Revenue and Customs v The Rank Group

A party to proceedings before a Value Added Tax and Duties Tribunal who is dissatisfied in point of law with the decision of the Tribunal may appeal under Section 11(1) of the Tribunals and Enquiries Act 1992 to the High Court. The Commissioners for Her Majesty’s Revenue and Customs (“HMRC”) appeal two decisions of the Tribunal (Theodore Wallace and AJ Ring FCA FTII) arising out of proceedings brought by the Rank Group PLC (“Rank”): the two appeals have been conveniently referred to as “the Bingo Appeal” and “the Slots Appeal” respectively.

The broad question in the Bingo Appeal is whether the VAT treatment of mechanised cash bingo breaches the principle of fiscal neutrality: and the core issue on the appeal is whether the burden lay on Rank to adduce evidence to prove not only that there was a difference in VAT treatment between similar (and apparently competing) products but also that the difference did as a matter of fact actually affect competition.

The broad question in the Slots appeal is again whether the principle of fiscal neutrality has been breached. But the core issues on the Slots Appeal are two prior questions: whether there was in law a difference in treatment between similar supplies (which involves a consideration of the meaning of the word “machine” in the phrase “the element of chance in the game is provided by the machine”) ; and if there was not, then whether the evidence established that there was in practice (if not in law) a difference in treatment. If either of those questions is answered affirmatively then the question arises (again) whether it was necessary for Rank to establish on the evidence actual competition between similar supplies.

It should be noted at the outset that the issues are now of historic interest because of legislative changes both to gaming and to tax law. But that is not to diminish the importance of the issues. The Rank cases could involve some £61 million pounds (quantum is an issue which has been left over for agreement or determination once the ground rules have been established): the Rank cases are themselves test cases with over 1,000 similar claims outstanding.

I will begin with the common background. Article 13(B) of the Sixth VAT Directive (Council Directive 77/388/EC) provided that member States:-

One of the general principles of community law is that of “fiscal neutrality”. This was described by the Court of Justice in the Isle of Wight Case [2008] STC 2964 at para. 42 as “a fundamental principle of the common system of VAT”, and by the Court in the French Drugs Case [2001] ECR I-3369 at para. 21 as “inherent in the common system of VAT”. The Court went on in para. 22 of that case to state:-

The Sixth Directive relating to the exemption from VAT of betting and gaming (but subject to limitations) was implemented in domestic law by the Finance Act 1972. It is with this domestic legislation that I am concerned. I must, so far as is at all possible, interpret it in a way that accords with Community law: Santex [2003] ECR I-1877 para. 63. Schedule 5 FA 1972 contained the exemptions from the general charge to VAT. Included in the exemptions was:-

The provision of any facilities for the placing of bets or the playing of any games of chance”.

Section 14 games took place on premises licensed for the purpose and for regulated maximum stakes. But by section 21 of the Gaming Act 1968 there was taken out of the scope of section 14 certain such games where the stake was below 50p and where any money prize was below £25 and any non-monetary prize below £500. Such games remained exempt from VAT (because they were not played by virtue of regulations made under section 14, but by virtue of the provisions of section 21).

The Notes to Items 1 of Group 4 of Schedule 5 to the Finance Act 1972 were then amended by the Value Added Tax (Betting, Gaming and Lotteries) Order 1975. This added to Note (1) (exclusions from the exemption) an additional sub-paragraph (d) which related to “the provision of a gaming machine”. This reversed the previous complete exemption of gaming machines (a change not itself inconsistent with the Sixth Directive according to Advocate General Jacob’s Opinion in Glawe (see paragraphs 10 and 23)): but for the purpose of bringing the activity within the charge to tax “gaming machine” was given a particular meaning. Games of chance played by machines that fell outside the definition remained within the exemption from VAT. The gaming machines that were brought within the general charging provisions to VAT were defined in a new Note (4). This was in these terms:-

It is constructed or adapted for playing a game of chance by means of it; and

A player pays to play the machine…either by inserting a coin or token into the machine or in some other way; and

The element of chance in the game is provided by means of the machine”.

It should be noted that Note (4) was not designed to bring within the charge to VAT all games of chance played by means of a machine. The only games of chance caught were those played on machines which had three features, namely, they were constructed for playing the game, they had a “slot”, and the element of chance was provided by means of them. So there were bound to be some machines that fell on one side of the definitional line, and some that fell on the other.

These provisions (albeit in re-enacted form) are those which govern the period to which the Bingo Appeal and the Slots Appeal relate. The provisions of the VAT Acts and the Statutory Instruments (“the fiscal stream”) cross refer to the substantive law of gaming (“the social stream”). This is apparent from the express reference (in Note (b) to Item (1)) to section 14 of the Gaming Act 1968. But it was common ground before me that it is apparent also from the definition of “gaming machine”, where the words used in the VAT legislation are similar to those enacted in Section 26 of the Gaming Act 1968. Section 26 is in these terms:-

In the preceding sub-section the reference to playing a game of chance by means of a machine includes playing a game of chance partly by means of a machine and partly by other means if (but only if) the element of chance in the game is provided by means of the machine”.

With that common background delineated I can address the Bingo Appeal.

The nature of ordinary bingo will be well known. As Lord Denning MR explained in R v Herrod [1976] 1 All ER 273 at 279, the variant called “prize bingo” is like ordinary bingo but played with a sophisticated apparatus.

The Bingo Appeal concerns mechanised cash bingo. Paragraph 4 of the Decision of the Tribunal dated 27 May 2008 contains a description of mechanised cash bingo derived from an agreed statement of facts. For the purposes of the Appeal I can summarise the findings as follows:-

The Court can be seen applying this approach in the AITC case [2007] ECR I – 5517. Under English domestic law VAT was charged on management services provided to an investment trust at a time when management services provided to an authorised unit trust (or similar vehicle) were exempt from VAT. Authorised unit trusts and open ended investment companies were variable capital funds which were obliged to buy back their units or shares from investors who wished to sell them; they had to be authorised by the Financial Services Authority. Investment trusts are, of course, limited liability companies quoted on the Stock Exchange in which an investor who wished to realise his investment sold his shares at a price negotiated on the basis of market supply and demand. Investment trusts were regulated in a different way. In its guidance to the English court the ECJ expressed the opinion that AUTs, OEICs and ITCs are three forms of special investment which spread risk, involving investment in securities through the intermediary of a collective investment undertaking which allows private investors to invest in wide ranging investment portfolios (and thus to reduce the stock market risk). The assessment of similarity is being carried out at a fairly high level of abstraction. This advice to the domestic court reflected the advice which had been given by the Advocate General (see paragraph 39 of the Opinion) that:-

It is pertinent to note that when making the comparison between the relevant goods or services the identity of the provider is not material. This appears from the decision in Linneweber (supra) at paragraphs 25 to 30. Herr Linneweber had permission to provide gaming and entertainment machines in restaurants and amusement arcades. The German authorities said that he had to pay VAT on his turnover. Operators of licensed public casinos did not have to pay VAT (though to obtain the licence the stakes and winning opportunities in the case of gaming machines installed in casinos had to be significantly higher than those in the case of gaming machines which were lawful outside casinos). The German authorities argued that the gaming machines operated by licensed casinos were not comparable with the gaming machines operated by Herr Linneweber. The Court took the view (see paragraph 25 of the judgment) that it was a general principle of community case law that the identity of the provider of the service and the legal form by means of which the provider exercised its activities were, as a rule, irrelevant in assessing whether products or services supplied were comparable. It applied that general principle to hold (see paragraphs 29 and 30 of the judgment) that in determining the conditions on which the operation of gaming machines was to be exempted from VAT Germany could not validly make that exemption dependant upon the identity of the operator of such machine. The principle of fiscal neutrality precluded a law which exempted from tax the operation of a gaming machine when it was carried out in a licensed public casino when the operation of the same activity by traders other than those running casinos did not enjoy that exemption.

So much for the exercise of comparison and the factors relevant to be compared. If as a result of the conduct of the exercise it is concluded that the relevant goods or services are identical or similar, then ordinarily they will be treated as being in competition with one another. That really follows from the nature of the exercise of comparing the two. But the principle is frequently articulated. I will confine myself to three citations. In Hoffman (supra) it was said (at paragraph 20) that the principle of fiscal neutrality:-

First, Mr Vajda QC was at one stage minded to argue that the element of “competition” to which the fiscal neutrality principle referred was “competition” in a competition law sense. Accordingly, when the court was saying that goods were “similar and thus in competition with one another” it was saying something about “demand substitution” or “supply side substitutability” and about “market definition”, and that those competition law concepts could be deployed to ascertain whether the goods really were in competition. But in my judgment the ECJ was not importing into the principle of fiscal neutrality the entirety of Community competition law. As the Advocate General’s opinion in the Isle of Wight case [2008] STC 2964 put it (at paragraph 19):-

This leads to the second point. Because the principle of fiscal neutrality precludes economic operators carrying on the same activity from being treated differently as far as the levying of VAT is concerned it follows that the application of the principle does not depend on particular local market conditions. That is made plain by paragraph 46 of the judgment in the Isle of Wight case (supra). Under the Sixth Directive local authorities were not to be considered taxable persons in respect of the activities in which they engaged as public authorities. The Isle of Wight Council provided off-street car parking facilities. The private sector also provided such facilities. The provision of parking facilities by the private sector was subject to VAT. Historically the Isle of Wight Council had accepted that it, too, was liable to charge and account for VAT. It then reconsidered its position and decided to claim exemption on the footing that parking was an activity in which it engaged as a public authority. The question was whether such a difference in treatment would lead to a significant distortion of competition. Because the Isle of Wight had paid VAT but (having reconsidered its position as a public authority) was now seeking repayment, it was in the interests of HMRC to argue that different VAT treatment would lead to a distortion in competition. It did so, submitting that what had to be assessed was the nationwide impact on the private sector generally of the exemption of local authorities from VAT on the provision of off-street parking. HMRC was, in effect, arguing that the question had to be answered by reference to a hypothetical market. The local authorities (who wanted to establish that their different treatment would not lead to an actual distortion of competition) argued that the question had to be answered on a public body by public body basis by reference to the area in which the particular body in question provided off-street parking. The UK Government argued that:-

This leads to the third point (to which I will return) which is that the evaluation of the activity may involve looking at matters from the perspective of the consumer (whose behaviour may be influenced by the VAT position) or by reference to the provider of the goods or services (whose economic operation may be affected by the VAT position).

I have so far considered the concept of “similarity” and the concept of “competition”, and how the nature of the enquiry as to “similarity” will tend to justify the inference that similar goods or services “will thus be in competition with one another”. The next stage in the argument is that if there is a difference in the VAT treatment of such similar goods or services then that will tend to distort “competition” between those goods or services. Two citations will suffice to demonstrate what I understand to be the rationale of community law on this point.

In the French Tips case [2001] ECR I-2667, VAT was, in France, charged on the total price paid by the customer in consideration of the services rendered. Under this rule VAT would be paid on any compulsory price supplement to a restaurant bill by way of a “service charge”. But gratuities spontaneously given by the customer would not be subject to VAT, because they were not “consideration”. By an administrative concession compulsory service charges were excluded from VAT (and so treated in the same way as gratuities), provided that four conditions were fulfilled (one of which was that the customer must have been informed in advance of the existence and percentage of the levy). The Advocate General pointed out (see paragraph 45 of the Opinion):-

In the AITC case (supra) management charges were exempt from VAT if supplied to AUTs or OEICs, but subject to VAT if supplied to ITCs. An investment trust had paid £2.7 million of VAT over a ten year period (so that the impact on an individual investor’s returns would have been minute and the disadvantage of the investment trust vis a vis competitor products relatively insignificant). The United Kingdom Government submitted that even if the existence of a competitive relationship was the decisive factor, the difference in tax treatment had no effect on it, because the amount of the extra tax burden borne by ITCs was in practice too small. The Advocate General advised that this argument be rejected because breach of the principle of fiscal neutrality:-

On the basis of these principles it therefore seems to me that a UK taxpayer who establishes that he provides goods or services which are similar (in the sense required by the principle of fiscal neutrality) to those supplied by another economic operator, and are thus in competition (in the sense in which that is to be understood for the purposes of the principle of fiscal neutrality), but whose services are treated differently for VAT purposes, has established that there is a distortion of competition (and that the principle of fiscal neutrality is engaged). He has established that he has been subjected, without justification, to a different tax regime.

But the principle of fiscal neutrality recognises that there may be cases in which different treatment is justified and does not violate the principle of fiscal neutrality. In my judgment it would be for the authority imposing the different treatment to establish the ground which justifies that difference, not for the taxpayer to eliminate all conceivable grounds of difference in order to show that his treatment was without justification. This is, of course, the way the arguments emerge in “infringement” cases before the ECJ, but it is also apparent from cases which are not led by the Commission.

The key authority on this question is the French Drugs Case (supra). In my judgment this provides an elucidation but not a radical restatement of the concepts of and reasoning in all of the earlier cases which I have cited (and those to which I have been referred but not cited). French law provided for differing VAT rates to be applied to medicinal products. Certain products were reimbursable under the social security system: they were charged at a rate of 2.1%. Other medicinal products were charged at a rate of 5.5%. To get onto the list of reimbursable items a medicinal product had to satisfy certain criteria relating to therapeutic effectiveness or cost effectiveness: and where two products equally satisfied those criteria, then preference was given to the branded product over the generic product. At an abstract level all medicinal products could be defined by reference to their curative or preventative properties and were, for that reason, “similar” products. But the Court held (at paragraph 25 of its judgment) that “reimbursable and non-reimbursable medicinal products are not similar products in competition with each other”. (It is not possible from that compound holding to discern whether the panel thought that the products were not “similar” (as that concept is to be understood in this context), or were not in “competition” (as that concept is likewise to be understood); or whether the panel thought there was any substantial difference between the two concepts; or whether there was a difference of view but all could agree on the compound formula).

The Court made two points. First, it said that whilst two medicinal products may have the same curative or preventative effect one might be regarded as too expensive (and therefore not to be included on the list of reimbursable products). It is at first sight difficult to see why that should justify the application of a different rate of VAT to the product price. I think (but am by no means sure) that the Court was saying that (irrespective of VAT treatment) price was of such importance that the two products could not be regarded as interchangeable. Second, the Court held that medicines on the list of reimbursable products “have a decisive advantage for the final consumer”. The decisive advantage was not the lower price by reason of the lower rate of VAT but the fact that the cost was reimbursable (and consequently it is not the lower rate of VAT which provides the reason for the consumer’s decision to purchase). The differing VAT treatment did not therefore infringe the principle of fiscal neutrality. The differing VAT treatment would have been relevant only if it had been shown that “a not insignificant quantity of reimbursable medicinal products” was purchased in circumstances where reimbursement was not possible (e.g. because there was no Doctor’s prescription). Consumers who would not be seeking re-imbursement would then be choosing between two medicines with the same curative properties but with different VAT rates applied to them. But that was not a case that had been argued. (I do not read this as a reference to actual local market conditions in which the individual supplier operated, but rather as a reference to the structure of the overall market).

In my judgment the French Drugs Case cannot stand as authority for the proposition that in order to engage the principle of fiscal neutrality a taxpayer must establish that the difference in VAT treatment distorts competition because it affects the consumer’s choice. That would be inconsistent with Linneweber, with the French Tips Case and with AITC. So far as Linneweber is concerned, the decision in Glawe (supra) had already established that the charge to VAT could not effect the statutorily prescribed proportion of the total stakes inserted by players which was allocated to the players’ winnings, so that even if players knew that one gaming machine was exempt and another was taxed, that could not affect their chances of winning or the amount they could win (and was therefore incapable of affecting the choice between exempt and non exempt machines). In the French Tips Case the consumer paid exactly the same amount for his meal whether or not the restaurateur could take advantage of the administrative instruction (and so to avoid VAT on the service charge). Once again, whether the bill was exempt or non exempt could not affect the consumer’s choice. In AITC the differing tax treatment of the management charges incurred by AUTs and ITCs had imperceptibly small consequences for consumers. In each of these cases there was a breach of the principle of fiscal neutrality because the provider of the exempt service was treated differently from the provider of the non-exempt service for reasons which had really nothing to do with the identification of the types of supply or service which should fall within the general charge to tax made by the Sixth Directive.

In my judgment in the instant case the Tribunal correctly understood the law concerning the relationship between the concepts of “similarity” and “competition”. It asked itself the correct questions and demonstrated no error of law in the answers it gave.

Games conducted under section 14 and those conducted under section 21 were interchangeable: they were all but identical. The “similarity” test is undoubtedly satisfied. At a conceptual level (by reference to the activity as such) the games were undoubtedly in competition with each other. It is no answer to say that Rank was in this instance supplying both games. The relevant competition is between supplies not suppliers. The identity of the supplier is irrelevant. The principle requires the competition position to be assessed in a notional overall market, not an actual local market. The supply and the consideration for the supply were exactly the same; but the differing VAT treatment led to a different financial outcome for the economic operator, and that suffices to distort competition. The fact that the player is not in fact influenced is immaterial (just as it was for the diner or the purchaser of collective investments). There is no discernible justification (in terms of the implementation of the Sixth Directive and the adoption of a common system of VAT) for the differing tax treatment.

I would therefore dismiss the Bingo Appeal.

I turn to the Slots Appeal. As I indicated in paragraph 3 above the key initial question is the true construction of the word “machine” in the phrase “the element of chance in the game is provided by the machine” (words which are found in the relevant sections of both the fiscal and the social stream of legislation). If a game of chance is wholly or partly played on a machine, and the chance in the game is “provided by the machine” then VAT is chargeable. But if the element of chance is provided by something other than “the machine” then the activity is exempt from VAT.

Slot machines constructed or adapted for playing a game of chance may be used for a variety of purposes. They may be pure gambling machines, or pure amusement machines, or machines which provide amusements with prizes. Rank had operated some pure gaming machines under section 31 of the Gaming Act (known as “Jackpots”) and some intermediate amusement machines under section 34 of the Gaming Act (known as “Amusements with Prizes”) where it recognised that the element of chance in the game was provided by the machine: and it paid VAT of £25.9 million. It had also operated other machines constructed or adapted for the playing of games of chance (“the comparators”) where it took the view that the element of chance was provided by something other than the machine, and it did not pay VAT on games played on such machines. Because of this difference in treatment Rank wished, on the grounds of a breach of the principle of fiscal neutrality, to reclaim the VAT which it had paid. Part of HMRC’s response was that the comparators were not themselves exempt as a matter of law, and that rather than re-paying VAT in respect of the gaming machines VAT should be assessed on games played on the comparators. In order to justify its claim to repayment on the ground of differential treatment Rank had therefore to establish that the comparator machines were in law exempt or were in practice treated by HMRC as exempt.

In discussing Re Herrod (supra) I noted that the element of chance in the bingo game played on the terminal (to use a neutral word) was provided by the interaction between the “gay lady” and the electric ball shuffler. What lies at the heart of the present appeal is a processor (to use another neutral word) which performs the same function for the terminal - a “random number generator” (or “RNG”).

The evidence before the Tribunal contained examples of terminals which utilised RNGs. One example (but not a comparator relied on) was the fixed odds betting terminal (FOBT) to be found at bookmakers’ premises. These were terminals that looked like traditional gaming machines and could be played for cash: they allowed a variety of simulated games to be played on them such as roulette, virtual horse- and dog-racing, golf and number games. The chance element that affected the result of the game was provided by an RNG. In Guidance issued in January 2005 HMRC had said:-

It will be observed that HMRC's guidance was founded upon the location of the RNG in relation to the terminal. The location of the RNG in relation to the terminal had been subject of comment in Smith & Monkcom’s “Law of Betting, Gaming and Lotteries” (Second Edition, 2001). In discussing mechanised cash bingo the editors had said:-

Rank contended that machines with an RNG which was operated and located remotely were comparable, as a class, to machines with an RNG operated and located internally. Given the number of machines within each class they did not provide particulars of every machine, but gave examples of particular models from particular manufacturers. The evidence before the tribunal demonstrated the existence of the following comparators:-

One provided to Stanley Casinos Ltd by Bally Gaming Systems which consisted of 30 “Cadillac Jack” machines (costing £150,000) and one server (costing £10,000). The software and the RNG were housed on the server which was situated in a room off the gaming floor. A verification hub and further software were located at a cash desk. The electronic player terminals were connected to these by a live network.

One offered to Rank by IDG-UK based on a simple network solution that did not require the use of expensive back-office computer. This consisted of half a dozen terminals linked to a Clerk Validation Terminal which housed an RNG: this generated separate random numbers for each machine, the machines themselves perhaps running separate games.

A machine offered by Astra in 2004 which had a separately powered RNG contained in a plinth on which the terminal itself stood, with the wire link to the RNG passing through a hole in the bottom of the terminal and serving that terminal alone.

A machine similar to that just described but with the RNG using the power fed into the terminal.

A further variant in which the RNG was not housed in a separate unit and was simply a component to be attached to the wall behind the terminal.

Machines in which the RNG serving that machine was housed in the same case as the terminal itself.

The Tribunal came to the conclusion that “the machine” to which the note referred was the terminal into which the coins or tokens were inserted. They further concluded that where a dedicated RNG was situated inside such terminal so as to be an integral part of it (as for example where the RNG did not have an independent power source), then the RNG and the terminal formed part of a single machine by means of which the element of chance in the game played on the machine was provided; but that if the RNG was not an integral part of the machine (for example, where the RNG was situated outside the terminal and served a number of terminals) then the element of chance in the game was not provided by means of the machine.

The tribunal expressed only a provisional view in relation to terminals fed by RNGs which had an independent power source and were ordered and supplied separately. But they considered the point to be material only if such machines were on the market before multi-terminal machines (since that might then affect the date upon which comparator machines came on the market, which would in turn impact upon both the period for which a disparity in VAT treatment existed, and upon an HMRC argument that there was no breach of the principle of fiscal neutrality because it acted with due diligence in responding to the development of such machines by amending the legislation). The Tribunal therefore directed the submission of further evidence on these matters and arranged for a further hearing.

The Tribunal also concluded that if as a matter of law comparator machines were subject to VAT (and that all machines were “gaming machines” whatever relationship they bore to an RNG) then HMRC had treated them as exempt in practice, and that treatment in itself constituted a breach of the principle of fiscal neutrality.

Before considering HMRC's argument that these conclusions were wrong in law I must rule upon a number of preliminary issues.

First, Mr Lasok QC submitted that the appeal was premature because there had been no definitive finding upon the full range of comparator machines: the Tribunal has simply issued an interim decision based on matters raised at the April 2008 hearing because a full decision on the matters under appeal could not then be made having regard to the amendments on that occasion made by HMRC to the case which it wished to run in opposition to the appeal. Mr Lasok QC submitted that the correct course would be to await the further hearing.

I do not accept this submission. In paragraph 53 of its Decision the Tribunal clearly decided that the language of what is now Note (3) to Group 4 in Schedule 9 of the VAT Act 1994 is not apt to cover a series of terminals linked to one RNG. It thereby held that the supply of a game of chance on such a machine was exempt from VAT and that there was a comparator class which was differently treated for VAT purposes when compared with others of Rank’s machines for the purpose of applying the doctrine of fiscal neutrality. It is true that the actual size of the class has yet to be determined because it may include single terminal machines with independently powered remote RNGs. But the question of principle is appealable.

I do, however, accept Mr Lasok QC’s submission in relation to the category of machines mentioned in paragraph 55 of the Decision (machines with a dedicated but independently powered and separable RNG) in respect of which the Tribunal has invited further evidence and convened a further hearing. It has taken no decision (even in principle) on such machines and may well decide that they (like machines incorporating dedicated RNGs) constitute single machines (and so are taxable in the same way as Rank’s section 14 machines).

Second, an issue arose as to the material I might take into account in construing the statutory provisions which impose limitations upon the scope of the exemption to the general charge to VAT. I have set out the relevant provisions, enacted in 1975, in paragraph 9 above. Mr Vajda QC sought to rely on the witness statement of Mr Phil Sears, Head of the VAT Grouping, Reliefs and Input Tax Team at HMRC. Mr Sears was explicit in saying that neither he, nor to his knowledge any other officials within HMRC, had any personal knowledge of the relevant authority’s thinking in 1975. But he gave “evidence” that before 1975 all gaming machines had been exempt from VAT, and that the exemption created practical difficulties for many smaller pubs and clubs as they became subject to the rules on “partial exemption” (I think, though this was not included in the witness statement itself, because such operators could not deduct input tax incurred on other activities from their exempt income from the gaming machines, which imposed on them an administrative burden). Mr Sears then stated:-

In my judgment it is impermissible to take into account any of that material in construing the enacted words. Mr Sears’s opinion of what must have been the thinking of the public authorities in 1975 is not admissible evidence going to any issue of fact that was before the Tribunal. An internal note by one civil servant to another (which has not been referred to in Parliament in the process of enactment) is no more admissible in construing a statute than one contracting party’s letter to his solicitor would be in construing a contract. The task for me is objectively to ascertain the intention of Parliament as apparent from the words which it used in the enactment and the mischief to which it was plainly addressed. I leave the suggested material out of account.

Third, at the hearing before me (but not before the Tribunal, in the grounds of appeal, or in the skeleton argument) Mr Vajda QC relied on an approach to statutory construction which he called “the always-speaking principle”. He relied upon the decision of the House of Lords in Quintavalle [2003] 2 AC 687. The Human Fertilisation and Embryology Act 1990 was passed to regulate the creation of human embryos in vitro. It prohibited the creation of cloned embryos through the substitution of the nucleus of a fertilised egg with the nucleus of an adult human cell. Between 1990 and 2001 a technique was developed known as “cell nuclear replacement”. The question was whether that new technique was within the scope of the statute and, if so, whether it was prohibited. In paragraph 9 of his speech Lord Bingham observed:-

Mr Vajda QC cited, as an illustration of this principle in operation, the decision in Victor Chandler International v Customs and Excise Commissioners [2000] 1 WLR 1296. The Betting and Gaming Duties Act 1981 forbade the distribution in this country of advertisements or other documents relating to the making of bets with off-shore bookmakers. In the late 1990s VCI, based in Gibraltar, proposed to broadcast on teletext the odds it was offering. The court held that a teletext screen was a “document” for the purpose of the 1981 Act.

I, of course, accept Lord Wilberforce’s authoritative statement. But insofar as its application to the construction issues arising on the 1975 Order depends upon a specific factual basis, I would observe that the Tribunal did not find (and was not asked to find) facts directly relevant to this issue. I accept that evidence led on other issues establishes that the precise comparator machines on which Rank relied did not exist in 1975: and I would take judicial notice of the fact that the specific technology which the RNGs utilised did not then exist. What is much less clear on the evidence is whether the random generation of numbers by the technology existing in 1975 (for example, by an electronic ball shuffler) was part of the policy background to the 1975 Order. In the absence of a proper factual foundation I have not found “the always-speaking principle” to be of great help.

I turn to the issues on the Slots Appeal. The first is whether the comparator machines were, as a matter of law exempt from VAT. HMRC argued that they were not, even where the RNG formed part of a physically separate unit (such as a clerk validation terminal) networked with terminals loaded with software supporting different games on different terminals. It was suggested that either the entire network was “the machine” (irrespective of how many player terminals there were served by the single RNG): or alternatively that each player terminal was “the machine” and that the RNG was a piece of “apparatus” which formed part of “the machine” (so that the clerk validation terminal or other RNG was not a separate machine but was a part of every terminal it served). Three approaches were suggested which lead to this conclusion.

The first was the functional approach. This approach argues that the physical location of the RNG and how it is connected to the terminal on which the game is played are entirely irrelevant. The RNG might (the example is mine) be located in Las Vegas and communicate its randomly generated numbers by satellite signal to a game terminal in Blackpool: but for the purposes of the Act the RNG which generates the determining event, and the terminal which houses the game the outcome of which is governed by that determining event are all one machine. That is so because it is being “used” as one machine, as it was designed to be (because you cannot play a game on the terminal without an RNG). To take account of where the RNG is, how it is powered and how it is connected to the terminal was described as “absurd”.

I do not accept this approach as correct. The function of the RNG cannot, of course, be ignored: it is inherent in the condition that “the element of chance in the game [be] provided by means of the machine”. But it does not follow that all other considerations are irrelevant. Looked at as a whole the definition of “gaming machine” has regard to physical characteristics. It was common ground that the relevant machines had to have a slot to take payment by cash or token, and that a payment for a game made at a remote cash desk in order to play a game on a terminal would not bring the terminal within the definition of “gaming machine”. That is not absurd: it is simply part of the definition. It is no more absurd to say that a determining event generated by a physically separate non-dedicated unit which governs the outcome of the game played on the terminal also takes the terminal outside the definition of “gaming machine”.

If one is to adopt a “functional” approach I consider it equally possible to say that functionally the terminal is a machine which, through embedded software, records the result of a game and presents it to the player, but that the outcome of the game is in fact determined by a separate machine generating a stream of random numbers (at perhaps 100 times per second) – as is the case with mechanised cash bingo which HMRC had never (until the very closing part of Vajda QC’s speech in reply) suggested was conducted on gaming machines. Or, indeed, an FOBT (the nature of which is to be considered at the adjourned Tribunal hearing).

The second approach may be called “the regulatory argument”. This takes as its starting point Section 26 of the Gaming Act 1968 which also defines a machine constructed for playing a game of chance in terms which include a requirement that “the element of chance in the game is provided by means of the machine” (though other elements of the definition are different from those used in the VAT legislation). This analysis therefore starts with the same words set in a different context in an Act passed for a different purpose. But HMRC submit that that is legitimate because of the agreement that the social stream and the fiscal stream of legislation must be read together. HMRC submit that Part III of the Gaming Act 1968 (which begins with section 26) has the object of controlling the sale, supply and maintenance of the gaming machines to which it applies, and seeks to define the circumstances under which and the conditions in which they may be used for gaming. Accordingly, it is argued, the definition must be approached with that object in mind and construed purposively so as to achieve that result. It makes no sense to seek to regulate machines under Part III where the outcome of the game is determined by an RNG dedicated to the terminal and situated within or adjacent to it, but not to seek to regulate under Part III machines where the outcome of the game is governed by an RNG situated remotely in some other piece of equipment or apparatus. Thus machines that fell outside Part III would not be covered by provisions that prevented access by children to such machines, and not subject to the limits on the number of machines per casino that are to be found in Part III. If it is to be assumed that Parliament intended to regulate substantially similar machines in the same way then a definition which places machines with an integral RNG on one side of the definitional boundary and machines with a remote RNG on the other side of the definitional boundary would not achieve the regulatory purpose. If the machines are to be treated as the same for the purposes of the social stream of legislation then they must be treated as the same in the fiscal stream of legislation.

I do not accept this argument. I begin with an acknowledgment that in accepting for the purposes of this Appeal (and without comment) the concession that the same phrase must be read in an identical sense in both the social stream and the fiscal stream of legislation I am conscious that there is a vast hinterland of regulatory, criminal and fiscal legislation relating to gaming by machines of which I have been referred to only a very small part. So, for example, if I accept HMRC’s argument as to the meaning of the relevant words in the Note to Group 4 of the VAT legislation I would in fact render criminal the activities of many gaming machine operators (thought to be lawful at the time) for a substantial period. That, of course, is not a reason for not giving the words in the VAT legislation their true legal meaning. But it is a reason to exercise caution when faced with arguments which start at so distant a point from the enacted words which I have to construe. In truth, and despite Mr Vajda QC’s very best efforts, I have found it very difficult to discern any coherent picture in the provisions regulating betting and gaming. I find myself aligned with the editors of Smith & Monkcom (supra) where they state (at paragraph B 10.8):-

Part III plainly contemplates that there will be some gaming machines that are within the scope of its provisions and some that are not (see for example the terms of Section 28(2) and its reference to “whether a machine to which this part of this Act applies or not”). The function of Section 26 is to define the boundary. The boundary is defined partly by reference to physical characteristics (the existence of a slot). There is no reason for thinking that all other physical characteristics are entirely irrelevant. “The machine” by means of which the element of chance in the game is provided has got to be the machine with the slot in it. The regulatory context helps me to decide that the argument that “the machine” is the system of terminals linked to a common RNG is wrong (because it would effectively mean that the restrictions on numbers of machines on any given premises for which Part III provides would be meaningless since the restriction would relate to the RNG in each system, to which vast numbers of playing terminals could be linked). But it does not help me to decide whether a machine which cannot be played on its own, but is dependant upon its connection to a remotely housed RNG (which itself also serves other machines providing different games) falls to be regulated under Part III, or under Part II and/or other legislation relating to lotteries and amusements. The question is not whether the machine shall be regulated or unregulated, but as to what system of regulation it is to be subject.

The third approach may be called “the linguistic argument”. HMRC argue (and I accept) that in section 26 of the Gaming Act 1968 the word “machine” in the phrase “playing a game of chance by means of the machine” and “the element of chance in the game is provided by means of the machine” includes “any apparatus”: see section 52 of the 1968 Act. HMRC submit that in the shorter Oxford English Dictionary one definition of “apparatus” is that which is necessary for the performance of some activity or function. In the expression “the element of chance in the game is provided by means of the machine” the word “machine” includes anything that is necessary for the performance of that function, so that the “machine” includes the RNG necessary to provide the element of chance. Where the RNG performs that function for several terminals “the machine” is either the system as a whole or the terminal plus the remote RNG.

This is a formidable argument. Mr Vajda QC bolstered it by reliance upon the “always speaking” principle as justifying a flexible interpretation of the concept of “the machine”.

I have already rejected the first of the suggested alternative constructions (namely that “the machine” means the entire network of terminals connected to a single RNG). It is accordingly the second on which I focus. I do not accept the argument because I consider that it places too great a burden on the terms of section 52. That section does not say that for the purposes of the Act each machine includes any apparatus used in connection with it. It simply says that the word “machine” includes within its scope “apparatus”, so as to render sterile any debate about whether a particular piece of equipment is a machine or a piece of apparatus. Thus in a multi-terminal system with a single RNG each element in the system could be called “a machine” or “apparatus”. But that does not help with the question whether the apparatus containing the RNG is part of the apparatus containing the slot.

In my judgment the Tribunal adopted the correct approach. In the phrase “the element of chance in the game is provided by means of the machine” the “machine” to which reference is made is obviously that which is constructed for playing the game of chance, that which the player plays and that into which a coin or token is inserted (reading the Note to Group 4 consistently with Section 26 of the Gaming Act 1968). Nobody argued that the resolution to the problem of the multi-terminal system utilising a single RNG lay in the words “by means of” and there was no examination of how the remotely generated number came to determine the outcome of the game played on the terminal. The argument proceeded on the footing that the element of chance had to be provided by “the machine” and the problem lay in identifying “the machine”. The “element of chance” is the determining event which governs the outcome of the game being played on the machine which has the slot in it and which the player is playing. Where the determining event is a random number there is I think no difference in principle between a human being selecting a numbered ball, an electric ball shuffler (such as that used in the National Lottery) producing a numbered ball or a microprocessor emitting a stream of numbers. It is a question of fact in each case whether that determining event is produced by “the machine”, and fine distinctions might have to be drawn. In my judgment the principle by reference to which those judgments have to be made is whether the outcome of the game may sensibly be regarded as determined by an external event which the machine records or is produced by the machine itself. Like the Tribunal I would hold that the random generation of a number in a separate unit which serves various player terminals (which may themselves be running different games) is properly regarded as an external event and not one produced by the machine that the player is playing. Like the Tribunal I do not think it is possible to elaborate further.

In the result I hold that there is no legal definition of the word “machine” which compelled the Tribunal, on the evidence which was before it, to conclude as a matter of fact that the element of chance in the game was always provided “by means of the machine”.

Once it is held that there are games of chance played on “gaming machines” (the supply of which is subject to VAT) and there are games of chance played on machines that are not “gaming machines” (which are not subject to VAT), it is apparent that a breach of the principle of fiscal neutrality may have occurred. Applying the principles outlined above I would again hold that there was such a breach. Games played on such terminals were interchangeable. The “similarity” test is undoubtedly satisfied. At a conceptual level (by reference to the activity as such) the games were again undoubtedly in competition with each other. The supply and the consideration for the supply were exactly the same; but the differing VAT treatment led to a different financial outcome for the economic operator, and that again suffices to distort competition. There is again no discernible justification (in terms of the implementation of the Sixth Directive and the adoption of a common system of VAT) for the differing tax treatment.

I would therefore also dismiss the Slots Appeal.

The Tribunal went on to hold that if they were wrong in holding that there was a class of comparator machines that was as a matter of law exempt from VAT then HMRC in practice treated multiple terminals linked to separate free-standing RNGs as exempt. Although the point does not strictly arise having regard to my conclusion on the first limb of the appeal, it was an alternative basis for the decision below and was fully argued on this appeal. I ought therefore to deal with it.

The doctrine of “fiscal neutrality” is applied not only by reference to the strict fiscal position at law but also by reference to “administrative practices”. HMRC submitted that for such a practice to be relevant in the context of fiscal neutrality the practice had to be clear and unambiguous, consistent and based upon full knowledge of the facts by the relevant public authority. As Mr Lasok QC good-humouredly pointed out, this formulation enables a public authority to act in a discriminatory way that distorts competition and undermines the principles of VAT uniformity provided that it does so in a way that is obscure, variable or based on ignorance.

Mr Vajda QC acknowledged that there was no authority containing a statement of principles supporting his submission as to the circumstances in which the doctrine was engaged: but he said it was implicit in the cases.

In the Metropol Case [2002] ECR I-81 the Austrian Government drew a distinction for the purposes of the deductibility of VAT between cars and other vehicles. In 1987 the Federal Minister for Finance issued a circular which said that minibuses did not fall within the restrictive tax provisions applicable to cars, and he set out the criteria by which minibuses could be distinguished from cars. The circular had no legislative force: but it formed the basis of “a consistent administrative practice” (as it was described by the court). The Sixth Directive enabled member States to maintain any rules of national law that excluded the right to deduct VAT which were actually applied by their public authorities. The court held that this provision “does not refer only legislative acts in the strict sense, but also to administrative measures and practices of the public authorities of Member State concerned”. Mr Vajda QC submits that this endorses his submission about the nature of a relevant “practice”. He proffered as a further example the decision in Danfoss Case C-371/07, which related to an administrative practice that had remained in force from 1978 until 1999. I agree that these cases illustrate what is sufficient: but they do not establish what is necessary.

Light is thrown upon that by the decision of the ECJ in the Commission v Greece Case C-489/06. What was in issue was compliance by Greek hospitals with EC rules relating to public procurement. Greek domestic law was compliant: but certain hospitals did not act in accordance with it. Greece claimed that repeated non-compliance by the contracting authorities was exceptional and could not constitute an infringement of the EC public procurement rules. The Court held (in paragraph 46) that:-

This is consistent with the position under domestic law. In Commissioners of Customs and Excise v National Westminster Bank PLC [2003] STC 1072 a car leasing company purchased vehicles from a manufacturer, but it also received payments from the manufacturer which it regarded as consideration for the bulk purchasing of cars. It paid VAT on the entire price of the vehicles, rather than treating the payment by the manufacturer as a discount on the full price, and paying VAT only on the net difference. It made a claim for overpaid VAT. Customs responded by saying that a repayment of VAT would amount to “unjust enrichment” since the overpaid VAT had already been passed on to the ultimate customer. Other car leasing companies had reclaimed overpaid VAT but not had the “unjust enrichment” point taken against them by their local VAT office. When NatWest complained about this difference in treatment Customs declined to readdress the question and stated that “our failure to consider unjust enrichment in appropriate cases is regrettable”. The hearing before Jacob J. concerned the availability of the defence of “unjust enrichment” in relation to the repayment claim. Jacob J. described the position thus:-

The Tribunal found that from March 2004 at the least there was public debate about the comparator machines; that on 14 July 2004 there was a meeting between the British Amusement Catering Trades Association and HMRC to discuss the comparator machines (which it was said had been around since 2001) at which it was agreed that these did not constitute “gaming machines”; that following that meeting HMRC wrote to BACTA (in terms approved by the head of the relevant VAT team and that had been the subject of some seven weeks’ consideration) to confirm that the comparator machines

In my judgement the Tribunal made no error of law in concluding on this evidence that there was a relevant “practice” under which some comparator machines were not treated as gaming machines (whatever the true position at law) and were treated as exempt from VAT. The acceptance that various operators did not have to pay VAT on supplies made by such machines was not a series of coincidental blunders by individual VAT offices but was part of a coherent and considered approach by the regulatory and fiscal authorities to what was a difficult question: and it informed the basis upon which tax was accounted for by operators and collected by HMRC.

Mr Vajda QC submitted that such published guidance could not be a relevant “practice” because HMRC had changed its position and had now issued retrospective assessments based on its revised view of the law, meaning that there would now not be a period in which the subject machines and the comparator machines were differently treated for VAT purposes. I do not accept this argument. First, retrospective assessments can only be raised in respect of those periods that remain open. It is too late to raise retrospective assessments in respect of some periods. Second, the supplies, some of which were subject to VAT and some of which were exempt from VAT (and which are on that account to be assumed to be affected by the disparity in treatment) long ago completed. Equalising treatment now cannot undo competitive distortion that occurred at the time. Third a subsequent change of position cannot be probative of what was the position at the time when the facts occurred.

I therefore hold that Tribunal’s alternative ground for its decision is also sound in law and would have dismissed any appeal against it.

For these reasons I dismiss both appeals: and in doing so acknowledge my indebtedness to Counsel for the clarity of the written and oral advocacy. I do not expect attendance of legal representatives when I formally hand down judgment. I will on that occasion adjourn consideration of all matters of costs and all other applications to a hearing to be fixed through the usual channels (and for that purpose extend any applicable time limits until the conclusion of the adjourned hearing).