenient, and I hope it may assist the parties going forward, now to consider further illustrative examples that were discussed in argument.

A Danish company is due to declare a dividend. On the dividend declaration date (D), X who neither owns nor has contracted to acquire any shares sells 100 shares in the company to Y, for standard settlement (T+2), so settlement is due on the record date. On the record date, X enters into a stock loan with immediate settlement (T+0), to borrow 100 shares in the company from Z, who owned 100 shares on date D and had not committed to any species of share trade prior to the stock loan to X. The stock loan and the sale complete (cash against delivery on the record date), so that 100 shares are transferred to Y. Which of X, Y and/or Z was a rightful recipient of dividends on 100 shares, for tax purposes?

Mr Onslow KC for the DWF Defendants, by whom this example was introduced at trial, contended that since Z was a dividend date shareholder who had not committed their shareholding to be sold to anyone else, for tax purposes Z was a shareholder of 100 shares and therefore a rightful recipient of dividends on 100 shares, liable to dividend tax. That is the tax position even though Y rather than Z was the record date shareholder in respect of what was previously Z’s shareholding and will receive a real dividend, in that they received a transfer of cum-div shares, and any payment made to Z referable to the dividend will be a manufactured dividend. I agree with all of that.

Mr Onslow KC submitted further that Y also “has on an accruals principles basis “received” a dividend for Danish tax purposes as a binding contract for the purchase of shares has been concluded prior to the end of the [dividend declaration date].” I disagree with that. When dividend entitlements accrued, at the close of the market on the dividend declaration date, Y had only a contractual right against X to receive shares from X that X was not then in a position to deliver. Therefore, Y did not have a final and binding right to a shareholding in the sense required by Danish tax law to be, for tax purposes, a current shareholder. No dividend date shareholder’s shareholding had been reduced by 100 shares, for tax purposes, in favour (indirectly) of Y. For Y to be considered a shareholder for the purpose of Danish tax would be to create an excess shareholding of 100 shares for that purpose which is an impossibility.

For completeness, I note that it makes no difference to the analysis or the result for Example C whether the successful settlement is gross or net. Either way, on the settlement date 100 shares are taken from Z by debiting its securities custody account (which ex hypothesi must have had at least 100 shares in it – there are no ‘negative balances’), and 100 shares are given to Y by crediting its securities custody account. The difference between gross and net settlement would be that in the former, but not in the latter, X’s securities account at its custodian would, in addition, be both credited and simultaneously debited.

The DWF Defendants introduced three variants on Example C. I consider each in turn below.

The facts are those of Example C, except that now the sale by X to Y is a forward sale for settlement on T+3 (= D+3, the day after the record date), and X and Z conclude the stock loan with immediate settlement on that date (D+3).

As in Example C, only Z was a shareholder for tax purposes when the dividend entitlement accrued so as to be liable for dividend tax. The difference now is the non-tax difference that in Example D, Z retains the real dividend – it is the record date shareholder as well as the dividend date shareholder – and will receive payment accordingly; any payment received by Y will be in respect of a manufactured dividend (as it was bound to be since Y’s purchase settled after the record date).

Here, the facts are those of Example C, but in addition Y on the record date enters into a stock loan as lender, with W as borrower, for immediate settlement (T+0). After successful settlement on the record date (whether effected gross or net) W holds 100 shares.

Again, as in Example C, only Z was a shareholder for tax purposes when the dividend entitlement accrued so as to be liable for dividend tax. The non-tax difference this time is that W, rather than Y, is the record date shareholder entitled to and in receipt of payment of the real dividend; any dividend-related payment received by Y (presumably from W as stock borrower) will be a manufactured dividend. As in Example C, and assuming there are no other transactions to take into account, Y becomes the shareholder for tax purposes on, but not before, the record date, since W has only borrowed Y’s shareholding acquired that day, and has not sold it on to a third party.

This time, finally, the facts are those of Example E and W = Z, with settlement on a net basis, meaning that no share transfer occurs. Thus, in Example F:

Again, as in Example C, only Z was a shareholder for tax purposes when the dividend entitlement accrued so as to be liable for dividend tax – it was a dividend date shareholder who had not committed their shareholding to be sold to anyone else. The non-tax difference is that in Example F, as in Example D but for a different reason, Z has retained the real dividend entitlement and is paid accordingly; and any payment to Y referable to the dividend under the stock loan terms agreed with Z will be a manufactured dividend. As in Example E, likewise assuming no other transactions to be considered and subject to one caveat, Y becomes the shareholder for tax purposes on, but not before, the record date, since Z now holds 100 shares and Y has a final and binding right as stock lender to a future delivery of 100 shares from Z.

My one caveat is that what I have just said assumes it is proper to treat Y as having delivered to Z pursuant to the stock loan, so as to have acquired the stock lender’s right to a ‘redelivery’; and that Z’s corresponding stock borrower’s obligation to ‘redeliver’ is in due course physically settled (since otherwise the whole transaction will fall to be recharacterised ex tunc as having been a cash-settled financial contract that generates no dividend tax liability).

In the light of the detailed discussion above, I return to s.69B(1) of the Danish Withholding Tax Act, which I set it out in translation at the start of this judgment (paragraph 8 above). By its plain terms, when applied to Danish dividend tax, s.69B(1) obliged SKAT to meet a repayment claim made by a party liable to Danish dividend tax under s.2(1)(6) of the Act, or s.2(1)(c) of the Danish Corporation Tax Act, on dividends received by that party from which tax had been withheld under s.65 of the Act, if the tax withheld exceeded the tax permitted to be levied from that party by an applicable DTT. Where such a claim was made to SKAT, its obligation under s.69B(1) was to repay the excess tax amount it had levied by the s.65 withholding.

For that purpose, and so far as material, dividends were anything distributed by a Danish company to its current shareholders or members (s.16A(2)(1) of the Danish Tax Assessment Act); and Danish dividend tax liability was imposed, by s.2(1)(6) (natural persons) or s.2(1)(c) (corporate entities), on non-Danish residents who obtained or received dividends falling within that definition.

Interpreted and understood in accordance with the principles explained by Prof Laursen (and Mr Bachmann), dividends were obtained or received by a party, for the purposes of Danish tax law, if and only if a right to dividends accrued to them. They were referred to, in that case, as having been the rightful recipient of the dividends in question.

As a matter of Danish tax law, a right to dividends accrued on the date when it was declared, and more specifically (in that regard following a customary rule of Danish securities law), for Danish listed companies, it fell to be treated as accruing at the close of trading of the Danish stock exchange on that date. That right accrued to those who were then shareholders for tax purposes in the company declaring the dividend.

Consistently with all of that, at the material time s.65(1) and s.66(1) of the Danish Withholding Tax Act required the company declaring the dividend to withhold 27% of the total dividends (s.65(1)) and pay that amount (s.66(1)), as a liability created by the declaration of the dividend and falling due then for payment to SKAT by the company shortly thereafter (ibid).

For tax purposes, as under Danish company law, the aggregate holdings of all shareholders at any given moment would be, and could never exceed, 100% of the share issue at that moment. In that regard, I prefer Prof Laursen’s and Mr Bachmann’s evidence to Mr Mikelsons’, to the extent that he maintained any contrary view.

It follows that so long as the company had fulfilled its withholding obligation under s.65(1), i.e. so long as it had distributed 73% of the declared dividend – and whether or not it had made any payment to SKAT – tax of 27% was withheld from all dividend entitlements that accrued upon the dividend declaration in question. In that case, therefore, tax at 27% was withheld from all shareholders for tax purposes to whom a right to dividends accrued, i.e. all rightful recipients (for tax purposes) of dividends.

That means for this case, and subject to SKAT’s logically prior allegations of sham transactions, the criterion that will matter for the Main Trial is that the tax refund applicant must have been a shareholder for tax purposes when rights to dividends accrued, i.e. at the close of the market on the dividend declaration date. If they were such a shareholder, then:

SKAT proposed that s.69B(1) of the Danish Withholding Tax Act contained four requirements, namely that the applicant (i) had a Danish dividend tax liability (ii) in respect of dividends they had received (iii) from which tax had been withheld (iv) in excess of Denmark’s entitlement to tax the applicant under an applicable DTT. The conclusion stated and explained in the previous paragraph involves that being a shareholder for tax purposes when a dividend was declared in respect of which the company fulfilled its withholding obligation under s.65(1) of the Act was both necessary and sufficient for all of requirements (i) to (iii). In his argument for SKAT, Mr Graham KC resisted that conclusion, contending that, irrespective of the shareholding requirement, since “tax is only withheld by Danish companies on behalf of the company’s shareholders who are liable to tax … and … no tax is withheld on behalf of recipients of manufactured dividends, … the defendants’ case in the wider matter is … doomed”. To similar effect, as he later put it, “if your Lordship is with me on requirements 1, 3 and 4, then it’s going to be extremely difficult to see how, whatever differences there may be on requirement 2, it can take the defendants very far.”

I consider that to be a false argument. It was possible to be both a shareholder for tax purposes when a dividend is declared, and therefore liable to Danish dividend tax on the declared dividend, and also the recipient of a manufactured dividend (whether in that half of Mr Graham KC’s premise he was referring to the receipt of a payment or the accrual of a right to a payment). The fact that “in the wider matter” (meaning when we get to the facts next year) an applicant whose refund application I may consider received a manufactured dividend (if that be the position) does not ‘doom’ the application to a finding of invalidity. Rather, validity will turn on whether the applicant was a shareholder for tax purposes when the dividend referenced in the application was declared, so as to satisfy (all of) requirements (i) to (iii) (assuming I am right that it will always be the case that the company complied with its withholding obligation in respect of the referenced dividend), and whether an applicable DTT entitled the applicant not to be taxed at 27%, so as to satisfy requirement (iv). That means, if I am also right that all applicants I shall consider were entitled not to be taxed at 27% on Danish dividends under applicable DTTs, that the case on validity will ultimately concern only SKAT’s allegations of sham transactions, and requirement (ii).

Thus, the effective invalidity allegation against the impugned claims is going to be, I think, that in each case the applicant, it will be said, was not a shareholder for tax purposes when the dividend referenced in the application was declared by the Danish company in question, and therefore had had no right to any dividend and no dividend tax liability, and therefore had no right to a dividend tax refund.

Interpreted and understood in accordance with the principles explained by Prof Laursen (and Mr Bachmann), in preference to Mr Mikelsons’ evidence to the extent he maintained different views, those shareholders, i.e. the shareholders for tax purposes when rights to dividends accrued, were:

That is to say, in more compressed fashion, those liable to Danish dividend tax at the material time were the dividend date shareholders, except where their shareholdings were then committed (directly or indirectly) to others, in which case, and then to that extent only, the others to whom they were then committed, subject to the impact ex tunc of any subsequent failure to perform any such commitment by a share transfer.

I have entertained doubt whether the true position might not be simpler, namely that the shareholders for tax purposes, liable to Danish dividend tax, are the dividend date shareholders except where their shareholdings are at that time only borrowed, in which case, and to that extent, their stock lenders instead. If the statement I have set out in paragraphs 306-307 above of the overall effect of Prof Laursen’s analysis were considered too complex or impractical of operation to be correct, my finding would be that it has not been shown that Danish tax law departs from Danish company law as regards who is treated as a shareholder except for the specific case of shares that are only borrowed at the moment by reference to which the question is asked. That would result in the simpler proposition stated at the start of this paragraph.

Come what may, I reject the defendants’ contention that a contract to sell shares makes the buyer, without more, a shareholder for tax purposes.

By his tongue-twisting question-begger, “Does a short seller sell shares?”, Mr Jones KC in truth meant to ask whether a short seller of Danish shares is always treated by Danish tax law as transferring shares to their buyer when the sale contract is concluded, so that the buyer always becomes from that moment a shareholder for tax purposes capable of being liable to dividend tax in respect of a dividend later declared by the company. The short answer is no, as SKAT has claimed.

The parties liable to Danish dividend tax in respect of a dividend declared by a Danish listed company are the shareholders in the company, for tax purposes, at the close of the market on the dividend declaration date. That is because, as I have concluded:

Under Danish tax law, the shareholders in a Danish listed company at any given time, for tax purposes, are the shareholders of record at that time, except where their shareholdings were then contractually committed (directly or indirectly) to others, in which case, and then to that extent only, the others to whom they were then ultimately thus committed, subject always to the unravelling impact of any subsequent failure to perform some such commitment by a share transfer.

The conclusion of a contract for the sale of shares in a Danish company does not make the buyer a shareholder, for tax purposes, under Danish law, if at the time the contract is concluded the seller holds no shares that could be transferred to the buyer nor has any right by virtue of a contract to transfer to the buyer a shareholding then held by another. The position of such buyers was a significant focus of the trial because SKAT will be contending at the Main Trial that the tax refund applications about which it complains were made by applicants who, if their transactions were real (not sham), were in that position throughout the dividend declaration date for the dividend by reference to which they presented a claim to SKAT. I consider it appropriate, therefore, to assist the parties by spelling out as consequential findings that:

In the light of those overall conclusions, and the detailed analysis in the main body of this judgment, my individual answers to the Validity Issues ordered to be determined at this trial are set out in the Appendix that immediately follows.

What were the requirements of a valid application for a refund of WHT from SKAT between August 2012 and July 2015 (the “Relevant Period”)?

During the Relevant Period, under what circumstances would an applicant be liable to taxation pursuant to section 2 of the WHT Act or section 2 of the Danish Corporation Tax Act?

What constituted “receipt” of a “dividend” declared by a Danish company for the purposes of the Danish tax law rules applicable to the making of a refund application during the Relevant Period?

Subject to issue 3(d), what constituted “ownership” of “shares” in a Danish company for the purposes of the WHT Act, the Corporation Tax Act, and the making of a refund application, during the Relevant Period?

If it was a necessary requirement for a valid application for a refund of WHT during the Relevant Period that tax had been withheld (i.e. issue 1(c) above), was this requirement satisfied by tax being withheld by a Danish company at source and if so what connection did such withholding of tax need to have to the applicant for a refund of WHT?

If it was a necessary requirement for a valid entitlement to a refund of WHT during the Relevant Period that tax had been withheld in respect of dividends received by the applicant at a rate that exceeded the rate permitted under the relevant DTT (i.e. Issue 1(d) above), what did this requirement entail? If so, for the purposes of the DTTs, what constituted a “dividend” “paid” by a Danish company “to” an applicant?

During the Relevant Period, did an applicant have the right to make an application for a refund of WHT directly under the DTTs without the need to satisfy the requirements set out in answer to Issue 1 above?

If relevant and admissible, what was the significance of whether the applicant was the “beneficial owner” of a dividend declared by a Danish company to the validity of an application for a refund of WHT?

Is the answer to any of the questions at 1 to 8 above affected by:

Is the answer to any of questions at 1 to 8 above affected by: