Revenue and Customs v Dempster (t/a Boulevard)

This is an appeal by the Commissioners for Her Majesty’s Revenue and Customs (“HMRC”) against a decision (“the Decision”) of the London VAT Tribunal (Howard Nolan and Shahwar Sadeque) promulgated on 3^rd May 2007, which allowed the appeal of Mr Noel Dempster against HMRC’s decision by letter dated 9^th September 2004 to refuse his claim for an input VAT credit in the sum of £156,998.89 in respect of his accounting period 1^st January to 31^st March 2004.

Mr Dempster’s claim was based upon two alleged supplies of computer software by Abacus Business Systems Limited (“Abacus”) in January 2004, supported by invoices from Abacus, dated respectively 12^th and 26^th January, and which differed from each other only in their respective dates and invoice reference numbers. Their otherwise identical contents may be summarised as follows. Each of them confirmed an order from Mr Dempster, trading as Boulevard, from an address in Brixton, London for:

In order to decide whether to accept or reject Mr Dempster’s claim, HMRC had sought and obtained from him a compact disc purporting to contain a copy of the software which was the subject matter of his claim. That CD had written on it:

The reasons given by HMRC for refusing Mr Dempster’s claim were, first, that they were not satisfied that taxable supplies had occurred in accordance with the claimed transaction, and, secondly, that the contents of the SmartSoft Disc were inconsistent with the description of the relevant goods contained in the supporting VAT invoices. HMRC adhered to both those grounds in the course of Mr Dempster’s appeal to the Tribunal. In their Statement of Case HMRC amplified the first of those grounds by alleging, in paragraph 21, that:

The Tribunal rejected both HMRC’s grounds for refusing Mr Dempster’s claim. As to the first, in summary, they concluded that although Mr Dempster was in many respects a thoroughly unsatisfactory witness, it had not been established that he knew that the transactions which gave rise to his claim were shams in the relevant sense, so that even if they had in fact been part of an attempted VAT fraud (as to which they made no finding one way or the other), he was not disabled from pursuing a VAT credit in respect of supplies of computer software which they found had taken place.

As to the second ground, I have had some difficulty in discerning precisely the basis upon which the Tribunal rejected HMRC’s case, due to an infelicity in the language of the critical part of their Decision. The appeal was argued by Miss Haynes and by Miss Choudhury who appeared for Mr Dempster on the basis that the Tribunal had decided that, as a matter of fact, the VAT invoices relied upon did sufficiently describe the software actually supplied.

HMRC’s grounds of appeal, as developed in submissions, may be summarised as follows. As to the first issue (sham), the Tribunal’s decision was vitiated by errors of law both as to the nature of a sham, and as to the type or extent of knowledge necessary to visit Mr Dempster with the consequences. Alternatively, they submit that the Tribunal’s decision in relation both to sham and knowledge was perverse. As to the second issue (misdescription of goods) they simply submit that the decision was perverse.

There was, as between counsel, no discernible difference as to the legal definition of sham, or as to its consequences. Subject to one caveat by Miss Choudhury, to which I shall return, both invited me to treat as a sufficient definition the following description given by Diplock LJ in Snook v London and West Riding Investments Limited[1967] 2 QB 786 at 802:

To that general principle Miss Choudhury reminded me of an exception, identified by Arden LJ in Hitch & ors v. Stone (Inspector of Taxes) [2001] STC 214, at 234, namely that it is not invariably necessary to show, in relation to multi party transactions, that every party to it knew it was a sham. In my judgment that exception to the general requirement to show knowledge is of no relevance to the present case.

As to the effect of a finding of sham in relation to a transaction relied on for the purposes of a claim to a VAT credit, it was (inevitably) common ground that if a purported supply was fictitious, in the sense that the documents relied upon purported to identify a supply which in fact did not take place at all, then no VAT credit could be claimed: see for example McNicholas Construction Co Ltd v. CCE [2000] STC 553, per Dyson J at paragraph 50. There was however an issue as between counsel’s submissions on the question whether any element of sham in the documentation of a transaction is sufficient to deprive it of VAT consequences, regardless whether that element is of itself relevant to the question whether the underlying transaction constituted a taxable supply.

Miss Haynes submitted that once it is shown that a transaction was a sham in any respect, then subject to the question of knowledge, it could not be relied upon for the purposes for a claim for a VAT credit. Miss Choudhury submitted that for a transaction to be deprived of its ordinary VAT consequences, it must be a sham in some relevant sense, relevant that is to its VAT consequences. To the extent that it matters, I am inclined to agree with Miss Choudhury’s narrower submission. For example, the parties may, for some reason unconnected with its VAT consequences, deliberately misstate the date of a transaction in the documents recording it. In my judgment it is only where the difference between the truth and the misstatement affects the VAT consequences of the transaction (for example by removing it from one accounting period to another) that such a misstatement is of relevance to the VAT outcome.

Of greater materiality to the present case was Miss Choudhury’s submission that it is important to distinguish between, on the one hand, a sham (in which the parties do not intend to create as between them the rights and obligations recorded in the documents) and a transaction in which the parties’ intended rights and obligations are correctly recorded in the documents, but which is capable of being described as uncommercial or artificial, for example where it forms part of a pre-arranged composite transaction the only purpose of which is to obtain a tax advantage. This distinction is precisely encapsulated in the following paragraph of the judgment of Arden LJ in Hitch v. Stone (supra):

The same distinction appears in the speech of Lord Goff of Chieveley in Ensign Tankers v. Stokes [1992] STC 226 at 245 to 247, in which he concluded that a composite transaction which could fairly be described as involving a “conjuring trick”, artificiality and self-cancelling elements, and which in the same case Lord Templeman described as involving “play acting” should nonetheless not be called a sham, in the sense relevant for present purposes.

I accept Miss Choudhury’s submission. In my judgment the critical question whether a transaction which, as documented, appears to attract certain VAT consequences is to be deprived of those consequences because it is a sham, is whether the rights and obligations expressed in the documents, to the extent relevant for VAT purposes, are different from those which the parties intended to confer and incur.

The parties’ differences as to the nature of the knowledge requirement broadly reflected the differences which I have already described, in relation to the legal nature of a sham. Miss Haynes submitted that, regardless of the precise respects in which the underlying transaction departed from that portrayed by the documents, it was sufficient to show that the party sought to be affected knew that in some more than de minimis respect there was a departure, whereas Miss Choudhury submitted that it was necessary for HMRC to show, in the present case, that Mr Dempster knew that the documents failed to reflect the underlying transaction in a relevant respect. Again, I prefer Miss Choudhury’s submission. Indeed, it seems to me questionable, in relation to a two party transaction, whether, without that knowledge, there can be a finding of sham at all, in the relevant respect.

To this general proposition there is an obvious exception. Where the applicant for the VAT credit knew (actually, or in the sense of being wilfully blind to the fact) that he was participating in a tax fraud, then the extent of his knowledge of the respects in which (if at all) the documentation failed to record the intended rights and obligations of the parties will become irrelevant. Fraud unravels all, and VAT is no exception. Furthermore, the question whether a claimant for a credit is to be found to have had Nelsonian knowledge, either that he was participating in a tax fraud, or of the relevant respects in which the documents relied upon were a sham, may be considerably affected by his knowledge that other parts of the transaction, not relevant for VAT purposes, failed to reflect the reality.

The Tribunal did not set out in any length its own understanding of the relevant legal requirements in relation to the sham issue. On the contrary, they complained in paragraph 3 of the Decision that they had not actually been addressed on the legal requirements, but stated their understanding that:

The Tribunal had well in mind the general fraud exception to what I have described as the requirement to show knowledge of the relevant respects in which (if it be the case) the transaction as documented constitutes a sham, as appears from this extract from the following paragraph of the Decision:

In order to decide whether the Tribunal’s view that the critical question was whether, to Mr Dempster’s knowledge, the actual transactions were dealing in something “merely feigning to be the claimed computer software” involved an error of law, it is necessary to explore the facts in a little more detail. The Tribunal found, contrary to HMRC’s pleaded case, that Mr Dempster had in fact made supplies of computer software to Spectrum, and duly despatched it via DHL as courier: see paragraph 59 of the Decision. Implicit in that finding is the conclusion, as a matter of fact, that Abacus made a supply and delivery of identical computer software to Mr Dempster, rather than, as was implicit in HMRC’s case, direct to Spectrum, if at all. There was no issue but that Spectrum had paid the purchase price to Mr Dempster, and that Mr Dempster had paid the price identified in Abacus’s VAT invoices to Abacus, inclusive of the VAT that was the subject of the claim for a credit. None of these facts were, or could be, challenged on this appeal.

The Tribunal heard evidence from Mr Dempster which, had they accepted it, would have suggested that the transactions in issue constituted the first stage in a new business development of his, into the purchase and sale of computer software, after a business career previously devoted to dealing in footwear. Mr Dempster’s evidence suggested, if true, that the transactions in issue had been ordinary commercial purchases and sales contracted by him in the market place. The Tribunal had no hesitation in rejecting this explanation, from a witness who they described as difficult, unsatisfactory and unreliable, both because he said things which were plainly untrue, and pretended, in their view untruthfully, to be unable to recollect important aspects of the relevant chain of events.

The Tribunal concluded that, far from being transactions in the ordinary commercial course of events, the successive sales by Abacus to Mr Dempster and by Mr Dempster to Spectrum were pre-arranged and highly artificial transactions, giving rise to only two possible analyses: either that Mr Dempster had throughout been a knowing party to a VAT fraud, or that he had been tutored to play a small walk-on role in a pantomime, without knowing, or having reason to suspect, its improper purpose. I can do no better than set out the Tribunal’s own summary of the process by which it reached that conclusion:

The summary in paragraph 52 leads us to one or other of two conclusions. Either the Appellant was a knowing party to transactions that were designed to avoid VAT, or the Appellant was tutored for a small role in a transaction where he should appear to be conducting trading transactions but in fact would be performing a scripted pantomime part. It is possible that the Appellant did not know why he was being offered the marvellous opportunity to make a £44,000 profit for virtually nothing. He might have thought that there was some entirely non-tax-related reason why Abacus had to obscure the fact that they were genuinely selling software to a Canadian client. There could have been any number of curious non-VAT related reasons why he would need to be a ‘dummy in the middle’, of other genuine transactions. But the one summary of the facts that we absolutely reject is that the Appellant commenced a new genuine business, or that he thought for a moment that he had commenced a genuine new business.”

As I have said, the Tribunal rejected HMRC’s primary case that there had been no computer software sale at all by Abacus to Mr Dempster (or by Mr Dempster to Spectrum), and were therefore left only with HMRC’s alternative case, which they summarised as follows, in paragraph 60:

Having asked themselves the questions which I have described, the Tribunal decided first that they could not (and by implication did not need to) decide whether in fact the software was genuine, or a fake: see paragraph 69 of the Decision. They rested their conclusion on the sham issue on a finding that, if the software was a fake, Mr Dempster neither knew nor turned a blind eye to that fact: see paragraphs 70 and 71.

It is evident from a reading of the Decision as a whole, and in particular paragraphs 2 and 76, that the Tribunal reached that conclusion without enthusiasm, mindful of the fact that there was evidence before them which could justify the opposite conclusion, and that they felt constrained in effect to acquit rather than convict Mr Dempster of having the relevant guilty knowledge because the critical assertions to that effect had not been put to him in cross examination. This latter point is apparent from the passage in paragraph 62 of the Decision which I have quoted above, but also from paragraphs 71 and 76.

Miss Haynes challenged that approach as wrong-headed. She submitted that the whole thrust of her cross examination of Mr Dempster before the Tribunal had been directed to establishing that his account of the transactions was untrue, and that they were to his knowledge parts of an artificial pantomime rather than on genuine arm’s length commercial terms. She submitted that since, by necessary implication (albeit not in express terms), HMRC’s Statement of Case before the Tribunal had constituted a case of dishonesty, it was unnecessary for it to be put specifically in cross examination to Mr Dempster either that he was a knowing party to a VAT fraud, or that he knew, or turned a blind eye to the fact that, the software which he traded as an intermediary between Abacus and Spectrum was fake or worthless.

I emphatically disagree with that submission. First, the Tribunal’s summary of what was not put in cross examination is stated with clarity on no less than three occasions in the Decision and I was provided neither with a transcript, nor notes (whether by the Tribunal itself or by the parties) of the cross examination with which to be in any position to conclude that the Tribunal’s summary of the cross examination was other than fair and accurate. Secondly, it is a cardinal principle of litigation that if serious allegations, in particular allegations of dishonesty are to be made against a party who is called as a witness they must be both fairly and squarely pleaded, and fairly and squarely put to that witness in cross examination. In my judgment the Tribunal’s conclusion that it was constrained, notwithstanding suspicion, from making the necessary findings of knowledge against Mr Dempster (necessary that is to permit the consequences of the alleged sham to be visited upon him) was nothing more nor less than a correct and conventional application of that cardinal principle.

More generally, I have been unable to discern any error of law in the Tribunal’s approach to the sham issue. Once the Tribunal had concluded (in my judgment correctly) that the course of the cross examination prevented them from reaching any general conclusion that Mr Dempster was a knowing party to a VAT fraud, the only basis upon which the case of sham could be made good against him was, as they correctly identified, that the software actually supplied, both to Mr Dempster and by him to Spectrum, was, as they put it, fake. The only documents recording the sale by Abacus to Mr Dempster were Abacus’s VAT invoices. There was plainly no element of sham so far as concerned date (even if that were relevant) or price, and the Tribunal had concluded that there was a transaction of sale and purchase which, although not part of a continuing business, was a transaction designed to yield a profit for Mr Dempster and therefore fairly and squarely within the VAT net. As Miss Choudhury put it, the identity of the subject matter of the transaction was therefore the only relevant respect in which the VAT invoices could be said to have involved sham. Shorn of the allegation that the whole pantomime was to Mr Dempster’s knowledge a VAT fraud, the fact that his attempt to display it as part of an ongoing business was, viewed separately, both untrue and a sham is in my judgment neither here nor there, since the VAT invoices did not themselves purport to suggest whether the transactions were one-offs, in a career otherwise dedicated to the trading of shoes, or part of an ongoing business in computer software. The only relevance of those wider issues was as part of the analysis of the question whether Mr Dempster turned a blind eye to the relevant facts, and this the Tribunal had well in mind and dealt with in detail.

Nor in my judgment can the Tribunal’s conclusion on the sham issue be categorised as perverse. In this respect, Miss Haynes relied upon the trenchant findings of the Tribunal that, contrary to his own evidence which they disbelieved, Mr Dempster had in fact at the very least been involved in the poor performance of a scripted part in a pantomime rather than in genuine commercial transactions. How could they, she asked, then have failed to conclude that he was a knowing party to a dishonest sham in every relevant respect?

In my judgment the long answer to that rhetorical question is to be found in the detailed analysis of the correct answer to the two alternatives posed in paragraph 53 of the Decision, set out in paragraphs 56 to 71, which in my judgment displays no hint of perversity. The short answer is that the key aspects of HMRC’s alternative cases were simply not put to Mr Dempster in cross examination. Accordingly I reject HMRC’s challenges to the Tribunal’s decision on the sham issue.

It was common ground between counsel that by reason, in particular, of Regulation 14(1)(g) of the VAT Regulations, a claim for an input credit may properly be refused by HMRC if the relevant VAT invoices fail to provide “a description sufficient to identify the goods or services supplied”.

The Tribunal concluded that Abacus’s VAT invoices (which I have quoted in full at the beginning of this judgment) did provide a sufficient description, compliant with Regulation 14(1)(g). HMRC challenged this conclusion solely on the grounds of irrationality. The test therefore is whether any reasonable Tribunal, properly directing itself as to the relevant law, could have reached that affirmative conclusion.

In the context of the Tribunal’s finding that there was a supply of computer software of some kind by Abacus to Mr Dempster, HMRC’s case in relation to Regulation 14 was that Abacus’s invoices failed sufficiently to describe the goods supplied in two relevant respects. Both limbs depended upon the assumption that the software actually supplied conformed in all respects with the SmartSoft Disc supplied by Mr Dempster to HMRC during their consideration of his claims.

The first point taken was that, whereas Abacus’s VAT invoices described the software as “educational”, there was no educational content in the content of the SmartSoft Disc sufficient to answer that description. The second point was that, educational or not, the software supplied cannot have been (as described in the VAT invoices) V Ware 01 licensed software. On the contrary, it was SmartSoft software.

The sheet anchor of HMRC’s case before the Tribunal was that it was clear that the subject matter of the supply (by Abacus to Mr Dempster and by Mr Dempster to Spectrum) was the same software as was contained in the SmartSoft Disc. This is what Mr Dempster had asserted during the investigation of his claim. Furthermore, Mr Dempster’s case before the Tribunal did not resile from that assertion. I was shown the skeleton argument of Mr Timothy Brown who appeared for Mr Dempster before the Tribunal, which, accepting that proposition, sought to show how the software on the SmartSoft disc could be both educational and properly described as V Ware 01 licences.

Besides the SmartSoft Disc, which was in evidence by way of a photograph of its outward appearance, and an expert report as to its contents, there was also put in evidence for the Tribunal a promotional brochure supplied by Abacus to Mr Dempster, produced by a company named ITS Limited, and which, while being startlingly uninformative about the content of V Ware’s software offering, did refer by way of a bare heading to educational materials.

The Tribunal dealt very briefly with HMRC’s point that the software actually supplied was misdescribed as educational. They said this, at paragraph 74 of the Decision:

So far as I can discern, that reasoning did not depend upon a decision whether the software actually supplied was that to be found on the SmartSoft Disc, or some other (now unavailable) software broadly in conformity with the Brochure. Questions as to the proper breadth to give to the use of an adjective in an invoice, and as to the applicability or otherwise of that adjective, thus construed, to computer software seem to me to be precisely those matters of mixed fact and law which the Court of Appeal has recently emphasised should be left to the good sense of the experienced and specialised tribunal of fact, and with which, on appeal, this court should be very slow to interfere: see Able (UK) Ltd v HMRC [2006] EWCA Civ 1207.

On the assumption contended for by HMRC, both on this appeal and before the Tribunal, namely that the software actually supplied was that to be found on the SmartSoft Disc, it is to be noted that the items identified in the report obtained by HMRC included a “Mastermind” computer game. It does not seem to me irrational to describe software including such a game as, at least potentially, educational, in the sense that players of that game might well educate themselves in the process. If on the hypothesis that the software actually supplied was some other software broadly corresponding with the Brochure, then save for the fact that Mr Dempster was unable to provide a copy of it to the Tribunal, it is hard to see how the Tribunal’s analysis can be faulted.

I turn therefore to HMRC’s alternative point, namely that the software actually was SmartSoft software and not V Ware. To this point the Tribunal devoted a more detailed analysis. Although no case had been advanced that anything other than the contents of the SmartSoft Disc had been supplied by Mr Dempster to Spectrum, they expressed themselves as far from convinced that this was the software supplied in fact. It will be recalled that the Tribunal had, in the earlier part of its Decision directed to the sham issue, expressly declined to decide whether the software actually supplied was genuine or not, preferring to deal with that part of the case on the basis of Mr Dempster’s lack of knowledge. In relation to the alternative case based on an inadequate description of the goods in the VAT invoice, Mr Dempster’s knowledge was wholly irrelevant, as they duly observed.

Paragraph 75 of the Decision contains the whole of the Tribunal’s analysis of, and decision on, this point. Most of it consists, as the Tribunal acknowledged, of speculation. They noted that all the relevant documentation, including Abacus’s invoices, Spectrum’s orders and Mr Dempster’s invoices to Spectrum, together with a written confirmation by Spectrum of receipt of the bulk of the goods, given, upon inquiry, to HMRC in November 2004 uniformly identified the subject matter as V Ware 01 licenses. The Tribunal expressed (and explained) an inability to understand why, even if there was a fraud, and even if the fraud was less than skilful, the parties should have made the foolish and unforced error of failing accurately to describe the software actually supplied. If there was no fraud, then the uniformity of all the documentation, including that emanating from Spectrum, uncoupled with any protest after delivery that Spectrum received the wrong goods, would make the misdescription even more improbable.

The Tribunal’s decision on the point is expressed in the final sentence in paragraph 75, as follows:

HMRC’s criticism of that analysis was that, faced with apparent common ground that the software actually supplied was the material in the SmartSoft Disc, the Tribunal allowed speculation about the background to the relevant supplies to get in the way of the hard fact that by no process of analysis could that material be described as V Ware. Mr Dempster had simply failed to discharge the burden of showing that the goods matched the description in the invoices.

There is much force in that submission and, had I been deciding that question myself on the materials available to me, I might well have concluded that HMRC’s analysis was correct on the balance of probabilities. But the Tribunal are the arbiters of fact, and their conclusion can only be successfully appealed if it was reached by an error of law, or was one which was perverse; i.e. beyond reason. Furthermore they heard three days of evidence, albeit that much of it came from a witness (Mr Dempster) who they described as unsatisfactory and untrustworthy.

The Tribunal were not prepared to assume, merely from its being common ground, that the material actually supplied was in all respects precisely the same as that in the SmartSoft Disc, for reasons which they fully explained, nor that, even if it was, it could not properly have been re-labelled V Ware, on sale by Abacus to Mr Dempster, for on-sale to Spectrum. They considered it necessary to weigh that analysis against what they clearly regarded (as an experienced specialist tribunal) as the improbability that all the parties (including any shadowy orchestrator of the pantomime) simply concurred in a misdescription that even a planned fraud would have rendered unnecessary.

I can find nothing irrational in the process whereby the Tribunal reached the conclusion that such a misdescription was highly improbable, even though it does not lead me to the same conclusion. In the end it was for the Tribunal to balance the probabilities and improbabilities of the competing analyses, and to attribute such weight to each factor in that balance as they reasonably thought fit. It is clear that they directed themselves correctly as to the burden of proof.

The result is therefore that this second limb of the appeal also fails, so that the appeal must be dismissed. I reach that conclusion with no greater happiness or assurance than did the Tribunal, as they made very clear in the Decision. I also accept that this is a case in which there was reason to suspect a fraud of which Mr Dempster was not wholly innocent, but for such a serious allegation to be made good, it needed to be fairly and squarely put in cross-examination, and it was not. It is consistent with the European jurisprudence which governs VAT that an innocent participant in another’s fraud is not thereby to be deprived of any beneficial VAT consequences of the transaction in which he participated, if otherwise effective for that purpose: see Kittel v Belgium (2006) Case C 439/04.