DFS Furniture Company Ltd v Revenue & Customs (Rev 1) [2009] UKFTT 204 (TC)

FTT-Tax
DFS Furniture Company Ltd v Revenue & Customs (Rev 1)
[2009] UKFTT 204 (TC) · 2009-08-07
[1]This is an appeal of two parts. The first part concerns whether advertising and certain other inputs were direct cost components of the Appellant’s supplies both of sofas and of insurance intermediary services. The second relates to whether or not the Respondents “approved” a special method for the apportionment of the Appellant’s non directly attributable VAT.[2]This is a decision in principle on these issues as they relate to appeals against assessments made on the Appellant, DFS.[3]DFS is a well known supplier of sofas. (In this decision we have used the term “sofas” to encompass the variety of types of upholstered sitting furniture which DFS supplies.) There will be few who have not seen its television and other adverts. But if you go to one of the Appellant’s stores and decide to buy a sofa then you may be offered, and you may agree to take, one of two types of insurance. The first type is anti stain insurance: that speaks for itself. The second is Personal Protection Insurance (PPI) and is only offered if you have bought a sofa on interest free credit terms: it is insurance against illness or unemployment which could affect your ability to repay the credit. About 20% of customers take anti stain insurance and about 30% take PPI.[4]DFS does not itself provide the insurance but arranges for the customer’s application to be accepted by a third party insurance company. Those insurance companies pay commission to DFS for the introduction of the customer to them. Thus if the result of your visit to a DFS store is that you take both a sofa and an insurance policy, DFS will have made two VATable supplies: a taxable supply of the sofa to you and an exempt supply of intermediation to the insurance company.[5]DFS incurs costs in its business on which it bears input VAT. That input VAT is recoverable (and offsetable against the output tax it has charged and owes HMRC) in accordance with rules in the relevant EC Directive (the relevant directive at the time was the Sixth Directive) and domestic legislation. Broadly, unless HMRC (we use HMRC to include the previous manifestation of the Respondents as the Commissioners of Customs and Excise) permit or direct a “special method”, the input tax is fully recoverable if the input is a direct cost component of a taxable supply, not recoverable at all if it is a direct cost component of an exempt supply, and if it is neither – if it is “residual” – will fall to be recovered in accordance with a fraction determined by the legislation.[6]There is no doubt that DFS’ costs of making or buying sofas are direct cost components of its taxable supply of the sofas. This appeal concerns other costs. In the first part of the appeal DFS argues that, in particular, its advertising costs (“think sofas: think DFS”) are a direct cost component only of its sales of sofas – that its advertising is directly used only in selling sofas and not in insurance intermediation. In the second part DFS argues that HMRC had approved a “special method” of VAT recovery which permitted almost all its input tax to be recovered (including all that on its advertising).[7]The law and the relevant facts in relation to each part are almost wholly distinct, and we shall therefore deal with the two parts separately.[8]We heard oral evidence from Mr Massey who joined DFS as its deputy managing director in 1988; Mr Barnes, who joined DFS as its Finance Director in 2003; Mr Todhunter, who joined DFS as its Company Secretary in 1983, was financial controller from 1988 to 2003 and remains with the company as company secretary; Mr Gibson, who was HMRC’s lead officer for DFS from 1996 until 2008; Mrs Harrison the officer of HMRC who from 2006 conducted the discussion with the company about partial exemption; and from Mr Hall, who had been an officer of HMRC between 1987 and 1997and gave evidence as to the practice and procedure of HMRC but had no direct involvement with DFS. We also had four bundles of copy documents. From that evidence we make the findings of fact in the first section of Part I of this decision, and the “Events since 1986” section in Part II, (together with any facts related above). PART I : Attribution Our findings of fact[9]During the period relevant to the appeal DFS operated about 77 retail stores. Through 73 of them it sold sofas and other upholstered furniture, and through four of them cabinet furniture. DFS’ sales were not made in this period through its website. Mr Massey told us that DFS had a telephone ordering service. On reviewing his evidence it was not clear to us that this was available or used during the period relevant to the appeal. We conclude that in that period telephone sales were not a significant part of DFS’ activity. .[10]DFS had three factories in the UK which manufactured sofas. Its own factories supplied 15-20% of the furniture it sold. The balance it bought from external manufacturers.[11]A typical showroom will display some 80 different models. Each model will be available in various forms: four, three and two seater sofas, sofa beds, chairs, footstools or corner units. And each will be available in three or four different coverings (fabric or leather) and each fabric in perhaps 60 different colours. Endless sofas.[12]DFS are and were a one product specialist upholstered furniture business. They aimed to be able to offer sofas to appeal to 80% of the market: as Mr Massey said, perhaps not to those who live in castles or cardboard boxes, but to all the rest of us. Endless sofas.[13]A sofa showroom is divided into 60 or 70 room sets each set displaying a different model. The displays will be complimented by table lamps or pictures. DFS do not aim to sell theses accessories, but if a customer buys a sofa and says “I want that lamp too”, the lamp will be sold.[14]The interactions between the sales staff and customers will vary in nature, but generally the customer’s first decision, after perhaps discussion about prices, styles and fabric and colours, and after perhaps more than one visit and more than one discussion, will be to decide upon the style and form of sofa he or she wants and to settle on the fabric and colour. At this stage the customer will generally sit down (probably on his or her choice of sofa) with the salesman and an order form. The salesman will start to fill in the order form with the details of the furniture being ordered. Around this time there may also be a discussion about extras such as cushions, the care of the sofa, care products warranties and anti stain insurance; and also about delivery dates and the method of payment.[15]The customer will generally be offered the choice of paying by cash immediately (or in part on delivery) or some form of interest free credit. Interest free credit is available on all goods if the customer is accepted by the relevant finance company: of those who seek credit [a very substantial proportion] are approved.[16]Whilst we believe that there will be times when anti stain insurance is discussed before a customer settles on his order, we believe it likely that in the substantial majority of cases anti stain insurance will be discussed after the customer has decided to buy, and a final decision as to whether to purchase the insurance will be taken only after the details of the furniture order have been entered on the order form (although normally the order form will be signed, and the customer formally committed, only once it is complete with an entry in relation to anti stain insurance – if it is being taken up.)[17]Likewise, although there may be cases in which PPI may be discussed before the decision to buy is made, we believe that in the substantial majority of cases it will be discussed and settled upon as part of the decision as to how payment will be made. PPI is relevant only if some sort of credit is taken. Mr Massey described the PPI decision as a final option at the end of the process. We accept this: the structure of the order form and the credit application would generally be followed by a salesman, and PPI comes on that form after matters such as the details of the sofas ordered; a salesman’s focus will be on securing the order for the furniture (for which he or she receives commission) before selling PPI (for which he or she receives less commission); whilst he or she is likely to refer to interest free credit in leading a customer to a purchase it is in our view less likely that the additional cost of PPI will be raised at an earlier stage.(Although we note again that the customer will sign an order incorporating sofa specifications and other matters such as PPI only at the end of the process.)[18]Salesmen receive commission on completed orders. They earn commission on the sale of furniture. [The commission varies with the type of sale.] They also earn commission on sales of PPI and anti stain insurance.[19]After completion of the order form and (in [those] cases where credit is taken) the credit details form, a copy is given to the customer and copies are retained for processing by DFS. There is an office area in the sales floor of most showrooms where processing takes place. Processing results, in the case of credit sales in the credit application being forwarded to the finance company, and in the case of cash sales, and credit approved orders, in instruction being sent for the manufacture of the sofas. Anti stain insurance and PPI acceptances will be recorded in DFS’ systems at this stage. Orders are fulfilled by the manufacture and delivery of sofas. They are not sold from stock. Anti stain insurance[20]Anti stain insurance is not offered on leather sofas. 62% of sales are of leather sofas. So anti stain is potentially relevant only to 38% of sales. About 84% of those customers who buy fabric sofas take up the offer of anti stain insurance. Thus only about 32% of all customers take up the insurance. The decision to take up anti stain insurance is recorded by ticking a box on the order form.[21]DFS has offered anti stain insurance since 1997. On 29 September 1997 it entered into an agreement with the insurer Pinnacle Insurance limited and the claims handler and administrator, Guardsman UK limited. The agreement describes DFS’ wish to provide a scheme under which its customers will have the opportunity of purchasing anti stain insurance, and for the purposes of the scheme appoints Pinnacle and Guardsman. DFS is not expressly appointed agent of Pinnacle, but Pinnacle agrees to accept all insurance sales made under the scheme. Guardsman is appointed to deal with all claims handling. DFS is to collect the premium and account for it to Pinnacle. Insurance rates are to be agreed between DFS and Pinnacle, and Pinnacle agrees to pay DFS commission at rates to be agreed between them.[22]Since the inception of the agreement DFS has applied an anti stain chemical spray to those insured sofas suitable for spraying (not all fabric sofas are). The agreement does not expressly impose any obligation upon DFS to apply the spray, although there is a clause under which DFS agrees that the intellectual property rights in the Guardsman anti stain product will remain with Guardsman, and the terms and conditions of the insurance policy refer to the benefits of Guardsman’s fabric protector. Mr Massey said that DFS regarded itself as obliged to spray, and that Pinnacle worked on that basis too. We find that, by the time of the periods relevant to this appeal, DFS were contractually bound to apply the spray to all eligible sofas which were insured.[23]Where the anti stain spray is to be applied, it will, in the case of sofas manufactured by DFS be applied at its factory, and in the case of the sofas from third party suppliers, be applied at the stores before despatch. Each of the stores has a warehouse area into which deliveries of sofas are received and checked, and from which they are dispatched. There is a small area a few times the size of an average sofa in which the spraying is done and in which the relevant equipment is stored. It occupies a small fraction of the warehouse area.[24]The following table indicated the take up of anti stain insurance. Furniture orders Fabric: 38% Leather: 62% Anti stain take up 84%:yes; 16% No No insurance offered[25]Thus anti stain insurance is sold on about 32% of orders. Of that 32% about one third are sprayed in the factory, one third at the store, and one third are not suitable for spraying. Interest free credit and PPI[26]PPI is relevant only where customers take credit. It provides insurance against disability, unemployment or death. DFS entered into agreements with a number of credit providers under which it agreed to provide to the credit provider applications for credit arising from particular stores. DFS was not expressly appointed to act as agent for the credit provider. On receipt of the application the credit provider would accept or decline to give credit to the customer. The customer would pay for the PPI as part of his or her monthly payment.[27]We did not see a copy of an agreement between DFS and a PPI provider, but we accept that DFS was entitled to receive and received commission from each insurer in respect of the insurance placed with it.[28]Interest free credit was available in the relevant period on “straight terms” - where monthly payments started immediately - or on deferred terms - where monthly payments started after an interval. In each case the customer became indebted to the credit provider for the amount of his or her sale and the credit provider made a payment to DFS. That payment would be made net of a subsidy borne by DFS to cover the cost of the credit – that subsidy would thus vary with interest rates and the term of the credit. This mode of payment gave rise to the Primback issues which we discuss in the second part of this decision.[29][Special arrangements applied to those customers who elected for credit on deferred terms and then repaid their loan before the first interest payment became due.][30]When a salesman had determined that a customer would buy on credit, he or she would discuss PPI with the customer. Since January 2005 DFS has been regulated by the FSA in respect of the sale of PPI and was required to make an assessment and recommendation. This was done by filling in part of the order form in which the age and employment status of the customer were detailed and a formal recommendation made as to suitability of the policy. Formal training is given to the salesmen in relation to the provision of advice on PPI.[31]The following table indicates the take up of PPI as a proportion of orders: Cash Sales : [less than half of all sales] Orders with Credit sought: [more than half of all sales] Credit declined: [small fraction] No PPI: [more than 1/3 of all sales] PPI: [less than ¼ of all sales] Deferred: [majority of PPI] Straight: [relatively small part of PPI] [Settled early] [Not settled early][32]The following table indicates in broad terms the financial weight of DFS’ supplies of sofas and insurance intermediation. The figures in the later columns are taken from Mr Barnes’ evidence, and those in the first columns from documents in the bundle before us and relate to the 2007 year. We find that these were approximately the financial effects in the period of the appeal although there were variations from year to year. The figures are not intended to be precise findings of amounts, but rather indications of broad relative proportions (thus for example the exempt income is a total of 4.5% of all income in the table; in fact it varied between 3.8% and 7.6% in the period 2000 to 2008): Total gross income %age of total gross income % of total gross margin % of gross margin on orders with insurance Sales of Sofas £604m 96% 94% 80% PPI Commission £10m 1.5% 1% 4% Anti stain commission £19m 3% 5% 16% The final column indicates the split of the contribution to gross margin of the various components in the case of sales where some insurance is taken up. The penultimate column indicates the split of gross margin across all sales. We note the relative unimportance of insurance intermediation income. The take up of anti stain insurance and PPI as a percentage of sales varies from store to store depending on the area in which the store operates and the skill of the staff and their willingness to sell the insurance products. The Adverts[33]We saw videos of DFS’ television adverts, copies of its poster adverts and copies of booklets and direct mailing literature. (We should mention that since the hearing at least one of us has seen DFS adverts in newspapers and on television. We have not taken any formal note of those adverts but they were substantially the same as those we were shown.)We noted the following features of those adverts:(1) none made any reference to any form of insurance;(2) all displayed pictures of sofas;(3) the television adverts displayed many different styles and coverings of sofas generally in quick succession. The booklets contained pictures of many different shapes, colours, and styles of sofa;(4) almost all made reference to the availability of interest free credit in some form;(5) almost all expressly or implicitly urged the customer to come to DFS and buy a sofa quickly – for example by indicating that there was £250 off, or sofas were half price, for a limited period;(6) one brochure and one letter indicated that coffee tables or lamps might be available, but this was incidental to the overwhelming emphasis on almost every sort and shade of sofa you could imagine. Another indicated that scatter cushions and free delivery were available;(7) they contain DFS’ logo, and generally the catch phrase “think sofas, think DFS”;(8) the addresses of DFS stores were listed on the back of the brochures; and(9) DFS’ website address was given.[34]If the question before us was: what did the adverts relate to? We would have no hesitation in answering that they related only to sofas and (but only to a minor extent) to cushions, tables and lamps, and did not relate to insurance. If we had not thought it before seeing the advertisements, we would have thought afterwards:
“think sofas, think dfs”
. The object of the adverts was clearly to encourage the public to come to DFS and buy a sofa - tomorrow.[35]Posters generally show only one sofa. If there is a poster campaign different sofa may be shown on different posters. Poster campaigns are undertaken only in conjunction with television advertising.[36]Advertising is conducted mainly in campaigns. There will be television campaigns for example close to bank holidays. The sofas to be featured in a campaign are chosen by the marketing department. The chairman is closely involved. The spending on advertising and marketing is substantial, amounting to [a material proportion] of the value of the sales of sofas.[37]Mr Massey said that a customer will come to a store because he or she likes the products they see in the adverts. We think that is likely and that it may be either because they conclude from the adverts that they like a particular sofa, or a particular type of sofa, or because they conclude that there may well be a sofa there they may well like.[38]Sales performance across all stores is monitored closely. A Product Print is produced by the accounting system showing the sales of each model of sofa ranked by the value of sale in a given period. The print does not refer to PPI or Anti stain income. A copy of the print for sales on 1 and 2 April 2006 was before us. We accept Mr Massey’s evidence that these prints were used by the marketing department to measure the success of an advertising campaign by comparing the relative value of total sales, and the relative values of sales by type of sofa in periods before, during and after the campaign, and examining whether the sales of sofas featured in the adverts went up in the league table. We accept that if during a campaign the sales of a featured model did not go up then that model would be dropped: the aim is to promote “winning” ranges of furniture rather than less successful lines.[39]We think it likely that the effect of an advertising campaign was to increase sales, and to increase sales of sofas of, or similar to, the types featured in the adverts. We reach the first conclusion because it would be unlikely that a commercial enterprise would spend so much on advertising if it did not have that effect and the second conclusion from the scrutiny of the Product Print by the marketing department during the campaign.[40]Mr Massey accepted that one of the functions of advertising was to get people into the stores. Mr Barnes implicitly accepted in correspondence with HMRC that the proportion of sales in which insurance products were taken was not affected by an increase in sales. Thus increased sales led to increased intermediation income[41]Head Office Activities(1) The arrangements with Pinnacle and the PPI providers were long standing and stable. Mr Massey had occasional meetings with the insurers.(2) The FSA regulation of the PPI business gave rise to a committee, the Treating Customers Fairly committee, which met monthly in a smallish room at head office. Among other things it monitored the proportion of sales on which PPI had been taken at the various stores.(3) The accounting system recorded and reported insurance sales and commission as well as sofa sales.(4) The training of staff for FSA purposes would we believe be managed or overseen from Head Office. The Relevant Law Summary of the applicable principles[42]We were referred to BLP Group plc v C&E Comms [1995] STC 424, Midland bank plc v C&E Commrs [2000] STC 501 , Abbey National plc v CCE [2001] STC 297 , Dial a Phone v C&E Commrs [2004]STC 987 , C&E Commrs v Southern Primary Housing Association Limited [2004] STC 209, St Helens School Northwood v HMRC [2007] STC 633 , and Mayflower Theatre Trust Ltd v HMRC [2007] STC 880. We draw the follwing principles form those cases:(1) Article 2 of the First Directive provided that VAT should be chargeable “on each transaction” after deduction of the amount of VAT “borne by the various cost components”. In this formulation the question is whether a particular cost is a cost component of a particular transaction;(2) Article 17 of the Sixth Directive provides for deduction of input VAT “insofar as” the inputs “are used for the purposes of”, or “ are used for” taxable transactions;(3) sections 25 and 26 VATA 1994 and reg 101 to 109 SI 1995/2518, and their respective predecessors are designed to implement those provisions.(4) we must base our conclusions on objective considerations: the ultimate aim of the trader is irrelevant ( BLP) ;(5) we should treat the “cost component” and the “used for” tests as equivalent and may apply either or both (see BLP , and paras[18] and [24] in Dial a Phone );(6) if there is a direct and immediate link between input and output we should treat the “used for” test as satisfied ( BLP ). In Dial a Phone Jonathan Parker LJ (at [28]) indicated that it was clear from the judgements in BLP and Midland that there was no material difference between the “direct and immediate link” test and the “used for” test;(7) In Midland further guidance was sought from the ECJ on the meaning of “direct and immediate”. Little additional guidance was given. But at [29] and [30] the Court indicated that the input expenditure must be a cost component of the output transactions which “utilise” the inputs. The Advocate General in Midland did attempt to give some further help. He said at [29] that:
“The use of the two adjectives “direct” and “immediate” cannot but refer to a particularly close link between the taxable transactions…carried out by a taxable person…and the goods or services supplied. In particular the adjective “direct” means that there cannot be the appropriate link …where a third transaction takes place between them breaking the causal chain, or where the link is very distant in time…The adjective “immediate” denotes a particularly close temporal proximity between the two transactions…the…requirement is that the time which has elapsed between the two transactions should not be too long.”[Our italics] And, at [31]: “…there is always a direct and immediate link …wherever in the light of objective factors the [inputs] are used by the taxable person to carry out one or more taxable transactions…Moreover the link must be identifiable according to objective criteria; that generally means that the link should reflect the normal relationship between the two supplies, so that the second should follow the first not in mechanical way, but according to the normal and regular order of causal chains.” (8) it might however be a necessary condition for the existence of a link that the input was essential for the particular output: in Southern Primary , Jacob LJ said, at [33]: “One can look at it another way. There is nothing about the development contract as such which makes the land purchase essential. If the housing association had already owned the land or had bought it from some third party, the inputs of the development contract would have been just the costs of carrying it out. The fact that there were commercially linked land transactions does not mean that those transactions are directly linked to the costs of the development contract. One would not say that the cost of buying the land was a cost of the development contract itself, It follows that the input tax on that cost is not a cost of the contract.”
We note the use of “essential”. Jacob LJ seems to be saying that if the input is not in some way “essential” then it cannot be a cost component. In other words, being essential to the output may be a necessary condition (although not a sufficient condition) for the necessary linkage; (9) in determining whether there is a direct and immediate link the existence of an intermediate taxable transaction should not of its own lead us to conclude that there is no such link ( Dial a Phone ); (10) we should not treat the existence of a closer link to A as precluding a conclusion that there is a (sufficiently) direct link to B. In Abbey , the advocate General said that the issue was:
“whether the taxed output is a cost component of the a taxable output, not whether the most closely linked transaction is itself taxable …the conclusion to be drawn from [BLP] is that the question to be asked is not what is the transaction with which the cost component has the most direct land immediate link but whether there is a sufficiently direct and immediate link with taxable economic activity ….it [remained clear from BLP] that the ‘chain breaking effect’ …of an exempt transaction will always prevent VAT used for such a transaction from being deductible…The need for a “direct and immediate link” thus does not refer exclusively to the very next link in the chain but serves to exclude situations where the chain has been broken by an exempt supply.”[Our italics]. In the judgement of the Court no reference is made to the sufficiency of the direct and immediate link other than to recite the Commission’s contention to that effect. Instead the court merely restates the direct and immediate test. In Dial a Phone Limited Jonathan Parker LJ at [74] accepts the test propounded by the Advocate General: “in other words the quest is not for the closest but for a sufficient link”
. We note that the Advocate general said “ sufficiently direct …link”, and that is the formulation used elsewhere in his judgement by Jonathan Parker LJ; (11) we should bear in mind that a direct link is a particularly close link. (In Abbey the Advocate General (footnote to [35]) noted his agreement with the sentiments expressed in the first of the passages set out at (7) above despite comments which led to Jonathan Parker LJ’s conclusion at (10) above By his comments the Advocate General did not, it seems to us, indicate that the only way a chain could be broken was by an exempt transaction, or that an intervening transaction could not be among the features which led to a conclusion that a link was not (sufficiently) direct; (12) on their own, neither close commercial linkage, the fact that an input enabled an output to take place, nor the fact that the output would not have taken place but for the input should lead us to conclude that the test is satisfied ( Southern Primary ). It seems to us, however, that these may be factors which could support a direct link even though they are not sufficient conditions for a link; (13) “use” is not physical use but some form of economic use ( St Helen’s School at [76] ); (14) a sufficient nexus may exist without the input being a cost component of the price charged in a narrow sense ( Mayflower [42] ); (15) conversely the mere fact that the input cost may be reflected in the prices charged may not suffice to render the inputs cost components ( Dial a Phone [20]); (16) input tax will be residual if either it is on an input which is a cost component of both exempt and taxable supplies or if it is input tax on overheads ( Abbey ); (17) but before considering whether an input is an overhead we should first examine whether there is an appropriate link rather than first considering whether a cost is an overhead ( Mayflower) . At [33] Carnwath LJ says:
“…The special treatment of “overheads” or “general costs” serves a particular purpose in the VAT system, for those inputs which would not otherwise be brought within the calculation. It should not be extended beyond that purpose.”; (18) In Abbey , the ECJ also explained that the overheads rule must apply not always to the business as whole but was capable of applying to identifiable parts of the business: “[39] that rule must apply also to the costs of [inputs] which form part of the overheads relating to a part of a taxable person’s economic activities which is clearly defined and in which all the transactions are subject to VAT, since those goods and services thus have a direct and immediate link with that part of his economic activities.”
. Other Cases[43]We were referred to three cases illustrating the approach taken by other tribunals to this issue.[44]In Britannia Building Society VATD 14886, the tribunal considered four television adverts screened for the society. The adverts concerned the activities of potential customers, borrowing for a mortgage, the virtues of Britannia savings products, saving with the Britannia, and Britannia customers buying domestic products. Each advert lasted no more than a minute. 99% of Britannia’s outputs were exempt. The tribunal found that the adverts were intended to illustrate the wide diversity of Britannia’s financial services and were not “product focussed” but were intended to maintain the Britannia image in the minds of the general public; none was designed to illustrate any particular aspect of the services but all were produced…with the intention of promoting Britannia as a group (paras 21 and 22). It held that the commercials were designed to and “did” inform, and remind, viewers of the Britannia group and its variety of services and were intended to move the group away from the traditional building society image: they were not attributable to particular supplies.[45]Mr Thomas says that the case shows that you must unlock yourself from the four corners of the advertisement and look at its wider role in the business. He accepts that this may involve two stages, the first looking at what the advert does, and the second looking at the importance of the advert to the nature of the business.[46]Mr Cordara fastens on the finding that the adverts were not product specific but for generic products. He suggests that in any event only if such products were sold outside the UK would they not be exempt.[47]It seems to us that what the tribunal did in Britannia was not to focus on the detail of each moment the adverts, but to look at the general impression the adverts left in the mind of the ordinary viewer: what the advert “did”. Specific services were mentioned in the adverts but what the adverts did was “to inform, and remind” the viewer of the Britannia and the range of its products in addition to those associated with a conventional building society. That approach to adverts seems to us to be important and correct. They are not statutes whose words are to be carefully analysed. The use to which an advertisement is put in a business must, in the absence of extraordinary factors, depend upon the effect it has or may be expected to have on a potential customer.[48]Town and Country factors Ltd VATD 19616 concerned whether the supply of screens and the displays on those screens were attributable only to the exempt over the counter (“OTC”) betting outputs of the appellant or generally to all supplies made in its betting shops and elsewhere. The screens provided sports – and in particular racing – broadcasts and betting odds. The tribunal found that whilst the most direct and immediate link was to the supply of OTC betting, there was also a sufficient link to the other supplies made by the appellant. People could come to and remain in the betting offices to watch the screens and there was nothing to stop them walking out afterwards without receiving a supply. Whilst the tribunal agreed that the inputs were different in nature from purely general overheads because of their close link to betting they were in that respect no different from advertising expenditure. By creating an exciting atmosphere in the shop they encouraged the take up of the appellant’s supplies. “By regarding the screens as in part advertising the [shops] the link is to supplies leading directly from the advertising.”[49]The tribunal made clear that the question was, not how the consumer used the screens but how the appellant used them. At para 14 the tribunal noted that this was not a case of an input being a cost component of particular goods, and for that reason it would concentrate on the direct and immediate link rather than cost component test. At [18] it commented that in that case there was no real distinction between a direct and immediate link to exempt and taxable supplies and being a general overhead.[50]Mr Thomas says the approach adopted by the tribunal in that case is that which the Commissioners urge in this: that the advertising is used to attract people into DFS’ stores; once there they may buy nothing but they are offered the full panoply of DFS taxable and exempt supplies. Mr Cordara says that in that case the screens were just like heat and light – general overheads attracting people into shops rather than to buy sofas.[51]We note that the screens attracted people to the shops rather than to particular supplies made by the shops.[52]In Royal Agricultural College VATD 17508, the tribunal held that advertising by a college which was aimed at attracting students to the college and which on occasion referred to the college “experience” was not directly linked to the supplies made at the college shop and bar. There was no clear reference in the adverts to the shop and the bar. The tribunal found that the advertising was directly and immediately linked to the supply of education by the college. The link that to the shop and the bar was subsequent and indirect, and not sufficient.[53]The absence of the direct and immediate link in Royal Agricultural College , is similar to the absence of any link between the production costs and the shop and bar sales in Mayflower (see [40]).[54]Last in this connection we note the other significant context in which the ECJ used and applied the adjective “direct” in translating the use of “for” in the Sixth Directive. That was in relation to the question of whether a supply was made “for consideration”. In Apple and Pear Development Council v C&E Commissioners [1988] STC 221 , the ECJ restated that “for” required a direct link between the supply and the consideration, and found that in that case there was no direct link because the taxpayer derived benefits indirectly from the services provided by the supplier to the industry as a whole and might not benefit at all from certain services, and because there was no link between the level of benefit and the charge made. Although the ECJ’s later visits to the question of “for consideration” (in Tolsma and Town and Country Factors ) were directed to more to the question of reciprocity, we find the approach of the Court in Apple and Pear to “direct” of interest. The inputs at issue[55]Four classes of input tax were debated before us. Advertising, Head Office, Factories, and Stores. We address each in turn Advertising Costs The Parties’ arguments.[56]Mr Thomas put HMRC’s case in the alternative. His first alternative is that the advertising costs were directly linked neither to sofas nor to intermediation and were cost components of neither supply. They were therefore overheads, and fell into the residual category.[57]His second basis is to say that there is a direct and immediate link between the advertising and both the sale of sofas and the intermediation. As a result the inputs again fall into the residual category.[58]HMRC’s Statement of Case indicated that it agreed that there was a direct and immediate link between the advertising and the sofas. In other words it did not espouse the overheads analysis of Mr Thomas’ first alternative. However the overheads argument was raised by his skeleton argument which was served several days before the start of the hearing. We were not minded to refuse to consider the argument in those circumstances.[59]In relation to the “overheads argument” Mr Thomas says that the advertising is not directly linked to the sale of sofas (or to intermediation) because:(1) the advertising cost is not built into the supply of a sofa in the same way in which raw material is incorporated into say a theatre programme;(2) the activity is not “incorporated into an identifiable output” (see the Advocate general in Abbey at [42]);(3) the advertising is not used in the actual sale of a sofa to a customer. The customer does not benefit from the advertising when he gets the sofa;(4) the advertising is a cost of getting the customers into the store. Once there they are offered a suite of supplies. There is no link to any part of that suite of outputs.[60]In relation to his second basis Mr Thomas says that there is a direct and immediate link both to the sofas and the intermediation because:(1) DFS has two separate streams of outputs: sofas and intermediation. The advertising serves both.(2) This is not a single stream like BLP in which the benefits of one transaction were to be used in a later ones – where the intervening transaction broke the link – but two chains each of which were served by customers coming to the store attracted by the adverts(3) Objectively it can be seen that the advertising is exploited to make both streams of supply. Even looking solely at the adverts, they showed specific sofas but it was clear that their purpose was wider than to advertise a single sofa.(4) Just looking at the advertisements does not tell you how they are used in the business. The question is what was the use of the costs, not what is the nature of the adverts. Objectively the function of the advertising was to generate customers in stores. Objectively they are used therefore for all the supplies made at the stores.(5) In Dial a Phone the adverts did not refer to the insurance for which the customer paid but only to free insurance. There was no reference in the adverts to the supply which gave rise to the intermediation income. Yet the tribunal and the Court of Appeal were able to conclude that the adverts related to both activities. The same should be true in DFS’ case.(6) There is no scope for the application of a de minimis approach at this stage in the analysis. The test is not concerned with the degree of use but with linkage to use.(7) The relative importance of the supplies made by DFS as a result of customers coming to the stores was irrelevant to whether there was a sufficient link. In any event, the figures showed that the income from the intermediation was not insignificant. It had sufficient importance. In that context the take up rate of insurance was misleading. What mattered to the evaluation of the intermediation activity was the way the advertising was used to offer the insurance, and more were offered insurance than took it up.(8) There was no scope in this case for the concept alluded to in Southern Primary that an input became “exhausted” by the first supply for which it was used. The advertising was not consumed by the sale of a sofa.[61]Mr Cordara says:(1) BLP showed that the nexus required was a narrow one,(2) Midland showed that more than a consequential relationship was required;(3) Southern Primary showed that a “but for” test was not enough;(4) BLP shows that it is not enough if the input benefits or causes the output;(5) The taxpayer’s motive or ultimate aim is not enough;(6) What is required is a particularly close link;(7) It is sufficient if the input is necessary or essential for the output;(8) Advertising is essential to the sale of the sofa (although not to its manufacture). The relevant economic activity is the sale of sofas not their manufacture.(9) The nature of the advert was at the core of the reasoning in Dial a Phone ; in that case the trader also had an obligation to push the insurance; the relationship was tighter;(10) In DFS’ case there are two supply chains. They are separate: it can no more be said that the wood used in the manufacture of the sofa was a cost component of the intermediation than it can be said that the advertising is a cost component of the intermediation;(11) It is not good enough to say that the advertising generally attracts people to the stores and thus to the insurance; you have to point to precisely how the advertising is used in the intermediation;(12) Intermediation starts after the customer has chosen the sofa, not when the customer sees the advert. The advertising has been used by then.(13) Mere enhancement of brand awareness is not a sufficient link to the branded supplies - otherwise any advert mentioning a name or brand would be a cost component of all supplies made.(14) Merely sitting on a sofa, or deciding to buy a sofa or being in the store does not provide a particularly close link to the intermediation.(15) The appropriate parallels are to the bar sales in Mayflower , and to Royal Agricultural College .[62]Mr Cordara also suggests that there is a, normally unspoken, identification in the decisions in this area, of a paradigm transaction – a normal transaction of which the relevant outputs are examples, and that the quest tacitly becomes a search for those inputs which normally form part of that type of business. There is some hint of this approach in Midland (at [31]) and Southern Primary . Our Conclusions - Advertising costs[63]It seems to us that a trader may use advertising in a range of ways. He may use it to advertise his name, his business as a whole ( Britannia ), a sector of his business, specific types of products, or a single item (a pink 2005 Ford Mondeo Mark X Estate with 24,500 miles on the clock). At one end of that spectrum there will generally be no direct link to any particular transactions, at the other end there may be some link to the single item being sold. The nature of the advert and the conduct and nature of the business will provide the objective criteria which determine the directness of the link. But in each case the question arises as to how the advertising is used for the relevant transaction. No direct and immediate link to supplies.[64]If the cost of advertising is not a cost component of the sales of sofa or the provision of intermediation, then it will fall into the overheads pot. It seemed to us that whether or not advertising could be said to be used for particular transactions depended upon what the legislation meant by “transaction” in the phrases “used for taxable transaction” in Art 17 , and “On each transaction” in Art 2.[65]If the “transaction” was construed as simply the provision of a sofa or the provision of the intermediary services, then, for the reasons in the following paragraph, it seems to us that the advertising was not used for what was supplied; if on the other hand the transaction was the broader economic activity of selling the sofa or involving bringing together customer and insurer, then we believe that the advertising could potentially be said to be used for that activity. This distinction exists by reference both to the “used for” test and the “cost component” test: both hinge on the nature of the output transactions. A consideration of the use of the “economic cost” is not able to dispel the difficulty because that looks at the input rather than the nature of the output. Nor does Carnwath LJ’s warning in Mayflower that the special treatment of overheads should not be extended beyond its purpose point in any one direction: the first step in his analysis is to determine whether there is a direct and immediate link with an output transaction; the purpose of the overheads rule starts to operate after the decision has been made as to the nature of “transaction”.[66]On the first basis it seems to us that advertising costs would not generally be cost components of an output transaction and are not in this case. They are not analogous to the “raw materials” of the programmes in Mayflower in the same way that the production details were; they are not used in the provision of the goods sold in the same way as the service of designing the goods sold is used in their supply or training may be used in selling insurance; there is nothing in what is supplied which is changed or improved as a result of the advertising; neither is there a clear nexus to every particular sale: the sofa sold (or intermediation supplied) will be, and will cost DFS, the same whether or not its sale is as a result of the advertising rather than personal referral or a previous purchase. The advertising is not “essential” to the supply of a sofa in the same way in which the land was not essential to the building contract in Southern Primary . The advertising costs are used to create demand, not to make a supply. The advertising bears to the supplies the same relationship(i) which the production costs in Mayflower bore to the increased theatre bar and shop sales – increasing those sales but not affecting the nature of what was supplied; and(ii) which the land expenditure in Southern Primary bore to the development contract –enabling it to take place but not affecting the inputs required to carry it out.[67]On the second basis it seems possible for advertising to be used for selling activity because it is intimately bound up with getting that activity going.[68]Were it not for the authority of the Court of Appeal decision in Dial a Phone , we would favour the first approach. That is because the reference to “taxable transactions” in Art 17 appears to be a reference back to the heading of Art 5, and the language of the items in that Article (for example item 1:
“supply of goods shall mean the transfer of a right to dispose of tangible property”) concentrates on the transfer of what is disposed of not the activity surrounding its disposal. That approach is supported by the approach of Jacob LJ in Southern Primary : he asks whether the acquisition of the land was “essential” to the building contract, not to the activity of winning that contract. There he is looking at the specific outputs. It also echoes his approach in that case at [35] where he says: “…if one applies the ‘fundamental principle’ that VAT applies to each transaction by way of production or distribution…one is driven to ask whether the land purchase price is a cost component of the development contract…”
That looks transaction by transaction, not to the economic activity surrounding the transactions. On this basis we would agree with Mr Thomas’ first approach and find that DFS’ advertising costs were overheads and therefore residual.[69]However the acceptance by the Court of Appeal in Dial a Phone (and Mayflower see [34]) that the advertising costs there could be cost components of the output transactions made by Dial a Phone indicates to us that the second approach must be the correct one. There advertising did not affect what was sold but was found to be a cost component of the transactions undertaken by the appellant. That indicates to us that the test is by reference to the activity of making the sales rather than to what is sold.[70]On this basis it seems to us that the costs of advertising are a cost component of the activity of supplying sofas. That is for the following reasons:(1) The adverts are plainly directed at the sale of sofas. There is nothing in the nature of the business of the Appellant which suggests that this is not an object of the advertising;(2) The very fact that the Appellant – a commercial enterprise selling sofas -incurs very considerable expense in placing adverts which are directed at the sale of sofas indicates objectively that the adverts are directly linked to the selling, and the costs of the adverts are directly and immediately linked to the sales activity: no enterprise would incur such costs if they were not for the sales activity.(3) The link between the advertising and the sales activity is free from any intermediate steps and is directly rather than indirectly connected to and bound up with that activity.(4) The advertising is “essential” to the sales activity in the sense that it increases that activity Are the advertising costs components of the insurance intermediation?[71]It is clear to us that there is a link between the advertising and the intermediation. The issue is whether that is a direct and immediate link. The existence of a direct link to the selling of sofas does not prevent there being also a direct link to the intermediation. The issue is whether that is a direct and immediate link (a sufficiently direct and immediate link).[72]We agree that the content or effect of the advert may not be determinative, and the role an advert plays in a business is relevant. An examination of the facts might, for example, reveal that the advert was for a sprat but was used to sell a mackerel. That might be the case if the result of the adverts was consistently to increase the sale of mackerel rather than sprats, the success of the campaigns were measured by reference to mackerel sales and, the business designed the adverts by reference to factors which increased mackerel sales.[73]The following factors may support but do not compel a conclusion that there is a direct link:(1) that advertising increases sales of sofas and increased sales of sofas increase the sale of insurance and that increases intermediation income: in other words that some insurance sales would not have taken place but for the advertising, and the advertising is commercially linked to intermediation activity;(2) because intermediation is albeit a small proportion of DFS turnover and profit, it is nevertheless significant in absolute terms, and DFS might be expected to find and use tools to increase that income;[74]The following factors suggest that there is no direct link:(1) the advert attracts customers to buy sofas; only once they have decided to buy a sofa will they consider insurance. A purchase of a sofa could follow directly from an advert; intermediation required a separate act by DFS (and by the customer). There is thus an intermediate step between the advert and the intermediation which makes the link between the two more indirect. The extra step does not prevent a direct link but makes the link less direct;(2) advertising creates a demand for the sofas and enables their supply; the sale of the sofa creates the possibility of the demand for the insurance; the intermediation supply is made only if that possibility is turned into an actuality. That does not happen in the majority of cases. There is an intervening uncertainty. The link to the intermediation activity is weakened or made less close by the additional uncertainty - and indeed the probability - that there will be no intermediation supply. (In our view the intermediation supply is the actual bringing together of the customer and the insurer, not the attempt to do so. It thus takes place in less than half the sofa sales.) ;(3) whilst we accept that the question is not how the consumer uses the advertising but how the taxpayer uses it, in our view an objective appreciation of the effect of the advert is relevant to an objective appreciation of what the advert is used for. That appreciation may be affected by the role which it may be shown that the advert played in the business and the way in which the business worked. As we have found, the adverts had the effect of getting customers to want to come to DFS to buy a sofa (or to think about buying one);(4) the way in which DFS selected sofas to go in the adverts and monitored the results of its advertising indicated that it was using the advertising for the effect the adverts had, and did not indicate another use;(5) the comparative unimportance (rather than secondary financial importance) of the intermediation income indicated a link which, objectively, was not particularly close: it gave no reason to believe that the advertising was directed to earning that income.[75]On balance the considerations pointing away from a particularly close or direct link persuaded us that the advertising was not used by DFS for the activities relating to intermediation.[76]We did not find the fact that the effect of the adverts (or DFS’ admission that their function) was to get people to come to the store relevant. That is because the stores were plainly places where people were to be sold sofas. Going to a store was necessary to buy a sofa from DFS: it was merely a step in the process, and was not part of a plan to use a sprat to catch a mackerel. And going to a store enabled insurance to be sold only if a sofa was bought.[77]There was much discussion before us of the decision in Dial a Phone . Dial a Phone’s business was the marketing of mobile phones, airtime and the arranging of insurance. It was obliged to promote insurance through its agreement with Cornhill, and received (exempt) commission income. It advertised its business. The adverts referred to the phones and airtime packages available, to free accessories and to the availability of three months’ free insurance, but made no mention of paid for insurance. Its business was conducted through call centres which followed a script. The first part of the script lead the caller to take an airtime package. Having signed up to that, the script would go on to lead the caller to sign up for insurance. The caller got the free insurance if he signed a (cancellable) deferred start direct debit mandate for paid-for insurance. About 66% of customers signed up for the insurance. The Court of Appeal found that the following findings of the tribunal were well nigh irresistible inferences from the facts:
“49.The[arrangement] of insurance is not ancillary to the advertising. Free insurance is not used merely to attract customers to sign airtime service contracts for the phones. It is clearly intended to attract new customers to Cornhill as well and the Appellants have a direct financial interest in the customers staying with Cornhill on the completion of the three free months. It is for that reason that the Appellant not the phone service providers are funding a proportion of the free three months insurance with Cornhill… “51.The advertisements relate both to the Appellant’s intermediary service introducing customers to mobile phone airtime providers, and to their insurance intermediary service introducing customers to insurance business with Cornhill. All insurance is paid for.”
[78]It is clear to us that the tribunal was concerned first to set aside an argument that because the advertised insurance was free, what was advertised did not relate to what was supplied. It did that by saying that the clear purpose of the free insurance was to get customers to take up the paid for insurance. Thus the adverts could be seen clearly to be linked to the paid for insurance. It did not reach this conclusion by reference to the non insurance parts of the adverts. The nature of that discussion indicates to us that the tribunal was concerned with the way in which the content of the advert was linked to the supplies made.[79]Jonathan Parker LJ, having noted that the search was for a sufficient link, said:
“[75] It follows that it matters not that the insurance intermediary services may be viewed as being in a commercial sense secondary to the making of taxable supplies, or even that they may be provided only after a taxable supply has been made, provided a sufficient direct and immediate link exists between them and the marketing and advertising costs.”
[80]He went on to say that he did not view the insurance intermediary services as secondary in nature, noting that it represented as substantial proportion of Dial a Phone’s income. He continues: (1) “…Insurance with Cornhill was part of the package of services advertised by DaP. To my mind the fact that the advertisement referred only to the initial free three month period of insurance is hardly surprising, and says nothing as to the relative importance to DaP of the insurance element of the package...”.[81]It seems to us that(i) in these passages Jonathan Parker LJ is saying that on its own the secondary importance of the intermediation would not prevent the adverts from being directly linked to it, but(ii) the very fact that he discusses the relative importance in the following paragraph indicates that he did not consider that the relative importance of the intermediation was wholly irrelevant. His language also makes clear that it is not irrelevant that the link to the Cornhill insurance was part of “the package of services advertised by DaP” ie the part of the advert referring to free insurance.[82]Our conclusion is not, we believe, inconsistent with the approach taken by the tribunal in Dial a Phone . First of all, in that case the content of the advertisement was clearly regarded by the tribunal as relevant to the purpose for which it was used (the tribunal’s discussion as to whether the fact that only free insurance is mentioned shows that). We too regard the content and effect of the advertisements as relevant to the objective determination of how they are utilised. In Dial a Phone the reference in the advertisement to free insurance linked the adverts to the paid for insurance. That link was from(a) the content of the adverts, and(b) from the underlying commercial arrangement with Cornhill. In our case (a) is not present. The link in our case derives solely from the way in which DFS’ business is conducted – namely that a sale of a sofa may sometimes lead to intermediation. The absence of (a) makes the link less close. Second, the sales process of Dial a Phone led seemingly inexorably in each customer’s case to the question of insurance (paras 29 and 30 of the tribunal’s decision suggest that almost all the customers took the free insurance although 45% cancelled the direct debit mandate so that commission was received only in respect of about 66% of contracts sold). In DFS’ case there are contingencies in the process which interrupt the link: PPI is relevant only if credit is taken, anti stain insurance is relevant only to fabric sofas. Those contingencies result in a large proportion of sales being made without any intermediation activity. These factors lead us to the conclusion that although there is a link between the adverts and intermediation in DFS’ case it is not as clear and close as it was in Dial a Phone and not sufficiently close to be a direct link.[83]Like Dial a Phone the intermediation provided by DFS took place after, and was and less commercially important than, the primary sale activity. Those factors would not on their own have prevented us from finding the presence of a direct link in DFS case. But other factors weighed in the balance too.[84]In Town and Country , the television screen expenditure brought the customers in and helped them stay so that they would be tempted to take supplies from the appellant. Objectively the effect and the purpose of DFS’ advertising is to get people to come and buy a sofa, - tomorrow- rather then to get them into the shop. They have to go to the shop to buy a sofa but that is a consequence of what the adverts do rather than what they are used for. There were no objective considerations in Town and Country which pointed away from linkage to the totality of the appellant’s activities. In DFS’ case: the way it monitored sofa sales, what the adverts did , and the lack of any strong, or certain, link to the insurance sales do not suggest a sufficiently direct link to anything other than sofas.[85]In Britannia the tribunal found that the function the adverts played was to inform and remind the viewer of the variety of Britannia’s activities. The function of DFS’ adverts is, in our view, to inform the viewer of the variety of DFS sofas and to tell the viewer to buy one quickly. It is a consequence of the sale of sofa that more insurance may be sold, but in our view the role the adverts played in DFS’ business was to get people to buy sofas. We do not say that the adverts were exhausted when the sofa was sold, but that objectively their function stopped with the sofa.[86]The link in Royal Agricultural College between the advertising and the college bar and shop sales was in our view less direct and immediate than the link in DFS’ case, because the bar and shop sales followed later and less mechanically from the student’s enrolment than the intermediation followed from the sale of a sofa. But in DFS’ case the objective link of the advert to the first transaction – the sofa sale – was stronger, and in both cases there was lack of certainty that the second supply would follow the first – proportionately or at all.[87]We agree with Mr Thomas that there is not in DFS’ case a single stream like that in BLP in which an intervening exempt transaction breaks the connection. But the insurance is not be sold unless a sofa is sold and that weakens the link. That weakness was there in Dial a Phone but the reference to insurance in the adverts and the automatic pushing of insurance to all customers provided a strong direct link to the intermediation activity. We also agree that both sofa selling and intermediation are served by customers coming to the store, but objectively the nature of the advert and the evidence of DFS’ business suggest to us that it is used to sell the sofas rather than to get customers into the store. Stores[88]Mr Cordara says that the fact that a small area is used in the stores for spraying with the anti stain product is not relevant. By that time the intermediation activity has ceased and the spraying cannot be part of the intermediation process. The situation is similar to that of the later legal costs in Midland . Neither is the activity of the staff in the stores – a few minutes talking about insurance with a customer – enough to link the inputs for the stores with the insurance activity. You have to provide shelter but it is peripheral.[89]In the stores sofas are displayed, sold, received from manufacturer or factory, sprayed, and dispatched, insurance products are sold, and the related paperwork for all these activities is processed and input to DFS’ computer system.[90]In our view the inputs used to provide and maintain the stores are residual and are not attributable to any identifiable part of DFS’ business.[91]If the test were whether those inputs were used in the narrow activity of the transfer or provision of a sofa, or the bringing together of customer and insurer, we would find that they were not so used – that there was not a direct link between for example, the electricity for the lights and those actions – and accordingly that the inputs were overheads and hence residual. If on the other hand the test is whether the inputs were used in the activities of selling and intermediating ( the test we believe was used in Dail a Phone ) then in our view the inputs were directly linked to both those activities (and were therefore residual): they were necessary and commercially essential for those activities, not remote from them, physically close to them, and part of the cost of performing them. We also take account of the facts that: there was a commercial link between input and activity, that but for the stores costs the other activities would not have taken place, and that the maintenance of the stores benefited those activities; we do not treat those factors as sufficient or necessary conditions, but as part of the fact pattern which leads to the conclusion that there was a very close link between input and activity. In reaching that conclusion the comparative size of the output activities is to our mind irrelevant: it played a part our conclusion in relation to advertising costs because it was relevant to assessing the role the advertising played in the business. Its size suggested that intermediation was less likely to be a purpose at which the advertising was aimed and so for which it was used; the use of the stores is by contrast principally physical, and objectively eclipses other considerations.[92]We also regard the use of the stores for the application of the anti stain product as relevant. Midland did not lay down a rule that an input always had to precede an output; it merely stated that generally it would. The application of the anti stain product was, we have found, a contractual requirement of the sale of the anti stain insurance, it was directly and immediately linked to that sale and necessary for it. The obligation to apply the anti stain product was incurred when the insurance sale was made; that it was applied later is immaterial.[93]We do not regard the residual costs as being attributable only to one identifiable part of DFS’ business. DFS had two streams of business but the conduct of those streams was mingled: the same people sold, accounted for and managed the sale of both the sofas and insurance in the same places. The accounts we saw indicated no accounting separation of the activities, other than the separate identification of one of the lines of intermediation income. The accounting data included insurance sales in assessing store performance. Factories[94]Mr Thomas says that the factories were used for spraying the anti stain product as well as for making sofas. The inputs for the maintenance and provision of the factories were therefore used in the intermediation. Mr Cordara repeats the argument that the intermediation has finished by the time the anti stain product is applied.[95]It seems to us that the factory inputs are used in supplying the intermediation. The factories are used for the spraying and the spraying is used for the intermediation. The spraying is part of the intermediation service which DFS provides: it is necessary for it, it is closely linked to it, and there is nothing which breaks the link between the use of the factory and the intermediation. For the reasons set out above we do not believe that the direct link between the two is destroyed by the order of the events. We accept that the use of the factories for spraying will be minimal in comparison to their use for manufacture of sofas, but the test is whether there is a direct link, not whether the use is not de minimis. Head office[96]Mr Thomas points to the liaison with the insurance providers carried out at Head Office, and the interaction with the regulator (the FSA) carried out by head office in relation to the sale of insurance. All of that required the facilities of head office.[97]Mr Cordara says that only slight use is made of head office. There was a financial products committee, but that only required an occasional room. The arrangements with the insurance providers were well established and required little input.[98]In our view the Head Office costs are not directly or immediately linked to the sale or selling of sofas or intermediation. They are therefore residual because they are overhead costs. The inputs incurred to create and maintain the head office functions are used to provide setting and support for monitoring, planning, managing, accounting for, and directing all the aspects of the business. They are only indirectly used for carrying out those activities. PART II The Special Method The Relevant Law[99]The EC background to this part of the appeal lies in para 5 of Article 17 of the Sixth Directive: “As regards goods and services to be used by a taxable person both for transactions covered by paragraphs 2 and 3, in respect of which value added tax is deductible, and for transaction in respect of which value added tax is not deductible, only such proportion of the value added tax shall be deductible as is attributable to the former transactions. “This proportion shall be determined in accordance with Article 19 for all the transaction s carried out by the taxable person. “However, Member States may:(a) authorise the taxable person to determine a proportion for each sector of his business…(b) compel the taxable person to determine a proportion for each sector of his business(c) authorise or compel the taxable person to make the deduction on the basis of the use of all or part of the goods and services(d) authorise or compel the taxable person to make the deduction in accordance with the rule laid down in the first sub paragraph, in respect of all goods or services used for all transactions referred to therein(e) provide that where the value added tax which is not deductible by the taxable person is insignificant it shall be treated as nil.”[our italics][100]Our italics highlight: the way in which member states may act, the possibility of apportionment by reference to business sector or use, and the authority to set a de minimis level. Para (d) gives permission for a method to be imposed under which rather than trying to isolate particular inputs, all inputs are apportioned.[101]Article 19 provides for the allowable proportion of the residual input tax (the phrase we use to describe the tax which is not attributable only to taxable, or only to exempt outputs) to be (broadly) taxable turnover divided by total turnover.[102]The UK’s primary legislation implements At 17(5) in section 26(1) VATA 1994 (and its predecessors). It provides that the allowable input tax is to be determined by regulations. By section 26(2) the regulations are to be made “for securing a fair and reasonable attribution of input tax to” taxable supplies and supplies assimilated thereto.[103]The detail is thus left to regulations. Different regulations with differing methods and calculations of attribution have been made since 1973, and, in the period from 1986 to 1973 – the period relevant to the arguments in this part of the appeal - four different sets of regulation applied successively. Rather than set them out here we deal with them in the history we set out below, since their terms may have affected the nature or interpretation of the actions of the parties. However, at this stage we note that each set of regulations provided for a standard method of apportionment, but also permitted a special method if such a method was “allowed or directed”, or, after 1992, “approved or directed”.[104]We use “standard method” to refer to the standard method prescribed at any time, and “special” method for any other method permitted under the regulations. We use “exempt input tax” to refer to the input tax attributable in accordance with the regulations to exempt supplies. Events since 1986 .[105]Some of the findings we make here are irrelevant to our eventual conclusion, but they are set out in case, if there is an appeal, our path to our conclusion is found wrong. 1986[106]In 1986 Mr Todhunter was the Company Secretary of Northern Upholstery Furniture Group Limited (which we understand to have been the previous name of the Appellant). Mr Bear was the officer of HMRC who looked after the Appellant’s VAT affairs on the ground. Mr Hanson was Mr Bear’s superior. Mr Bear visited the Appellant regularly and had a cordial relationship with Mr Todhunter.[107]At this time DFS was not arranging insurance for its customers. Its supplies were substantially taxable supplies only. But it embarked upon the development of a site at Droitwich. The site was bigger than DFS needed for the store it wanted there, and a second building was built on the site which it decided to rent out. At that time the renting out of a building was an exempt supply and there was no “option” to make it a taxable one.[108]Mr Todhunter discussed this development with Mr Bear at the time of one of his control visits. Mr Bear helped Mr Todhunter write a letter to HMRC suggesting a special method for the calculation of its irrecoverable input VAT. The letter said:
“Dear Mr Bear, “As explained during your control visit on 5 September 1986 the company will be making exempt supplies in relation to rental income etc [we find that “etc” was not meant or understood materially to broaden “rental income”]. “Due to the uncertainty that the Company’s exempt outputs will exceed the limits set out in para 16 of VAT notice 706 and the complexity of applying the direct attribution method on an ongoing basis, it is requested that an alternative method be adopted of retrospective direct attribution. “At the end of each VAT quarter and the VAT year the exempt supplies will be checked against the above limits. If the limits are exceeded, using the entries in the nominal accounting ledger accounts, the individual invoices falling into the following categories:- (a) goods and services directly related to exempt supplies and (b) goods and services not directly related to taxable or exempt supplies “will be identified and their relevant input VAT analysed and the disallowable VAT calculated. “Your assistance is most appreciated”
[109]Mr Bear corresponded about this with his superior, Mr Hanson. His note to Mr Hanson indicates that he thought that the inputs related to the exempt supplies would be readily identifiable and auditable. He makes no mention of inputs attributable to both taxable and exempt supplies or of what would nowadays be termed overheads. Mr Hanson’s reply touches on this last issue: he says “the proposed partial exemption system is somewhat unusual” and then he asks whether the proposed system will identify inputs which are not attributable to exempt income and continues “-or is anything which is not directly attributable to exempt income regarded as being attributable to taxable income?”. Mr Bear replies that “inputs not directly attributable to taxable or exempt supplies have already been provisionally identified (mainly head office expenses) and… will be sorted into the appropriate i/p tax column”. He anticipates some arguments over the fine detail.[110]On 8 December 1986 Mr Hanson wrote to Mr Todhunter giving approval to the method he sets out in the letter (we call this the Hanson Method): “To use this method you must deduct all input tax on goods and services directly related to your taxable supplies, and not deduct any of the input tax on goods and services directly related to your exempt supplies. Any remaining input tax which is not directly related to taxable or exempt supplies (such as input tax on your general overheads) should be apportioned using the standard method.” (We note the difference between Mr Bear’s sorting into columns and Mr Hanson’s fraction.)[111]We also note some ambiguity in Mr Hanson’s language. His language deals with input tax directly related to taxable supplies, and directly related to exempt supplies, and then with input tax not directly related to taxable or exempt supplies, but does not clearly deal with input tax related to both taxable and exempt supplies. It is possible that in his use of “directly” he means to exclude that possibility because any input will gravitate only to one of the three poles, or (in our view the more likely interpretation) that he intends such input tax to fall within the third category. We also note that there is not in his language the incorporation of the words “exclusively” or “wholly” which appear in later statutory contexts.[112]Mr Hanson then indicates that an annual adjustment must be made and explains that if DFS does not qualify as a fully taxable person it must “recalculate the amount of residual input tax you are entitled to deduct for the tax year on the lines described in paragraph 26 (a) of Notice 706” and declare or claim the difference. If on the other hand DFS qualifies “as a fully taxable person under the outputs rules in paragraph 13 of Notice 706” it will be entitled to full deduction for the whole year and can reclaim any over payment.[113]This correspondence must be seen against the background of the then applicable regulations and HMRC’s guidance in the then current version of Notice 706 (…and possibly its predecessor). The then relevant regulations, SI 3985/886, provided for no direct attribution by a person making taxable and exempt supplies, but for a standard method under which a single fraction of the total input tax would be deductible. That fraction was the value of taxable supplies divided by the value of total supplies (we call this the “standard fraction”). The imposition of that method appears to be authorised by para (d) of Art 17(5) of the Directive.[114]The 1984 version of Notice 706 reflected these rules. Paragraph 16 (referred to in Mr Todhunter’s letter) described the provision in the regulations setting de minimis limits for exempt outputs: if they were below these limits the taxpayer could treat itself as fully taxable. There was also a de minimis applicable to exempt input tax.. Paragraph 20 set out a Direct Attribution method which could be used with approval from HMRC as an alternative to the standard method in the regulations. That Direct Attribution method was in substance the method approved in Mr Hanson’s letter.[115]Mr Hanson refers in his letter to the calculation of residual input tax allowance “on the lines described in paragraph 26(a) of Notice 706” and to the output tax rules in “paragraph 13” of that notice. These appear to be references to the previous version of Notice 706: whilst Mr Todhunter’s references refer to the 1984 version, Mr Hanson seems to have been working from the 1983 version (but since the principle difference between the two seems to be only that before 1984 there were two possible standard methods it does not seem to us that there was any material misunderstanding of the rules by Mr Hanson at the time.).[116]Mr Todhunter confessed himself to have been (and still to be) no expert on VAT. We think his self assessment was correct. He explained that his understanding of the method which had been approved was this: (2) DFS would identify the inputs which related directly to the Droitwich building letting and disallow them; (3) DFS would disallow items in head Office expenses which related to the joint Droitwich site. These would be items such as landscaping and security for the site; (4) other head office expenses would remain fully deductible as they had been previously – (effectively they were treated as relating solely to the previous and continuing taxable outputs so that a pre-existing attribution of input tax would not be disturbed when another stream came into being save to the extent that the new stream added new inputs).[117]We call this the Todhunter Method. We have noted some ambiguity in Mr Hanson’s letter: we note below a submission by Mr Cordara that the Todhunter Method was a permissible and proper interpretation of the method prescribed by Mr Hanson when viewed in the light of contemporary legal understanding. However for the purposes of our defined term the “Hanson Method”, we mean a method under which input tax directly immediately linked with and exclusively used: for taxable supplies was allowable, for exempt supplies not allowable; and for both or for overheads to be apportioned using the standard fraction.[118]Mr Todhunter says he regarded the method which he was to operate as requiring no fractional disallowance of items such as “general overheads” input VAT. It seems to us possible that Mr Todhunter came to his understanding of the method from his discussions with Mr Bear. Mr Bear’s note to Mr Hanson, in which he refers to the sorting of non directly attributable inputs – mainly head office expenses - into columns, suggests that Mr Bear may have considered that the method did not involve the use of a fraction to disallow part of the input tax on head office expenses generally but rather the identification of specific items attributable only to Droitwich or which were common to Droitwich and the main business. We believe that he communicated that approach to Mr Todhunter. We accept that such was Mr Todhunter’s understanding of what had to be done. It was not however the method which had been approved by Mr Hanson, for Mr Hanson had referred in his letter of approval to the fractional standard method approach to the attribution of such inputs.[119]We find it likely that the application of the method applied in relation to the Droitwich outputs produced an amount of exempt input tax which was below the then de minimis threshold. . 1987[120]In 1987 the 1985 regulations were amended. The amended regulations substituted for the standard fractional attribution of all input tax, a direct attribution method along the lines of Article 17(5) of the Directive, but with attribution of input tax which was not used “wholly” in making taxable supplies or “wholly” in making exempt supplies on the basis of use rather than by reference to the standard fraction, although provision was made in regulation 30(2) for the Commissioners to allow the use of the standard fraction to determine the allowable part of that residual tax. The Commissioners were also empowered to allow other special methods.[121]In 1987 too DFS started to provide PPI to its customers and to make exempt supplies of insurance intermediation or arrangement to the insurers. This requires us to consider further provisions of the regulations.[122]The 1985 regulations had provided a disregard in relation to certain securities transactions incidental to a taxpayer’s business. From 1 April 1987, as a result of SI 1987/510, regulation 31 of the amended 1985 regulations provided a new disregard (we call it the “insurance disregard”) which encompassed the arranging of insurance:
“there shall be treated as attributable to taxable supplies any input tax attributable to exempt supplies of the following descriptions— … (c) any services comprised in item 3 of Group 2 of Schedule 6 to the Act;… [but this exclusion shall] not apply in the case of a taxable person who carries on the business of, or a business similar to, any of the following: ;… (c) an insurance company, agent or broker;”
. Item 3 of Group 2 of Schedule 6 to the (Value Added Tax) Act 1983 exempted “the making of arrangements for the provision of any insurance or reinsurance …”[123]Thus provided that DFS did not carry on the business of an insurance agent, the supply it made to the insurers of arranging insurance transactions between them and its customers was to be treated as a taxable supply for the purpose of the standard method. We believe it unlikely that Mr Todhunter was aware of this provision.[124]In addition the 1987 amendments changed the de minimis provisions so that the de minimis disregard applied (and the taxpayer would be treated as fully taxable) only if exempt input tax was below specified levels; the disregard which had previously operated by reference to the level of exempt outputs was removed. The disregard was available if exempt input tax (as calculated under the method applicable to the trader) was less than any of: (5) £100 per month (6) both £250 per month and 50% of input tax; or (7) both £500 per month and 25% of input tax.[125]DFS made its VAT returns on the basis described in para 113 above. To the extent that this method can be said to encompass consideration of the exempt PPI income we call the method used the “extended Todhunter Method”. No further disallowance of input tax was made in those returns to reflect the earning of exempt PPI intermediation income. It is likely that Mr Bear or his successor audited those returns. We also find it likely, given the cordial relationship between Mr Bear and Mr Todhunter, that Mr Bear was aware of the PPI income. In this year too we find that the method produced de minimis exempt input tax.[126]We make the findings in the paragraph above despite an answer which Mr Todhunter gave to a leading question from Mr Cordara. At the end of Mr Todhunter’s evidence he had explained that the approach he had adopted was his understanding of what Mr Bear had explained to him. He than said “the actual approach never came into operation because the figures were below the limit, so we never had to do the calculation.” . Mr Cordara then asked “So Mr Bear would never have known, if I can lead you, necessarily what you were thinking?”. Mr Todhunter replied “No.”. It seemed to us that Mr Todhunter said No because he thought he ought to agree with Mr Cordara and didn’t really think about the question. His answer did not make sense. Mr Bear had discussed checking the operation of the method with Mr Hanson. He was assiduous. We find it unlikely that he would not have examined carefully the operation of the newly agreed method. The fact that Mr Todhunter produced a de minimis figure would not have dissuaded a careful officer from checking to see whether or not it was truly de minimis. Mr Bear might possibly not have known what Mr Todhunter was thinking but it is clear to us that he would have known and checked what he was doing . 1989[127]In 1989 the VAT legislation in relation to grants of land was changed, and taxpayers were given the right to elect to disapply the exemption applicable to such supplies. On 20 October 1989 Mr Todhunter wrote to Mr Featherstone, who was by then HMRC’s officer dealing with DFS, opting to tax the Droitwich (and other) rentals. As a result DFS ceased to have exempt income from land from that time onwards.[128]It remained, however, in receipt of exempt insurance arrangement income. Under the insurance disregard that income could be treated as taxable income for input tax recovery purposes if DFS did not carry on the business of an insurance agent. 1992[129]In 1992 the regulations changed again, and the use based criterion for the operation of the standard method in relation to residual input tax was replaced by the standard fraction. (This method was therefore similar to the Hanson method.) The regulations maintained the insurance disregard. The de minimis limits for exempt input tax were changed: the limit applied only if the exempt input tax was below £600 a month (reg 33A). The standard fraction was to be rounded up to the nearest integer. At the same time the wording of the provision relating to the use of special methods was changed. Regulation 31 provided that:
“(1)Subject to…, the Commissioners may approve or direct the use by a taxable person of a method different from [the standard method], save that where the use of a method was allowed prior to 1 August 1989 there shall not be included in the calculation (if the method would otherwise allow it) [self supplies]”
[130]This was the change from “allowing” to “approving” a special method. 1995[131]In 1995 the 1992 regulations were replaced. The standard method and the rounding up to the nearest integer were retained. The de minimis limit was increased to £650, and the insurance disregard was retained in regulation 105. 1996[132]In 1996 Mr Gibson took over from Mr Featherstone as HMRC’s officer dealing with DFS on the ground. DFS’ affairs were now to be handled by the Large Business unit of which Mr Gibson was a part. His superior officer Mr Galloway also came to be involved with DFS at the same time. In the following years Mr Gibson made a number of visits each year to DFS. We shall deal later with the detail of those visits.[133]At the time of the handover Mr Gibson had some discussion with Mr Featherstone about DFS and the nature of its business. He was told that it had some exempt income. He was not told that a special method had been approved in 1986. 1997[134]In 1997 DFS started receiving commission for anti stain insurance sales.[135]The anti stain spray was purchased by DFS. The purchase cost bore VAT. In 1997 Mr Todhunter instructed the lady who was in charge of the purchase ledger, not to enter any input VAT in relation to the purchase of the anti stain product and to treat the gross amount of the invoice as the expense. The result of this was that for all relevant subsequent years, no VAT input tax credit was produced by the accounting system in respect of the anti stain chemical and no input tax credit was claimed in its VAT returns. We refer to this the “blocking” of input tax.[136]No other relevant input tax disallowance was made. As a result for this and all relevant subsequent years DFS claimed deduction for all its input with the exception of the anti stain spray input tax (and certain other statutorily barred inputs such as car costs which are irrelevant to this appeal). 1999[137]In 1999 the regulations were amended to remove the insurance disregard in what was then regulation 105 of the 1995 regulations. As a result from 9 March 1999 incidental supplies of insurance arrangements could no longer be treated as taxable for the purposes of the partial exemption calculation[138]The Primback re- adjustment . In August 1999 DFS made a claim for repayment of over paid VAT which was connected with the Primback case.[139]DFS sells goods on interest free credit. If the goods cost £100, the customer becomes indebted to the finance company for £100 and pays off that indebtedness without interest. DFS provides a subsidy of, say, £10 to the finance house to cover the financing cost. The finance company account to DFS for the difference, £90. The question arose in these circumstances whether the consideration for VAT purposes for the sale of the goods was £100 or £90. This issue was litigated by Primback Ltd. In 1996 the Court of Appeal decided that VAT was due on £90 rather than £100 (see Primback Ltd v CCE [1996] STC 757). Accordingly from 1996 onwards DFS made VAT returns (and reclaims) on the basis of consideration of £90. (In the meantime the House of Lords referred the issue to the ECJ which, in 2001, held that the Court of Appeal was wrong and the VATable consideration was £100.)[140]In 1999 DFS realised that it had been calculating its claims under the 1996 Court of Appeal decision wrongly. Effectively, because it received £1 of commission on the PPI sales it had treated its sofas as having been sold for £91, on which VAT would be payable; it should have treated the sofas as sold for £90 and treated £1 as consideration received for an exempt supply. (In actuality its mistake was even more explicit: it had added back VAT on insurance commission in calculating the adjustment to be made for the Primback claim). These calculations were seen by Mr Gibson and referred expressly to the insurance commission.[141]So in July 1999 Mr Todhunter rang Mr Gibson, talked through the basic transaction of a finance sale with commission and the logic of the correction needed. On 30 July he wrote to Mr Gibson saying:
“We have discovered an error in the basis of the [Primback] calculation by including the VAT adjustment for insurance commission, included in the payments received from the finance house”
He enclosed a Voluntary Disclosure form indicating that the error was “due to inclusion in error of insurance received from the finance house”, and a schedule headed “Insurance Commission received in error” setting out the adjustments month by month. The claim related to the period October 1996 to 31 March 1999, and was for £2,146,102.[142]Mr Gibson was about to go away on holiday. He asked Mr Moore, a more junior officer in the office, to process the claim. Mr Moore did so and communicated his finding to Mrs Sorsby in a nearby office of HMRC. She authorised repayment and the authorisation was countersigned by two other officer in that office. Neither Mr Moore nor the other officers had any detailed knowledge of DFS’ operation. The repayment was made in full in September 1999.[143]We note that, if DFS’ intermediation income was not required to be ignored under the insurance disregard in reg 105, then under the then standard method, relevant input tax in relation to the whole period would have been required to be apportioned; if the insurance disregard did not apply, then the repeal of that regulation in March of that year meant that it was (only) for the last 20 days of the period for which this claim was made that intermediation income could have given rise to a need to apportion relevant input tax if the standard method (or the Hanson method) had applied. 2004[144]In January 2004 two members of HMRC’s anti avoidance team visited DFS and interviewed Mr Todhunter. They made manuscript notes of the answers he gave to questions on a questionnaire. After the meeting the questionnaire was typed up. Taking the manuscript notes together with the questionnaire we find that the officers were told that: (8) DFS offered its customers PPI and anti stain insurance; (9) the insurance was provided by third party insurers; (10) DFS received commission from the PPI insurers which Mr Todhunter thought was treated as exempt; (11) no commission was received on the stain guarantee policies, but contradictorily, that they were treated as exempt; (12) (again rather inconsistently with both (3) and (4)) everything was standard rated except the anti stain guarantee; and (13) almost in total contradiction to all the above, that DFS was not partially exempt.[145]As a result of the officers being told at an early stage in the discussion that DFS was not partially exempt, the section of the questionnaire on partial exemption was not completed. However in our view the questions in that section would have revealed no more of relevance to this appeal than the information related above.[146]Against the question on the questionnaire “Have you …been partially exempt?” the typed answer appears “no - de minimis and treated as fully taxable”; the manuscript notes simply say “not PE”. It is possible that the extended answer was given orally by Mr Todhunter and recalled on formally completing the questionnaire. Having heard Mr Todhunter we do not think he would have given or been able to give so technical an answer, but he said that as far as he was concerned at the time they were not involved in partial exemption because they never had to do a calculation. It also seems possible to us that this phraseology may have come from a review of the standing record prepared by Mr Gibson (see para 158 below.).[147]Mr Todhunter accepted that if he had thought that DFS was operating a special method he would have told the officers that in response to the question. 2005[148]In 2005 a new regulation 102(5) was substituted providing that an approval or direction for a special method should only have effect if it was made in writing in the form of a document identifying itself as such. The amendment had effect for methods approved or directed after that date.[149]The regulations in relating to rounding up were also changed. If residual tax was less than £400K per month the rounding up to the nearest integer was retained; otherwise the fraction was to be rounded to two decimal places.[150]Also in October of this year Mr Gibson asked one of his colleagues, Mr French, to audit the purchase accounting system. Mr French’s audit threw up a credit note received by DFS in relation to the anti stain product, the VAT on which had not been recorded in the system, and which therefore had not resulted in a decrease in DFS ’ input tax credit. Although Mr Todhunter’s evidence was not initially clear on this point, we accept his later clarification that he explained to Mr French the reason for this: no credit had been taken for the input tax on the original invoice because DFS blocked that input because it related to the exempt anti stain output (see para 132 above) and therefore no adjustment for the credit should be made.[151]In November 2005 Mr Gibson wrote to Mr Todhunter about the 2005 Budget changes to the rounding up rules in applying the standard method fraction. He said “It is known that your group makes exempt supplies. I understand that in the past you have considered yourselves de minimis under the standard method. I am writing to inquire as to whether the 2005 budget changes…may have an effect on your partial exemption situation from 1 April 2005” 2006[152]In January 2006 Mr Todhunter replied to Mr Gibson’s letter. He indicated that all input tax was directly attributable either to taxable or exempt supplies and that “no standard or special method is applied, and that therefore there can be no rounding”. He set out the levels of directly related exempt input tax and of the exempt income from PPI and anti stain commission.[153]Mr Todhunter (and HMRC) had forgotten about Mr Hanson’s letter of 1986 by this time.[154]After this there was more correspondence, meetings, and the assessments now under appeal. Mr Gibson’s visits[155]Mr Gibson was the lead officer for DFS from 1996 until his retirement in 2008. Mr Gibson had joined HMC when he left school and remained with them until his retirement. He sometimes found it hard to recall the detail of his work over the last ten years or so of his tenure, and we got the impression that in the closing years of his career he may not have been the terrier he was in his youth. But we also believe that, rightly, in our view, he did not regard DFS as tricky or careless or likely to get involved in complex avoidance schemes. As a result he viewed them as low risk. In so doing he overlooked the possibility that DFS may have been muddled or confused or honestly pursuing a mistaken or inappropriate practice.[156]Mr Gibson was the lead officer for another ten or so large companies. None of them had a substantial proportion of exempt income, and he had little experience of partial exemption. The other officers in his office were in a similar situation although each region had a partial exemption specialist to whom queries could be referred. Mr Gibson arranged two or three visits a year to audit DFS. He would draw up a plan of work for each year. The plan would involve the audit of different parts of the system which produced DFS’ VAT returns. So for example for 1999/2000 (not an untypical year) 26 man-days of work were planned which were allocated to auditing:(a) the VAT return system(b) the purchase ledger(c) the retail sales system(d) the nominal ledger; and(e) a general review.[157]The audit work was designed to check the accuracy of the build up of the data for the VAT returns. There would be an investigation and recording of the way in which the systems worked, an evaluation of the controls which were intended to ensure the coherence and accuracy of the system; and tests on the operation of the system by for example checking that an invoice recorded on the system had been accurately and properly recorded. The checking of the VAT return system would involve chasing the figures on the return back to the system documents from which the entries came. If one was happy that the figures on the return came from those systems and the systems should and did work properly, one could have confidence in the figures on the return. There was also some checking that uncreditable VAT had not been claimed.[158]Before each visit Mr Gibson would write to DFS letting them know what would be audited, and at the end of the visit write with a report on the result of the audit. These letters consistently indicated that no major problems had been found. Thus for example in 2001 Mr Gibson wrote to Mr Todhunter and said:
“Following my visit of this week, during which I carried out an audit of your VAT returns, I am writing to inform you that no errors were disclosed from the checks imposed by me. I would like to thank you for your help during the visit.”
[159]In the course of his audits Mr Gibson would have seen nominal ledger entries indicating that substantial amount of insurance commission had been earned from two different sources, and that these had been treated as being exempt. He did not give further thought to these figures. He also saw accounts in which the commission income was described. His review of the P&L account in 2004 concluded that there were no obvious undeductible costs.[160]During the course of his audit of the purchase ledger Mr Gibson did not become aware that the input VAT on the anti stain spray had been blocked. Mr Gibson was not expressly told that that VAT had been blocked.[161]Mr Gibson did conduct checks to ensure that input VAT relating to cars and business entertainment had been blocked. He also investigated whether any input VAT was attributable to a share issue. He did not investigate whether input tax was attributable to any of the exempt commission income.[162]Documents prepared by Mr Gibson in 2001 and 2003 as part of the standing record for the trader indicate that “the trader has some exempt income from finance commission but is [usually] de-minimis for partial exemption.” Mr Gibson said that the fact that there had been no partial exemption calculations in DFS’ records had led him to this conclusion, and that he intended it to mean that the expectation was that the de minimis would apply, rather than that he had checked it. We noted earlier that Mr Todhunter had said that the exempt input tax in early years was below the limit. It seems to us likely that this had led to an understanding by HMRC’s officers dealing with DFS that its exempt input VAT was de minimis. That understanding had been passed to Mr Gibson by his predecessor, Mr Featherstone. Thus Mr Gibson’s conclusion derived also from that source.[163]Because he understood that DFS were de minimis in terms of partial exemption, Mr Gibson did not conduct a special partial exemption audit. He regarded partial exemption as a low risk area for DFS. Had he realised that DFS were blocking the VAT on a substantial amount of input tax (on the anti stain spray), he would have regarded partial exemption as an issue and conducted a partial exemption audit.[164]In our judgement, Mr Gibson looked at partial exemption through the wrong end of the telescope. Rather than asking: has the trader exempt outputs? and then: is there input tax related to this, and if so is there any input tax which should not be allowable, and if so is it de minimis? he started by asking the last question without considering how one would determine what the disallowable input tax would be. He simply did not think about the question of the attribution of inputs to exempt outputs. Auditing partial exemption meant to him, not determining whether there were exempt outputs and if so whether was disallowable VAT, but checking the accuracy of a partial exemption calculation done by the business. His focus was on chasing the figures rather than on applying the law. He may have been lulled into a false sense of security by the comment by his predecessor that DFS were de minimis, but despite being aware of the nature and size of the exempt income did not apply his mind to the issue further.[165]Mr Todhunter. We have said that we thought that Mr Todhunter’s description of himself as no VAT expert was correct. His evidence as to whether DFS had residual tax, whether what it was doing was operating a special method or the contemporary standard method, and as to whether it was partially exempt was confused and at times contradictory. In our judgement this stemmed from a lack of understanding of these terms. We found his evidence in this regard neither reliable nor helpful. His evidence however in relation to what was done and who said what was clear and we accepted it. Summary: Our conclusions on methods applied and applicable.[166]From 1987 onwards DFS was in receipt of significant exempt income. The only relevant input VAT it blocked was that relating to the anti stain spray (from 1997). 1987-1988[167]In this period DFS had exempt property and PPI income. If the insurance disregard did not apply then DFS’ returns as prepared under the Todhunter method would not have been consistent with(a) the standard method, because inter alia there was not attribution of head office expenses which in our view must have been used in part for PPI purposes and thus cannot have been used “wholly” for taxable supplies, or(b) the Hanson Method for much the same reason.[168]If the insurance disregard did apply, the same conclusion applies:(a) the failure to apportion head office expenses - to recognise that some must have been used partially for the Droitwich income - was inconsistent with the standard method; and(b) that failure was also inconsistent with Hanson Method. 1989- 1996[169]In this period the only exempt income was from PPI. If the insurance disregard applied, the intermediation supply would have been treated as taxable for input tax recovery purposes. As a result under any standard method DFS would have had no exempt input tax and DFS’ returns would have been consistent with the result of applying(a) the prevailing standard method,(b) the Hanson method (if different) (save perhaps in relation to record keeping) and(c) the Todhunter method.[170]The issue of whether or not the insurance disregard applied was not debated before us in detail (we note that in InsuranceWide – see below - there were three days of argument on the meaning of insurance agent), but on the facts as we understand them an officer of HMRC might have concluded that it applied, although we come to a different conclusion (see paras 207ff below). We discuss below the results which follow if the insurance disregard did not apply. 1997-1999[171]This was the period in which anti stain commission was also received. If in this period the commission was not received in the course of a business as an insurance agent the insurance disregard would have applied to this income too, but DFS’ returns would not have been consistent with the result of applying either(a) the standard method or(b) the Hanson method (if different), because of the blocking of the anti stain spray inputs which would not have been required if the insurance disregard applied. The returns were prepared under the extended Todhunter method. 1999-2005[172]In this period the insurance disregard could no longer apply. As a result the commission income could not have been treated as taxable for input tax recovery purposes. The size of the anti stain spray input tax on its own would have exceeded any de minimis. Thus in this period DFS’ returns would not have complied with the applicable standard method or with the Hanson method, but were prepared under the (extended) Todhunter method. DFS ’ Knowledge[173]It seems to us that for most of the period from 1986 to 2005 DFS’ knowledge can be equated with that of Mr Todhunter, although we did wonder whether the auditors had considered the issue at any time.[174]Mr Todhunter knew how input VAT was being treated in DFS’ returns.[175]We do not believe that in 1986 and probably until 2005 (if at all) that Mr Todhunter fully understood the difference between a standard method and a special method. He thought however that he understood Mr Bear’s instructions and that he was acting in accordance with them. In later years he continued to follow what he thought those instructions to be. He knew he was operating a method but did not know whether it was standard or specially allowed or approved.[176]Mr Todhunter was therefore neither under the impression that HMRC had permitted a special method, nor under the impression that they had not.[177]We do not believe that Mr Todhunter had sufficient knowledge of the VAT system in 1987 and 1997 or sufficient knowledge of how HMRC conducted audits to have been able to conclude that HMRC would or would not, or should or should not, specifically have checked the effects of the receipt of exempt income. But from 1987 onwards he was under the impression that there was nothing wrong with what DFS was doing.[178]Whilst he did not hide anything from HMRC he did not know enough to put squarely to HMRC such facts as would, if HMRC had said “that’s fine”, have enabled him to be confident that what he was doing had been approved.[179]A person with a much greater knowledge of VAT and the working of HMRC may well have concluded on balance that HMRC were happy with DFS’ treatment of exempt inputs because it would in our view be reasonable for such a person to suppose that a reasonably competent officer would have been aware of the exempt income and would have considered whether input VAT should be blocked. Such a person (and indeed a person with reasonable VAT knowledge and only a general acquaintance with HMRC’s workings), however, would have been puzzled by HMRC’s failure to raise the issue and could reasonably have been expected to have put the matter squarely before HMRC.[180]In our view HMRC had enough knowledge of relevant facts to have been able on reasonable investigation to conclude from 1999 that DFS’ exempt input tax was not being computed in accordance with either the applicable standard method or the Hanson method. The Parties’ Arguments[181]Mr Thomas says: (14) because of the application of the insurance disregard in the period up to 1999, the events in that period have no bearing on the question of whether a special method was permitted in that period in relation to intermediation income; (15) after 1999 an input tax restriction would have been applicable only if exempt input tax amounted to more than 1% of DFS’ turnover ( because of the rounding up of the standard fraction to the nearest integer). That calculation was never done. Mr Gibson did not know that it exceeded 1% of turnover. The issue is not whether Mr Gibson knew the size of the intermediation income, but whether he addressed his mind to the question of whether it exceeded 1 %. He did not. If Mr Gibson never gave the issue any thought it is difficult to see how it can be said that the Commissioners approved a method: they could not unconsciously exercise a statutory power, particularly one which required a judgement to be made. There was no evidence that in 1999 or thereafter HMRC even considered partial exemption; (16) the Hanson Method was a “three pot” method. The Appellant had operated a “two pot” method. The Appellant did not seek approval for a two pot method instead. A variation of the three pot method was not considered by the Appellant or the Commissioners. The application of the insurance disregard meant that there was no need for any variation of the Hanson Method to be discussed or agreed; (17) anti stain spray input tax blocking was not discussed with and was not obvious to HMRC. Mr Gibson did not know about the two pot method. He cannot have judged it acceptable; (18) Mr Todhunter’s evidence was that the old method became the standard method in 1992. That evidence was consistent with the Hanson Method being the allowed special method; (19) the audit letters from Mr Gibson cannot be treated as giving an all clear. There function was not to do that. They made no reference to input tax recovery or residual tax. They are not blanket acceptances of DFS returns; (20) the evidence in relation to the Primback Voluntary Disclosure did not indicate any consideration by HMRC of a special method. Mr Todhunter made no reference to a Special Method in his precursor conversation with Mr Gibson; (21) the evidence in relation to the 2004 anti avoidance visit by HMRC indicates that at that time Mr Todhunter did not think that there was a special method in operation. The Appellant’s early discussions with HMRC in 2005 and other evidence before the tribunal indicate that DFS did not in 2005 or before then consider itself to be operating a variation on a special method. The argument for a variation of a special method is a retrospective analysis. All that happened was that DFS operated its own view of the standard method and that that view was wrong; (22) Mr French’s awareness of the anti stain blocking in October 2005 cannot found “approval” because from April 2005 approval had to be in writing. In any event Mr French did not consider partial exemption; and (23) DFS did not understand itself to be operating a special method. DFS were operating the standard method incorrectly. No variation of any method was sought or obtained, The Appellant’s case is that HMRC, even in the absence of a ruling which could bind them in judicial review proceedings, are bound to accept what DFS did by reason of an implied approval from which they are not permitted to resile.[182]Mr Cordara says: (24) viewed against the mindset of the period, the Todhunter Method is a proper interpretation of the letter from Mr Hanson. We expand on this below; (25) the Wellington test is an objective test of conduct:(a) it requires only deliberate conduct by the taxpayer and awareness by HMRC;(b) it is not necessary that either party believed itself to be allowing agreeing or varying something;(c) erroneous belief is irrelevant;(d) the question is whether HMRC were aware, not whether they must be taken to have exercised a power. Once they have sufficient facts to know what is being done the line is crossed;(e) it is not necessary for the taxpayer formally to have informed HMRC of what it was doing; (26) the approach adopted by DFS was deliberate and clear. Mr Todhunter thought he was doing what he had to do, and he understood what he was doing even though he may not have understood it was a special method; (27) HMRC knew DFS had substantial exempt income and that it was reclaiming all its residual tax. There were no fraction calculations. From 1999 onwards it was clear that the 1% rounding de minimis did not apply; (28) the Wellington test is as applicable to “approval” as it is to “allowing”. The principle of legitimate expectation must be observed by member states when the exercise powers conferred on them by community directives (see para [57] Gemeente Leusden [2007] STC 776), and is therefore relevant to the meaning of “approval”.[183]We noted above some possible ambiguity or gap in the language of Mr Hanson’s letter. Mr Cordara says that we must guard against reading the letter in the context of the present state of the law in which “direct” has become a special term, and we now understand the distinction between overheads and other inputs. BLP he says was decided 9 years after Mr Hanson wrote his letter: it was in BLP that “direct” emerged as a touchstone of the required nexus. Mr Hanson wrote against a background in which there was no culture (or judicial clarification?) surrounding input tax attribution. He notes that Mr Hanson did not say “exclusively” used, but “directly related”. In that context it is possible to read “direct” as meaning most closely related. On that basis inputs used for an antecedent taxable stream of outputs remained most closely connected and thus directly related to those outputs, and so remained allowable. Mr Cordara points to the arguments of HMRC in MBNA in which just such an argument was run: only the additional inputs occasioned by a new supply should be apportioned. Thus even more recently he says the approach taken by Mr Todhunter was alive. Discussion[184]In this section we discuss(i) the meaning of “allow” and “approve”,(ii) whether a method was allowed or approved before 1999 whose approval had effect thereafter, and(iii) whether a special method was approved after 1999. (I) Meaning of “allow” and “approve”.

(a) allow

[185]In Wellington Private Hospital Limited v CCE LON/92/2284, the tribunal considered the meaning of “allow” in regulation 30(2) of the 1995 regulations. The Chairman considered that “allow” should be given a wide meaning and that two features had to be present:
“Firstly, the taxpayer must have knowingly adopted or sought to adopt a special method. Secondly, Customs must have been aware of what the taxpayer was doing or seeking to do.”
[186]In that case the taxpayer had indicated to Customs that it would use one method of determining its unallowable input tax but had actually used another. The tribunal found that the method it actually used could not have escaped the notice of the visiting Customs officers, and that accordingly the method actually used had been allowed. In that context, it is clear to us that when the tribunal said “the taxpayer must knowingly have adopted …a special method” it intended those words to encompass that the taxpayer had adopted a course of conduct which amounted to a special method. In other words that the taxpayer did not have to know that the way it computed its deductible input tax was different from the standard method so long as it knew what it was doing. So formulated we gratefully adopt the same test. That is because there is nothing in the word “allow” which suggests that he who is allowed to do something needs to know the effects of what he is allowed to do. .[187]We note that it is an objective test of awareness, not a test of whether HMRC knew what they were doing. Thus if in permitting something they made a mistake of law that does not prevent their cognisance of what was happening constituting allowing. The question is whether HMRC were aware, not whether it may be inferred that they knew what they were doing.

(b) approve

[188]Mr Cordara says that “approve” must be viewed in the context of the Directive which permits member Sates to “authorise or compel” the use of methods other than that in Art 19. He says there is a contrast to be drawn between the active nature of compelling and the more passive nature of authorising which infects the proper construction of “approve or direct” in the domestic legislation. He notes that in Wellington the decision indicates that the Commissioners used the words allow and approve interchangeably. The contrast is to be drawn domestically between compelling and approving. Approve should be construed along the lines indicated by Wellington .[189]“Approve” to our minds has different connotations from “allow”. “Approve” suggests some demonstration of consent, whereas “allow” encompasses simply letting something happen. We accept Mr Thomas’ argument that the 1992 change from “allow” to “approve” indicates that some different concept was intended, and that the contrasting use of allow and approve in regulation 31 of the 1992 regs reinforces that indication. Reference to the Sixth Directive does not in our view help Mr Cordara: “authorise” suggests an act of giving authority rather than letting something happen. It is different from compelling, but because it is different does not mean that it means allowing. If the domestic legislation is to be construed as an implementation of the Directive, then we would construe “approve” as requiring some act of giving authority.[190]Mr Thomas makes four further submissions: (29) it is ultra vires the powers conferred on the Commissioners to give approval to a method which does not reasonably or fairly attribute input tax on the basis of use. And accordingly that any purported approval given of a method which is unfair or unreasonable is void. We note that the Appellant objects to this argument being considered because it was not addressed in HMRC’s Statement of Case.; (30) he says that before any act on behalf of the Commissioners can be said to be approval:(a) the Commissioners must have considered the issue;(b) the Commissioners must have known ( rather than reasonably been able to infer that ) the method was different from a standard method;(c) the act must be one which, objectively examined, evinces an intention to give approval; (31) whilst approval by a course of conduct might be possible, such conduct must be clear and unambiguous; and (32) the method operated (or to be operated) by the taxpayer must be understood by the taxpayer to be different from the standard method before it can be said that that method has been approved.[191]Mr Thomas says that the purpose of the relevant regulations was not to insulate a taxpayer who has wrongly applied the standard method from the retrospective assessment which is visited on any other taxpayer who has wrongly calculated its VAT, but to achieve a fair and reasonable attribution of inputs to taxable supplies by reference to use. It would be wrong to import into the regulations some form of gloss on the meaning of approval based on doctrines of reasonable expectation because that would go against the grain of legislation which envisages an act of approval by the Commissioners.[192]Mr Cordara says that there is scope for the application of the doctrine of legitimate expectation because one is dealing in an environment in which the rules can be altered by the exercise of administrative powers to authorise or compel.[193]We accept that the Act confers a power to make regulations for the purpose of securing a fair and reasonable apportionment on the basis of use (we shall call this a Fair basis or method below). We accept that the regulations must be viewed in that light and that such purpose permeates the regulations. We also accept that a method could be unFair as much because it overly favoured HMRC as because it overly favoured the taxpayer. But we do not accept that the use of the power given by the regulations in a manner which results in an unFair method is void. That is for the following reasons: (33) the regulations permit HMRC to terminate the use of an approved method (see reg 102(3) 1995 Regs, reg 31(2) 1992 Regs. reg 36(2) 1987 Regs). Those provisions suggest that it is not necessarily intended that an approved unFair method is void; (34) since 2003 the regulations have provided that if a method does not achieve a Fair attribution, adjustments must be made (see reg 102A and C 1999 Regs as amended). That at least hints that it is possible that an unFair method may have been approved or directed as well as that a method may have become unFair; (35) a right of appeal is given against the imposition of an unFair method or the failure to approve a Fair Method. Such right is not limited to whether or not HMRC acted reasonably but extends to an appraisal of the method. That suggests that the proper remedy for the taxpayer against an unFair method is to appeal (with the corresponding remedy for HMRC being that in (1) above). Warren J said in St Helen’s School Northwood ltd v HMRC [2007]STC 633 at [27]:
“…if the tribunal thinks that both the existing method and the proposed method are unfair and unreasonable, it could not allow the appeal even if it considers that the proposed special method is less unfair and unreasonable that the existing method.”
. There Warren J seems at least to be assuming that an unFair method imposed by the Commissioners is not void.[194](In our view the same considerations and conclusions apply in relation to “allowing” in relation to this issue.)[195]We do not accept Mr Thomas’ point (2)(a) and (b): the regulations cannot have been intended to require an investigation into the actual knowledge and state of mind of HMRC before a taxpayer could be confident that approval had been given. So construed the regulations would be unworkable and impractical: no one would be able to rely even on a plain letter from HMRC saying “we approve” without extensive further investigation. In our view the test must be an objective one: the question must be whether HMRC have acted so as to convey their approbation of the method. But these issues are, in the sense we indicate below, indirectly relevant to that objective determination.[196]In our judgement approval is given by HMRC to a method used or to be used by the taxpayer for calculating recoverable input tax if HMRC conduct themselves in such a way that their conduct would convey to a reasonable man in the circumstances of the taxpayer(a) that HMRC knew what method was being used or proposed,(b) that they had considered it,(c) that they had agreed to it, and(d) that their conduct constituted the communication of those elements.[197]As a result what HMRC actually knew, and what the taxpayer actually understood are irrelevant.[198]Mr Thomas’ ultra vires point re-emerges in our formulation in a different form. If what HMRC apparently approved is so odd, so contrary to the spirit of attribution based on use that “there must have been some mistake”, then it could not be said approval had been given because it would be clear that the objectively obvious mistake meant they had not truly approved.[199]Our formulation leaves no room for the views of an expert in HMRC’s procedures or practice. We assume a reasonable person who is acquainted with the law and who knows what was said to and by and done to the taxpayer by HMRC. He may assume reasonable competence on the part of HMRC and an understanding by them of the law, but the detail of how HMRC normally work or what they would normally consider internally is irrelevant. For this reason we have taken no notice of the evidence of Mr Jeremy Hall in this respect.[200]HMRC’s approval must be of a method which is “other than” the current standard method (see 30(2) 1985 regs, 30(5) 1987 regs, 31(6) 1992 regs etc). If what the taxpayer does in the circumstances is consistent with the prevailing standard method and also with a differing method, then unless it appears that consent was given to the programme for, or basis of, the calculation (the set of steps to be undertaken rather than what figures have entered the calculation) then approval of what has been done cannot be approval of an “other method”. Put another way, the logic of the method which has been approved must be clear.[201]If HMRC make a mistake as to the law then that does not in our view vitiate their approval so long as, objectively along the lines we have indicated, approval has been given. (II) Was a special method approved before 1999 which continued thereafter?[202]In the context of the arguments advanced before us this heading raises the following questions which we address under the corresponding headings below: (36) Does a method allowed before 1992 transmute into an approved method in 1992? (37) Can a method which has been allowed or approved cease to be so treated because it is not or has not been used? (38) When, by reason of a change in the regulations, a special method becomes the same as the standard method, but then the regulations change again, does approval given before the first change survive so as to remain effective after the second change? (39) Did the insurance arrangement disregard apply at any time between 1987 and 1999? (40) Was the Todhunter method or an extended version of it allowed or approved in any part of this period? (1) Does a method allowed before 1992 transmute into a method approved thereafter?[203]We consider that it does.[204]In 1992 the regulations substituted “approval” for “allow”. The regulations came into force on 11 March 1992. In the April 1992 version of Notice 706 HMRC explained the new standard method. In paragraph 19 there was a box which dealt with certain transitional matters. The second and third paragraphs in that box read:
“Conversely, the new standard method was formerly a special method. If you have been using it in the past and wish to continue using it in the future, you also need take no action as formal approval is no longer necessary. “If you have been using any other approved special method in the past, and wish to continue using it in the future, you may do so subject only to any review of your method which the local VAT office may carry out.”
[205]In our view that third paragraph constituted approval by HMRC of any method previously allowed (subject to later review if any). The use of the words “approved special method” does not dissuade us from this conclusion. The preceding paragraph indicates an elision of the concepts of approval and allowing, and a broad approach to the language of the regulations. The intention is to our minds clear. (2) Can a method which has been allowed or approved cease to be so treated because it is or has not been used?[206]Reg 36 of the substitution into the 1985 regulations made by the1987 regulations provides:
“(1)Save as the Commissioners may otherwise allow or direct and subject to paragraph (2) of this regulation a taxable person using any method shall use it for at least two years. “(2) The Commissioners may at any time notify a taxable person that the use of any method allowed under regulation 30(5) is terminated…from such future date as may be specified in the notice.”
[207]Regulation 31(2) of the substitution made by the 1992 regulations provides:
“(2) A taxable person using a method as approved or directed by the Commissioners under paragraph (1) shall continue to use that method unless the Commissioners approve or direct the termination of its use.”
[208]Regulation 102(3) of the 1999 regulations provides in the same words as those above.[209]These regulations indicate to us: (41) if a method is not used in a computation because for example the outputs which gave rise to its use are no longer made, the obligation and right to use the method continues and will revive if those outputs recommence; but (42) on the other hand, if a method was allowed or approved and then the taxpayer uses a different method, the allowed or approved method will lapse only when and if the different method is allowed or approved. (3) When, by reason of a change in the regulations, a special method becomes the same as the standard method, but then the regulations change again, does approval given before the first change survive so as to remain effective after the second change?[210]In our view it does not in the period up to 1992 in which the “allowing” words were used. That is because in the interim period it will cease to be an “other method”, and thus cease to be something which is allowed or approved by the Commissioners for the purposes of the regulations.[211]In the approval world the issue potentially arises only in relation to the replacement of the 1992 regulations by the 1999 regulations. Regulation 3(2) of the 1999 regulations provided that anything done under or for the purpose of the revoked regulations “shall be continued…for the purpose of the corresponding provision of these regulations”. Thus an approval given under the 1995 rules would continue to be an approval under the 1995 regulations. Further the standard method did not change. (4) Did the insurance arrangement disregard apply at any time between 1987 and 1999?[212]The disregard does not apply “in the case of a taxable person who carries on the business of, or a business similar to…an insurance …agent or broker”. That raises two questions(a) did DFS carry on activities which were or which were similar to those of an insurance agent or broker? and(b) did those activities mean that DFS was conducting a “business” for these purposes?[213]In relation to the first question, we have been assisted by the judgement of Evans-Lombe J in InsuranceWide.com Services Limited v The Commissioners of Her Majesty’s Revenue and Customs [2009] EWHC 999 (Ch) , which was given on the fourth day of the hearing before us but not referred to by the parties despite a later invitation to do so. The judge held that the introduction service performed by a taxpayer in channelling would-be assureds to insurers with a view to their making a contract of insurance was sufficient to render a taxpayer an insurance agent or broker within Art 13B of the Sixth Directive and/or Sch 9 Group 2 Item 4 VATA 1994. It seems to us that there should be no difference between the meaning of insurance agent or broker in the relevant regulations and that in Sch 9 Group2 Item 4, and that the reasoning in this judgement leads to the conclusion that DFS were in relation to PPI and anti stain insurance acting as an agent or broker notwithstanding that it may not have had power to commit the insurance companies to accepting the insurance.[214]In relation to the second question, we note that it is to be asked in the context of a taxpayer who is already acknowledged to be carrying on a business, and thus that merely carrying on those activities in the course of that business is not enough. However,(i) the context suggests that the fact that a taxpayer carries on a business which includes activities other than those specified will not prevent the disregard from applying,(ii) the regulation does not refer to a “separate business”, and that suggests to us that one is not required to ask whether the activities could, divorced from the other activities of the taxpayer, stand on their own as a business, but rather to ask whether the activities were such as taken together were a business. Thus for example if an estate agent sold his business car one would expect to find that the estate agent did not carry on the business of selling cars even though the car was sold in the course of his business; but if a garage proprietor mended cars but also sold cars from his forecourt, one would expect to find that he carried on the business of selling cars even though it was not a stand alone business.[215]In this context the indicia of “business” accepted by Gibson J in CCE v Lord Fisher [1981] STC 238 provide some guidance:(a) was the activity a serious undertaking earnestly pursued,(b) was it actively pursued with recognisable continuity,(c) did it have a measure of substance measured by the value of its supplies,(d) was it pursued in a regular manner on sound and recognised business principles,(e) was it predominantly concerned with making taxable supplies,(f) were those supplies of a kind which are commonly made by those who seek to profit from them?[216]Those factors suggest to us that the arranging of the insurance by DFS was such a business. It was an adjunct to the main business of DFS but was not not purely incidental or occasional. It was pursued seriously, regularly and in a businesslike way. The income derived was substantial in absolute terms even though it was less than about 5% of the turnover of the company.. The position seemed clearer in relation to the period after 1997 when anti stain insurance was sold. Then the company had also to arrange the spraying of the sofas. It was an activity commonly conducted with a view to profit which involved the making of supplies.[217]Thus we conclude with some confidence that from 1997 onwards that the disregard did not apply, and that between 1987 and 1997 that it is fairly clear that it did not apply.[218]There is one further question which can be conveniently dealt with at this stage, and that is how HMRC would have considered the position between 1987 and 1997. In the 1992 version of Notice 706, HMRC set out their view on the application of the disregard. At paragraph 15 they say:
“When you arrange insurance…transactions regularly and on a large scale the question arises as to whether you are carrying on a separate business of, or similar to, one of the descriptions mentioned above. The answer to this question is ultimately a question of fact but the amount of exempt input tax incurred may be an indicator. If, for example, a retailer selling freezers offers insurance against damage to the contents of the freezer, it is highly likely that the brochures describing the insurance together with the proposal forms will be provide by the insurance company and the retailer will incur little, if any, input tax related to the insurance cover over and above that incurred on the overheads of running the shop. However similar insurance may be offered by a larger retailer with a chain of shops. In this case, because of the scale of the operation, the insurance paperwork could be administered centrally, perhaps at head office, and there may well be identifiable input tax on office equipment, computers etc. This could suggest that there is a separate business of, or similar to that of an insurance agent or broker.”
[219]We do not share HMRC’s view as to the central importance of exempt input tax to this question - other than as one indicator of the substance of the activity - but it seems to us that an officer who was aware of the PPI insurance and of this statement would have given serious consideration to whether the disregard applied in the period 1987 to 1997, and if he was aware of the anti stain spray inputs would have been likely to conclude after 1997 that the disregard did not apply. The emphasis on exempt input tax may well have persuaded an officer to conclude in the 1987-97 period that the disregard did apply. (5) Was the Todhunter Methodor an extended version of it allowed or approved before 1999?[220]DFS had exempt income from Droitwich from 1986 to 1989. In that period it applied the Todhunter method to compute its creditable input tax. In that period Mr Bear visited. Mr Todhunter said, and we accept that Mr Bear was quick to point out errors. Mr Bear was involved in setting up the method. His note to Mr Hanson suggests that he had in mind the need to check its operation. We find that he did check, and did see, what DFS was doing and that he allowed it. What DFS was doing was something which amounted to a special method. The possible ambiguities in Mr Hanson’s letter, the general points Mr Cordara makes about the less developed state of notions of direct and exclusive attribution, and the antedating of the legislative introduction of “wholly”, taken together with Mr Bear’s correspondence with Mr Hanson, convince us that the Todhunter method was that which Mr Bear saw and allowed. We find therefore that HMRC allowed the use of that method as respects the Droitwich outputs.[221]The Todhunter Method was designed in the context of Droitwich. It operated by searching the expenses ledger for those costs applicable to both Droitwich and the rest of the business and apportioning them. The search was for additional expenses which arose as a result of Droitwich.[222]The philosophy or logic of the method applied to Droitwich does not translate automatically into the PPI world. Whereas Droitwich was a separate activity divorced from the main business of DFS, PPI was intimately bound up with that business. It does not seem to us that the method which was applied to Droitwich had to be extended to apply to PPI, because what DFS did in relation to PPI failed to examine the general costs of the business into which PPI was integrated. Thus if the insurance disregard did not apply what DFS did in relation to PPI was not a simple application of the Droitwich Todhunter method but a modification or extension of that method.[223]However, PPI started in 1987 just after the settling of the Todhunter Method in relation to Droitwich. We find it likely that in these circumstances that Mr Bear or his successor at that time examined what was done and allowed it.[224]It is possible that such allowance was in the context of a view that the insurance disregard applied. If the disregard did apply then there was nothing new to “allow”. Thus there was no allowing of a method which ould apply to PPI type income.[225]However, if the disregard did not apply then, even if HMRC took the view that it did apply, there would in our view have been an allowing of the method adopted. That method was an extension of the Droitwich method. That is because “allow” as indicated in Wellington imposes an objective test by reference to HMRC’s conduct not a test by reference to what it thought. It does not seem to us that allowing the Todhunter method to be extended to cover the PPI income was an obvious mistake.[226]We call the Todhunter method, as extended to cover insurance commission income the “extended Todhunter method”.[227]Thus if the disregard did apply then there was between 1987 and 1992 an allowing of the extended Todhunter method. If the disregard did not apply then all that was allowed was the Todhunter method confined to Droitwich style outputs.[228]Anti stain commission income was first received in 1997 – in a period in which approval rather than allowing was key. In this period Mr Gibson was the lead officer for DFS. We discuss below whether he separately approved DFS method in that period.[229]It is clear to us that the extended Todhunter method was used in relation to the anti stain income. The nature of that income was not in our view so different from the PPI commission income that the principles of the extended Todhunter method could not be said to have applied to it. There was if the insurance disregard did not apply no separate method used in relation to the anti stain income. (III) Was a special method approved after 1999?[230]Our concern is whether such approval been given by the period 06/04 – the period relevant to the first assessment[231]After 1999 we make the following appraisal of what HMRC said and did and must have known: (43) it performed regular audits lasting some 20 or 30 days a year in total of DFS’ VAT systems (44) it would, in particular though its audit of the sales ledger and the 1999 Primback re-adjustment, have been aware that DFS had substantial exempt income; (45) it would, by January 2004 have been aware, from the anti avoidance team visits, of the nature of the exempt income; (46) it was not aware until 2005 (through Mr French) that anti stain spray costs were blocked; (47) its audits were directed to particular systems. The audit of some systems would not indicate a consideration of the proper claiming of input tax. Thus no conclusion could be drawn from for example the letter of 12 April advising that only two small errors had been found in the audit of the purchase ledger. On the other hand on 15 November 2000 it wrote indicating that the audit carried out on the VAT return system had revealed no errors. That audit had been presaged as including the scrutiny of the management and annual accounts. A letter of 7 November 2001 and 19 January 2003 relating to the audit of VAT returns indicated no errors. A letter dated 14 January 2004 indicated that no errors had been found with the “build up of your VAT returns”; and (48) DFS was not asked and did not explain how it dealt with input tax attribution.[232]The method DFS used in this period was not consistent with the standard method. In our view a reasonably competent officer could be expected to have considered whether DFS should have blocked some of its input tax either by reference to the standard fraction or otherwise. Such a consideration could reasonably be expected to have given rise to question about input tax blocking. But no such questions were asked. The fact that the questions were not asked would lead a reasonable person to conclude that no consideration was given to the issue. That is to be contrasted with the letters relating to the audit of the VAT returns. These letters could be read either as a statement that the system for producing the return had operated according to plan and numbers had been properly transferred to the returns form the relevant (audited) sources, or that the method by which the returns were compiled was accepted. The fact that the letters are in that sense ambiguous, coupled with the implication that no thought had been given to partial exemption, leads us to the conclusion that a reasonable person would not conclude that HMRC had considered and approved the method which DFS had adopted. Conclusions[233]We find that:(i) the insurance disregard did not apply between 1987 and 1999;(ii) in 1987 HMRC allowed the use of the Todhunter method as regards the land. That allowing had the effect of superceding the Hanson method which thereupon lapsed;(iii) in 1987 onwards HMRC “allowed” the use of an extended form of the Todhunter method to encompass the effects of insurance commission income;(iv) no notice was given by HMRC revoking that method;(v) in 1992 that method transmuted into an approved method;(vi) that approved method was not revoked during the appeal period;(vii) that method was therefore the correct method to use in that period.[234]We consider briefly below the effects which would follow if some of our findings were wrong. We fear that there are a considerable number of possibilities but the two below reflect those of our conclusions with which we had the most trouble.[235]If we are wrong about the insurance disregard in relation to PPI, then(i) the Todhunter method was used as respects property only in 1987, and in our view when PPI income arose the method was not extended to cover it: the method was specific to income from property and not to activities integrated with the sofa business -it was not a general method. The proper construction of what was allowed on this hypothesis would in our view to be to fill in the gaps with the standard method applicable from time to time;(ii) no method was allowed separately in respect of insurance commission income associated with the business;(iii) the failure to use the method after 1987 did not prevent its remaining allowed;(iv) it transmuted into an approved method in 1992;(v) in relation to anti stain commission income after 1997 we find that whether or not the insurance disregard applied in respect of such income no approval was given either before or after 1999 for a method which did not apply the standard method in respect of insurance commission income;(vi) the standard method would thus apply in relation to the insurance income in the period relevant to the appeal but with the prior use of the plain Todhunter method as respects any property income in that period.[236]If neither the plain nor the extended Todhunter method was allowed in 1987, then no method was allowed other than the Hanson method. That method became the standard method in 1992. As a result it lapsed at that time. Therefore there was no extant method at 1992 and no inherited method became approved. We found that no approval was given to any method after 1992 (whether or not the insurance disregard applied for the period 1992 to 1999). Consequently there would have been no approved method extant for the period of the appeal, and the standard method should have been used. Costs[237]We make no award of costs: there was something close to negligence over the years on both sides and the factual conclusions are finely balanced. Other Matters[238]If either party considers that there is commercially sensitive material whose presence in the publication of this decision would cause serious concern, we direct that they make application with their proposals for change within 10 days of the release of this decision to them. [The results of that application are reflected in this report.][239]The Appellant and the Respondents have a right to apply for permission to appeal against this decision pursuant to Rule 39 of The Tribunal Procedure (First-tier Tribunal) (Tax Chamber) Rules 2009. The parties are referred to "Guidance to accompany a Decision from the First-tier Tribunal (Tax Chamber)" which accompanies and forms part of this decision. CHARLES HELLIER TRIBUNAL JUDGE RELEASE DATE [of original decision]: 7 August 2009

Cited in 1 later judgment