[Home] [Databases] [World Law] [Multidatabase Search] [Help] [Feedback]
	First-tier Tribunal (Tax)
You are here: BAILII >> Databases >> First-tier Tribunal (Tax) >> DSG Retail Ltd & Ors v Revenue & Customs (Rev 3) [2009] UKFTT 31 (TC) (31 March 2009) URL: http://www.bailii.org/uk/cases/UKFTT/TC/2009/31.html Cite as: [2009] STI 1604, 11 ITL Rep 869, [2009] STC (SCD) 397, [2009] UKFTT 00001 (TC), [2009] UKFTT 31 (TC), [2009] UKFTT TC00001 (TC)

[New search] [Printable RTF version] [Help]

[2009] UKFTT 31 (TC)

TC00001

TRANSFER PRICING - captive insurance company insuring extended warranties for customers - whether business facilities provided at less than the arm's length price under s 770 TA 1988 - yes - whether provision made differs from the arm's length provision under Sch 28AA - yes

THE SPECIAL COMMISSIONERS

(1) DSG RETAIL LIMITED (2) MASTERCARE COVERPLAN SERVICE AGREEMENTS LIMITED (3) MASTERCARE SERVICE

AND DISTRIBUTION LIMITED Appellants

- and - THE COMMISSIONERS FOR HER MAJESTY'S REVENUE AND CUSTOMS Respondents

Special Commissioners: DR JOHN F. AVERY JONES CBE

CHARLES HELLIER

Sitting in public in London on 24 November 2008 to 12 December 2008

Jonathan Peacock QC and Francis Fitzpatrick, counsel, instructed by Reynolds Porter Chamberlain LLP, for the Appellant

David Goy QC, Michael Green QC and Rupert Allen, counsel, instructed by the General Counsel and Solicitor to HM Revenue and Customs for the Respondents

© CROWN COPYRIGHT 2009


 


REDACTED REPORT OF THE DECISION

Save where text appears in square brackets, this is the text of the decision of the Special Commissioners in this appeal. Certain detailed findings which relate to commercially sensitive information have, so far as compatible with a proper understanding of the Commissioners' reasoning, either been redacted or are reported in square brackets in a form which is intended to preserve confidentiality.

1. These are appeals against assessments or loss determinations for the accounting periods ended 30 April 1997, 1998 and 1999 for all three Appellant companies and referrals pursuant to paragraph 31A of Sch 18 to the Finance Act 1998 for the accounting periods ended 30 April 2000, 2001, 2002, 2003 and 2004 for all three Appellant companies (although no adjustment is now proposed in relation to the third Appellant). They relate to the transfer pricing between the Appellant companies and the Group's insurance subsidiary in the Isle of Man, Dixons Insurance Services Limited ("DISL"), which insured or reinsured the liability to customers for extended warranties. For the periods covered by the referral the question for determination has been agreed and this is dealt with at the end of our decision. The Appellant companies were represented by Mr Jonathan Peacock QC and Mr Francis Fitzpatrick, and the Respondents ("the Revenue") by Mr David Goy QC, Mr Michael Green QC and Mr Rupert Allen.

Introduction

2. The Appellant Group (we shall use this expression where reference to no particular company is intended, but we shall exclude DISL from such expression) of which the parent company is DSG International plc is the largest retailer of electrical goods in the UK, comprising Dixons, Currys and PC World. The retail company is DSG Retail Limited ("DSG"). It offers to its customers extended warranties relating to goods sold which continue after the normal one year manufacturer's guarantee for periods of an additional one to four years.
3. There are two separate periods to consider in relation to the facts. First, "the Cornhill Period" (May 1986 to April 1997) during which Cornhill Insurance plc ("Cornhill") was the insurer of extended warranties sold to customers and DISL reinsured 95% of its risk. That period is in issue only as to reinsurance policies written during 1996-97 (to which we shall refer as "scheme year 1996-97" to distinguish it from the accounting period 1996-97), the reason being that the tax issues for earlier accounting periods have been settled (and the Revenue make no claim in respect of earlier business that is reflected in the accounts for 1996-97 or later accounting periods). Policies written in scheme year 1996-97 will, of course, affect subsequent accounting periods up to 2001-02 as well. Secondly, "the ASL Period" (April 1997 onwards), when, in consequence of the proposal in the November 1996 budget to increase insurance premium tax ("IPT") to 17.5% with effect from 1 April 1997, the Appellant Group changed from insured extended warranties to service contracts issued to the customer by a non-group company, Appliance Serviceplan Limited ("ASL"), whose liability was wholly insured by DISL. Although the rates of IPT and VAT were the same, service contracts had the advantage over insured warranties that, while VAT was payable on the service contracts, VAT on the repair costs was deductible. (The Revenue did challenge whether IPT was payable following the change but the Tribunal found that the risk was outside the UK: see DSG International Insurance Services Limited v HMRC (2007) IPT Decision 0013) During the ASL Period, Cornhill continued to provide insured extended warranties for DSG Ireland Limited and it also insured mobile phones, theft and accidental damage and food spoilage loss, which were reinsured as to 95% by DISL.
4. There are also two separate periods to consider in relation to the law. During the first, comprising 1996-7 (the Cornhill Period and relevant only to scheme year 1996-97) to 1998-99 (part of the ASL period), the law was contained in s 770 of the Taxes Act 1988. In the second, affecting periods 1999-00 to 2003-04 (part of the ASL period), the law is contained in Sch 28AA of the Taxes Act 1988, introduced by the Finance Act 1999. Briefly, the parties are agreed that the main difference between them is that under s 770 one takes the facts of the transaction as they are and asks whether a price would have been charged for entering into the transaction if the parties had been at arm's length, whereas under Sch 28AA one asks whether the terms of the transaction would have been different if the parties had been at arm's length.
5. In asking these questions the parties are agreed that the time at which the arrangements are looked at to determine whether they are at arm's length is in 1993 (the time of the latest arrangements with Cornhill) and 1997 (the time of entering into the arrangements with ASL), and that hindsight cannot be used. (In finding facts relating to periods later than 1997 to give a complete picture of the transactions we are not implying that those facts can be used to determine whether any transfer pricing adjustment should be made.)
6. We had about 40 files of documents. We heard evidence from Mr S C Carroll (at the relevant time Group Development Director of DSG International plc), Mr G Budd (at the relevant time company secretary of companies in the Appellant Group and a director of DISL), Mr A J Moore (Finance Director, DSG Group Financial Services), Mrs D Winrow (DISL), Mrs C Douse (Willis Group plc), and Mr M P Rust (Deloitte & Touche LLP). We also heard evidence from the following expert witnesses called by the Appellants: Mr M Bezant (LECG), Mr C Beaton (Callum Beaton (Insurance Consulting) Limited); and the following expert witnesses called by the Revenue: Mr S Gaysford (Frontier Economics Limited), Mr P A Bawcutt (retired insurance consultant and the author of a book on captive insurance companies) and Mr C A Heneage (managing director of Lloyd's broker SBJ Global Risks Limited). Mr Bezant and Mr Gaysford also made a joint report, as did Mr Beaton and Mr Bawcutt, for which we are grateful. We were grateful to the Appellants' solicitors for providing us with the documents on a CD, which proved most useful to us when writing the decision. To avoid repetition where we say that a witness of fact (ie witnesses other than the expert witnesses) said something we accept this as a finding of fact unless the contrary is clear from the context.

Background and history

7. Dixons had offered extended warranties since the late 1970s, at the time mostly in relation to photographic goods, insured through Lloyd's. Currys, which sold white goods such as fridges, freezers, washing machines and dishwashers, had a larger extended warranties business than Dixons'; it was insured by Multi Guarantee Limited which went into liquidation in July 1982. Problems existed for other insurers writing extended warranty business in the early 1980s. In July 1982 Currys entered into an agreement with Orion Insurance Company plc ("Orion"), an independent insurance company, which provided extended warranties through Currys from then until 1987. Following the take-over of Currys by Dixons in 1984, the Appellant Group wanted to combine the two companies' extended warranty arrangements into one but in the light of the difficulties over Multi Guarantee Limited and the failure of other companies insuring extended warranties, the Lloyd's underwriters with whom they had previously dealt declined to take on more business and only two other insurance companies offered this business at the time.
8. From 1984 onwards the Appellant Group was well placed to assess the risks of extended warranties. Mastercare Service and Distribution Limited ("MSDL") was a Currys' subsidiary that did repairs (not only for Currys' customers but for the public) and had shops throughout the country. Dixons had a smaller similar operation that was combined into MSDL after the take-over. As part of this work MSDL kept data about the reliability of products.
9. A retailer has a considerable advantage in selling extended warranties. The Competition Commission said in its December 2003 Report on Extended Warranties on Domestic Electrical Goods:

"4.17 The principal means of distribution is via electrical retailers, with extended warranties being sold as a secondary purchase accompanying that of an electrical item. Effectively, therefore, it is difficult for other suppliers to compete because of the retailers' point of sale advantage. Additionally, since the consumer is primarily shopping for an electrical item and chooses a retailer with this purchase in mind, there is little competition between different retailers for the sale of extended warranties. We have found no evidence to suggest that electrical retailers, in their advertising and marketing activity, actively compete to sell extended warranties (or indeed extended warranties-plus-goods in combination).

…

12.33 The most significant element of the differentiation among different providers of EWs [extended warranties] is between those providers that sell EWs at the POS [point of sale] of DEGs [domestic electrical goods] - that is, high street DEG retailers, Internet and mail-order companies - and those that do not, that is, direct sales by insurance companies, most manufacturers' EWs, and sales through credit card and utility companies.

12.34 EW providers generally agree that selling an EW at the same time as the DEG is sold gives the firm a competitive advantage. The main explanations given by DEG retailers generally relate to the convenience factor involved for consumers. It was put to us that consumers value the opportunity to sort out all elements of the DEG purchase at the same time, including cover, and this makes EWs sold at POS a more attractive offer…".

The significance of the point of sale advantage is demonstrated by the Commission's figure that 77.1% by number and 81.1% by value (including free extended warranties, but excluding renewals) of extended warranties were provided at the point of sale.

10. The Commission also found at paragraph 2.336 that:

"…a scale monopoly situation exists in the supply of EWs by Dixons Group as agent of [ASL] and Allianz Cornhill. All three companies are also part of the complex monopoly situation and are persons in whose favour that situation exists."

Facts

(1) The Cornhill Period

11. The Appellant Group decided to offer extended warranties insured (or reinsured) by its own insurance company. On the advice of its insurance brokers, Willis Corroon plc ("Willis"), it was decided to form DISL in the Isle of Man. Since DISL was not authorised to write insurance in the UK the Appellant Group approached Cornhill to front the insurance by issuing the policy to the customer, retaining 5% of the risk, and reinsuring the remainder with DISL. Cornhill was at the time insuring extended warranties on electrical goods for other retailers, mainly the privatised electricity distribution companies. There was a conflict of evidence (and no evidence from Cornhill) about whether Cornhill retained 5% of the risk because it did not want to take more because of the risk involved, or whether this percentage was chosen because it was, as Mr Bawcutt said, the minimum conventional amount for a fronting insurer to take. Mr Budd said that it was the Appellant Group's intention that Cornhill should retain 10% of the risk. Mrs Douse said that Willis had a hard job persuading Cornhill to front an extended warranty programme of this size, and that 5% was enough for them as they already had a large portfolio of extended warranty business and were already the biggest player in that market. It may therefore be that Cornhill wanted to retain 5% of the risk because of size rather than concerns about the risk. As we did not have any evidence from Cornhill we make no finding of fact about the reason for Cornhill retaining 5% of the risk, or whether the Appellant Group wanted it to take more. There were suggestions made on behalf of the Appellants that Cornhill was attracted by the possibility of increasing its recognition by the public and selling other insurance products to the Appellant Group's customers but Cornhill had agreed with Dixons Group plc and Dixons Finance plc that the information disclosed to it would not be used for other purposes without the consent of those companies. Mr Carroll thought that Cornhill did market life assurance using the database derived from their insurance of customers of the Appellant Group before 1997. There are no references to these possible benefits in any of the documents and we do not make any finding of fact that this was a relevant consideration for Cornhill.
12. DISL was incorporated in the Isle of Man on 27 March 1986. The directors of DISL were Peter Drinkwater, a consultant to Willis, Martin Webster, formerly of Willis (managing director), David Campbell, a director of Willis in the Isle of Man, and Mr Budd, the company secretary of Appellant Group companies. Mrs Dorothy Winrow, Fellow of the Chartered Institute of Insurers, joined DISL in 1990 and became a director in 1991. She worked part-time and was not involved in the setting of premiums. Mrs Shimmin, who was mainly responsible for the investment side, became a director in 1994 or 1995. When the arrangements with Cornhill were first proposed, the Appellant Group said in a letter of 27 November 1985 to Cornhill that DISL would adopt a very conservative accounting policy, under which forecast underwriting profits would be held in reserve for three years, although from the accounts this does not seem to have been done, although there is no suggestion that its accounting policy was not conservative. The underwriting profit for each year was included in the profit and loss account and only part was transferred to a claims reserve. The notes to the accounts state that the claims reserve represented prospective underwriting losses on policies incepted but not on risk at the balance sheet date. Dividends were paid from 1988-89 onwards. DISL followed Cornhill's accounting treatment. The same letter agreed that DISL's capital of £5m would be paid up in cash. DISL is exempt from tax in the Isle of Man under the Income Tax (Exempt Insurance Companies) Act 1981. It was licensed to carry on reinsurance business by the Isle of Man regulator. The original application was on the basis of a 100% loss ratio.
13. In outline, in the Cornhill Period starting in May 1986 extended warranties were sold in stores operated by DSG through Coverplan Insurance Services Limited ("CIS") (then called Dixons Finance plc) as agent for Cornhill. For each class of products a gross and net contract price was fixed. The gross contract price was the price charged to the customer for the policy. Out of the net contract price was paid an administration fee to MSDL and the balance was paid as premium to Cornhill. CIS was entitled to the difference between the gross and net prices as its sales commission. Cornhill reinsured 95% with DISL. When the customer made a claim Cornhill paid MSDL for repairs, collecting 95% of the claim from DISL. By 1997 Cornhill was using 380 product groups for the purpose of calculating premiums.
14. In practice the arrangements operated by Cornhill originally setting the price charged to customers for warranties (and hence the premiums taking into account the commission to CIS which was negotiated with the Appellant Group) by reference to burning cost plus underwriting profit. Mrs Winrow thought that they did not do this yearly but built on the existing structure. This suggests that Cornhill were content with the loss ratios and did not need to consider the premiums in detail. Premiums increased by 5% in 1993-94, and by 5% in 1996-97.

Contracts

15. The contractual position between Cornhill and DISL was contained in a series of agreements made on 25 July 1986 as follows:

(1) Agreement between Cornhill and DISL setting out the terms under which DISL reinsured Cornhill's risk. DISL was required to maintain a solvency margin of at least 20% of the greater of gross premiums written or gross insurance liabilities. DISL was not permitted to carry on other business. (The Isle of Man regulator required a minimum capital of £100,000, so the contractual margin was more onerous.). Illustrative figures of the solvency margin calculation are given in paragraph 47 below.

(2) Treaty of reinsurance effective from 31 May 1986 under which DISL reinsured 95% of Cornhill's risk for 95% of premiums received. Cornhill was entitled to a reinsurance commission ("ceding commission") to be retained by it out of the 95% of premium payable to DISL of initially 4% (premiums up to £25m in any 12 month period), 3% (premiums between £25m and £50m) or 2% (premiums above £50m). Cornhill was also entitled to an administrative fee of £1.85 per policy. The Treaty was terminable on six months' notice.

(3) Trust deed under which all premiums received by DISL were to be paid to the trustee and were to be used to satisfy DISL's obligations under the Treaty of Reinsurance. DISL was entitled to withdraw amounts in excess of 105% of its liabilities under the Treaty as certified by an actuary. Actuarial reports were obtained annually from Watson Wyatt in this connection.

(4) Agreement between Cornhill, CIS and other Appellant Group companies under which CIS agreed to act as agent for Cornhill for the sale of insured warranties. The agreement set out in a schedule the difference between the gross premiums (paid by the customer) and the net premiums (paid to Cornhill), the difference being paid to CIS as its sales commission. The gross and net premiums were to be agreed monthly between Cornhill and CIS and were to be set for each product so far as possible so as to make an underwriting profit for the product without subsidy from other products. Premiums were to be paid to Cornhill weekly within 28 days of the end of each trading week. The agreement could be cancelled on six months' notice.

(5) These agreements were entered into on the same day and were intended to operate as a single arrangement.

16. The following changes were made to the financial arrangements. These were negotiated between the Appellant Group and Cornhill rather than by DISL:

(1) From 1 May 1987 to 30 April 1988 Cornhill's ceding commission rates were reduced to between 2% (premiums above £50m) and 2.5% (premiums up to £50m). From 1 May 1988 to 30 April 1989 they were reduced to 2% (premiums above £50m) and 2.25% (premiums up to £50m). From 1 May 1989 the rate was further reduced to 2%. From 1 May 1990 it was again reduced to 1.5%, and the agreement instead of being terminable on 6 months' notice was continued until 30 April 1993. The 1.5% ceding commission is based on the 95% of the premium going to DISL, which is 1.425% of the whole premium. This results in DISL receiving 93.575% of the premium for taking 95% of the risk, and Cornhill receiving 5% of the premiums for taking 5% of the risk plus an additional 1.425% of the premiums as ceding commission (we consider below whether this should be looked at as 6.425% of the premiums for 5% of the risk).

(2) At the same time (1990) a retrospective profit commission became payable from Cornhill to CIS (but not by DISL to Cornhill) from scheme years 1986-87 to 1989-90, in respect of profits accounted for in 1990-91 and subsequent years, of the whole of the underwriting profit (based on its retained share of 5% of earned premiums less its 5% of claims) in excess of 20% of earned premiums, and half of the underwriting profit between 10% and 20% of earned premiums. Assuming, as was always the case, that the loss ratio was less than 80%, this is the same as the whole of the profit over 15% of earned premiums (ie premiums allocated to the accounting year) and we shall in future refer only to 15% in this and other similar provisions (although at the hearing a loss ratio of up to 85% was referred to we believe that this should be 80%, being the figure such that 15% of premiums are retained^[1]). The calculation was based on 5% of earned premiums and calculated separately for each accounting year with the profit commission payable three months after the end of the accounting year. Payment was made in respect of accounting periods in which premiums were allocated arising from sales made up to 1989-90.

(3) At the same time, from 1990-91 a profit commission payable by DISL to Cornhill was introduced of the underwriting profit (defined as 93.575% of premiums, less 95% of claims) in excess of 15% of 93.575% of premiums. Similarly a profit commission payable by Cornhill to CIS was introduced of the excess of the underwriting profit (defined as 98.575% of the premiums paid to Cornhill, ie DISL's share of 93.575% plus Cornhill's 5%, over total claims) in excess of 15% of 98.575% of premiums. The effect was that Cornhill received DISL's profit commission (based on 93.575% of premiums and 95% of claims) and passed it on together with its share (based on 5% of premiums and 5% of claims). As before, both were expressed as the whole of the profit over 20% of premiums and half of the profit between 10% and 20% of premiums, which we shall continue to refer to as the whole of the profit in excess of 15% of premiums on the basis that the loss ratio was below 80%, as it always was. Both of these profit commissions were payable within six months after the end of all policies written in respect of the scheme year i.e. 5½ years from the end of the scheme year, when the results were known. It should be noted that although extended warranties could be for 1 to 4 years following the end of the manufacturer's guarantee period this profit commission was paid 5½ years after the scheme year in respect of all warranties. The cause of this renegotiation was an unsolicited offer for the Appellant Group by Kingfisher in December 1989. As part of wanting to show improved profitability to fight off the bid the Group reviewed the underwriting experience (which was then only 3½ years old) and concluded that the claims would be lower than had been expected. Mr Carroll, who negotiated the changes on behalf of the Appellant Group, would have preferred a higher initial commission for CIS but Cornhill would not agree to this (no doubt because of the security offered by the larger amounts tied up in the trust fund where larger premiums were paid to DISL) and the profit commission was the compromise. Once the profit cap of 15% was introduced Mr Carroll said that it was no longer necessary for the Appellant Group to negotiate so hard with Cornhill over CIS's initial commission because any under-payment would be recovered as profit commission.

(4) Also from 1990-91 by a deed of 31 December 1993 the minimum assets kept in the trust was changed from 105% to 100% of liabilities but liabilities now included the profit commission. By letter of 23 December 1994 the liability for profit commission was to be taken to be £[**]m for 1995, £[**]m for 1996 and nil thereafter, so that the 100% came to exclude the profit commission. Reports from 1997 onwards were given on this basis (but reports relating to the ASL Period were on the basis of 120% of liabilities as required by those arrangements, again excluding profit commission). Although not dealt with by the witnesses, there is a note by Cornhill explaining that the removal of the profit commission was because of the effect it had on the Appellant Group's (including DISL) gearing ratio.

(5) By letter of 5 March 1993 Cornhill told DISL that it had agreed in principle to extend its arrangements with the Appellant Group until 30 April 1998 and asked if DISL were willing to make the same extension to the reinsurance treaty.

(6) By letter of 7 October 1993 from Cornhill and signed by Mr Carroll on 15 October 1993 on behalf of CIS (then Dixons Finance plc) which was the company responsible for the extended warranty business, Dixons Group plc (the parent company) and Dixons Stores Group Limited (later DSG Retail Limited), from 1993-94 the profit commission payable by Cornhill was increased to the underwriting profit (based on 98.575% of premiums less 100% of claims) in excess of 12% of 98.575% of premiums. No change was made to the profit commission payable by DISL to Cornhill, which continued at the excess of the underwriting profit (based on 93.575% of premiums less 95% of claims) in excess of 15% of 93.575% premiums. The effect of this change is set out in paragraph 17 below. It was agreed in correspondence on 23 December 1994 that profit commission was payable by Cornhill only when it had received DISL's share, and that if any new taxes were introduced that were not provided for in the agreement any increased costs would be dealt with so as to "maintain the relative profit participation of the parties."

(7) By the same letter, from 1993-94 an interest rate floor and cap was introduced under which had the effect that if the London Inter-Bank Bid Rate was less than 7% pa CIS made it up to that rate, and if it were more than 8% pa Cornhill had to pay the excess to CIS. The cash balances for this purpose included the ceding commission expressed as 6.51788% of 98.575% of premiums [equivalent to 5% plus the commission of 1.5% of 95% = 6.425% of the whole premium], less claims. The adjustments were made by adjusting the profit commission. It was also provided that the agreement between CIS and Cornhill could not be terminated before 30 April 1998.

(8) On 18 November 1993 a new Treaty of Reinsurance was entered into between Cornhill and DISL having effect from 1 May 1993 which was not terminable before 30 April 1998. The Treaty provides for the continuation of the profit commission payable by DISL to Cornhill based on the excess underwriting profit over 15% of 93.575% of premiums.

(9) When DSG Ireland Limited was formed in late 1996 as the sales company for Ireland that company operated in place of CIS in relation to such sales. Cornhill's agreement with DISL was amended to include these sales.

(10) The sales, administration and Irish agreements were brought to an end with effect from 3 May 1997 in connection with the making of the ASL agreements.

The effect on Cornhill of the 1993 changes

17. Assuming that the loss ratio is under 80%, during the profit commission arrangements operating from 1990 to 1993, on a total premium of 100, Cornhill will make an underwriting profit of 0.75 [0.15*5], and DISL an underwriting profit of 14.036 [0.15*93.575]. In 1993 this changed for Cornhill, leaving DISL's position unchanged, so that Cornhill now made an underwriting loss of 2.20725 [0.12*98.575 - 0.15*93.575], the calculation of which is explained following the table below, a situation that Mr Beaton found surprising as an underwriter will aim for an underwriting profit. However, Cornhill also receives ceding commission of 1.5% of 95 (=1.425) and interest of between 7% and 8% on the 6.425 less claims. The Revenue gave an illustrative figure for the amount of interest earned by Cornhill, which the Appellants did not dispute, based on a claims ratio of [a percentage, X%, which was less than 80% and in line with previous loss ratios] and an average interest rate of 7.5%, amounting to [**], which is dependent on estimates not only of the timing of claims but the mix of the length of policies. It is only this element of interest that depends on the amount of claims, the other items being constant (for a claims ratio under 80%), as explained below. In summary, Cornhill makes a pure underwriting loss of a fixed amount that is turned into a profit by interest at a minimum and maximum rate, the amount of which is affected by the amount and timing of claims. In addition it receives a fixed amount of ceding commission.
18. [The Commissioners set out a table containing the results for Cornhill and DISL for the two periods. The decision contains a second table on the alternative basis (as favoured by the Appellants) that the 1.5% Cornhill ceding commission is included as a payment for assuming its risk. The Commissioners explain that:

(1). The loss ratio they use, of X%, is the same ratio as that used by the Appellant Group in the figures in the Table described in paragraph 31 below. It does not affect the amount of the underwriting profit or loss for any loss ratio below 80% but will affect the amount of investment income.

(2). As explained earlier, the pure underwriting profit or loss figures are constant for loss ratios of under 80 because the effect of the claims ratio cancels out. Assuming a premium of 100 and total claims of C, Cornhill's underwriting profit is:

5 - 0.05*C - (98.575 - C - 0.12*98.575) + (93.575 - 0.95*C - 0.15*93.575)

The first bracket is the profit commission payable by Cornhill, and the second bracket the profit commission receivable from DISL. All the other items cancel out leaving 0.12*98.575 - 0.15*93.575 = (2.20725). DISL's profit is 0.15*93.575.]

19. The table shows the significant effect of the change made in 1993 to Cornhill's position while leaving DISL's unchanged. Based purely on the figures in the table, with a loss ratio of [X]% (which is relevant only to the calculation of the investment income), in the period 1990 to 1993 Cornhill earned proportionately 1.20 times more than DISL if the ceding commission is not taken into account or 1.76 times if it is (or 1.37 times if the margin is applied to the 6.425), which is the expected result in view of the ceding commission. After 1993 Cornhill's advantage is completely reversed and DISL earns proportionately 18.5 times more than Cornhill if the ceding commission is not taken into account, or 1.64 times if it is (or 2.11 times if the margin is applied to the 6.425). We make no finding on which party's approach is to be preferred as we consider it unnecessary for us to do so.
20. We are not convinced that the above shows a true comparison because it fails to take account of the different capital required by Cornhill and DISL in order to earn their respective margins. The following are our own views not based on any submissions from the parties. [The decision then contains a calculation of the capital which each of DISL and Cornhill were required to provide, and, on the basis of the returns calculated in the table, the Commissioners conclude that] DISL earned 10.64^[2] times Cornhill's return on capital. In principle we regard this as a more meaningful comparison but we emphasise that these figures are not agreed and the precise figures should not be taken to be findings of fact.
21. We find that these changes had the effect intended by the Appellant Group of squeezing Cornhill's profits and it was not expected by the Appellant Group that DISL would change the terms of its profit commission payable to Cornhill to reflect the increase in the profit commission payable by Cornhill to CIS. This is demonstrated by the fact that Cornhill had proposed to extend the Treaty of Reinsurance on the previous terms (see paragraph 16(5) above) before it had finalised its agreement with the Appellant Group. If Cornhill had wanted to preserve its negotiating position against DISL it would not have proposed the continuation of the current arrangements with DISL even before it was legally committed to the Appellant Group or would have negotiated a concurrent change in profit commission with DISL. The different margins earned by Cornhill and DISL indicate, so far as they can validly be compared, that a third party was prepared to make a lower margin then DISL, but at this stage we are stating the facts and not drawing any conclusions about whether Cornhill can be used as a comparable.

Accounting treatment and loss ratios

22. For the purposes of this appeal, in relation to insurance written in scheme year 1996-97, being the only year under appeal for the Cornhill Period, it is the arrangements that existed following these changes in 1993 that are relevant. By 1993 there had been experience of loss ratios for the seven scheme years starting with 1986-87, of which the first two years were closed years for which the results were finally known.
23. In DISL's accounts no premiums were treated as earned in the first year when the goods were covered by the manufacturer's guarantee; thereafter the premiums were treated as earned according to a profile based on each type of product and the period of cover. [By reference to figures taken from DISL's accounts for the periods ending 2 May 1987 to 26 April 1996 the Commissioners conclude that] the average loss ratio for those years as shown in the accounts was [very close to the X% used in the calculations above. The figures showed however a loss ratio for 1988 of 82%, but a fairly sharp decline thereafter, with subsequent ratios lying within 10% of the average.]. We comment in the next paragraph about the falling figure for loss ratios in the first few years
24. The loss ratio for any accounting period as shown in one year's accounts is a mixture based on several years' estimates. For example, it will comprise estimates for claims on contracts sold in the year and in the previous five years. One can trace the changing estimate of loss ratios for a single year's sales (a scheme year) in Watson Wyatt's actuarial reports for each year. The reports we saw started at the time of the ASL Period. [The Commissioners then set out a table drawn from those reports for the years to March 1997, 1998 and 1999, showing for those years the loss ratios shown for the results of the underwriting years 1991-92 to 1996-97. The Table showed that], for example, for 1996-97 there was a loss ratio for the current year, which had not then ended, even though the manufacturer's warranty applied for all sales during the year and the only claims can have been for matters not covered by that warranty. This means that the first year's ratio is entirely (or almost entirely) a forecast. The Watson Wyatt reports show that the claims frequency for the previous year is used for the latest year. The reduction in loss ratios as time progresses seen in the table (and the first few years' accounts in the table in the previous paragraph) is typical, and Mr Beaton described it as what one hopes would happen. We understand that these ratios take account of the premiums received in the year with the amount and timing of claims being estimated, and the cost of claims being that in the previous year adjusted for inflation, until one reaches the final year when all figures are actual ones. The early years' higher loss ratios are therefore a reflection of a conservative estimate of the amount and cost of claims, which time shows to have been too conservative, and do not reflect an actual change in loss ratios in the period.
25. These overall loss ratios for a year comprise a considerable spread of ratios for different products. Taking 1991-2 as an example, as at 6 March 1997, when the figures were final, while the overall ratio was [close to X]% the individual ratios varied between 5% (games consoles for 2 years following the manufacturer's guarantee, to which we shall refer as 1+2) and 285% (frost-free fridge-freezers 1+4). But the majority were much closer to the mean than these outlying examples.
26. We saw earlier figures in DISL's records at 18 December 1996 which showed a similar pattern: [The Commissioners then set out a table showing combined loss ratios at fixed months' intervals after each year end which showed for the years 1991 to 1995 a pattern of loss ratios for each year diminishing as the length of the period after the end of the relevant year increased]. We understand that DISL calculated these by comparing the earned premium with actual claims (plus an estimate of claims incurred but not yet reported) ie the opposite method from that adopted by Watson Wyatt. We saw a graph of actual claims paid against net premiums which for each year rose approximately in a straight line from about 10% (7 months after year end) to close to X% for 1991 at 49 periods after year end and corresponding figures for later sales years, for example 1992 had reached just under [**]% after 46 periods.
27. The above table shows the latest information available to DISL around the time that the change to the ASL Period was being considered. DISL would also have had available similar information relating to earlier scheme years, which we have not seen.

(2) The ASL period

28. As has been mentioned above, the factual situation changed from 13 April 1997 in consequence of the increase in levels of IPT to 17.5% from 1 April 1997 that had been announced on 26 November 1996. This increase in IPT rates had serious implications for the Appellant Group, reducing the total premiums by 15% (17.5/117.5). Mr Carroll estimated that the annual cost would be about £[**]m but if they changed to service contracts that were subject to VAT rather than IPT the ability to recover input VAT would reduce that [cost to about a third of that sum.]. The Appellant Group was therefore under great time pressure to bring the new system into operation. Originally it was hoped that Cornhill would continue to be involved as owner of the company providing service contracts, with DISL continuing to reinsure 95% of the risk, but Cornhill decided in early February 1997 that neither it nor its holding company was willing to be involved in non-insurance business, which was confirmed in a letter of 14 February 1997. Next DSG's insurance brokers, Willis, were asked if they would form a wholly-owned subsidiary to issue service contracts taking 3% of the risk and insuring the balance with DISL (a similar profit commission arrangement as had operated with Cornhill was proposed by Mr Carroll on 25 February 1997 with DISL's profit cap being 15% and Willis's 13.25%, with Willis receiving an insurance commission of 0.7%, which would have had the effect of eliminating the 3% risk). It was assumed by all parties that if Willis were involved DISL would be the insurer. Willis were interested in principle, but they declined as they did not want to take any risk as it was not their core business and also that they did not want to have to consolidate the business in their accounts. Mr Carroll considered that Willis did not want to take any risk because of the uncertain nature of the risk but we consider that this is unlikely to have been the reason. A 50/50 company between Willis and another shareholder was also considered and rejected by Willis on 18 March 1997 on the basis that they did not want to retain any risk. The Appellant Group also approached other possible providers of service contracts but without success.
29. By 2 April 1997 it was agreed with Willis that an Isle of Man company, ASL, would be formed, owned as to 40 per cent by Willis and as to the remainder by two individuals found by Willis, Mr Malcolm Byrne (a former director of Powerhouse, an electrical retailer who had run an extended warranty business at Eastern Electricity) and Mr Peter Samuel, a lawyer, whom none of the witnesses from the Appellant group or Mrs Douse of Willis seemed to have met (and he was referred to as Michael Samuel by Mr Carroll). The chief executive was Mr Merveyn Miller who was well known to Willis and was a director of other Isle of Man insurance companies. ASL's share capital was £100,000. It had tax exempt status in the Isle of Man, which was necessary to the maintenance of DISL's exempt status there; this explains the use of an Isle of Man company which was liable to corporation tax in the UK, although Willis gave an shareholder's undertaking on 7 May 1997 that it would be tax resident in the Isle of Man. An incentive fee of £250K was agreed with Willis if they completed the new arrangements by 23 April 1997 (with a reduction of £10,000 for every day's delay, including weekends).
30. While we accept that there were commercial advantages in presenting ASL as an independent company to customers, just as there had been with Cornhill fronting the insurance in the Cornhill Period, in that the effect of the trust arrangements—which were enforceable by ASL—was that the customer was not exposed solely to the Appellant Group's performance and credit risk, another reason for the introduction of ASL was presumably to avoid the CFC legislation applying to DISL. Neither the trust arrangement nor ASL would have been needed if the provider of the warranties was a substantial independent entity. There were suggestions by some, but not all, witnesses that the Appellant Group would not have been able to find alternative insurers to DISL (or reinsurance for DISL) in the market. There was no evidence that the Appellant Group ever tested the market in this respect, and so we decline to make a finding of fact that the insurance market could not have taken DISL's place. It was the Appellant Group's intention that DISL should continue with its existing role but now as insurer rather than reinsurer. This is demonstrated by the fact that a note of a meeting on 17 January 1997 (which was when it was thought that Cornhill would issue the service contracts) identified the need to change DISL's licence to that of insurance. It was ascertained by 13 February 1997 that this was not a problem and so long as an application was made by 20 March 1997 it could start business by 1 April 1997. The licence was duly obtained and we were shown regulations requiring a solvency margin that assets exceeded liabilities by 15% of the previous financial year's premiums receivable (if greater than a margin of £150,000). (There was discussion in connection with Mr Bawcutt's evidence of how one calculated the capital required when policies are for a period of up to five years. It seems that one takes the total premiums shown in the previous year's accounts, which will relate to policies sold in that period, and the length of the policy is relevant only to the calculation of total liabilities, which will be referable to premiums for several periods: illustrative figures are given in paragraph 47 below.) A proposal made by Mr Carroll to the Finance Committee of the Board on 25 February 1997, at the time when a 50/50 Willis company was proposed, stated that it enabled DISL's tax exemption worth £[**]m a year to be continued.
31. A schedule prepared by Mr Carroll on 23 April 1997 and sent to Mrs Winrow shows the effect of the change from IPT to VAT as follows:

	Insurance	Service agreement
Sale price	100.00	100.00
IPT	(2.44)
VAT		(14.89)
CIS commission	([A])	([a])
Net to Cornhill	[B=100-2.44-A]
Net contract price		[b=100-14.89-a]
Administration fee	([C])	([C])
Premium to DISL	([D])	([d])
Retained by Cornhill	B-C-D
Retained by service co		b-C-d

All the figures in the service agreement column up to the last line are consequential on the tax change. For example, the CIS commission is virtually the same percentage of the after-tax sale price ,and so it was assumed that that percentage would be constant; this determined the premium paid to DISL. Against the last item in the table Mrs Winrow had written ".688% of net contract price after admin fee." It is close to the 0.68% figure negotiated shortly afterwards with Willis (see next paragraph). Mr Carroll was not only adjusting the previous figures for the change from IPT to VAT but was also effectively determining the premium paid to DISL (or at least, the maximum premium that could be paid) as the residual figure after the payment to ASL. This also demonstrated by the fact that a later part of the schedule is as follows (the first two columns of figures are under the same headings as above):

Cornhill/service company profit and loss account

Net contract price	[B]	[b]
Administration fees	(3.90)	(3.90)
Net contract price less admin	[E=B-3.90]	[e=b-3.90]
DISL premium	[D]	[d]
Retained income	[E-D]	[e-d]		-[*]%
Claims costs	([F])			-100%
Profit commission
Receivable	[G]	[g]		-5%
Payable	([H])	([g])		-18%
Net operating profit	LOSS(E-D-F+G_H)	[e-g]		-175%

DISL profit and loss account

Premium	[D]	[d]		[]
Claims2	([K])	([k])		[]
Profit commission payable	([G])	([g])		[]
Net operating profit	[D-K-G]	[d-k-g]		-8%

Dixons consolidated profit

DISL	[D-K-G]	[d-k-g]	[]	[]
CIS	[A+H]	[a+g]	[]	[]
Admin	3.90	3.90	(0.00)	0%
Total	[total]	[total]	[]	[]
Dixons	[as above]	[as above]
Cornhill/service company	[loss as above]	[as above]
Total	[total]	[total]

Cornhill's loss is after taking its ceding commission into account but before investment income, which is constructed differently from the table in paragraph 18 above. If, consistently with the table in paragraph 18 above, the above table had started with the premium excluding ceding commission (2.69) the loss would have been 1.18 which is in the same ratio as the loss of 2.207 on a premium of 5 in the table in paragraph 18 above, thus showing that the basis of our calculations and those prepared contemporaneously are the same.

32. Willis, on behalf of ASL, negotiated a fee of £[F] pa for ASL. We saw a spreadsheet prepared by Mr Carroll on 3 April 1997 which starts with a sales figure for extended warranties of £[*]m which resulted in a net contract price of £[**]m, which was presumably based on an estimate for the current year. From this was deducted the ASL fee of £[F+2K] leaving an insurance premium to DISL of £[**]m, of which 0.68% is £[just under F]. A summary section shows £[F+2K] receipt at the time of sale plus £2K after the end of run-off, making a total profit of £[F+4K]. We deduce from this that the difference between the gross and net contract prices was arranged to be the figure such that when £[F]K was deducted from it, this amounted to 0.68% of the resulting figure (being the insurance premium to DISL). The spreadsheet also notes that target net sales proceeds (ie the insurance premium) was £[**]m and the target net operating result £[F]. We infer that it was agreed that ASL would receive 0.68% of the insurance premium paid to DISL, although this figure was not included in any of the agreements or even in correspondence.
33. A further initial payment was made to ASL by increasing the net contract prices by £10 for the first 41,000 contracts to produce £410K (made up by the £250K incentive fee (see paragraph 29 above) plus a £160K contribution to expenses). Although the documents were signed on 6 May 1997 and operated from 11 May 1997 the incentive fee was in fact paid in full. Willis also received a finder's fee for introducing the business to ASL the terms of which were varied from 6 May 1998 in correspondence on 11 January 1999. This was based on ASL's turnover and was on a sliding scale with the rate increasing with the turnover. In ASL's accounts to 31 December 1998 a finder's fee of £[**] was paid to Willis and a further £[**] accrued, making a total of £[**]; in the 1999 accounts the corresponding figures were £[**] and £[**]; in the 2000 accounts £[**] and £[**]; and there are figures in the 2001 budget.
34. On 28 November 2000 Mr Budd wrote a memorandum to the Group Executive Board pointing out that ASL would earn about £[E]m in the current year, and saying that he had renegotiated a reduction of ASL's income from 1 May 2000 to £[65% of E] pa in return for an extension of the agreement until 30 April 2006. This was effected by changing the percentage from 0.68% to 0.51%. Again this was not recorded in any agreement. Indeed, Mr Budd's contemporary manuscript note said "Nothing in writing."
35. There was disagreement between the parties about whether ASL had substance or was making it look as if it had substance by, for example, conducting correspondence and faxes with DISL even though for the first three years they occupied the same floor of the same building. We saw a memorandum of 24 April 1997 drafted by Mr Sidders (then head of tax at the Appellant Group) setting out exactly what should be discussed at meetings and how regularly they should take place. A meeting of 12 August 1998 attended by representatives of ASL and the Appellant Group recorded "Format of meetings to be client/supplier relationship. Expect some aggro and have own written Agenda and shopping list!"; "ASL team to be more pro-active and stretch DSG teams. ASL to be more demanding and testing,"; "ASL will need to be firm with requirements of Agreements"; and "No evidence is currently being shown that Dixons are being given a hard time…". Later documents showed an approach by DISL which was consistent with those requests. There is therefore some evidence that ASL's activity was contrived, as is also suggested by its effectively receiving a fixed fee which could be increased only by increased sales of warranties. We do not consider it necessary to make any findings on this particular issue. As far as we are concerned ASL exists as a company in its own right in the sense that there was no evidence that ASL acted merely as agent or trustee for any member of the Appellant Group or that its actions or its assets or liabilities were otherwise those of any person other than itself (if the Revenue did not think it was we would presumably not be hearing a transfer pricing case but a CFC case) but its importance is demonstrated by its income of 0.68% (later 0.51%) of the premium paid to DISL, designed initially to equate to a fixed income of £[E] (later reduced to £[65% of E]). ASL's willingness to accede to requests from the Appellant Group to act in certain ways might however be indicative of the continuance of an arrangement under which ASL complied with a desire on behalf of the Appellant Group to ensure that DISL was the insurer of the extended warranties on satisfactory terms.
36. We do, however, accept Mr Carroll's and Mr Budd's evidence that the Appellant Group genuinely wanted an outside provider so that it could tell customers, the Office of Fair Trading ("OFT") (which had expressed concern about the move to service contracts as being less secure for customers), the British Retail Consortium (whose Code of Conduct Mr Budd was involved in drafting) and commentators who were writing critical articles about extended warranties, that the benefit of the warranties would continue even if the Group failed. The Appellant Group had to give annual reports to the British Retail Consortium on consumer protection, which were passed to the OFT. While the strength of Cornhill was very different from that of ASL, a company with a capital of £100,000 receiving net payments for service contracts of some £[**]m pa (figure for the year to 31 December 1998) after commission, there was still advantage in claiming the existence of this separation even though this was optical rather than real as the customer's protection was through the trust arrangements even allowing for the fact that it was ASL rather than the customer who was the direct beneficiary of the trust.

Operation of the ASL arrangements

37. During the ASL Period, warranties were sold in stores operated by DSG through Masterplan Service Agreements Limited ("MCSAL") (or DSG from 2001) as agent for ASL. As before, for each class of products a gross and net contract price was fixed. The gross contract price was the price charged to the customer for the policy. MCSAL (or DSG) was entitled to the difference between these prices as its commission. The [Commissioners then set out the MCSAL (or DSG) total commission as average percentages of the gross contract price in the years ended 30 April 1998 to 2003]. Out of the net contract price was paid an administration fee to MSDL, and the balance was paid to ASL which retained a percentage (initially 0.68%) of the premium paid to DISL, paid an administration fee to MSDL, and insured the whole of its liabilities with DISL and took no risk itself (other than on DISL's ability to pay, which was secured by means of the trust arrangements). When the customer made a claim ASL paid MSDL for repairs claiming on DISL. MSDL administered the service contracts including evaluating claims, providing customer support, arranging repairs or replacements and monitored repair costs. The trust arrangements enabled payment from the trust fund directly to MSDL.
38. The ASL arrangements started on 11 May 1997, although contracts were sold by MCSAL from 13 April 1997 and were put into the new arrangements when they started, and so economically the ASL Period started on 13 April 1997.

Contracts

39. The outline of the contractual position was as follows:

(1) Sales agreement of 6 May 1997 between ASL, MCSAL and DSG, under which MCSAL was appointed ASL's agent for selling service contracts.

(a) The commencement day is 11 May 1997. The contract continued unless either party gave six months' notice to expire on or after 30 April 2002. This was subsequently extended and at the same time MCSAL's commission was increased by about £1.5m pa.

(b) ASL was responsible for setting gross and net contract prices. The initial prices for each product were set out in a schedule to the agreement and could be changed on one month's notice but not on more than three occasions a year. Changes could be made only after consultation with MCSAL which had "the right to approve such changes, such approval not to be unreasonably withheld or delayed"; we interpret this as requiring the consent of MCSAL to a change subject to the requirement that such consent should not be unreasonably withheld or delayed. If the gross contract price is reduced the net had to be reduced by the same monetary amount; if the gross is increased the net could be increased only with MCSAL's approval. We were shown some anomalies in the schedule of gross and net contract prices, for example that the net price (£51.86) was larger then the gross price (£38.72) for printers 3+1 costing less than £250. The agreement provided that in such case the higher net price still had to be paid by MCSAL to ASL. These prices were discussed at meetings of a joint Marketing Committee.

(c) Further commission was payable to MCSAL on receipt of profit commission from DISL (see paragraph 39(3)(b) below) of 84.42%*[net contract prices + ASL's margin] - claims under the insurance contract, equals 84.994% of [premiums - claims]. This is roughly the same as the payment by DISL of 85%*[premiums - claims], but the rounding of 84.42% gives a small margin retained by ASL. The effect is that ASL retained its margin of 0.68% in full and did not pay 85% of it in profit commission. ASL also agreed to pay a volume bonus to MCSAL if net contract prices up to 30 April 2002 exceeded £[**]m.

(d) From 29 April 2001 DSG Retail Limited acquired the business of MCSAL and replaced MCSAL as the sales agent. References below to MCSAL should be taken to include references to DSG from that date.

(2) Administration and repair agreement of 6 May 1997 between ASL and MSDL which provided for repairs to be carried out by MSDL and for MSDL to administer the service contracts on behalf of ASL. MSDL was entitled to an administration fee of £1.56 plus £0.78 (both plus VAT) per contract sold. MSDL agreed to fulfil ASL's obligations under the service agreements, maintain administration documents, provide management information, respond to customers' enquiries in accordance with performance standards, and provide a customer complaints handling service in relation to complaints made against ASL.

(3) By an insurance contract of 6 May 1997 operating from 11 May 1997 DISL insured the whole of ASL's liabilities to holders of service contracts.

(a) The premium payable was calculated as the number of products in a particular category times a rating factor for that category and duration times the number of years of cover times the sum insured as set out initially in the schedule. The rating factors were initially based on the premiums charged by Cornhill in the Cornhill Period. Rating factors were agreed for new products. Mrs Winrow said that these were intended to simplify the rating process as they did not need to set a separate net premium for every individual product. Originally there were 56 product categories compared to the 380 categories at the end of the Cornhill Period; this had increased to 240 by 2004.

(b) A profit commission of the excess underwriting profit over 15% of premiums received by it (if the loss ratio was less than 80%) was payable by DISL as before with payment within six months after the end of five years.

(c) The contract was terminable on 6 months' notice to expire on or after 30 April 2002. This was subsequently extended to 31 May 2006.

(4) By a trust deed of 6 May 1997 between DISL, Coutts (Isle of Man) Limited and ASL, DISL was to pay all moneys received to the trustee to be held on trust (a) to pay the income to DISL, (b) to discharge all liabilities of DISL to ASL, (c) to reimburse DISL for payments made to ASL otherwise than out of the trust fund, and (d) for DISL.

(a) DISL was entitled to take money out of the trust fund on an actuary's certificate to the extent that it exceeded 120% of its liabilities to ASL (excluding the profit commission). The Watson Wyatt reports to which we have referred above were prepared in connection with such withdrawals. If the assets were less than 120% of liabilities DISL had to pay to the trustee the lesser of the shortfall and the amount of income that had accrued at the date of the certificate. (Mr Peacock reads this as including income previously paid to DISL but we consider that it may mean income that has accrued and has not been paid out to DISL on the basis that income can be paid out only within 120 days of a certificate and so income paid to the trustee after that must stay in the trust until the next certificate; the point was not argued and it is not necessary to resolve this.)

(b) Royal Bank of Canada (IOM) Limited was appointed trustee on 25 February 1999 and RBC Trustees (Guernsey) Limited on 3 May 2001.

(5) A further trust deed related to the keeping by MSDL of a trust account for payments of VAT. The trust fund was held on trust (a) to pay the VAT for which ASL was liable, (b) to pay to MSDL the VAT on payments to it under the administration and repair agreement, (c) to pay to MCSAL VAT on the profit commission to which it was entitled, and (d) the balance for ASL.

(6) These contracts were entered into on the same day and were intended to operate together.

(7) The agreement with Cornhill (see paragraph 15(1) above) was still in force and so Cornhill's consent was required to DISL taking on this new business in order to protect Cornhill in relation to claims under policies written by it in the past, liability under which continued until 2002, and reinsured with DISL which were still in force. A Cornhill document of March 1997 noted that DISL's liability to Cornhill at April 1996 was £[**]m of which £[**]m was estimated profit commission. Net contract prices were initially set by ASL at the same rate as when Cornhill were the insurer so as to demonstrate to Cornhill that the business was not less favourable to DISL than it had been. Cornhill allowed its historical data to be used by ASL for this purpose. The solvency margin required by Cornhill was reduced from 20% to 16.25% in October 2000.

Continued Cornhill business during the ASL Period

40. In addition to the ASL business insured by DISL, and the run-off of its reinsurance entered into during the Cornhill Period, DISL continued to reinsure 95% of Cornhill's risk in the following businesses:

(1) As mentioned above, Cornhill continued the existing arrangements relating to the full range of products with DSG Ireland Limited from 3 May 1997 but on different terms from that date. As before, 95% was reinsured with DISL. The ceding commission retained by Cornhill was increased to 1.75% and no profit commission was payable either by DISL or Cornhill. The agreement dated 25 June 1998 (applying from 3 May 1997) provided that DSG Ireland Limited could require the net contract price (there defined as net premium) to be reduced so that the premiums did not exceed 125% of expected claims plus administration fee expenses (ie a loss ratio of 80%), and Cornhill could increase the net contract price to achieve that ratio if it would otherwise have been lower. The commission retained by DSG Ireland Limited (the difference between the gross and net contract prices) went up from 2000

(2) The Link business. This is dealt with in paragraph 112 below.

(3) In addition, during the ASL Period Cornhill continued to provide extended warranties on mobile phones, 12 months theft cover on products covered by the service contracts and also cover against food spoilage. ASL did not want to include these in the service contracts in case they were considered to be insurance.

(4) The existing reinsurance treaty with DISL was extended first to 30 April 1998, then to 31 May 2002, and to 30 April 2007.

41. During the period from 1997 to 2004 the split of total DISL premium receipts was 20% from Cornhill and 80% from ASL. The split of underwriting profit was 40% and 60% respectively (this takes into account earned premiums and profit commission, and so some of the earned premiums and profit from Cornhill business up to 2002 relates to the Cornhill Period, which may not reflect a meaningful comparison). The Revenue did not seek to make any transfer pricing adjustments relating to any of the continuing Cornhill business.

Premium rate setting by ASL

42. In the first year the price charged by ASL and the premiums paid to DISL after ASL's commission were set to correspond with the premiums charged by Cornhill in the previous year although Mr Bawcutt pointed out that these were not adjusted for the absence of the ceding commission or the reduced cover of accidental damage and theft that continued to be insured by Cornhill. Although Mrs Douse thought that premiums were benchmarked in the market she was not personally concerned with this, and Mr Heanage said that Willis did not have a dedicated extended warranty broking team; in the absence of other evidence we do not accept that this was done. In any event, she doubted whether the market could have absorbed this volume of business and the market was not tested. In setting premiums Mrs Douse said that DISL did not take investment income into account and Cornhill had not done so in the Cornhill Period either. She considered that this was how the insurance market worked and was influenced by insurance companies accounting for underwriting separately. We interpret this to mean that insurers try to make an underwriting profit before taking investment income into account but they must be aware of the effect of investment income. So far as Cornhill was concerned it was bound to make an underwriting loss before interest on any loss ratio up to 80% (see paragraph 17 above) and so investment income must have played some part in premium setting because they must have been budgeting for an overall profit. A much later letter from Cornhill dated 12 October 2000 said "the original deal for the extended warranty account was structured between Cornhill and DFS and referred to a fairly complex profit share. The agreement was struck on the basis that Cornhill would be accruing significant investment income on the unearned premium reserves we were keeping on our net line." We do not therefore accept that investment income played no part in premium setting by Cornhill. In general, Mrs Douse and Mr Beaton were agreed that an underwriter would try not to take investment income into account, but Mr Bawcutt said that brokers would argue that it should be taken into account particularly where the premium was held for some time before claims and would have achieved a reduction in premiums. It was put to him that he had not taken into account the trust arrangements, the likely amount of investment income and the competitive market in which DISL operated, but we find that by continuing the premiums set by Cornhill DISL was budgeting to receive a similar level of investment income in completely different circumstances from Cornhill's where DISL was not relying on the investment income to cancel an underwriting loss, as Cornhill did.
43. There was never any renegotiation of premium rates although adjustments were made in some areas. In the 1998-99 scheme year these premium ratings were maintained with a small variation in a few items with a loss ratio of over 100%, which would change the total premium by £176 assuming the mix of products remained constant. New rating factors were agreed for new products, or new products were slotted into an existing appropriate rating factor.
44. In April 1999 Watson Wyatt were asked to report on the premium rate structure as DISL no longer had input from Cornhill. Their report of 14 July 1999 lists as one of the constraints that had been discussed with DISL that the premium should be within [P]% to [P+1]% of the gross price charged to the customer, which they said equated to a projected loss ratio for the 1999-00 scheme year of [close to X%]. This [P]% figure is also referred to in a meeting between ASL and DISL of 26 May 1999 as the figure to which DISL's premium was intended to return in the light of the reduction in retail prices of electrical goods, and a matter which Watson Wyatt would consider. It is a similar figure to the [DISL premium percentage] that was equivalent to the previous figures when adjusted for the change from IPT to VAT shown in the table in paragraph 31 above. Mrs Winrow said that this was a coincidence but we find that the parties were aiming to keep close to the figures in that table, which is not a commercial way of setting premiums. The report dealt with the position if the premium was the same as the gross contract price (although it made clear that it was not intended that this should be done, but this demonstrated what the theoretical maximum premium would be); and also what the premiums would be if the constraints were fulfilled in all cases (which it was also not suggested should be done in practice, but this demonstrated the cases in which it was not possible to reduce loss ratios below [X+1]% because otherwise the premium for the products concerned would be over 110% of the price paid by the customer). The report suggested changes to the premium structure that were considered most effective in terms of meeting the stated criteria (which resulted in a projected claims ratio of [X+3.4]%), and a number of additional such changes (which resulted in an initial projected claims ratio of [X+3.3]%). In fact Mrs Winrow said that only rates for televisions were adjusted in the light of the report. The report suggests that DISL had imposed its own restraints, such as DISL's premium being [P]% of the gross contract price and a loss ratio of [X+1]%, rather than that it was obtaining advice about fixing premiums according to market criteria. It seems that the figure for DISL's premiums as [P]% of the gross contract price was not in fact achieved. [Substituting the actual commission percentages paid to MCSAL (or DSG from 2001) (see paragraph 53 below) in the table in paragraph 31 above the Commissioners set out DISL's premium as a percentage of the gross contract price in the years ended on 30 April 1998 to 2003.]. Mrs Winrow agreed that they had a target loss ratio, which the report shows was in the region of [X+1]% to [X+3]%.
45. The premium rating factors were negotiated during the year to which they related and the premium adjusted retrospectively. For example, minutes of a meeting of 24 May 2000 between ASL and DISL noted that Mr Wilcock of ASL said that there were still unresolved issues relating to 1999-2000 (which ended on 30 April 2000) that prevented him from returning the signed schedules. The minutes of the 1 September 2000 meeting noted that the issues had now been resolved and Mrs Winrow requested that ASL return a copy of schedule 1 of rating factors dated 20 July 1999. It therefore seems that the schedule was not signed on 20 July 1999 but on about 1 September 2000, and we infer that the figures were not agreed until after the year to which they related when the sales and prices for the year were known. We did not see the schedule for that year and do not know what date it contained. Net premiums were increased by 1.25% for 1999-00 and as a result of changes to the rating factors total net premiums were forecast to fall by 10% but in the event the diminution was more than made up by increased sales, the amount of which was known by the time the rating factors were finalised, and so there was no real reduction in premiums. We also saw a schedule taking effect from 3 May 1998 dated 6 November 1998; and one taking effect from 19 May 2001 dated 21 November 2001.
46. An inflationary increase in premiums of 1.25% was negotiated for 2002-03 when the repair rates were increased by 2.5%.

Accounts and loss ratios

47. By the start of the ASL Period DISL had substantial retained profits and had shareholders' funds of £[**]m at 27 April 1996 and almost £[**] at 3 May 1997. The Isle of Man regulator's required solvency margin (when DISL's business was only reinsurance) was £100,000 and the auditors certified that it had assets [substantially] over this figure of £[**]m. We do not have the calculation for the following year when the regulator's solvency margin was changed now that DISL was carrying on insurance but the calculation in the auditors' certificate at 1 May 1999 shows the excess of assets over the solvency margin as 15% of premiums over shareholders' funds both as shown in the accounts to that date (rather than the premiums in the previous year's accounts which was in the regulations we were shown). A similar calculation at 3 May 1997 [would show a very substantial excess of shareholders' funds over 15% of premiums ]
48. DISL's accounts made provision for the profit commission once earned premiums less claims paid exceeded 15% of written premiums. Its underwriting reports, which were prepared by Mrs Winrow from data supplied by MSDL, showed, in addition to this, a figure for ultimate projected profit commission in the year starting with the current year. [The Commissioners took the example of DISL's underwriting report at 1 February 1997. The then current year's profit commission figure as a percentage of the total of the profit commission and the net underwriting profit, suggested] that at that stage it was probably expected that a high proportion of the total underwriting profit would be returned as profit commission: had there been an expectation that loss ratios would deteriorate we would have expected to see some separate provision (whilst from DISL's own perspective it did not matter whether it termed a provision "provision for profit commission" or "provision for extra claims," it was a subsidiary of Dixons which would be interested in the expected profit commission (which would be been returned to the Appellant Group through Cornhill); the lack of a separate provision therefore suggests that none was considered as necessary. The balance sheet figure in its audited accounts at 3 May 1997 was [several tens of millions of pounds]. Although was no evidence on this we infer that DISL considered that at least a substantial payment of profit commission was always likely.
49. [In order to record the evidence the Commissioners then set out information about DISL's accounts but none of this information (except possibly an estimate of the figures for the year ending 3 May 1997, which included a loss ratio very close to X%) would have been available when the ASL arrangements were being set up and so had not been taken into account in reaching their decision. The average loss ratio for this period as shown in the accounts was within 0.2% of X %. The average loss ratio for the whole period from 1987 to 2004 as shown in the accounts was within 0.4% of X%. These figures included all the businesses carried on by DISL, both insuring ASL and reinsuring Cornhill.]
50. The loss ratio for any accounting period as shown in DISL's accounts is a mixture based on several year's estimates and also of the different businesses carried on by DISL. For example, it will comprise estimates for losses on contracts sold in the year and in the previous five years, whether relating to insurance or reinsurance carried out by DISL. [The Commissioners then demonstrated how one could trace the changing loss ratios by looking at a single year's sales in Watson Wyatt's actuarial reports for each year. These showed the same progressive reductions in the ratios over time as in the Cornhill Period set out above.]
51. As before, the average ratios include a wide variation in ratios for individual products. [Example omitted]
52. These loss ratios were described as good by Mrs Winrow and extremely good by Mr Bawcutt. Mr Heneage said that he had never seen this level of profit in a commercial arrangement.

MCSAL's (or DSG's from 2001) initial and profit commission

53. [The Commissioners then set out a table showing MCSAL's (or DSG's from 29 April 2001) initial and profit commission as a percentage of the gross contract price.]

DISL's risk

54. We looked at the economic effect of the ASL arrangements. Suppose the total amount paid by ASL to DISL is 100. At the end of 5½ years the effect of the profit commission is that if claims are nil MCSAL receives 85, and if the claims are 1 MCSAL receives 84 and so on until claims are 90 when MCSAL receives nothing. Effectively the first 80 of claims are entirely for MCSAL's account. Only if claims exceed 80 does the cost move to DISL with DISL making an underwriting profit to the extent that claims are less than 100, and a loss if they exceed 100. One could say that the effect (leaving aside the 5½ years and the effect of the investment income) is the same as if ASL has "insured" the first 80 of claims with MCASL and taken out a stop-loss insurance with DISL if claims exceed 80. Or, perhaps more accurately since DISL has taken the whole risk to ASL (again leaving aside the 5½ years and the effect of the investment income), that DISL has taken out a "reinsurance" with MCASL for the first 80 of claims for a premium of 85, and has retained the remaining risk for the balance of the premium of 15.
55. The witnesses, particularly Mr Beaton and Mr Bawcutt, disagreed about the degree of risk to which DISL was exposed, and both counsel emphasised the point of view relevant to his case. While risk is highly relevant to transfer pricing we consider that the loss ratios achieved in the past, Watson Wyatt's actuarial reports, and DISL's projections of profit commission payable, are of much more assistance to us than the impressions of witnesses in quantifying the risk undertaken by DISL in relation to existing products; the impressions of the witnesses were more relevant in relation to (a) the expectation of changes in products, (b) what manufacturers might do in the face of widespread component problems, and (c) the possibilities of widespread failures. All insurance involves risk although in this type of insurance the maximum liability would be the replacement cost of the product (possibly more than once) in contrast to some classes of insurance where payment for, say, personal injury is not limited. Here the major unknown related to the reliability of new products, the number of which was particularly pronounced with electronic goods from the mid 1990s, and the risk of inflation in repair costs. These are potentially five year contracts during which the reliability of new goods is unknown, as are their repair costs and whether further developments in such goods will make them uneconomic to repair. Mr Gaysford pointed out that the manufacturer, at least in the short term, would deal with failures for which the manufacturer was responsible, by recalling and replacing products, which would mitigate the risk of all of a class of products failing. The price of electronic products was falling (and was expected to fall) during this period and so the point at which an item was beyond economic repair was reached sooner. New products tended to be marketed at first at premium prices because some purchasers were prepared to pay high prices for new products, following which the price would drop to as little as half over the next three to five years. On the other hand, Mr Gaysford and Mr Heneage pointed out that the pace of change could be expected to bring benefits. If a product was overtaken by a new one, for example the replacement of video recorders by DVD recorders, the price of the old product dropped quickly to virtually nil and so making replacements was cheaper. If the replacement cost is greater than the repair cost because of falling prices of products the customer can be given a replacement at lower cost, which is likely to suit the customer. The Watson Wyatt reports also comment on the advantages and disadvantages relating to the risks attached to new products, without clearly coming down on one side.
56. DISL must have been alive to these issues. We therefore prefer to base our findings on the degree of risk adopted by DISL in its audited accounts and in Watson Wyatt's actuarial reports which must reflect the degree of risk actually perceived by DISL. In determining the loss ratios for each product in the first year, these adopted a prudent approach that is borne out by the loss ratios in virtually all cases reducing each year until the end of the five years (see the tables [referred to] in paragraphs 24 and 26 above for the Cornhill Period). Looking at the position in April 1997 DISL had the ratios for all previous year's sales available to it (many would have been closed years and some in the course of the five years in various stages of increasing certainty) going back to 1986. Looking at the table most of the then known loss ratio percentages were [within 5% of X%, apart from] one, the latest year, [which was at X +10]% (which would have been X+3% excluding mobile phones, which were introduced as a new policy on 1 October 1996 for which there was little data and accordingly an additional contingency had been reserved). Although much was made of the "mono-line" nature of DISL's insurance the averaging effect of different loss ratios on different products in a range comprising a wide variety of types of domestic electrical goods had in the past always resulted in a fairly even ratio and so we give limited weight to this factor. The reports set out separately the risk attached to new benefits but the amounts of these are small in relation to the whole, although reaching 2.5% of claims at 7 March 1998. When giving their certificates Watson Wyatt added contingency margins of between 15% and 25% in quantifying their estimate of liabilities and the 20% margin required by the trust deed in the ASL Period was added on top of these. Mr Beaton agreed that the Reports were, as one would expect, given on a conservative basis. We consider that these figures show the perceived risk on a prudent basis that had continued since the beginning. It would have been DISL's expectation in 1997 that these loss ratios would not have been expected to be exceeded by a significant margin, particularly as the mixture of products involved made it unlikely that excessive claims would have arisen on many of these at the same time.
57. In determining the degree of risk it is also relevant that the group accounts, which were the accounts of interest to shareholders, recognised the entire profit from the insurance of extended warranties in the year of sale, of which the Competition Commission said in their Report on the supply of extended warranties on domestic electrical goods within the UK of December 2003 that the Appellant Group was the principal exception to the spreading treatment adopted by most retailers. In its separate accounts MCSAL and (from the period ended 30 April 2002) DSG recognised the profit on the profit commission over the life of the contract but starting after three years, so that, for example, the profit for scheme year 1997-98 was recognised in the MCSAL accounts starting in the year ended 30 April 2001. The underwriting profit must have therefore been considered sufficiently certain for an estimate satisfactory to the auditors to be made in the first year in the group accounts and in MCSAL's and DSG's separate accounts after the third year, which will have reflected only two years' claims experience, although many of the warranties will have been for 1+3 years or less. DISL's treatment of the profit commission in its audited accounts and underwriting reports (see paragraph 48 above) is also relevant. Finally, the Appellant Group's inclusion of profit commission in the projections for the financial effect of the change from insurance to service contracts (see paragraph 31 above) is to the same effect. We take account of this, in relation to accounts after 1993 or 1997, only to the extent that it gives some indication of understanding at the earlier (relevant) date
58. Support for this approach is given by the fact that Cornhill agreed to the 1993 changes to the profit commission arrangements. With a loss ratio of [X]% as shown above, if one sets the underwriting loss against the interest Cornhill makes a profit of [..] on a premium of 5, leaving it with the ceding commission of 1.425. We are reluctant to calculate the effect of changes in the loss ratio on the interest because it depends on assumptions both of when the claims are incurred and the mix of length of policies. But it is clear that if Cornhill shows only a very small profit of [**] ([**]%) on a premium of 5 any significant increase in the loss ratio will remove this profit (because the interest will be lower) and will mean that it no longer benefits from the whole of the ceding commission, which we infer that Cornhill would have wished to do as (at least the substantial part of) its return for providing its solvency margin. This indicates to us that Cornhill would have been prepared to enter into these arrangements only if they were fairly sure about the loss ratios not being likely to change significantly (Cornhill were not taking any significant risk in relation to DISL's ability to pay because of the trust arrangements).

Law

The Section 770 regime

59. For relevant accounting periods ending before 1 July 1999 the provisions of ss 770 to 773 TA 1988 applied and provided so far as relevant to these appeals:-

"Transactions between associated persons

770. Sales etc. at an undervalue or overvalue

(1) Subject to the provisions of this section and section 771, where any property is sold and -

(a) the buyer is a body of persons over whom the seller has control or the seller is a body of persons over whom the buyer has control or both the buyer and the seller are bodies of persons over whom the same person or persons has or have control; and

(b) the property is sold at a price ("the actual price") which is either -

(i) less than the price which it might have been expected to fetch if the parties to the transaction had been independent persons dealing at arm's length ("the arm's length price"), or

(ii) greater than the arm's length price,

then, in computing for tax purposes the income, profits or losses of the seller where the actual price was less than the arm's length price, and of the buyer where the actual price was greater than the arm's length price, the like consequences shall ensue as would have ensued if the property had been sold for the arm's length price."

(2) Subsection (1) above shall not apply -

.....

(d) in relation to any other sale unless the board so directs.

(3) Where a direction is given under subsection 2(d) above all such adjustments shall be made, whether by assessment, repayment of tax or otherwise, as are necessary to give effect to the direction."

Section 773 (4) provided:-

"(4) Sections 770.... shall, with the necessary adaptations, have effect in relation to lettings and hirings of property, grants and transfers of rights, interests or licences and the giving of business facilities of whatever kind as they have effect in relation to sales, and the references in those sections to sales, sellers, buyers and prices shall be deemed to be extended accordingly."

60. It was common ground that DSG, MCSAL and DISL were persons within s 770(1)(a). No express admission appears to be made by the Appellants that MSDL falls within that section, but the way in which Mr Goy put the Respondent's final case no allegation appears to be made that MSDL gave any facility to DISL to which the provisions should apply. No suggestion was made that Cornhill or ASL fell within the subsection.
61. It was not disputed that directions were made by the Respondents under s 770(2)(d).
62. It is clear to us that "giving business facilities of whatever kind" is intended to encompass a wide variety of things. As the Special Commissioners said in Ametalco [1996] STC SCD 399, these words are a wide description, and this approach was adopted in Waterloo v IRC [2002] STC SCD 95. It is not in our view limited to a contract between the described persons. But one must be able to identify a "facility", something which confers some form of benefit or advantage. That benefit or advantage need not be an immediate pecuniary advantage but could be the hope or expectation of profit or other advantage.
63. Thus in relation to s 770 we have to decide (a) whether there was the giving of any facility to DISL by any of the Appellants; (b) if so what price was actually paid for it; and (c) what would have been the price paid if DISL and that Appellant had been at arm's length.

The Schedule 28AA Regime

64. By s 108(1) FA 98, ss 770 to 773 were substituted by a new s 770A. Section 770A gave effect to Sch 28AA. That schedule applied to "provision" made or imposed between relevantly connected persons. The substitution took effect for corporation tax purposes for accounting periods ending on or after 1 July 1999 (see SI 1998/3178), and was expressed to take effect "in relation to provision made or imposed at any time." Thus it applied to a "provision" whether it was made before or after 1 July 1999. Relevant to these appeals are the following provisions of that schedule:-

"1 (1) This Schedule applies where -

(a) provision ("the actual provision") has been made or imposed as between any two persons ("the affected persons") by means of a transaction or series of transactions, and

(b) at the time of the making or imposition of the actual provision -

(i) one of the affected persons was directly or indirectly participating in the management, control or capital of the other; or

(ii) the same person or persons was or were directly participating in the management, control or capital of each of the affected persons

(2) Subject to paragraphs 8, 10 and 13 below, if the actual provision -

(a) differs from the provision ("the arm's length provision") which would have been made as between independent enterprises, and

(b) confers a potential advantage in relation to United Kingdom taxation on one of the affected persons, or (whether or not the same advantage) on each of them,

the profits and losses of the potentially advantaged person or, as the case may be, of each of the potentially advantaged persons shall be computed for tax purposes as if the arm's length provision had been made or imposed instead of the actual provision.

(3) For the purposes of this Schedule the cases in which provision made or imposed as between any two persons is to be taken to differ from the provision that would have been made as between independent enterprises shall include the case in which provision is made or imposed as between any two persons but no provision would have been made as between independent enterprises; and references in this Schedule to the arm's length provision shall be construed accordingly."

2 (1) This schedule shall be construed (subject to paragraphs 8 to 11 below [which are not relevant in these appeals] in such manner as best secures consistency between-

(a) the effect given to paragraph 1 above; and

(b) the effect which, in accordance with the transfer pricing guidelines, is to be given, in cases where double taxation arrangements incorporate the whole or any part of the OECD model, to so much of the arrangements as does so.

(2) In this paragraph "the OECD model" means-

(a) the rules which, at the passing of this Act were contained in Article 9 of the Model tax Convention on Income and on Capital published by the Organisation for Economic Co-operation and Development; or.

(b) any rules in the same or equivalent terms.

(3) In this paragraph "the transfer pricing guidelines" means-

(a) all the documents published by the Organisation for Economic Co-operation and Development, at anty time between 1^st May 1998, as part of their Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations; and

(b) such documents published by that organisation on or after that date as may for the purposes of this schedule be designated, by an order made by the Treasury, as comprised in the transfer pricing guidelines.

Meaning of "transaction" and "series of transactions"

3 (1) In this Schedule "transaction" includes arrangements, understandings and mutual practices (whether or not they are, or are intended to be, legally enforceable).

(2) References in this Schedule to a series of transactions include references to a number of transactions each entered into (whether or not one after the other) in pursuance of, or in relation to, the same arrangement.

(3) A series of transactions shall not be prevented by reason only of one or more of the matters mentioned in sub-paragraph (4) below from being regarded for the purposes of this Schedule as a series of transactions by means of which provision has been made or imposed as between any two persons.

(4) Those matters are -

(a) that there is no transaction in the series to which both those persons are parties;

(b) that the parties to any arrangement in pursuance of which the transactions in the series are entered into do not include one or both of those persons; and

(c) that there is one or more transactions in the series to which neither of those persons is a party.

(5) In this paragraph, "arrangement" means any scheme or arrangement of any kind (whether or not it is, or is intended to be, legally enforceable)."

65. Paragraph 1(1) speaks of provision between two persons. For the schedule to apply those two persons must be identified. Whilst the transaction or series of transactions by which the relevant provision is made or imposed may encompass transactions between persons other than those two persons, the identified provision must be between two persons only. There is, in our view, no scope for reading "two" as "two or more".
66. There is no further definition of "provision" in the schedule. But we note that transaction includes informal arrangements and understandings and series of the same. That suggests that "provision" is not limited to formal or enforceable arrangements but may be of a similar nature, since it may made by means of such informal arrangements or understandings. Article 9 of the OECD Model Tax Convention contains the phrase "conditions made or imposed between the two enterprises in their commercial or financial relations", and the Appellants note that para 42 of the Explanatory Notes to clause 106 of the Finance Bill 1998 stated that the term provision is analogous to that phrase. The obligation to construe para 1 in such manner as best secures consistency with the OECD model indicates that "provision" should be given a similar meaning to that which should be given to "condition" in the model. There seems to us however, to be nothing in the model which indicates that "condition" should be restricted to formal or enforceable arrangements.
67. It can be by means of a "series" of transactions that a provision is made or imposed. "Series" normally carries the connotation of order, of some correspondence between the natural numbers and the members of the series. But by para 3(2) that meaning is extended. A number of transactions can constitute a series so long as they are entered into in pursuance of the same arrangement.
68. The Appellants did not dispute that DSG, MCSAL, and DISL were persons any two of which fell within para (1)(b)(ii). No such admission is made relation to MSDL but Mr Goy did not argue before us that MSDL's profits should be adjusted.
69. Unlike s 770 no direction is required from the Board before schedule 28AA takes effect: it applies automatically to the computation of the taxable profits of affected companies.
70. Para 2(1)(b) speaks of giving effect "in cases where double taxation arrangements incorporate the whole or any part of the OECD model, to so much of the arrangements as does so". (Para 14 defines "double taxation arrangements" as meaning "arrangements having effect by virtue of section 788"). There are two possible readings of this provision: under the first it applies only where the parties to the provision are resident in jurisdictions between which there is a double tax treaty which incorporates the model or any part of it; under the second it applies generally and independently of whether or not there is a relevant double tax treaty between those jurisdictions because its effect is to apply the meaning that would apply if there were such a treaty in force to interpreting Sch 28AA. The question is whether "in cases" refers to cases of particular provision or to any cases where there are double tax treaties.
71. This issue was not argued before us and the assumption of both parties was that the second approach was correct: the OECD model was to be applied in interpreting the schedule. In our opinion the second approach is the correct one. It seems to us that "in cases where double taxation arrangements incorporate" is not referring to a specific double tax treaty: if the draftsman had intended the model to be applied only where there was a double tax treaty relevant to the parties, the natural way to draft it would have been to set out the condition at the start of the paragraph. Although there is a treaty with the Isle of Man it was not mentioned by either party and we presume that it does not apply because DISL is exempt from tax in the Isle of Man.
72. Like s 770's reference to "the" arm's length price, Sch 28AA uses the definite article, referring to "the" provision which would be made as between independent parties. Clearly there will be circumstances where independent parties could have made one of a number of different provisions, but the legislation requires one to be fixed upon.
73. In summary in relation to Sch 28AA: (a) we have to identify whether in relation to any Appellant there was any provision made or imposed between it and DISL, (b) if so, we have to determine whether the same provision would have been made had that Appellant and DISL been independent, (c) if not, then we must determine the provision which would have been made, and (d) we must then determine for each relevant year after 1 July 1999 what effect the making of the different provision would have had on the taxable profits of that Appellant.

Contentions of the parties on the law

74. Mr Peacock contends:

(1) No business facility was provided to DISL in the Cornhill Period in the form of the opportunity to enter into the insurance arrangements because the warranty business was regarded as high risk. In particular:

(a) DISL's position before and after 1993 was exactly the same. The effect of squeezing Cornhill's profits was to put more profit into the Appellant group in the UK.

(b) DISL's profits in 1990 were £[**]; in 1991 £[**]; in 1992 £[**]; and in 1993 £[**], thus showing considerable volatility, with a downward trend looked at in 1993.

(c) One cannot compare Cornhill with 5% of the profits and 95% reinsured, with DISL having 95% of the profits and no reinsurance.

(2) At the time of the start of the ASL Period there were no insurers in the market prepared to take on risk of this size; DISL was the only possible insurer. No opportunity to enter into the arrangements in the ASL Period was provided by any of the Appellants; the decision to engage DISL was made by ASL. It is not possible to adjust the consideration between DISL and ASL and then between ASL and MCSAL, being two transactions between unconnected parties.

75. Mr Goy contends:

(1) A business opportunity was provided to DISL by DSG for s 770 in the form of the opportunity to enter into an attractive insurance contract. DSG was the sales company without which there would have been no extended warranty business. DSG would have required payment by an independent company to enter into the same arrangements in the form of an initial payment and a profit commission.

(2) In the period relevant to Sch 28AA the provision was made by the 1993 and 1997 arrangements. There is no need to consider any further provision in the form of new prices or the extension of the contracts because if the initial arrangements had been at arm's length so would any subsequent variations. The provision was made by MCSAL (and DSG after 2001) by means of a series of transactions pursuant to the same arrangement (being the various contracts entered into) on terms different from those that would have existed had the parties been at arm's length. At arm's length MCASL would have been paid a further initial payment of 15% of the net contract price and a profit commission. Alternatively a payment would have been made to DSG.

The Differences between Section 770 and Sch 28AA

76. We should comment on three differences between the s 770 and sch 28AA provisions. First, the s 770 regime requires the arm's length price to be determined for the facilities actually provided. By contrast para 1 of Sch 28AA requires the identification of a provision between relevantly connected persons and the determination of what provision would have been made between independent persons. That notional provision could be different, not only in price, but in its terms from the actual provision made.
77. Second, the Sch 28AA regime, by virtue of para 2 requires effect to be given to the Transfer Pricing Guidelines as they apply to treaties following the OECD model. The incorporation is not wholesale: it merely requires the schedule as a whole to be interpreted in such a way as secures consistency between para 1 of the schedule and the OECD model in accordance with the Transfer Pricing Guidelines. There is no incorporation of the OECD model in s 770. But it seems to us that in determining the arm's length price, the approach of the OECD model is a useful aid which we should apply in the absence of any other guidance as they are the best evidence of international thinking on the topic.
78. Third, s 770 requires one to determine the price which would have been paid if the parties "had been independent parties dealing at arm's length". It seems clear to us that those words do not require any adjustment to be made in setting the price to the actual characteristics of the parties other than their independence. The actual assets, business and attributes of each party remain constant and may be relevant to the determination of the arm's length price. The language of para 1(2)(a) is different: "differs from the provision which would have been made between independent enterprises". It is at first sight possible that those "independent enterprises" may not be enterprises which do not share the same attributes as the actual parties to the provision. But it is clear to us that that interpretation is not consistent with the OECD model (see in particular the emphasis on comparability in the extracts below which presupposes looking at the actual characteristics of the enterprises between which provision has been made) and therefore that para 1(2)(a) should be interpreted as requiring consideration of what provision independent enterprises sharing the characteristics of the actual enterprises would have made.

The Overlap between the two regimes.

79. It is in theory possible that a transaction which took place before 1999 could result in adjustments under both the s 770 regime and the Sch 28AA regime, and that those adjustments could be different and result in either double taxation or non taxation. If for example in 1997 a facility (for example a licence) was provided for 10 years at £10 per year when between arm's length parties a premium of £50 would have been charged, then the provider's profits might be increased by £50 in 1997. After 1999 the question is asked "what would the arm's length provision have been?" and suppose the only answer is a payment of £15 per year. As a result £5 per year is added to the provider's profits. In such a case the provider's profits would be increased in total by 8 years times £5 = £40 plus the original £50. There is nothing in Sch 28AA which permits any account to be taken of a previous s 770 adjustment, but it seems to us that where there is a range of provisions which could have been made between independent enterprises some of which might give rise to double counting or omission and others which would not, those which do not should be preferred. Likewise, in applying Sch 28AA in successive years to the same provision, if there is a choice of possible provision, that choice should be exercised consistently.

Was there provision of business facilities to DISL in 1993?

80. We consider DISL's position in 1993 to determine whether any of the Appellants gave any business facility to DISL. While Mr Peacock is correct in saying that DISL's position did not change in 1993, its position relative to Cornhill did change. Mr Carroll, on behalf of DSG (and other companies), negotiated the squeeze in Cornhill's position with Cornhill. Cornhill was in a position to squeeze DISL, but did not do so, with the result that Cornhill's position relative to DISL was disadvantaged. Cornhill's agreements with DSG and DISL were both due to end on 30 April 1993 and could presumably be terminated thereafter on six months' notice, the notice applying before the term was extended to 30 April 1993. Cornhill asked DISL on 3 March 1993 (before it was legally committed to the Appellants but after it had agreed terms in principle) whether it was willing to extend the reinsurance treaty on the same terms (including the 15% profit cap) until 30 April 1998. On 7 October 1993 Cornhill agreed the terms with DSG (including the 12% profit cap and the cap and floor on investment income). On 18 November 1993 Cornhill agreed the new reinsurance treaty with DISL. If Cornhill had been considering its own interests there is no reason why it should have made such agreements without squeezing DISL in turn. That it did not do so can be explained only by inferring that it had agreed with DSG that it would not do so, and we find that this is what happened. We have compared the resulting returns to Cornhill and DISL in paragraphs 19 and 20 above.
81. We are not persuaded by Mr Peacock's points that Cornhill with 5% cannot be compared directly to DISL with 95% and no reinsurance, although any such comparison must take account of the different capital requirements, as we have done in paragraph 20 above. Because this is a quota share arrangement the two are otherwise comparable. Nor do we consider that the volatility in profits of DISL is relevant as this is dealt with by the profit commission.
82. The net effect is that the Appellants locked Cornhill into a five-year contract that squeezed Cornhill's profits, and indirectly locked DISL into a five-year contract under which its position was unchanged. We find that this is the indirect provision of a business facility by DSG to DISL for which DISL made no payment. Since Cornhill was at arm's length with the Appellant Group it is clear that DISL's relationship with DSG was not. We find that the opportunity was provided by DSG in whose stores insurance was sold. Accordingly we find that by providing DISL with the opportunity of entering into the new reinsurance treaty at the time DSG indirectly provided DISL with business facilities in 1993, that s 770 is potentially applicable.

Provision

83. There was no real dispute between the parties that the contractual framework described at paragraph 39 above could be a series of transactions for the purposes of para 1(1) of Sch 28AA. Each document was an arrangement and the fact that one or more of the parties as between whom it was alleged that there was provision was not a party to one or more of them was (as a result of para 3(3)) irrelevant.
84. In order for those transactions to be a series for the purpose of para 1(1) there must either be some sequence in which they can be put, or therey must be entered into in pursuance of, or in relation to, an arrangement. It seems to us that they were entered into in pursuance of the understanding between DISL, ASL and members of the Appellant Group which followed the negotiation with Willis in April 1997. Not only did these entities know, at the time the agreements were signed, that they would all take effect together but they were planned and seen as a series of interlocking agreements.
85. Mr Goy put the Respondents' case in relation to the nature of the alleged provision in two ways. His first basis followed his argument under s 770: he said that DSG had procured for DISL the opportunity to enter into the contract with ASL. That procuration was the making of provision between DSG and DISL. Mr Goy's second basis was his preferred one. On this basis he says that provision was made between MCSAL (and after 2001 DSG) and DISL by the series of contracts described above. Mr Goy accepted that on this basis in arguing that the provision was the series of transactions he was reading "by means of" in para 1 as including "consisting of".
86. Mr Peacock says that (1) there is a difference between a provision and the means by which it is given effect; (2) para 1 requires you to identify the provision and the persons as between which it is made or imposed; (3) there must be evidence for any provision and the particular two parties between which it is made or imposed; (4) on the evidence before us there was no support for any provision between DSG and DISL; (5) given that the extended warranty business was regarded as that of Dixons Finance, and that Mr Carroll signed letters indicating that he was acting for Dixons Finance and Dixons Group Plc, then, if there had been any provision, it was likely to have been between one of those companies (or an intermediate holding company) and DISL rather than between DSG or MCSAL and DISL:; and, if so, such provision was outside the scope of this appeal.
87. In relation to the ASL Period it seems to us that there was a provision made or imposed as between DSG and DISL. That provision was the arrangement that DISL would insure the extended warranty business written in DSG's stores on particular terms. That provision was not directly between DSG and DISL but in our view can properly be described as being "as between" them. The provision may have been imposed by someone other than DSG, but it took effect as between DSG and DISL. That arrangement was imposed by means of the series of contracts. The series of contracts was not itself the provision which took effect between those entities, but the means by which the arrangement was given effect. We conclude that such a provision was imposed between them because:

(1) when the IPT changes occurred the documentary evidence noted elsewhere in this decision indicated that the Appellant Group wanted to replace the Cornhill insurance structure with a structure involving the same parties (ie including DISL) and on the same terms, but with service contracts rather than insurance. What was intended was an arrangement which involved DSG and DISL;

(2) the evidence of the negotiations leading up to the final structure contained no realistic indication that anyone other than DISL would replace DISL;

(3) the negotiations and the arrangement into which they evolved and to which they gave effect encompassed an expectation of the premium income which would accrue to DISL ie of the terms on which DISL would participate;

(4) the evidence indicated a desire to retain the tax benefits of the DISL structure.

88. The only evidence before us that there could have been provision between other members of the Appellant Group (other than MCSAL - as to which see below) and DISL in relation to the extended warranty business was:

(1) in relation to Dixons Finance, that Mr Carroll acted on occasion on its behalf, and that to some extent the extended warranty business was regarded as that of Dixons Finance. But there was no evidence that Dixons Finance was practically or legally involved in any part of the extended warranty business or that anything it did or might have done affected any interest it might have in that business. The only role we can see it possibly performing in any arrangement with DISL is facilitating or procuring for DISL the arrangement as between DSG and DISL;

(2) in relation to Dixons Group PLC, that Mr Carroll signed on occasion on its behalf, and that it would generally have been in the overall interest of the top company in a group to retain profits within the group, and to retain them in the most tax efficient manner. But again that would limit the possible provision made between that company and DISL to the procuring for DISL of the arrangement as between DSG and DISL.

89. But, having identified a provision between DSG and DISL, whether or not procured or facilitated by a further provision by Dixons Finance or Dixons Group and DISL, the question for us is then whether that provision was different from that which would have obtained if the parties had been at arm's length, and if so what effect would that have on DSG's profits. It seems to us that if one assumes an unconnected DSG, then it is likely that its profits would be the same in relation to any provision between it and DISL whether it had itself made that provision or it had made it on the prompting of another person.
90. Where a taxpayer is a party to two or more non-arm's length provisions with different counterparties but which have the same effect and which are given that effect by the same series of transactions, then it is possible to read sch 28AA to require each such provision to be separately adjusted for without regard to the other or others because the only assumption required in determining the taxpayer's profits is that it is independent of the particular counterparty, and there is little room for applying para 1(2) to provisions in the plural deriving from separate bilateral provisions found under para 1(1). However, not only would such an approach result in double taxation, but it is plainly contrary to the whole spirit of the OECD Guidelines with regard to which para 1 must be construed. It seems to us therefore that at the very least in a case where the other bilateral provisions consist of procuring the taxpayer to enter into a provision between that taxpayer and another, no separate or fractional adjustment falls to be made for each provision, but instead a single adjustment in respect of the provision between the taxpayer and that other should be made which reflects the full effect of the non-arm's length nature of that provision. For this reason the existence or otherwise of provision between Dixons Finance or Dixons Group and DISL which consisted of procuring the arrangements between DSG and DISL in circumstances where such other provisions did not involve the conferring of payment or benefit on DSG by such other party neither deters us from our conclusion that there was an arrangement between DSG and DISL nor persuades us that any adjustment to our determination should be made to reflect such provisions.
91. Mr Goy advanced MCSAL as the party to a provision between it and DISL made by the series of agreements in the period for which it was such a party. It was MCSAL which contracted with ASL, and MCSAL which sold the extended warranties. But MCSAL sold these products in DSG's stores through the agency of DSG's employees under a contract to which DSG was a party (see paragraph 39(1) above). We had no details of the relationship between MCSAL and DSG in relation to its activities in DISL's stores, other than the recital in the 1997 agreement stating that DSG had agreed to allow MCSAL to market repair contracts to DSG's customers, but it seems to us likely that either MCSAL passed its profits on such business back to DSG, or that if MCSAL and DSG had been at arms' length it would have done so. In any event there must have been some arrangement or understanding between MCSAL and DSG, and that arrangement would be a transaction in the series between DSG and DISL. To concentrate solely on a provision between MCSAL and DISL ignores the provision or arrangement which must on the evidence before us have existed between MCSAL and DSG. We do not believe that any adjustment can be made between MCSAL and DISL in the years relevant to this appeal. If we are wrong and there was no provision between MCSAL and DSG, so that the point of sale dealing could be said independently to have belonged to MCSAL, then our conclusions as to any adjustment applicable to DSG apply, up to 2001, to MCSAL instead.
92. In relation to the application of Sch 28AA to the residue of the Cornhill Period, it seems to us, for the reasons set out above in relation to s 770, that there was a provision made or imposed between DSG and DISL which consisted of the arrangement that DISL would reinsure the insurance written by Cornhill in DSG's stores, and that that provision was effected by means of the series of agreements made in 1993 including the five-year extension to the reinsurance treaty.
93. For an adjustment to be made under Sch 28AA the provision must be different from that which would have been made as between arm's length parties. We accept Mr Goy's contention that if the 1997 provision had been at arm's length, then other provisions between these parties made in subsequent years would have been at arm's length. Thus there is no need to consider whether such other provisions give rise to other adjustments. We now proceed to consider what that provision would be. In the discussion which follows as in the argument before us our focus is on the profits which each party to an arrangement under which extended warranty business was insured or reinsured could be expected to make if they were independent. That discussion is a necessary precursor to the determination of the terms on which an arrangement between an independent DSG and DISL relating to that business would take place. The discussion of comparables thus looks not specifically at a comparable arrangement to that identified above, but to entities conducting comparable business.

OECD Guidelines

94. The OECD issued revised Transfer Pricing Guidelines for Multinational Enterprises and Tax Administration in 1995. They are expressly referred to in Sch 28AA and although they do not apply in terms to s 770 they existed in all the years under appeal and are the best evidence of international thinking on the topic. The guidelines stress the importance of looking for comparable transactions so long as they are comparable, the need to strive to make adjustments to create comparability if at all possible, and that more than one method may be considered:

"1.15….In order for such comparisons to be useful, the economically relevant characteristics of the situations being compared must be sufficiently comparable. To be comparable means that none of the differences (if any) between the situations being compared could materially affect the condition being examined in the methodology (eg price or margin), or that reasonably accurate adjustments can be made to eliminate the effect of any such differences. …

1.17 As noted above, in making these comparisons, material differences between the compared transactions or enterprises should be taken into account. In order to establish the degree of actual comparability and then to make appropriate adjustments to establish arm's length conditions (or a range thereof), it is necessary to compare attributes of the transactions or enterprises that would affect conditions in arm's length dealings. Attributes that may be important include the characteristics of the property or services transferred, the functions performed by the parties (taking into account assets used and risks assumed), the contractual terms, the economic circumstances of the parties, and the business strategies pursued by the parties…".

The guidelines list as factors determining comparability (1) characteristics of property or services, such as the nature of the services; (2) functional analysis, reflecting the functions that each party performs taking account of assets used and risks assumed; (3) contractual terms; (4) economic circumstances, such as the relative competitive positions of the buyers and sellers; (5) business strategies such as market penetration schemes.

"2.6 The CUP [comparable uncontrolled price method] compares the price charged for property or services transferred in a controlled transaction to the price charged for the property or services transferred in a comparable uncontrolled transaction in comparable circumstances. If there is any difference between the two prices, this may indicate that the conditions of the commercial and financial relations of the associated enterprises are not arm's length, and that the price in the uncontrolled transaction may need to be substituted for the price in the controlled transaction.

2.7 Following the principles in Chapter I, an uncontrolled transaction is comparable to a controlled transaction (ie it is a comparable uncontrolled transaction) for purposes of the CUP method if one of two conditions is met: 1. none of the differences (if any) between the transactions being compared or between the enterprises undertaking those transactions could materially affect the price in the open market; or 2. reasonably accurate adjustments can be made to eliminate the material effects of such differences. Where it is possible to locate comparable uncontrolled transactions, the CUP Method is the most direct and reliable way to apply the arm's length principle. Consequently, in such cases the CUP Method is preferable over all other methods.

2.9 In considering whether controlled and uncontrolled transactions are comparable, regard should be had to the effect on price of broader business functions other than just product comparability (i.e. factors relevant to determining comparability under Chapter I). Where differences exist between the controlled and uncontrolled transactions or between the enterprises undertaking those transactions, it may be difficult to determine reasonably accurate adjustments to eliminate the effect on price. The difficulties that arise in attempting to make reasonably accurate adjustments should not routinely preclude the possible application of the CUP method. Practical considerations dictate a more flexible approach to enable the CUP method to be used and to be supplemented as necessary by other appropriate methods, all of which should be evaluated according to their relative accuracy. Every effort should be made to adjust the data so that it may be used appropriately in a CUP method. As for any method, the relative reliability of the CUP method is affected by the degree of accuracy with which adjustments can be made to achieve comparability."

This last paragraph highlights the need to make every effort to make adjustments rather than to discard a possible comparable simply because there are differences. The last sentence of 2.11 indicates that it may be necessary to combine the CUP method with other less direct methods.

95. The guidelines give a secondary role to other methods where traditional transactions methods cannot be reliably applied alone or exceptionally cannot be applied at all (para 3.1). One such other method is the transactional profit method and in particular the profit split method. The Guidelines make the point in para 3.2 that it is unusual to find profit as a condition "made or imposed" in the relevant transactions, but this is an exceptional case because so long as the loss ratio is under 80% the underwriting profit for Cornhill and DISL is set, and within a limited range the investment income return is also set for Cornhill.

"3.5…the profit split method seeks to eliminate the effect on ;profits of special conditions made or imposed in a controlled transaction (or in controlled transactions that are appropriate to aggregate under the principles in Chapter I) by determining the division of profits that independent enterprises would have expected to realise from engaging in the transaction or transactions. The profit split method first identifies the profit to be split for the associated enterprises from the controlled transactions in which the associated enterprises are engaged. It then splits those profits between the associated enterprises on an economically valid basis at approximately the division of profits that would have been anticipated and reflected in an agreement made at arm's length. The combined profit may be the total profit from the transaction or a residual profit intended to represent the profit that cannot readily be assigned to one of the parties such as the profit arising from high-value, sometimes unique, intangibles. The contribution of each enterprise is based upon a functional analysis as described in Chapter I, and valued to the extent possible by any available reliable external market data. The functional analysis is an analysis of the functions performed (taking into account assets used and risks assumed) by each enterprise. The external market criteria may include, for example, profit split percentages or returns observed among independent enterprises with comparable functions….

b) Strengths and weaknesses

3.6 One strength of the profit split method is that it generally does not rely directly on closely comparable transactions, and it can therefore be used in cases where no such transactions between independent enterprises can be identified. The allocation of profit is based on the division of functions between the associated enterprises themselves. External data from independent enterprises is relevant in the profit split analysis primarily to assess the value of the contributions that each associated enterprise makes to the transactions, and not to determine directly the division of profit. As a consequence, the profit split method offers flexibility by taking into account specific, possibly unique, facts and circumstances of the associated enterprises that are not present in independent enterprises, while still constituting an arm's length approach to the extent that it reflects what independent enterprises reasonably would have done if faced with the same circumstances.

3.7 Another strength is that under the profit split method, it is less likely that either party to the controlled transaction will be left with an extreme and improbable profit result, since both parties to the transaction are evaluated.

3.8 There are also a number of weaknesses to the profit split method. One such weakness is that the external market data considered in valuing the contribution each associated enterprise makes to the controlled transactions will be less closely connected to those transactions than is the case with the other available methods. The more tenuous the nature of the external market data used when applying the profit split method, the more subjective will be the resulting allocation of profits.

…

c) Guidance for application

3.11 If the profit split method were to be used by associated enterprises to establish transfer pricing in controlled transactions, then each associated enterprise would seek to achieve the division of profits that independent enterprises would have expected to realize in a joint venture relationship. Generally, conditions established in this manner would have to be based upon projected profits rather than actual profits, because it is not possible for the taxpayers to know what the profits of the business activity would be at the time the conditions are established.

…

3.15 There are a number of approaches for estimating the division of profits, based on either projected or actual profits, as may be appropriate, that independent enterprises would have expected, two of which are discussed in the following paragraphs. These approaches—contribution analysis and residual analysis—are not necessarily exhaustive or mutually exclusive.

3.16 Under a contribution analysis, the combined profits, which are the total profits from the controlled transactions under examination, would be divided between the associated enterprises based upon the relative value of the functions performed by each of the associated enterprises participating in the controlled transactions, supplemented as much as possible by external market data that indicate how independent enterprises would have divided profits in similar circumstances. In cases where the relative value of the contributions can be measured directly, it may not be necessary to estimate the actual market value of each participant's contributions.

…

3.19 A residual analysis divides the combined profit from the controlled transactions under examination in two stages. In the first stage, each participant is allocated sufficient profit to provide it with a basic return appropriate for the type of transactions in which it is engaged. Ordinarily this basic return would be determined by reference to the market return achieved for similar types of transactions by independent enterprises. Thus, the basic return would generally not account for the return that would be generated by any unique and valuable assets possessed by the participants. In the second stage, any residual profit (or loss) remaining after the first stage divisions would be allocated among the parties based on an analysis of the facts and circumstances that might indicate how this residual would have been divided between independent enterprises. Indicators of the parties' contributions of intangible property and relative bargaining positions could be particularly useful in this context.

…

3.21 One approach to a residual analysis would seek to replicate the outcome of bargaining between independent enterprises in the free market. In this context, the basic return provided to each participant would correspond to the lowest price an independent seller reasonably would accept in the circumstances and the highest price that the buyer would be reasonably willing to pay. Any discrepancy between these two figures could result in the residual profit over which independent enterprises would bargain. The residual analysis therefore could divide this pool of profit based on an analysis of any factors relevant to the associated enterprises that would indicate how independent enterprises might have split the difference between the seller's minimum price and the buyer's maximum price."

Possible comparables

96. A number of comparables were put forward, first for MCSAL, on the basis that Mr Rust could not find comparable insurance companies to DISL and it was contended that if MCSAL's commission rate was arm's length, so must the premium paid to DISL. This approach was supported by Mr Bezant, although this proposition was contested by Mr Gaysford on the basis that it depended on the premium paid by the customer being fixed in a perfect market, which was not the case here where the price might be described as a monopoly price. We accept that in looking for comparables to MCSAL one needs to find one with a point of sale advantage or be able to adjust for its absence.
97. On the use of commission rates Mr Peacock contends for Mr Rust's and Mr Bezant's approach that if the retailer's commission rates are the same as in the comparables this means that DISL's share must also be the arm's length figure.
98. Mr Goy contends that commission rates on their own are not a good guide. If the gross price is high because of the retailer's monopoly position the commission rate will be higher even though the insurer gets the same premium. Similarly, if the gross price is high both the insurer's profit and the commission may be higher than in an arm's length situation.
99. We agree with Mr Goy that it is better to consider DISL's profit directly rather than indirectly by considering commission rates unless it is possible to adjust for the point of sale advantage, in which case it might be a useful tool. There was no information on which we could make such an adjustment.

Orion

100. The first possible comparable is the agreement made in July 1982 following the collapse of Multi Guarantee between Currys and Orion, which continued until the start of the Cornhill Period. No copy of the agreement can be found but Mr Rust saw a copy when preparing his transfer pricing report in 2000. Mr Rust records that the following terms applied:

(1) Currys acted as Orion's agent in the sale of extended warranties for up to five years on brown and white electrical goods through its retail outlets. Currys received a commission of the difference between the gross and net premiums.

(2) The agreement covered Great Britain [we suspect that this should be the UK], the Isle of Man and the Channel Islands.

(3) Currys did not administer claims; this was done by Blackwall Green, Currys' then insurance brokers.

(4) The agreement could be terminated on 90 days' notice.

(5) Agreements were for a maximum of 1+5 years.

(6) Currys could not operate another extended warranty scheme.

101. Mr Rust calculated that the middle 50% figures for commission under the Orion agreement was [Q]%, or [**]% on a basis that took into account the commission on the weighted mix of sales. Based on the same sales mix MCSAL's commission for 1989-90 and 1990-2000 would have been an average of [Q-3]% initial commission and [**]% discounted profit commission, total [**]%.
102. Mr Peacock contends that Orion is a good comparable as it was an agreement between a major retailer of domestic electrical goods and a third-party insurer. Any increase in the price of warranties to the customer between 1990 and 1994 as shown in the OFT Report of December 1994 (Mr Rust believed that the multiple electrical retailer referred to in Table 3 was the Appellant Group) was inflationary only.
103. Mr Goy contends that Mr Rust did not take account of the bargaining position of the parties in considering the comparability of the Orion arrangements. The arrangements were negotiated 15 years before 1997 when we are considering them and the position was quite different by then in that loss ratios for a long period were known. The changes made to the Cornhill contractual position shows that more profits were being made by insurers than was expected in the 1980s. The OFT Report covers only the changes in prices between 1990 and 1994, whereas we are concerned with the whole period since 1982.
104. We conclude that Orion is a potential comparable because it covered a similar range of goods for Currys and operated until the start of the Cornhill Period and Currys had the point of sale advantage possessed by the Appellant Group. Against treating it as a comparable is the fact that the agreement started in 1982, when there was little data about loss rates available, and ended in 1987, when the market for extended warranties was very different from 1993 and 1997 with which we are concerned. In particular there was limited information about the likely claims at the time it was entered into. It is also likely that the bargaining position of retailers improved with the consistently low loss ratios shown as more data became available. Since the agreement was terminable on 90 days' notice Currys might have renegotiated its terms if it had continued, as Mr Bawcutt pointed out. By failing to do so they might be taken to consider that the terms remained at arm's length. In contrast it was noted that the arrangements with Cornhill, set up later than the Orion contract, were renegotiated in 1990, introducing a retrospective profit commission taking away a substantial proportion of the profit [calculation omitted], which might demonstrate that the retailer was by then in a much stronger position, and the Appellant Group's position was improved again in 1993. But most importantly this seems to be a case of parties with more equal bargaining power because (we assume) that Orion was a substantial insurer and so it was an agreement made between equals. This is a factor for which we have no evidence enabling us to make adjustments.

National Satellite Services Limited

105. The second possible comparable is the agreement between CIS (then Dixons Finance plc) and National Satellite Services Limited ("NSS") made on 31 July 1990 and effective from 1 June 1990. CIS (then Dixons Finance plc) acted as agent for NSS in selling insurance contracts on behalf of Cornhill. CIS provided administration services.
106. The following terms applied:

(1) NSS sold satellite TVs and the agreement was restricted to customers who purchased or repaired satellite equipment. However, it covered a wider range of all types of electrical goods listed in the agreement. Insurance was available to customers within 365 days of buying the product, and it was provided that where the product had been purchased from DSG an EPOS [electronic point of sale] number had to be given. We understand this to mean that so long as the customer purchased satellite equipment from NSS the agreement could cover the purchase of insurance for other electrical goods purchased within a year previously (whether from the Appellant Group or elsewhere). Mr Carroll said that principally the products were satellite dishes and that NSS were not expected to comb the customer's house to see if there were any other goods on which they could sell policies. He could not recall whether most of the policies related to satellite dishes. Mr Rust thought that Mr Carroll's recollection was wrong. In view of the terms of the agreement we find that the agreement covered principally satellite equipment but might in addition cover other electrical goods purchased within the previous year, which might include goods purchased from the Appellant Group.

(2) Cover was for three years from one year after purchase.

(3) Policies were sold at the time of installation of satellite equipment.

(4) NSS was prohibited from marketing other extended warranty insurance.

(5) MCSAL had complete discretion whether or not to accept policy applications.

(6) NSS did not administer claims.

(7) NSS received a commission of the difference between the gross and net premiums set out in the agreement for the full range of electrical goods. These could be changed by agreement of the parties. Mr Rust calculated that the middle 50% the commission paid to NSS was in the range [**]% and [**]% (for comparison MCSAL's commission is higher as set out in paragraph 53 above).

(8) The agreement was terminable on one week's notice.

107. The agreement was terminated by DF in May 1991 after only 295 policies had been sold generating a net premium to Cornhill of £14,000.
108. Mr Rust pointed out that the sales might have been small but it was the expectation of the parties at the time it was entered into that mattered. There was no evidence of the expected amounts but we infer that the expectation was considerably higher than the actual sales otherwise the Appellant Group would not have thought it worthwhile to make the agreement. It may also be (although there is no evidence of this) that an additional attraction to the Appellant Group was the possibility of insuring goods purchased from it during the previous year.
109. We do not regard NSS as a comparable on the basis that the agreement related principally to satellite equipment (even though it might in addition include other electrical goods), that only 1+3 policies were covered, that the warranties were not sold in stores but at the time of installation of satellite equipment, and that the agreement was terminable on one week's notice. This last item shows that the parties were not contemplating a long-term relationship and would presumably have been expected to negotiate a longer-term one after some experience of operation had been obtained (although in fact the agreement was terminated after almost a year's operation). Neither could we adjust for the differences: we had no data on the expected loss ratios for satellite dishes and could not see how to adjust to reflect the difference between selling a warranty in store and selling it as part of installation.

Office of Fair Trading Report

110. The Office of Fair Trading's report "Extended Warranties on Electrical Goods" of December 1994 gives the commission rates paid by three unidentified retailers in 1994 as 30%, 32.5% and 42%. Mr Heneage also said that a typical sales commission was around 30%.
111. We pay no attention to the rates quoted in the OFT report for unidentified retailers because there was no evidence as to whether the rates agreed were between arm's length parties or what the other terms of deals were. Mr Heneage's rates are no doubt true as an average but we are concerned with the circumstances of the relationship between the Appellant Group and DISL.

The Link

112. The Appellant Group originally sold mobile phones and related items in stores called The Link. On 7 April 1997 the business was hived-down to the Link Stores Limited which became a joint venture between the Appellant Group (60%) and BT Cellnet (40%). This business was insured by Cornhill (and continued to be so insured during the ASL Period) and reinsured by DISL. CIS appointed The Link Financial Services Limited as its sub-agent to sell the insured warranties as agent for Cornhill. The contract is complicated in respect of the commission rates paid to The Link and the parties are not agreed on the effect. We summarise the terms as follows. Initially the commission rate was 34% subject to adjustment if there was a change in any of the component elements which determined the proportion of commissions in respect of warranty business to total sales of the Link business. In the event of such a change the parties were to negotiate in good faith to reflect the change. In making a change the auditors were to have regard to determining the commission rate to be the percentage as maintains the proportion the commission receivable from sales of warranties by The Link to the profit from all sales of warranties on the same products sold by the Appellant Group (we are not certain whether this profit was intended to include the profit of DISL) ("the profit percentage") such that this does not fall from one quarter to the next.
113. The Appellant Group's interpretation is that this means that if the Appellant Group's profit from such warranties after claims is [X]% of the Link's gross warranty sales, the ratio of 34/[X] ([..]%) must be maintained. This can result in the 34% commission rate going up or down.
114. The Revenue's interpretation is that the profit percentage must not fall, not that it must remain constant. In addition, the Revenue contend that the formula is intended to deal with changes in penetration rates, so that an increase should lead to a higher commission rate to reward the agent's increased performance. The reference to the auditors taking the profit percentage into account is only one factor to be taken into account, so that the profit percentage is no more than a limitation in the fair alteration to reflect changes in penetration rates. Accordingly, if there is increased profitability the profit percentage would otherwise fall and so, provided penetration rates have increased, the commission rate must be increased to keep the profit percentage constant. If profitability decreased but the commission rate remained constant, the profit percentage would otherwise increase. No adjustment would be required but the auditors could increase the commission rate to reflect increased penetration; but if penetration rates had decreased the prohibition in decreasing the profit percentage potentially limits the reduction that the auditors could determine. Thus this protects The Link even where both profitability and penetration rates have fallen. Since the alteration is from one quarter to the next any increase is permanent and one should not look at the initial percentage as the Appellant does.
115. We consider that the ratio of commissions on warranty products (not the gross sales proceeds of such products) to the whole of the Link's sales could change for a number of reasons: (a) change in commission rates; (b) changes in the relative prices of the products; (c) changes in the proportion of product sales which were accompanied by a warranty sale (which is what the Revenues seem to have in mind with their phrase "penetration rates"); and (d) the sale of other products in relation to which warranties were not offered. We note that profitability is defined as the profit made by the whole of the Appellant Group - not the Link. The agreement seems to be making sure that the Link gets a decent slice of the Appellant Group's warranty profits on phones rather than offering protection against lack of profitability within the Link. We consider that the Link is entitled to the higher of:

(1) x% of the vendor group's profits from phone warranty business, where x is fixed on a lock step starting with x=34% *first period phone warranty receipts/Appellant Group's profits on phone warranty sales (whatever that is), and

(2) y% of the Link business warranty receipts

where y is the higher of (a) 34% or if there has been a change in "penetration rates" a fairly adjusted replacement for 34%; and (b) the Group commission percentage ie the rate of commission earned in other parts of the Appellant Group - excluding the Link - on the sale of phone warranties.

It follows that we agree with the Appellant's interpretation that this amount could be both higher and lower than 34% of Link business warranty receipts.

116. The Revenue also say that the Appellant Group's figure [X]% overall profit is unrealistic so far as mobile phones are concerned in view of the higher loss ratios which would mean that the 34% represented a higher proportion of the profit. The projected loss ratio for 1996-97 was 115%. The figure in Watson Wyatt's report for mobile phones (which covered sales other than The Link) for 1996-97 was 107% at 1 March 1997 and 120.3% as at 7 March 1998.
117. Mr Goy again makes the point about the different negotiating positions of the parties. While the Link may be a bargain between equals, this was not the position with DISL. The agreement related only to mobile phones which had a worse loss ratio than the mix of products insured by DISL.
118. We conclude that The Link is not a comparable on the basis that it dealt only with mobile phones on which the risk was in 1997 clearly higher than the average relating to goods sold by DSG. It was also not clear whether the promotion of BT's products in the Link stores was a benefit taken by BT to some extent in lieu of the payment of additional warranty commission. Nevertheless one might draw the inference that in April 1997 when the agreement was made 34% was the minimum commission which would be ceded to an arm's length counterparty for this type of business. For business which was expected to be less risky one would therefore expect a greater commission. But we do not have the data to make the necessary adjustment.

DSG Ireland Limited

119. The Irish business started in late 1996 and was insured by Cornhill during the ASL Period with DSG Ireland Limited acting as sales agent. The warranties applied to the same types of goods as in the UK. As mentioned in paragraph 40(1) above, the parties aimed to adjust the net contract price to give a loss ratio of 80%. [The Commissioners then set out the commission rates for 1998-99, 1999-00,2000-01, 2001-02 [missing]; 2002-03, 2003-04, 2004-05 and 2005-06; and premium income from this business in the first year, 1997 and 1998 which they compare to the Appellant Group's UK premiums.] Cornhill would, on the basis of a 80% loss ratio, make a return of £20,000 pa (5% of 80% of £2m).
120. Mr Peacock contends that this is a good comparator as it relates to the same type of products. Although smaller than the UK business the size would not have been known in advance.
121. Mr Goy contends that the 80% loss ratio is the type of figure one would expect at arm's length and this agreement does not assist the Appellants as a comparable.
122. We conclude that DSG Ireland Limited is not a comparable on the basis that Cornhill contracted for a prospective loss ratio of 80% which is much higher than the Appellant Group's loss ratios on non-Irish business. Also this was part of a deal between connected parties - with Cornhill in the middle.
123. The following were also proposed as comparables for DISL.

Domestic & General Group plc

124. Domestic & General Group plc ("D&G") is a quoted company providing domestic appliance breakdown insurance. The majority of sales are through manufacturers although some point of sale business is generated by smaller retailers. Contracts were for 3 to 5 years. D&G handles claims administration and has a significant number of employees dealing with this aspect: 1,435 call centre staff in 2003. It had little debt finance.
125. Mr Bezant considers that D&G is a comparable company to DISL. He calculated the return on equity for both DISL and D&G after making certain adjustments, principally for dividend payments by D&G. The figures show a similar pattern with D&G making a higher return on equity than DISL until 2001 and consistent rates thereafter. He attributed the differences to DISL providing only underwriting services, while D&G provides claims administration, which as routine operations carrying less risk he would expect to command a lower profit margin, which we do not accept for the reasons given below.
126. Mr Bezant and Mr Gaysford agreed that the calculation of return on capital is widely accepted and widely used measure of the profitability of a business and that the use of a cost of capital based on industry benchmarks is a standard approach to assessing a normal level of profit. They agree that return on equity is a suitable starting point to compare the financial performance of DISL and D&G. The correctness of Mr Bezant's figures for DISL are agreed (although his methodology was slightly different from Mr Gaysford's); Mr Gaysford did not check Mr Bezant's figures for the D&G because he did not consider that they were relevant.
127. Mr Gaysford considered that return on equity for D&G is highly imperfect as a comparator for the profitability of DISL because D&G is a major provider of "off-the-shelf" extended warranty products to many small retailers and manufacturers who do not have their own extended warranty business. In the case of manufacturers they do not sell warranties at the point of sale in stores as does the Appellant Group. D&G also undertakes a number of activities not undertaken by DISL. He believes that D&G's capabilities as the largest independent provider of "off-the-shelf" extended warranty products is consistent with it being able to earn returns in excess of the normal level. Its bargaining power is very different from DISL's. He would not expect the claims handling aspect to command a lower return as it is part of the overall package. While we agree with Mr Bezant that a claims handling service on its own may be expected to be less profitable than underwriting, we agree with Mr Gaysford that for D&G it is the combination of underwriting and claims handling that enables D&G to make the profits it does.
128. Mr Bezant contrasts Mr Gaysford's comments with the extensive assumptions that Mr Gaysford makes in his method of calculating DISL's profits. He also notes that Schroder Salomon Smith Barney ("SSSB") in a valuation of DSG in 2001 specifically identified D&G as the most comparable insurance company to DISL, in preference to both another specialist insurance provider and to general insurers. The comparison made by SSSB related specifically to the cost of equity for DISL rather than its profitability. The return on equity for both companies was approximately 25% compared to other insurers' returns on equity in the range of 8% to 10%. SSSB valued DISL broadly in line with D&G. Mr Gaysford points out that their valuation was on the basis that there were no material intra-group charges or other unusual factors affecting the profits. In other words, he considers that in doing so they assumed away the problem with which we are dealing. SSSB also produced an alternative valuation of DISL based on its net assets and the value of the remainder of the contract which provided a significantly lower valuation. He considers that the SSSB report supports his cost of equity estimate (as opposed to the return on equity which depends on the nature of the business and its market position). Mr Bezant notes that D&G has made a high return on equity for an extended period, which appears to be inconsistent with Mr Gaysford's assumptions. Mr Bezant considers that the functions carried out by D&G in addition to insurance were broadly routine in nature and could be replicated by competitors such that D&G's observed return on equity should not persist over time. D&G as an extended warranty insurer is a better comparable for DISL than the insurance market generally. Mr Gaysford believes that D&G's brand, reputation, experience and organisational capabilities as the largest independent provider of "off-the-shelf" extended warranty products are consistent with its being able to earn returns in excess of the normal level. These capabilities are different from those required by a company like DISL underwriting extended warranty business of a major retailer. Mr Bezant considers that Mr Gaysford's comments do not render D&G less comparable than general insurers. Accordingly Mr Bezant and Mr Gaysford substantially disagreed over the comparability of D&G.
129. We are impressed by, and accept, Mr Gaysford's arguments about the significant differences between D&G and DISL, the most significant difference being in terms of bargaining power. D&G as a large public company dealt with manufacturers who did not have their own extended warranty business without a point of sale advantage and with smaller retailers. While its customers may be large in themselves their bargaining position is not solely related to their size. D&G was offering a complete service that the customers could not have obtained for themselves in the market. Customers would place considerable value on D&G's ability to offer a complete package of services and on its scale, experience and brand. DISL, on the other hand, was contracting with the largest electrical retailer in the UK which had the point of sale advantage, which D&G and its customers did not. The Appellant Group had the scale and expertise that would make it economic to undertake its own assessment of risk and to identify alternative providers and switching providers would be economic. It had a strong brand that had no need for an external brand to support its warranties. A credible option would have been for the Appellant Group to have hired its own actuarial staff and provided the service in-house. DISL was entirely dependent on the Appellant Group for its business. Mr Gaysford did not consider that there was any method of adjusting D&G's profits to eliminate this difference in bargaining power.
130. Mr Goy also pointed to the very different level of expenses in D&G, indicating that it had to work much harder than DISL to make a profit.
131. It would, at least in theory, be possible to adjust for the difference in capital, which is far higher for DISL, and the difference in the scope of the activities, which are much wider for D&G, and more easily for other factors, such as D&G showing the gross premiums before commission in its accounts, and DISL the net premium. But we accept Mr Gaysford's argument that it is not truly comparable because of the bargaining position of the Appellant Group with a notionally independent DISL which is dependent on the Appellant Group for all its business, compared to that of many smaller retailers with D&G in which the power is in the hands of D&G as a large public company, and of its manufacturer customers who did not have the point of sale advantage of the Appellant Group. Mr Gaysford did not consider that it would be possible to adjust for this difference. Even though D&G and its manufacturer customers did not, like the Appellant Group, have the point of sale advantage, there was no evidence from which we could, and we see no way to, adjust for the difference in bargaining power. For this reason, although no doubt D&G is the closest insurance company that there is to DISL, we do not consider that D&G is a suitable comparable to DISL or that there is a method of adjusting for the differences.

Cornhill in relation to these transactions

132. Since DISL took 95% of the risk and Cornhill 5% one might expect Cornhill to be a good comparable to DISL. We have endeavoured in paragraph 20 above to take account of the different amounts of capital required by each of them, on the basis of which we found that DISL earned a return on capital of 10.64^[3][4] times that of Cornhill. Mr Beaton said that a fronting insurer expects to do proportionately better than the reinsurer, which is the opposite of the case here.
133. Mr Peacock contends that since premiums are set by Cornhill the Revenue cannot argue that there was a benefit to DISL in 1993 and at the same time say that Cornhill was a good comparator in 1997. It is not clear that the terms applying to 5% of the risk with the remainder being reinsured, should be applied to 95% of the risk without any reinsurance. Even if Cornhill is a good comparator it was allowed to earn profits from 1986 until 1993 when the 12% cap was introduced, and so DISL should be treated in the same way and a 12% cap applied only after 7 years because Cornhill had 7 good years before the cap was introduced.
134. Mr Goy contends that the effect of the 1993 arrangements with Cornhill is that Cornhill are effectively left with only the ceding commission of 1.5% which represents a normal return on the capital of 20 needed to support a premium of 100.
135. We consider that potentially Cornhill provides a suitable comparator for the Cornhill Period because as a quota share reinsurer one would expect DISL to be in the same position as Cornhill. However, Cornhill clearly had a better bargaining position than DISL and as in the case of the other possible comparables an adjustment would be impossible to make on any reasonably accurate basis, and so the conditions for applying the CUP Method are not met (paragraph 2.7 of the Transfer Pricing Guidelines). Our conclusion is that the only assistance that can be obtained from Cornhill's position is that it puts a ceiling on what DISL could expect to make, when the different solvency requirements are taken into account. We attempt to demonstrate how this might be done in our conclusions below.
136. Our calculation of Cornhill's return on capital of [**] is an example of the operation of the approach favoured by Mr Gaysford (see below) that an arm's length profit for DISL would be a market return on the capital necessary to support its underwriting.
137. We see no merit in Mr Peacock's suggestions that Cornhill might be a comparable if its 12% profit retention were taken into account but only after seven years, which is the period from the start until 1993 when the 12% cap was adopted for Cornhill on the basis that Cornhill as comparator had additional profits for this period which was not taken away until after 6 years. It is not the 12% on its own that is relevant but the combination of DISL's 15% profit retention and Cornhill's 12%, resulting in a certain pure underwriting loss for Cornhill. The fact that this was only adopted after seven years ignores the fact that it was in operation in the year in question.
138. We do not consider the fact that at the start of the ASL period Cornhill required DISL to follow the premiums previously set by Cornhill as much assistance in determining whether DISL's profits were at arm's length. Following the 1993 arrangements Cornhill's only interest in the amount of the premium was that the interest should be greater than the fixed underwriting loss (on the basis that the loss ratio would not exceed 80%) and that there was enough in the trust fund adequately to cover expected claims. Since the interest rate was fixed between limits the only variable was the timing and amount of claims. This is not a normal case of an insurance company fixing its premiums. Its return was mainly the ceding commission of 1.425% which is a fixed percentage of the premiums. Cornhill was protecting its position by requiring its premiums to be followed.

Arm's length price

139. Mr Peacock contends that there are sufficient comparables not to reach the point of applying a profit split method. Mr Goy contends that there are no comparables and so one must apply a profit split method for which he relies on Mr Gaysford's evidence, except that in the end he proposed an additional 15% initial commission to MCASL plus a profit commission of the whole of DISL's return above a normal return on its capital.
140. We have to consider scheme year 1996-97 in the Cornhill Period which affects the results of DISL's accounting periods up to 2001-02, of which the arms length price for accounting periods up to 1988-89 is governed by s 770 and later ones by Sch 28AA. For the ASL Period starting in 1997-98 the arm's length price is governed for the first two years by s 770 and for later years by Sch 28AA.

The Cornhill Period

141. We concluded above that the most assistance that the terms that Cornhill agreed in 1993 can provide is a ceiling on the calculation of the profits that DISL can be expected to make when adjusted for the different solvency requirements. Some of the differences can be adjusted for but not the different bargaining position of the parties. Accordingly, we cannot use the CUP Method and we need to apply another method. However, it is relevant that the result negotiated with Cornhill was that it was certain to make an underwriting loss that was offset by a capped investment return, leaving it mainly with its ceding commission, which can be seen as a return on the capital allocated to the solvency margin required for writing this business. To quote para 3.6 of the Transfer Pricing Guidelines, measuring the return on capital "reflects what independent enterprises reasonably would have done if faced with the same circumstances." Accordingly we do not treat this period separately and the reasons that lead to our decision that DSG should take a further profit commission are equally applicable to the Cornhill Period. However, this must be viewed in 1993 at a time when there was less information available about past loss ratios and this factor needs to be taken into account.

The ASL Period

142. Mr Bawcutt proposed a method of calculating the arm's length return on 20% pa of the capital necessary to meet the solvency margin and treating any profit above this as excessive. This was criticised by Mr Peacock as being based on DISL's accounts that included the continuing Cornhill business and on other grounds. The basic idea of a return on the required capital is the same as Mr Gaysford's which in our view takes more relevant factors into account. Accordingly we shall concentrate on Mr Gaysford's evidence.
143. Mr Gaysford explained the economics of a commercial transaction under which each party will look to gaining a share of the benefits arising from the transaction. While Mr Peacock contended that Mr Gaysford's evidence was economic analysis divorced from reality we consider that it is necessary to take economic analysis into account in view of the Transfer Pricing Guidelines's requirement of using an "economically valid basis" in applying the profit split method. (That a profit split based on return on capital is found in the real world is demonstrated by the terms agreed with Cornhill.) Mr Gaysford said that a party's ability to gain a share depends on the total value generated by the transaction and its bargaining power. For many activities in the economy the value created is rarely beyond the normal rate of return. Competition (where it exists) will eventually force higher returns down to the normal level. The ability to make greater profits (referred to below as economic profits) arises where competition is limited, for example monopoly situations, legal protection from competitors through intellectual property rights, and the existence of a strong competitive advantage preventing competitors from competing equally or as effectively. Where economic profits do arise these are distributed between the parties according to the ability of each party to protect itself from normal competitive forces and hence according to the party's bargaining power. Here economic profits arise because of the point of sale advantage to the Appellant Group, not only as the advantage enjoyed by the retailer but also from their position as the largest retailer of domestic electrical goods in the UK. Its network of repair services was unlikely to give it a major source of competitive advantage since there was a competitive market for repair services. Similarly, claims administration was similar to the activity involved in other types of insurance activity and could be contracted out, which was unlikely to create significant economic profits. Past claims data could represent a source of competitive advantage as competitors without this data were disadvantaged, as the Competition Commission concluded. Here the Appellant Group possessed greater point of sale advantages over other retailers of domestic electrical goods and had long-term data about claims. This eliminated the bargaining power DISL possessed by reason of its possession of the same data. He instanced as an analogy contracts between retailers and store card providers, in which the Competition Commission had found the bargaining position was enjoyed by the retailer with the store card provider earning normal rates of return (although the magnitude of the rate will vary from industry to industry).
144. Mr Gaysford's view (which we accept subject to the point made in the following paragraph) of the relative bargaining position of the parties (assuming that DISL were independent) was that the Appellant Group had significant bargaining power; in fact it had almost all the long term bargaining power. This arises principally from the point of sale advantage, but also from the low cost of putting together a deal with an insurer; the low cost of switching insurers at a later date; access to data on past claims; and as the largest retailer with a strong brand it had no need for an external brand. By contrast, the insurer has to provide actuarial know-how, which is not scarce and could be hired by the retailer, and a financially sound business, both of which could be provided by other insurance businesses. All these advantages applied to the Appellant Group who as the largest electrical retailer in the UK with the point of sale advantage were in a particularly strong negotiating position. DISL was dependent for all of its business on the Appellant Group. Mr Bezant disagreed and considered that since there were no competitors in the market DISL enjoyed a strong bargaining position. This was the principal difference between him and Mr Gaysford. Mr Rust also disagreed, saying that "DISL was the only game in town as an insurer," which we accept only in relation to the short term, dealt with in the next paragraph. So far as the long term is concerned the market was never tested and we see no reason why another company similar to DISL could not have been created even if there were no other suitable insurers in the market and even though it would have required [substantial capital]
145. Both Mr Rust and Mr Bezant pointed to one advantage that DISL had (which we also accept), that was taken into account only to a limited extent by Mr Gaysford. When the announcement of the IPT change was made on 26 November 1996 the Appellant Group had to do something quickly to reduce the cost. Moving to service contracts saved it [many million pounds] pa compared to continuing with insured warranties and so any delay in making the change beyond 1 April 1997 was expensive. The Appellant Group needed (the assumed independent) DISL's systems for handling the existing business, its existing reinsurance licence (which could easily be changed to an insurance licence in less than two weeks), and the fact that its knowledge of the business did not require it to spend time getting to grips with the claims data. We consider that an independent party would have made the most of this position. But it was a short term advantage. If it had been overplayed by giving the Appellant Group a small share of the saving of moving to service contracts, the Appellant Group might have had to accept this in the short term but it would have made it likely that they would switch insurers as soon as it was able. An independent insurer would have taken into account that if it behaved reasonably it was more likely to have a continuing business for which it was dependent entirely on the Appellant Group, and so it would not press the timing advantage to the limit. We did not have any evidence about how long such an advantage would last but as an analogy the original Appellant Group-Cornhill-DISL arrangement was terminable on six months' notice, which may be the minimum period for which independent parties would set up such an operation; this seems to indicate that the advantage would have lasted six months plus the time taken to identify another insurer, before which it would not have been prudent for the Appellant Group to have given the six months' notice to terminate its arrangement with DISL. Another indicator of the importance to the Appellant Group of the time pressure was that Willis only committed itself to the proposal on 2 April 1997, by which time the IPT change was already in force, and that the Appellant Group agreed to pay Willis a £250K fee if they completed the arrangements by 23 April 1997 with a reduction of £10,000 per days's delay (including weekends), and also contributed £160K to their costs. ASL's role would be easier to create compared to DISL's which required a larger capital and an insurance licence and systems capable of handling the work (though not the administration of claims). We consider that in the light of our decision generally this is a matter on which we should leave the parties to see if they can agree a suitable reward for DISL in respect of this aspect.
146. On the basis that there were no comparables Mr Gaysford sets out a profit split approach that depends on the comparison of the actual profits with the level of profits implied by a normal rate of return on investors' capital, this being the level of profit expected from a competitive tendering process. Mr Bezant agreed with Mr Gaysford that "the comparison of returns on capital with a benchmark cost of capital is commonplace in many forms of contract negotiation. And that the estimation of residual profits over and above a normal rate of return, and allocation of those profits in line with the bargaining power of two contracting parties is a standard application of economic theory." However, Mr Bezant questioned whether this was a proper application of the OECD Guidelines, particularly as Mr Gaysford had not produced any functional analysis of the facts and circumstances pertaining to the DISL agreement. Mr Bezant also agreed that profit split methods are referred to in the hierarchy of methods identified by the OECD guidelines, but made the point that these methods tend to involve the allocation of residual profit margins, not the use of return on capital methods. Mr Gaysford considered that his method was in accordance with the OECD Guidelines, relying on para 3.5 quoted above by saying that the words "profit" or "return" can be defined in a number of ways and the correct application of these methods would be to start with the definition of profit or return as it is usually defined in the industry in question. Both Mr Bezant and Mr Gaysford agree that the calculation of return on capital is a widely accepted and widely used measure of the profitability of a business, and that the use or a cost of capital based on industry benchmarks is a standard approach to assessing a normal level of profit. Mr Bezant questions whether Mr Gaysford was in a position to conclude that DISL would make a normal level of return as an insurer in the absence of a functional analysis; Mr Gaysford considers that his report contains the appropriate analysis. Mr Bezant agreed that since both D&G and DISL had little or no debt return on equity is a suitable starting point to measure their financial performance. Accordingly there was a large measure of agreement between the two experts with their main disagreement being whether Mr Gaysford's method was in accordance with the OECD Guidelines.
147. The most common approach to calculating a normal return on capital is based on the capital asset pricing model (CAPM) which is used in the commercial world and by economic regulators and competition authorities in the UK. Since DISL has no debt funding the appropriate measure is the opportunity cost of equity, the rate of return that an investor would require in order to invest in share capital of a comparable business. The CAPM formula for this is:

R_E = R_F + ß_i(R_M - R_F)

Where: R_E is the cost of equity;

R_F is the rate of return on a risk-free investment;

R_M is the rate of return on investing in the equity market as a whole;

(R_M - R_F) represents the equity risk premium--the difference between average returns on equity investments and the risk-free rate;

ß_i is a measure of the riskiness of company i''s activities, relative to the market as a whole.

(1) For R_F Mr Gaysford took the annual average nominal par yield on 10-year British Government securities between 1997 and 2004. In doing so he used hindsight which we consider should not be used.

(2) For the equity risk premium Mr Gaysford used estimates from the literature of: low 4.0%, middle 4.5% and high 5.0%. He used the middle of the range in his calculations although he pointed out that the effect of the profit commission was to reduce DISL's risk because any increase in claims has no effect on DISL until claims reach 80% of premiums.

(3) For beta he looked at three possible approaches: that for all non-life insurance companies (0.71), for selected insurers undertaking reinsurance as their main activity (0.63), and D&G (0.57). He used 0.6 in his calculations.

148. Mr Bezant agrees that the CAPM is a widely-used and reasonable approach to estimating the cost of equity. He agrees with the individual components of a CAPM estimate of the cost of equity for DISL. He favours the upper half of the 4% to 5% range. He agrees that the beta for D&G may be an appropriate estimate of a beta for DISL, but considers that the use of betas from other insurers has only limited application as the cost of capital assessment for DISL should reflect the specific circumstances of the extended warranty insurance market rather than an insurance industry average. Again there was a large measure of agreement between the two experts with their disagreement being restricted to the precise figures to be used in applying the CAPM.
149. Mr Gaysford then considered what capital an independent insurer would have on which the rate of return should be applied, on the basis that in competitive tendering an insurer would not be able to command a return on an excess amount of capital. He used two methods to achieve this, although he pointed out that it was not unusual to find competitive insurance businesses that hold levels of equity in excess of the regulatory requirement. First a "bottom-up approach," which was his preferred approach, under which in effect he assumed that any profits in excess of the capital needed to meet solvency criteria were paid out as dividends. For this he used 15%, 20% and 25% of premiums. We consider that since Cornhill contractually required DISL to have a solvency margin of 20% of the greater of gross premiums written or gross insurance liabilities, this should be used since in both 1993 and 1997 Cornhill was in a position to require this. We considered whether some allowance should also be made for the fact that competitive insurance companies do have equity in excess of the solvency requirement and concluded that the trust arrangements meant that additional equity was not necessary, as is demonstrated by the fact that Cornhill were satisfied with the contractual limits that it imposed, which were in excess of the Isle of Man regulator's requirements. Secondly, he considered a "top-down approach" under which year by year he excluded retained profits that would not have arisen under arm's length pricing. In doing so he took the excess profits he calculated from 1995 to 2004 and then adjusted the retained profits in 1995 by the same percentage. Since this necessarily uses hindsight we do not approve this method.
150. Mr Gaysford made calculations of DISL's excess profits using the above method. Mr Bezant and Mr Gaysford agree that each other's calculations are mathematically correct. Mr Bezant considers that Mr Gaysford's approach based on reported profit in DISL's accounts, for example reflecting the accounting treatment of unearned premiums, actual claims experience and revisions to future expected claims in relation to the long-term nature of the policies, may be inappropriate under the OECD Guidelines. Mr Bezant points out that Mr Gaysford's calculations are ex post calculations of an expected return on equity rather than an ex ante assessment of the terms of an arm's length agreement, as required by the OECD Guidelines. This is relevant given the need to price insurance risk over an extended period. Mr Gaysford considers that the existence of the profit commission is a method of clawing back ex post profits and is in accordance with the OECD Guidelines. As before, there is a large measure of agreement between the two experts, the disagreement being, as before, whether Mr Gaysford's method is in accordance with the OECD Guidelines.
151. We do not set out Mr Gaysford's calculations since we agree with Mr Bezant's objections that they used hindsight, and also because he was working from DISL's accounts which also included the continuing Cornhill business and the run-off of the Cornhill period business about which Mr Gaysford assumed had similar profitability, which is not the case, as is demonstrated by the agreement concerning the loss ratio of the Irish business.
152. Mr Gaysford considered that the profit commission should be used as a basis for adjustment. Mr Goy, however, contended in the end for an additional 15% up-front commission to MCSAL and the balance by way of profit commission, also to MCSAL in his preferred contention, or alternatively to DSG. We assume that Mr Goy chose 15% on the basis that this is the proportion of the premium expected to be retained by DISL as a result of the profit commission. We do not agree with this approach as an additional up-front commission changes DISL's risk and the amount of investment income. More importantly, a profit commission is what was agreed with Cornhill in circumstances where the Appellant Group wanted additional up-front commission in 1990, and it therefore reflects what independent parties in fact agreed. Nor is an additional up-front commission indicated by Mr Gaysford's analysis, and it is not a change that can be made in any case under the s 770 regime. Because of the uncertainties of insurance we consider that any adjustment should be by way of profit commission when the results are known. This seems to us to be compatible with both s 770 and Sch 28AA.
153. The principal disagreement between the two experts was in relation to whether Mr Gaysford's method was in accordance with the OECD Guidelines. In terms of the Guidelines we consider that Mr Gaysford is using a profit split method based on the total profit with a mixture of contribution analysis and residual analysis approach. This looks to the functions of the parties, DISL providing insurance in circumstances where Cornhill, an independent party, had previously agreed a return which to a large degree represented its capital employed; and the Appellant Group providing the whole business by virtue of its point of sale advantage in circumstances where it has particularly strong bargaining power. External data in the form of the cost of equity is used to assess the value of DISL's contribution rather than to determine directly the division of profit. It is a mixture of the contribution analysis and residual analysis in that no first stage return is allocated to DSG, which makes sense here since because of their bargaining position (a factor specifically referred to in para 3.19 of the Guidelines) all the residual profit will be allocated to them. The result, as demonstrated by the bargain made with Cornhill, replicates the outcome of bargaining between independent enterprises in the free market (para 3.21 of the Guidelines) with the only different result being the rate of return on account of the weaker bargaining position of DISL. We therefore consider that Mr Gaysford's approach is in principle in accordance with the OECD Guidelines. The only factor used by Mr Gaysford which was not in accordance with the Guidelines was that he used hindsight.
154. Our conclusion based on comparing DISL's position with Cornhill's in the Cornhill Period and the continuation of the payment of premiums to DISL at a similar rate in the ASL Period, is that the actual provision is not at arm's length because if DISL had been independent it would have had to pay something to DSG. Accordingly there should be a formula under which additional profit would be allocated to DSG from DISL's profits that will leave DISL with a profit calculated on the lines of Mr Gaysford's method, but on the basis that it should be applied without using hindsight, although the difference is not that great when a formula agreed at the start sets out the method of calculating the profit commission in retrospect, and the formula can use values determined in the future, such as the current risk-free rate of return that can be determined from time to time (though not going as far as Mr Gaysford's average for 1997 to 2004). Mr Gaysford's "bottom-up" approach would be used for the calculation of the normal rate of return. We express no view of the figure for beta. Rather than trying to write such a formula ourselves we believe that it would be better to adjourn so as to allow the parties to see if they can agree it, and if they cannot the appeal can be restored for us to do so. We would, however, make the following decisions in principle.

(1) The formula would have no effect on (a) the results in later years of business written in scheme years earlier than scheme year 1996-97, or (b) the continuing Cornhill Business (together "the excluded businesses"). The margin on the excluded businesses must be capable of being calculated separately because for (a) separate loss ratios are maintained, and for (b) there is no profit commission payable. Under s 770 any contractual commitments made with Cornhill and the trust arrangements would be taken into account; these could be adjusted under Sch 28AA to the extent they arose from a non-arm's length provision between DSG and DISL. The calculation of DISL's liabilities would include liabilities of the excluded businesses. DISL's expenses would be unaffected.

(2) The formula would therefore affect only the remaining businesses ("the businesses in issue"). The capital required by the agreement with Cornhill (until 2002, the end of the liabilities of policies written in scheme year 1996-97) or the Isle of Man regulator, whichever is from time to time the higher, needed to support all the businesses would be calculated. Part of the capital would be allocated to the businesses in issue on the basis of their premiums and liabilities. A rate of return on that capital would be calculated using Mr Gaysford's method.

(3) The additional profit commission would be payable at the same time as the existing profit commission. Under s 770 payment would be subject to any restrictions, such as the trust arrangements or contractual commitments to Cornhill that might necessitate postponement of payment, in which case interest would run to compensate DSG for the fact that DISL would be earning interest on the amount due. Under Sch 28AA in so far as any such restrictions were the result of non-arm's length provision as between DSG and DISL they could be adjusted and to that extent payment made on the due date for payment of profit commission. Under Sch 28AA the formula need not assume that ASL was the provider of the warranties: ASL could have been supplanted by an independent DISL which would have supplied the same comfort to DSG's customers as the interest in ASL held by Willis provided. However it seems likely that if ASL were excised, an independent DISL would require an additional return comparable to that received by ASL for the costs of its independent functions.

(4) In respect of the Cornhill Period the arrangements with Cornhill provide a ceiling on DISL's return on capital. Based on our illustrative calculations in paragraph 20 above, which we appreciate may not be correct, Cornhill made a return on the capital required of [**]%. One factor in this calculation is that it includes a loss ratio of [X]%, which was envisaged in 1997 but this needs to be adjusted to what was expected in 1993 if that is different and to allow for the fact that there was in 1993 less information available about past loss rations than there was in 1997 (paragraph 141 above).

(5) DISL's short-term bargaining position in 1997 should be taken into account in giving DISL a suitably higher return for a period, and we leave open the possibility that this is not a return calculated by reference to a notional amount of capital.

(6) Although the amount of profit commission is not known until 5½ years after the scheme year, MCSAL and later DSG recognised profit in their separate accounts (and accordingly paid tax on it) after three years, which we infer is good accounting practice and we consider that the same should be applied in giving tax effect to the formula. (While paying tax on a higher profit commission before it is received might give cash flow problems this is a retrospective adjustment to profits for the years 1997 to 2004 and so this is not a real problem for the past, but might be a problem if applied in the future, although we understand that service contracts came to an end on 22 March 2005, for which the final profit commission will not be paid until 2010.) The formula should contain a deficit clause enabling a deficit for a particular scheme year to be clawed back over three to five years, which Mr Bawcutt said was not uncommon; this will not be relevant to the years under appeal but could be in later years.

(7) We appreciate that the effect of such a formula is that there will be two profit commissions, the existing one payable to MCSAL and the other to DSG as a result of our decision. Our understanding is that the former cannot be adjusted in the years subject to this appeal even if it results in a provision not at arm's length.

155. Accordingly we adjourn and direct that either party can require by notice to the Tribunal that the hearing be restored.
156. Subject to that, we answer the questions for determination as follows:

(1) Raised in the context of s 770 and s 773 of the Taxes Act 1988 transactions between associated persons:

(a) In respect of each Appellant, whether business facilities of whatever kind have been given by that Appellant to DISL. Yes, by DSG only.

(b) Whether at the time such facilities were given DISL was a body of persons over whom that Appellant had control, or that Appellant was a body of person over whom DISL had control, or both DISL and that Appellant were bodies of person over whom the same person or persons had control. Yes as to DSG.

(c) Whether the business facilities were given at a price ("the actual price") which was less than the price at which they would have been expected to be given if the parties to the transaction had been independent persons dealing at arm's length ("the arm's length price"). Yes.

(d) What adjustment, if any, should be made in computing for tax purposes the income, profits or losses of the Appellants, so that the like consequences shall ensure as would have ensured if the business facilities had been given for the arm's length price? Adjourned.

(2) In the context of Sch 28AA:

(a) In respect of each Appellant, whether provision ("the actual provision") had been made or imposed as between that Appellant and DISL by means of a transaction or series of transactions. Yes as to DSG only.

(b) Whether at the time of making or imposing the actual provision that Appellant was directly or indirectly participating in the management, control or capital of DISL, or the same person or persons was or were directly or indirectly participating in the management, control or capital of that Appellant and DISL. Yes.

(c) Whether the actual provision differs from the provision ("the arm's length provision") which would have been made between independent enterprises. Yes.

(d) Whether the actual provision confers a potential advantage in relation to United Kingdom taxation on that Appellant. Yes.

(e) What adjustment, if any, should be made in computing for tax purposes that Appellant's profits and losses if the arm's length provision had been made or imposed instead of the actual provision? Adjourned.

157. We are conscious that this decision may contain figures that are commercially sensitive. In case this is so, we direct that the Appellants inform us within 10 days of the date of release of this decision and we shall consider any proposals for modification of the version of this decision that is released to the public. [This report of the decision takes into account the representations received from the parties.]

Postscript

158. As this is the first transfer pricing case to have come to the Tribunal we repeat our invitation for anyone concerned with the case to write to the Tribunal with suggestions about how the Tribunal might make changes to the procedure for such appeals. If we might put forward one suggestion of our own, without in any way intending any criticism of the parties, we would have found it helpful if the table in paragraph 18 above had been agreed (or any disagreement stated) and given to us at the beginning. The terms on which Cornhill operated are quite complicated and were still being argued about in two submissions following the hearing.
JOHN F. AVERY JONES
CHARLES HELLIER SPECIAL COMMISSIONERS RELEASE DATE: 31 March 2009

SC 3056, 3057, 3060/07

Note 1   With a loss ratio of 80% DISL initially retains 20% of premiums then pays 50% of the difference between 10% and 20% of premiums (=5%) in profit commission, leaving it with 15%.    [Back]

Note 2   Note. Since the original decision was issued to the parties the Respondents have pointed out an arithmetic error in this calculation, with which we agree, that means that this figure should be 16.05.    [Back]

Note 3   This assumes a loss ratio of [close to X%]% in the insurance column and [a slightly lower percentage] in the service agreement column, the latter being the same after taking the VAT recovery into account (including the claims costs of £[*]).    [Back]

Note 4   As explained in footnote 2 this should be 16.05.    [Back]