“Motor vehicles would have been bought from manufacturers, and not from finance companies or motor dealers.” 44. The source of KPMG’s information appears to have been Ms Eastwood, about a year before she swore her affidavit. The affidavit itself and Ms Eastwood’s witness statement made no reference to who supplied the motor vehicles to Brammer. I have mentioned above how Ms Eastwood dealt with the question in her oral evidence. I am left with the impression that Ms Eastwood’s recollection as to whether vehicles were supplied directly by the manufacturers or through franchised dealers is not reliable. I am not satisfied on the balance of probabilities that Brammer purchased vehicles in the Period through dealers rather than directly from the manufacturers. 45. On the basis of that finding, Brammer has not satisfied me that it accounted for output tax on bonuses received from manufacturers. If vehicles were supplied directly by manufacturers then it is likely bonuses were treated as a discount and no output tax would have been accounted for on such bonuses. That is sufficient to determine the appeal against Brammer. However, I shall go on to consider the other factual issues. 46. If Brammer did purchase fleet vehicles through dealers, the question is then whether Brammer followed the 1987 Policy and accounted for output tax on manufacturer bonuses. Brammer says that as a diligent and compliant taxpayer it would have applied the 1987 Policy correctly. Further, the 1987 Policy advised manufacturers when negotiating incentives to clearly explain to customers that tax would be due from them on the amounts paid. Mr Gibbon says that Brammer was a diligent taxpayer and I should assume that it applied the 1987 Policy. He submitted that there was no evidence that Brammer was non-compliant in relation to VAT. 47. There was evidence in the form of an HMRC summary document as to how certain manufacturers, including Vauxhall and Ford, treated non-dealer fleet bonus payments in the years 1973 to 1996. This was produced by HMRC’s Motor Trade Unit of Expertise and was dated2 November 2004 . It appears from the summary and I accept that Ford and Vauxhall both treated bonuses as outside the scope of VAT until 1988, and as standard rated thereafter. I accept that was the case, and I infer that Vauxhall from1 July 1988 to 1996 and Ford from sometime in 1988 to 1996 would have expected to recover input tax on non-dealer fleet bonus payments where vehicles were supplied by a dealer. In order to do so they would have required VAT invoices identifying VAT on the bonus payments. It is not clear on the evidence whether manufacturers might have used “self-billed” invoices for this purpose or required customers such as Brammer to produce VAT invoices. The 1987 Policy itself refers to the possibility of self-billing and appears to discourage that practice as follows: “… we are aware of case where manufacturers have, of their own volition, regarded these payments as consideration for taxable supplies and have consequently self-billed the VAT. This represents a possible loss to the revenue if no output tax is accounted for by the customer.” 48. Mr Gibbon relied on the acknowledged dictatorial relationship between manufacturers and dealers, and submitted that the same relationship would have existed between manufacturers and fleet purchasers. I do not accept that is the case and I am not prepared to make such an inference. Fleet purchasers would be in a different bargaining position to dealerships. However, even without the so-called dictatorial relationship, manufacturers would have required a VAT invoice in relation to bonus payments. On balance, I am just about satisfied that it is likely there would have been discussions between the manufacturers and high volume purchasers such as Brammer as to the need for a VAT invoice and the requirement for Brammer to account for output tax on bonus payments. In all the circumstances I am satisfied that it is likely a compliant trader would have accounted for the VAT on bonus payments. 49. The question then is whether Brammer in those circumstances would have accounted for output tax on bonus payments received. In other words, was it a compliant trader? Ms Free submitted that the burden was on Brammer to show that it was a compliant trader, and not on HMRC to show that it was not a compliant trader. It is difficult to know how either party in a case such as this could go about showing that Brammer was generally compliant or non-compliant in the Period. There is no direct evidence and Ms Eastwood accepted that she could not give any evidence as to Brammer’s VAT function. However, knowing that HMRC had a specific policy for manufacturer bonuses and given that manufacturers would be likely to have discussions with high volume purchasers, I am satisfied on the balance of probabilities that Brammer would have accounted for output tax on bonus payments. Put another way, it is unlikely that a large business such as Brammer would not have declared the output tax. 50. Overall, I am satisfied on the balance of probabilities that if Brammer had purchased from dealers throughout the Period it would have been aware of the requirement to account for VAT and would have done so. 51. The next question, if I had been satisfied that Brammer accounted for VAT on bonuses in accordance with the 1987 Policy, is whether I am satisfied as to the amount of VAT, or at least the minimum amount of VAT which was overpaid. 52. Brammer’s case and its calculations of the amounts said to have been overpaid are not based on the number and actual cost of non-commercial vehicles purchased by Brammer in the Period. Instead, its starting point is the value of vehicles purchased in the Period. Brammer uses data from management accounts to identify additions to fixed assets (motor vehicles) in the period 1989 to 1994 and in 1996. The 1988 and 1995 figures are based on averages from that data. The data for 1996 was referred to in correspondence from KPMG to HMRC in 2009, but is apparently no longer available. Mr Gibbon therefore submits that if HMRC do not accept the figure for 1996 then the same averaging exercise could be carried out for that year. 53. HMRC do not really dispute the total figures for additions to fixed assets in those years, but do say that the figures used by Brammer will include VAT where input tax has been blocked, that is in relation to non-commercial vehicles. They also say that Brammer’s case relies on an apportionment of the value of vehicles purchased between commercial and non-commercial vehicles. Brammer’s apportionment is based on the ratio of vans to cars in April 2000-01. Ms Free submitted that involves too broad a brush. It uses the ratio from April 2001 whereas the Period is from 1988 to 1996. It assumes that vans and cars are similarly priced. Ms Free submitted that the cost of a Ford Granada was likely to be much more than the cost of a Ford van. It also assumes that none of the cars purchased were pool cars where input tax was recovered. 54. There is some force in Ms Free submission on this point. The question is whether I consider that these are minor issues at the margin, such that the approach of Henderson J in Prudential Assurance Co can be applied, consistent with the principle of effectiveness. On balance, I do not consider that these are minor issues. It seems to me that Brammer’s assumptions for the purpose of calculating the amount of VAT overpaid give rise to a significant margin for error, especially applying a ratio from April 2001 to the Period which commenced in 1988. The calculations may well be the best assessment that can be made on the material available, but I cannot say that they result in a figure that is likely to be the amount of VAT overpaid by Brammer, or that is likely to be the minimum sum that has been overpaid. 55. HMRC also criticise other assumptions made by Brammer in calculating the amount of VAT which it is said was overpaid. I do not need to describe the precise basis of Brammer’s calculation other than in relation to two estimates that are in issue. Brammer takes the value of non-commercial vehicles it says were purchased and grosses this up for a dealer discount of 8.11% off the gross list price. This is said to be the discount a dealer would have given to Brammer. VAT is then deducted at the rate in force as well as car tax in certain years to give a net price. A manufacturer’s bonus rate of 15% is then applied to give the bonus payment received and the output tax included in that bonus is calculated. 56. HMRC say there is no reliable evidence as to the 8.11% dealer discount or the 15% manufacturer’s bonus. 57. The 8.11% dealer discount was identified by KPMG and apparently derives from a Competition Commission Report in March 2000 which identifies a weighted average fleet discount given by leading manufacturers between 1994 and 1998. There is a table in that report (Table 7.3) which identifies in what I shall call Part A, the percentage discounts from 4 manufacturers (described as suppliers) and dealers to fleet purchasers. In what I shall call Part B, it identifies the percentage discounts from 6 manufacturers only to fleet purchasers. The identity of the manufacturers has been redacted by the Competition Commission, presumably because of commercial sensitivity. 58. Mr Gibbon carried out an exercise to calculate the average manufacturer and dealer discount in Part A, which was 23.14%, and the average manufacturer only discount in Part B, which was 14.65%. The difference gives a figure 8.49% for the average dealer discount, but Mr Gibbon said that it was “broadly supportive” of KPMG’s figure of 8.11%. 59. The basis of the manufacturer bonus figure of 15% is a statement of KPMG in a letter dated30 March 2009 to HMRC that: “…it is recognised throughout the fleet industry that fleet bonus support has stayed at a consistent percentage … throughout [the Period]…The percentage has been calculated by using known values of bonus payments from the main suppliers … and we believe an average of 15% is fair and reasonable and consistent with HMRC’s views” 60. There is no direct evidence before me as to the rate of manufacturer bonus paid to fleet customers in the Period. Brammer relies on what is said by KPMG in that letter, and also on the Competition Commission Report which suggests that the average manufacturer discount in the period 1994 to 1998 was 14.65% with a range of 9.7% to 19.3% 61. Ms Free submitted that taking averages in this way is too speculative. What would be required would be evidence that specific manufacturers gave specific dealer discounts and paid specific bonuses in specific years. Some measure of averaging may be acceptable, but not to the extent Brammer seeks to employ. 62. I accept Ms Free’s submission that Brammer’s exercise is too speculative. The evidence of manufacturer discounts in the Competition Commission Report refers only to 6 unidentified manufacturers. In other sections of the report 10 manufacturers are identified. I do not know whether the figures in Table B relate to Ford, Vauxhall or Peugeot which are the vehicles Ms Eastwood says were purchased at various times in the Period. The range of discounts is wide. In theory, I could take lowest discount of 9.7%. However, 2 of the manufacturers in Table B have the notation n/a in some of the years. That could mean that information was not available, or it could mean that it was not applicable because that manufacturer did not pay a bonus in that year. In my view the exercise is simply too speculative. 63. The same can be said of Brammer’s calculation of the average dealer discount. The range in Table A which shows the combined discount is 16.3% to 35% for 4 unidentified manufacturers. 64. It seems likely that Ford and Vauxhall both paid bonuses to high volume purchasers in the Period, I say that because HMRC produced a summary showing how they treated such bonuses in the period 1973 to 1996. I do not know whether Peugeot paid such bonuses at any time in the Period. However, the evidence as to which makes of vehicle Brammer purchased during the Period was not clear. In her affidavit, Ms Eastwood thought that Ford and Vauxhall vehicles were purchased earlier in the Period and Peugeot vehicles later in the Period. In her witness statement she was certain that Fords were purchased in 1993, Vauxhalls in about 1994 and Peugeots in 1997, which was outside the Period. However, the Fleet Sheet shows Fords being purchased in 1996 and 1997 and Peugeots in 1997 and 1998. Given the passage of time I do not consider that Ms Eastwood’s evidence is reliable as to the identity of the manufacturer at any given time. Ms Hunt’s evidence does not take matters much further, save to say that in 1994, Ford vehicles were being purchased and Fords were also purchased at some stage prior to that. 65. Save that Ford Vehicles were being purchased in 1994 and 1996, I cannot be satisfied whether the vehicles being purchased at any other times in the Period were Ford, Vauxhall, Peugeot or indeed any other make. 66. Looking at the evidence as a whole I am not satisfied as to the amount of VAT which Brammer might have accounted for in relation to manufacturer bonuses, or that there is a minimum amount I can be satisfied would have been accounted for. 67. Mr Gibbon somewhat half-heartedly pointed out that the Respondents are able to make VAT assessments using averages and assumptions when there is a lack of records. In my view that is quite different.The Value Added Tax Act 1994 specifically permits HMRC to make assessments to best judgment. Ms Free pointed out that the Upper Tribunal in Vodafone Group Services Limited specifically distinguished the evidence required by a trader to support a repayment claim and the position of HMRC in making an assessment. In the words of the Upper Tribunal: “…it is not possible to treat a repayment claim as the mirror image of an assessment”