ROCHE PRODUCTS PTY LTD v FC of T

Members:
G Downes P

Tribunal:
Administrative Appeals Tribunal, Sydney

MEDIA NEUTRAL CITATION: [2008] AATA 639

Decision date: 22 July 2008

Justice Downes , (President):

Introduction

1. This is a novel case which considers the circumstances in which transfer prices, paid for acquisition of property by subsidiaries of multinational corporations, can be adjusted for income tax purposes.

2. Roche Products Pty Ltd is a subsidiary of the multinational pharmaceutical company, Roche Holdings Ltd of Basel, Switzerland. F Hoffman - La Roche Limited, also of Basel, is the main operating company in the Roche Group.

3. The Roche Group carries on the business of selling and supplying prescription pharmaceuticals, over the counter pharmaceuticals and diagnostic products including diagnostic equipment and reagents. The prescription pharmaceutical business is based upon research and development. The Roche Group has other associated activities, but they are not relevant to these proceedings. Roche Australia relevantly carries on business through three divisions, reflecting the divisions described above.

4. Like all multinational pharmaceutical companies the Roche Group largely confines its sales of prescription pharmaceuticals to sales through its subsidiaries. The same is also generally true of its other activities. As Roche Australia concedes, these sales are not arm's length sales.

5. The Commissioner of Taxation has assessed Roche Australia to income tax on the basis that amounts paid by Roche Australia to Roche Basel were more than the amounts which would be paid in arm's length transactions. The issue in this case is whether the Commissioner's assessments are excessive. I have decided that they were. I have substituted lower increases.

6. Testing whether the Commissioner's assessments are excessive requires the establishment of a benchmark for arm's length sales against which the actual prices can be tested. This is never an easy task. In this case it is particularly difficult.

7. Where there is a substantial free market for goods it will not usually be difficult to find at least a range of prices for arm's length transactions against which prices paid by a subsidiary to its holding company can be measured. However, in the present case, difficulties arise, because there is no substantial free market and because it is difficult to find any comparable sales.

8. Pharmaceutical companies rarely sell their products through third parties. That means that there is generally no free market in which the products in question are sold. It also means that there is generally no free market for even potentially comparable products. There are a few cases in which free markets for the same or similar drugs can be found. However, these markets are generally very small.

9. Patent protection accrues to pharmaceutical products in their generic names. However, the patent holders generally market their products under a brand name which is different. The purpose for this is the legitimate business purpose of developing goodwill or brand loyalty which attaches to the brand name and not to the generic name. When the product is no longer protected by patent the patent owner is able to extend its monopoly to some extent because no competitor entering the market can use the established brand name. This object is assisted by the fact that pharmaceutical products are generally not identifiable by appearance. The brand is the product.

10. The task of a pharmaceutical company launching a new product is to bring its therapeutic properties to the notice of the public and to the medical profession which will prescribe the product. Companies seeking to enter the market to sell a prescription pharmaceutical no longer covered by patent have different marketing problems. Their product will not appear to be the same as, and will have a name different to, the original product. The original product will be well known. The newcomer will need to link the established product with their product. It will need to attract pharmacists to recommend brand substitution to patients and to interest medical practitioners in recommending or permitting brand substitution.

11. The result is that while the best method to determine the arm's length price for the sale of particular products would seem to be to determine what such products are sold for elsewhere, this is a particularly difficult task with pharmaceutical products. First, comparable sales are hard to find. Secondly, when they can be found, they will generally relate to marketing processes linked to varying retail circumstances.

The taxation scheme

12. The Commissioner relies upon two alternative bases to support the assessments. The first is found in the Income Tax Assessment Act 1936(Cth). The second is found in article 9 of the Agreement Between Australia and Switzerland for the Avoidance of Double Taxation with Respect to Taxes on Income [1981] ATS 5 and article 6 of the Agreement Between the Government of Australia and the Government of the Republic of Singapore for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income[1969] ATS 14, which are given effect to by the International Tax Agreements Act 1953(Cth). The Swiss Treaty is contained in schedule 15 of that Act, while the Singapore Treaty is contained in schedules 5 and 5A. There is a question in this case, to which I will come later, as to whether the double tax treaties confer power on the Commissioner to assess income tax at all. For the present, however, I will assume that they do, while I undertake an assessment of the operation of the relevant provisions. The Singapore Double Tax Treaty is relevant because some goods were sold to Roche Australia from Singapore. However, the parties accept that the Treaties are relevantly the same.

13. Section 136AD of the Assessment Act applies to acquisitions of property and operates where "the Commissioner… is satisfied that the parties to the agreement… were not dealing at arm's length with each other in relation to the acquisitions" (s 136AD(3)(b)). In such circumstances, where the "consideration exceeded the arm's-length consideration" (s 136AD(3)(c)), the Commissioner may apply the section in which event "consideration equal to the arm's-length consideration… shall be deemed to be the consideration given… by the taxpayer" (s 136AD(3)).

14. Two relevant aspects of the operation of the section are that it applies to "acquisitions" and that it employs a test which operates by reference to "arm's-length consideration".

15. Article 9 of the Swiss Double Tax Treaty and article 6 of the Singapore Double Tax Treaty operate when an "enterprise" in one country participates in the "management, control or capital of an enterprise" in another country. In such a case, where "conditions operate… which differ from those which might be expected to operate between independent enterprises dealing wholly independently with one another, then any profits which… might have been expected to accrue… may be included in the profits of that enterprise and taxed accordingly".

16. It can be seen that these tests differ in two significant respects from the tests in s 136AD. First, the treaties are concerned with "profits" and not "acquisitions". Secondly, the treaty test refers to "independent enterprises dealing wholly independently" rather than parties "dealing at arm's-length".

17. Nevertheless, the parties spent little time dealing with the words of either set of provisions and effectively accepted that the same result would obtain whichever was applied. It was pointed out that the concepts of "independence" and "arm's-length" are almost interchangeable and that variations in acquisition costs for goods will have a direct effect on profits. This may mean that little point is served in my determining whether the double tax treaties authorise the Commissioner to make assessments of tax in accordance with their terms. For the present, I will deal with the matter primarily on the basis that I am applying s 136AD.

Method of assessing transfer prices

18. Because Roche Australia accepts that its relevant acquisitions were not at arm's length I can go straight to the question whether they exceeded arm's length consideration.

19. The obvious starting point is to look for actual arm's length transactions, preferably for the same goods in the same market. Where there are no arm's length sales of the same goods in the same market it may be possible to find very similar goods or a very similar market. Then, the question is whether the goods or markets are sufficiently comparable and whether any, and if so, what, adjustments can be accurately made to compensate for any differences. This approach is a common one for valuers, particularly real estate valuers, and is described in the multitude of cases following the decision of the High Court of Australia in
Spencer v The Commonwealth (1907) 5 CLR 418.

20. Dealing first with the prescription pharmaceuticals market, there are arm's length sales by Roche of some of the precise products which are under consideration, but the sales were broadly made at the end of the period of patent protection to generic wholesalers. The question is, whether these sales are comparable. Can they be made more comparable by adjustment? There are also arm's length sales of a Roche product to a subsidiary of a rival pharmaceutical company. These sales were early in the life of the drug and relate to a product which needed to be established in the market. To that extent the sales are closer to many of the sales in question here, but it is claimed that the drug in question is atypical. Problems of this kind may require the employment of other methods to arrive at arm's length values. However, retreats to other methods, while avoiding one problem, are prone to result in the substitution of other problems, possibly more serious. In general terms, problems arising from comparables being atypical might be met by looking at a greater number of potential comparables. This may even out the differences. The problem with this approach, however, is that the very evenness leads to an average which may not be a comparable at all. The legislation is concerned with actual arm's length consideration, not the averaging of a range.

21. These are just some of the problems which must be examined relating to the Prescription Division. Different problems arise with respect to the Consumer Division and the Diagnostics Division.

History

22. The history of a matter under review is usually irrelevant to its outcome. However, that is not so in this case. The evidence cannot be understood without an understanding of its history and the impact that it has had, right up to final submissions, on the way the parties support their claims.

23. Of course, the taxpayer bears the burden of proving that the assessments are excessive (s 14ZZK of the Taxation Administration Act 1953 (Cth). If the taxpayer does not discharge this burden then the assessments will be affirmed. The focus must accordingly be upon the evidence presented to the Tribunal and particularly the taxpayer's evidence. This is notwithstanding the decisions to the effect that the concept of onus is a common law concept applicable to litigation and not to administrative decision-making (see
McDonald v Director-General of Social Security (1984) 6 ALD 6 and subsequent cases). The reality is that if the Tribunal, on appeal, is not satisfied on the balance of probabilities that an assessment is excessive, the appeal will fail. In considering this it will have regard to all of the evidence. Attributing the consequence to a failure to satisfy a burden does not really add anything.

24. This might be thought to make the evidence and the reasoning of the Commissioner irrelevant on review. However, the evidence before the Tribunal in the present case and the way the matter has previously been dealt with are so inextricably linked that the history needs to be recounted.

25. On 2 September 1998 a Deputy Commissioner of Taxation wrote to Roche Australia care of its accountants to inform it that the Australian Taxation Office was intending to undertake a transfer pricing record review or audit of the company. The audit took a substantial time, concluding with the issue of amended assessments for the substituted tax years ending 31 December 1992 to 31 December 2002. Assessments issued in 2004. Amended assessments, giving effect to the objection decisions, were issued in 2005 and 2006. In making the assessments the Commissioner drew on material in an expert report prepared in December 2004 by Dr Deloris Wright of Lakewood, Colorado in the United States of America.

26. In August 2006 an expert report was prepared for Roche Australia by Dr Daniel Frisch of Washington, DC in the USA. In preparing this report Dr Frisch had access to material which had not been available to Dr Wright. In particular, he had access to material and calculations prepared for Roche Australia by Mr Murray Hammond who had been retained by Roche Australia to advise and assist in dealing with the transfer pricing issues. A statement containing material prepared by Mr Hammond was filed in the proceedings. In his report, Dr Frisch was critical of Dr Wright.

27. In 2007 a further report was furnished by Dr Wright which addressed Dr Frisch's report. In addition, Dr Brian Becker, also of Washington, DC, furnished the Commissioner with a report addressing Dr Frisch's report. Finally, Dr Frisch prepared a further report responding to the second Wright report and the Becker report. A final short report by Dr Frisch was presented at the hearing.

28. The significance of this chronology of events is that the reports were not based on the same material. The later reports had regard to additional material, particularly Mr Hammond's material. The reports of Dr Frisch and Dr Becker responded to this additional material by employing different methods to Dr Wright. This process was compounded by the emergence of further material at the hearing which caused the Commissioner, in his final submissions, to invite me primarily to act on that material, so far as the Prescription Division case is concerned, and to evaluate that material myself, although with some guidance from the methods explained by the experts.

29. I will explain what this means in a little more detail. So far as the Prescription Division is concerned, Dr Wright initially had no material relating to acquisitions which could be assessed as comparable. Accordingly, she proceeded by looking for companies with comparable activities and worked back from gross profit margins or mark-ups of such companies to an appropriate adjustment of income. Dr Frisch had access to material relating to the sales of some of Roche's pharmaceutical products to generic wholesalers. Dr Frisch and Dr Becker worked primarily from this material. During the hearing the Commissioner sought to rely on material relating to Roche Basel sales of a patented drug, Inhibace (cilazapril), for resale by the subsidiary of a rival pharmaceutical company. Some material relating to this product was available prior to the hearing, but there was further evidence about it during the hearing. The experts were aware of the product prior to the hearing, certainly Dr Frisch was, but their evidence about it was limited.

30. Notwithstanding the way the evidence developed and the way it was relied on changed, particularly by the Commissioner, I was never informed that the Commissioner no longer relied on any part of the earlier evidence, though I enquired whether this might be the position.

31. Accordingly, the evidence of all the experts was before me. I was taken to the evidence in the order in which it was prepared. The experts then gave short oral evidence and were cross-examined in accordance with the order suggested by the burden provision, namely Dr Frisch first, followed by Dr Becker and Dr Wright.

32. Approached from any perspective this is not a simple case. Approached from many perspectives, which is how it was handled before me, it is a very complex case.

33. Transfer pricing issues relating to taxation are apparently highly sophisticated and highly complex in the United States. Each of the experts is an economist specialising in the field. Their approach to the issues before me must have been coloured by their United States experience. At times I wondered why Australian experts could not have approached this matter with just as much skill as the experts from the United States but without some of the presumptions which their work must have led to. Unfortunately, none of the experts were either asked to, or did, directly address the provisions of either the double tax treaties or the Assessment Act. Had they done so my task might have been easier. There was, however, a good deal of reference to the Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations (1995) issued by the Organisation for Economic Co-operation and Development of which Australia is a member. It may now be appropriate to turn briefly to these.

The OECD transfer pricing guidelines

34. The relevant tax treaty provisions relied upon before me are based on Article 9 of the OECD Model Tax Convention. Like the tax treaty provisions the Convention refers to "independent enterprises". The Guidelines describe this as "the arm's-length principle" (Art 1.6). They equate arm's-length dealings with the conduct of independent enterprises.

35. The Guidelines begin by explaining the importance of comparison (Art 1.15) although material differences between what is compared should be taken into account (Art 1.17). They begin by identifying "traditional transaction methods" of which three types are identified (Art 2.1):

36. The name of the first method describes its content: comparing the controlled price between related parties with an uncontrolled price between independent parties. Various matters are to be addressed to ensure that the prices are comparable. These are relatively obvious and may not add anything to the test itself. For example, no differences between the transactions or enterprises should materially affect price, unless accurate adjustments can be made.

37. The resale price method takes the resale price of the product acquired at a controlled price and applies an appropriate gross margin to arrive at an arm's length price for acquisition.

38. The cost plus method begins with the costs incurred by the supplier selling at a controlled price and applies an appropriate mark-up to arrive at an arm's length price for sale.

39. The Guidelines describe these methods as the most direct way of establishing an arm's length price. They state that they are preferable to other methods (Art 2.49). The evidence in this case leads me to suggest that, when it is available, the use of comparables might be said to be preferable to the other traditional transaction methods.

40. Other methods are covered by the Guidelines. They include profit based methods. One of these is the profit split method by which total profits are divided between holding company and subsidiary. An alternative is the transactional net margin method. It is described as follows (Art 3.26):

"The transactional net margin method examines the net profit margin relative to an appropriate base (eg costs, sales, assets) that a taxpayer realizes from a controlled transaction (or transactions that are appropriate to aggregate…)."

The margin is ideally "established by reference to the net margin that the same taxpayer earns in comparable uncontrolled transactions" (Art 3.26).

The prescription pharmaceutical market

41. The prescription pharmaceutical market is highly regulated. Prescription pharmaceuticals cannot be marketed without approval from a Commonwealth agency, the Therapeutic Goods Administration. New entrants into the market will be subject to close scrutiny by this Authority as well as by its equivalents throughout the world. Marketing permission will not be granted without significant evaluation relating to safety and efficacy. This, in turn, requires animal trials, trials in healthy human volunteers and, most importantly, significant clinical trials. Some clinical trials are undertaken by Roche Australia although these have ceased to be relevant to these proceedings.

42. Most sales of prescription pharmaceuticals in Australia are made through the Pharmaceutical Benefits Scheme. Under this scheme patients pay a fixed price for drugs which are subsidised by additional payments to dispensing pharmacies under the scheme. Approval to market a prescription pharmaceutical in Australia does not automatically include it in the scheme. Two Commonwealth agencies are involved. The Pharmaceutical Benefits Advisory Committee decides if the drug merits being included and the Pharmaceutical Benefits Pricing Authority determines the price if it does. Drug companies are free to market approved prescription pharmaceuticals outside the scheme but, because the prices will generally substantially exceed the subsidised prices of the scheme, they rarely do. Drugs dispensed in hospitals are, however, provided outside the scheme.

43. One of the witnesses was Frederick Nadjarian. He has been the Managing Director of Roche Australia for more than 20 years. Mr Nadjarian gave evidence, which I accept, that, generally, two factors interact to determine the prices approved by the Pricing Authority. First, the Authority generally allows a markup of about 30 percent over stated cost. Secondly, where products with similar therapeutic properties were already on the market, it seeks to fix prices of new entrants at the same level as the lowest priced brand.

44. This is how the market operates for patented products during most of their patent life. At the end of the patent life of successful products, and sometimes before the end of that life, new companies enter the market with products containing the same active ingredients as the successful products. These companies are usually called generic companies because they cannot use the established brand name and because they often sell the product by its generic name. Sometimes they will sell under their own brand name.

45. Pharmaceutical companies selling under patent need to develop a demand for their product. In the case of a drug which has unique and positive therapeutic effects this will be easier than with a drug whose properties are similar to those of an existing drug. Consumer advertising is normally not permitted. Accordingly, promotion centres around informing medical practitioners and the health industry generally about the properties of the drug. This often involves direct marketing to medical specialists and general practitioners. This is known as detailing.

46. The issue for generic companies is different. The drug may be well established by its brand name. What is required is attracting medical practitioners either to prescribe the drug by its generic name or to authorise brand substitution as well as attracting pharmacies to dispense the generic preparation and to encourage customers to accept the generic. Detailing may be some part of this process but it is apparently primarily concerned with attracting pharmacists to dispense generics by offering them incentives such as rebates, discounts and the supply of bonus stocks. The entry of generics into the market place tends to drive prices down.

The business of Roche Australia

47. Generally speaking the business of Roche Australia is to establish, develop and maintain marketing outlets in Australia for its Swiss parent. That part of the business which is relevant to these proceedings is carried on through three divisions, namely, the Prescription Division, which imports and sells Roche prescription pharmaceuticals mainly through major drug wholesalers, the Consumer Division, which sells over the counter products and the Diagnostics Division which sells diagnostic equipment and products such as reagents.

The Prescription Division

48. The most significant division in Roche Australia is the Prescription Division. It markets Roche's major prescription pharmaceutical products in Australia. Some of the products are protected by patents. Others no longer have patent protection.

49. At the beginning of the period covered by the amended assessments Roche Australia marketed a number of core products. They included Valium (a nerve relaxant and sleeping tablet), Mogadon (a sleeping tablet), Bactrim (an antibiotic), Rohypnol (a sleeping tablet) and Lexotan (for anxiety and tension). Newer products were Roaccutane (isotretinoin) (for severe acne) and Hypnovel (an injectable sedative). The newest products were Aurorix (moclobemide) (an anti depressant), Rocephin (an injectable antibiotic) and Rocaltrol (calcitriol) (an osteoporosis treatment). Aurorix was launched in Australia in 1992 at the beginning of the audit period. It was protected by patent until 1997.

50. Roche Australia received the products in manufactured form. It did, however, perform some secondary manufacturing and packaging in Australia. This was, however, much greater in the Consumer Division than the Prescription Division. Most prescription drugs were received in finished form ready to be distributed.

51. Prescription drugs are generally marketed by being prescribed by medical practitioners. Sales representatives visit doctors and discuss three or four products with them. However, the time of doctors is limited and sometimes the discussions are superficial. Roche Australia, through Mr Nadjarian, developed a practice of representatives focussing on one product. This practice began with Rocephin. It was successful. Roaccutane, Aurorix and Rocaltrol were then included. The method was more labour intensive and required a larger sales team. It was more expensive.

52. Between 1993 and 1995 sales of Aurorix, Rocaltrol and Rocephin more than doubled and sales of Roaccutane grew by more than half. Thereafter, with the approval of Roche Basel, Roche Australia developed a marketing strategy which built on Mr Nadjarian's strategy by identifying and concentrating on products with potential.

The Consumer Division

53. The Consumer Division of Roche Australia sold medicines which did not require prescriptions. They are often known as "over the counter" products. Many of the products were imported although the local manufacture and packaging content was greater than in the Prescription Division.

54. Included in the Consumer Division were products named Rennies (indigestion), Interdens (oral hygiene), Aspro (headaches and pain), Aleve (pain relief) and Elevit (multivitamins for pregnancy). These products were imported. They were called in the proceedings "Category 1 products" and treated separately. The significance of this was controversial. It seems that the Category 1 products were identified for separate consideration during the audit period. Roche Australia says, however, that there never was any agreement or arrangement that their separate consideration was appropriate.

55. Other Consumer Division products in the Division included Berocca, a multivitamin precursor of energy drinks which was very successful until its sales began to be eroded by emerging energy drinks.

THE DIAGNOSTICS DIVISION

56. The Diagnostics Division sold diagnostic equipment and preparations, such as reagents, for use with the equipment. These were products used by hospitals and medical laboratories in providing pathology and other services.

57. Roche Australia followed the practice of selling or leasing its equipment at full value while permitting purchasers to source preparations such as reagents competitively. Other companies sold equipment at reduced prices upon condition that all preparations and reagents were purchased from them.

58. Most of the Roche products were old during the reassessment period although it also marketed Polymerase Chain Reaction technology which rapidly copied genetic material and which it had acquired in 1991. This technology showed potential.

59. The Diagnostics Division was never particularly successful. In 1994 Roche Basel suggested that it be closed down. Mr Nadjarian decided against this course in the belief that it would become successful and the hope that the Division could be sold as part of a global sale of the Diagnostics Division. The Division was restructured and began to show promise. By 1997 sales were twice what they had been in 1993 although this was affected by equipment write downs. In 1998 Roche acquired the Corange Group and transferred its Australian Diagnostics Division to Boehringer Mannheim Australia Pty Ltd, which was part of that group.

The experts' evidence

Dr Deloris Wright

60. The first report of Dr Deloris Wright was prepared in December 2004 to assist the Commissioner in determining the outcome of the objections. The report breaks up the activities of Roche Australia into the three divisions identified above.

Prescription Division

61. Dr Wright discussed the most appropriate method to employ. She acknowledged that "the best way to determine arm's length prices is always the comparable uncontrolled price… method because these prices represent the price that unrelated companies agreed upon under the same, or similar, circumstances with each party negotiating at arm's length and under no obligation to deal". However, she did not employ this method because she did not have access to any relevant transactions. The best transactions were transactions with one party from the Roche Group and the second best transactions were transactions between parties unrelated to the Group. Neither were available.

62. Next Dr Wright referred to the resale price method and determined that this was the most appropriate method to employ for the distribution and marketing activities of the Prescription Division. As will appear, however, because of the lack of information she had, she was not able to employ this method by reference to actual sale prices. Dr Wright decided that the cost plus method was the most suitable for the Prescription Division's involvement in clinical trial management and secondary manufacturing.

63. Dr Wright introduced her analysis with a reference to profit based methods including the transactional net margin method. She noted that transactional based methods were preferable to profit based methods. She said she used transactional based methods to arrive at her results and then verified this with the use of profit based methods.

64. Dr Wright divided the activities of the Prescription Division into three components: clinical trials, secondary manufacturing and distribution and marketing. I will outline the process she followed.

65. She first identified contract research organisations which specialise in the management of clinical trials. She identified these through three companies providing data bases on the internet for a fee. These threw up 591 companies. By a process of analysis she discarded 580. She gave very few details of this process. Eleven companies met her criteria. Five companies appear to have been added from past searches. Fourteen of the companies were in the United Kingdom.

66. Material was collected relating to the markup for each of the years 1992 to 2002 for the sixteen companies. The results ranged from +24.8 percent for one company in 2002 to -20.0 percent for another company in 1997. However these outliers were discarded in favour of the interquartile range of 3.7 percent to 11.3 percent with a median of 7.9 percent. She concluded that a markup in this range was arm's length.

67. It will be noted that not only does this method not use sales by Roche companies it does not use sales at all. It uses calculations of markup from the internet. The companies examined were companies whose business was providing clinical trial services pursuant to contract. That is not, of course, what Roche Australia did. Although Dr Wright attempted to find companies whose activities were comparable to those of Roche Australia it must have been difficult to know precisely how much was outsourced in each case she selected and whether it was comparable to the work Roche Australia carried out. It also occurs to me that different arm's length results may be obtained in a company providing only one of a group of services such as clinical trials, secondary manufacturing and distribution and marketing, compared with a company providing all of those services. The one service company may make a greater markup because it is a specialist while the multi service company may make a greater markup because of the comprehensive service it supplies. The important matter is not which one of these, or some other alternative, is true, but simply that the availability of such considerations damages the comparability of the figures.

68. Dr Wright then turned to secondary manufacturing. She made a similar search with the same databases for companies providing secondary manufacturing services to pharmaceutical companies or "Formulate, Fill, Finish" (FFF) services as she called them. She examined 1,440 companies and rejected 1,438 of them. The remaining two companies were a French company and a Canadian company. Only the French company yielded markup figures for the whole period. The range was from 2.6 percent to 30.2 percent with an interquartile range of 10.4 percent to 17.9 percent. The median was 12.6 percent. She concluded that a markup in the interquartile range was arm's length.

69. Many of the observations I have already made apply equally to this analysis. There is the added problem that the sample from which the range is taken is very small. The range of 2.6 percent to 30.2 percent is the range for the French company. Dr Wright accepts that company as a comparable. It follows that a very large markup range can be found in a company operating at arm's length.

70. Finally, Dr Wright examined the distribution and marketing function. She began with the distribution function. This time she was seeking to apply the resale price method. She first of all looked for comparable companies in Australia but failed to find any. Next she looked at 17 other countries which she felt were comparable to Australia. Again, she could not find companies which carried on functions sufficiently comparable to the activities of Roche Australia, partly because they did not carry out the detailing or direct marketing through medical practitioners. Dr Wright ultimately decided to draw her figures for the distribution and marketing function from different sources.

71. Starting with distribution Dr Wright looked at independent pharmaceutical distributors who purchased from unrelated manufacturers. She first excluded Australian companies as not useful. She then looked at the 17 countries she had previously identified. She established that operating margins (interquartile of 0.9 percent to 2.3 percent for 1992 to 2002) across the countries involved did not show significant variation. This satisfied her that she was on the right track.

72. Dr Wright next identified 1080 companies. All but 25 were rejected. However, companies isolated in prior searches were added to make the total 49. The 49 companies yielded an interquartile gross margin range of 6.1 percent to 9.2 percent with a median of 7.5 percent. The gross profit to selling, general and administrative ratio (Berry ratio) was calculated. This indicated that markup on cost was fairly uniform.

73. Dr Wright then stated that an adjustment needed to be made for "carrying cost of inventory and net accounts receivable". She did not have the information to make the adjustment but stated that it was usually small, although it could be quite large in the present case due to the extended payment terms which Roche Basel gave to Roche Australia. She also stated that adjustments needed to be made for functional intensity. The adjustment was made.

74. In the result Dr Wright arrived at adjusted gross margins of -7.7 percent to 13.4 percent an interquartile range of 7.6 percent to 9.2 percent and a median of 8.3 percent. Converted to operating margins the range was -14.3 percent to 6.8 percent, the interquartile range was 1.0 percent to 2.6 percent and the median was 1.6 percent.

75. It is to be noted that Dr Wright said she was employing a cost plus method to arrive at figures for clinical trial management and secondary manufacturing. Hence, she uses markup figures. For distribution and marketing she used the resale price method. Accordingly, she used operating profit. Nevertheless the operating profits she determined were derived from the gross margins. It is difficult to see that there is any real difference in the methods as applied by Dr Wright. Of course, the methods would be quite different if actual transactions, rather than information taken from financial records, were being used. The method used by Dr Wright is not really a transaction method but rather a profit based method, with all their disadvantages, as Dr Frisch pointed out.

76. Dr Wright next addressed marketing expense. She divided this role between selling and marketing. Selling included the actual detailing; marketing was more concerned with developing the market. She began with marketing. She chose eighteen countries including Australia. From 265 companies she rejected 258, leaving seven. Six of the companies were from the United States and one from Canada. They included well known international advertising agencies. This process yielded a markup range of 2.9 percent to 23 percent with an interquartile range of 9.0 percent to 15.9 percent and a median of 13.1 percent.

77. Next she turned to selling. She chose companies from a list of 1,069. They were all United States companies. They provided various services to the pharmaceutical industry including outsourced sales. These companies yielded a markup range of -1.7 percent to 16.5 percent with an interquartile range of 3.3 percent to 10.3 percent and a median of 6.6 percent. Dr Wright ultimately chose the top end of the range "because I believe that to be the appropriate mark up".

78. Dr Wright then brought together the components she had identified in an arm's length gross margin range of 43.2 percent to 45.8 percent with a median of 44.4 percent. As a reality check she calculated the interquartile range of operating margins which came out at 4.1 percent to 6.7 percent with a median of 5.4 percent.

Consumer Division

79. Dr Wright then turned to the Consumer Division. She was only asked to deal with the Category 1 products. She chose to use the resale price method. She employed her usual method of identifying companies from commercial data bases. She had already looked for independent companies carrying out distribution and marketing of pharmaceutical products without success. She next searched for independent distributors of fully finished unpatented products in Australia. Finding none, she looked at the 17 countries to which she had previously had resort when determining distribution margins in the Prescription Division. She determined to use this list of companies.

80. The results at which Dr Wright arrived are slightly different to her prior results because she used the years 1992 to 2002 for her first set of calculations and 1994 to 2004 for the second set. The results are accordingly close to, but not identical with, the figures she had obtained before. The unadjusted interquartile gross margin range is 6.0 percent to 9.1 percent with a median of 7.5 percent and for operating margins is 0.9 percent to 2.2 percent with a median of 1.6 percent. The adjusted figures for gross margins appear to be 60.9 percent to 62.7 percent with a median of 61.7 percent and for operating margins are 3.9 percent to 6.9 percent with a median of 5.6 percent. She acts on the operating margin percentages. I note that the adjustments are very large.

Diagnostics Division

81. Finally, Dr Wright dealt with the Diagnostics Division. As usual, she commenced by searching for comparables in Australia. She looked for independent distributors of diagnostic products. She found none. She went to the 17 countries. She found none. However, 10 of the 47 distributors of pharmaceutical products also distributed diagnostic products. She used the 47 distributors as her comparables.

82. The comparables yielded an adjusted overall gross margin range of 1.4 percent to 22.5 percent with an interquartile range of 16.5 percent to 18.1 percent and a median of 17.2 percent. The adjusted operating margin range overall was -12.5 percent to 8.6 percent with an interquartile range of 2.6 percent to 4.2 percent and a median of 3.3 percent. These figures were finally adjusted to gross margins of 59.2 percent to 63.7 percent with a median of 61.4 percent and operating margins of 4.5 percent to 9.0 percent with a median of 6.7 percent which is the range she chose.

Dr Wright's conclusions

83. At the end of her 2004 report Dr Wright produced a table which showed the effect of her conclusions on the revenue of Roche Australia. In simplified form it is as follows:

Prescription Division
Distribution & Marketing Low Medium High
Revenue 1992-2002 $1,491,122,944
Arm's length operating margin 4.1% 5.4% 6.7%
Operating profit $61,364,902 $80,037,151 $99,774,309
Clinical Trials      
Costs 1992-2002 $92,880,455
Arm's Length Mark Up 3.7% 7.9% 11.3%
Operating profit $3,428,087 $7,326,295 $10,464,742
Secondary Manufacturing      
Costs 1992-2002 $49,873,476
Arm's Length Mark Up 10.4% 12.6% 17.9%
Operating Profit $5,173,215 $6,301,309 $8,926,007
Total Operating Profit $69,966,204 $93,664,755 $119,165,058
Consumer Health Division      
Revenue 1994-2002 $109,276,500
Arm's Length Operating Margin 3.9% 5.6% 6.9%
Operating Profit $4,215,491 $6,087,904 $7,508,938
Diagnostics Division      
Revenue 1992-1998 $50,722,232
Arm's Length Operating Margin 4.5% 6.7% 9.0%
Operating Profit $2,286,891 $3,378,571 $4,548,375
Total Operating Profit $76,468,586 $103,131,231 $131,222,371

Dr Daniel Frisch

84. Dr Daniel Frisch prepared his first report in August 2006. He had seen a copy of Dr Wright's report. He had considerable further information available to him which came to form at least part of the statement of Mr Hammond. In particular, he had access to information relating to sales by Roche Basel to Roche Australia and by Roche Basel to independent wholesalers.

Prescription Division

85. Dr Frisch based his calculations on three products. Between 1996 and 2003 they were sold by Roche Basel to independent wholesalers in Australia. The products are Rocaltrol (calcitriol) Roaccutane (isotretinoin) and Aurorix (moclobemide). They were sold to Alphapharm Pty Ltd, Arrow Pharmaceuticals Ltd, Hexal Australia Pty Ltd and Biochemie Australia Pty Ltd. They are all so-called generic pharmaceutical companies. The patent for Aurorix expired in January 1997. The patent for Roaccutane had expired in 1987. Rocaltrol had no Australian patent protection.

86. Dr Frisch accordingly applied the comparable uncontrolled price method. The drugs were identical in their active ingredients to the drugs sold by Roche Basel to Roche Australia. Dr Frisch considered that the volumes sold were sufficient for the sales to reflect arm's length prices. However, some adjustments were necessary. First, prices paid were not always in the same currency. These had to be converted. Secondly, Roche Australia purchased the tablets in blister packs while the independent parties purchased 30 kilogram drums. Roche Australia paid by the pack; the independent parties paid by the kilogram. Thirdly, it was necessary to adjust for packaging costs. Fourthly, the payment terms were different. Finally, adjustment may have been necessary to account for any brand premium payable under the Pharmaceutical Benefits Scheme.

87. Applying his method to sales of calcitriol, isotretinoin and moclobemide Dr Frisch arrived at results which are included in Attachment 1. The results record the final figures which Dr Frisch adopted. The attachment also shows comparative figures for Dr Becker. The comparable is the highest price paid by the generic companies.

88. For calcitriol the highest annualised price paid by the generics was higher than the price paid by Roche Australia in every year. Dr Frisch concluded that the price paid by Roche Australia was lower than the arm's length price. The same conclusions applied to the 10mg tablets of isotretinoin and the 300mg tablets of moclobemide. It is to be noted, however, that figures were not available for generic sales in every year. The figures covered every year from 1998 to 2003 for calcitriol and from 1996 to 2003 for 20mg isotretinoin, but only from 2001 to 2003 for 300mg moclobemide. The results were mixed for the other two products, namely 20mg isotretinoin and 150mg moclobemide. The sale price to Roche Australia of 20mg isotretinoin was less than the highest generic sale in 1996 to mid 2000, higher between then and the end of 2002, but lower for 2003. For 150mg moclobemide the sale price was lower in 1997 and 1998, higher from then to the end of 2002, but lower for 2003.

89. When the figures are averaged Roche Basel overall paid less than the total arm's length prices for all the products surveyed. Dr Frisch argued that in complex business dealings actual arm's length prices will vary. He concluded that because the prices paid by Roche Basel were predominantly less than arm's length prices "Roche Australia's transfer prices for these products should be regarded as arm's length from Australia's perspective".

90. Rocaltrol, Roaccutane and Aurorix represented about 21.5 percent of Roche Australia's prescription sales. Dr Frisch calculated the gross margin percentage earned by Roche Australia for the products and compared that with the gross margin percentage for the other products. Dr Frisch found "no systematic difference between the… prices" and concluded that his finding that the prices for the comparable products were less than arm's length should be applied to all products sold in the Prescription Division.

91. Dr Frisch dealt separately with the income earned by Roche Australia for conducting clinical research on behalf of Roche Basel. He agreed with the approach taken by Dr Wright and relied on her determination and calculations. This shows that from 2000, when Roche Basel increased its fee from 1.5 percent to 5.00 percent the amount paid to Roche Australia was arm's length. Since the average was also within the range, Dr Frisch concluded that the fees paid overall were arm's length.

Consumer Division

92. Dr Frisch considered that it was not appropriate to deal only with the Category 1 products. This was because "a company that markets and distributes a manufacturer's products does not make decisions by considering each product or a small group of products in isolation". He concluded that the entire range of Roche Australia's products in the Consumer Division should be considered. He considered that the correct question was whether "an arm's length party [would] have been satisfied with the profitability of Roche Australia['s]… overall portfolio of products"? Notwithstanding the logic of this approach I note that it does not seem to be consistent with the essence of the enquiry which relates to acquisition prices for property.

93. Roche Australia's Consumer Division had an overall operating profit margin of 8.1 percent measured as the ratio of earnings before interest and tax (EBIT) to sales. Dr Frisch considered this to be a high rate of profit. He concluded that the overall operations of the division were such that the prices for the products, as a whole, were arm's length. An arm's length company, he said, would be prepared to carry on business with the unprofitable prices paid provided it could continue to receive the profitable lines at the same prices. Again, this does not seem to be the correct question. Well it might. But the question is not whether it would continue but whether the prices it was paying were arm's length.

Diagnostics Division

94. This Division was financially unsuccessful. The reasons were specific to Roche Australia and apparently included the practice of permitting purchasers of Roche equipment to source supplies of the products and reagents to be used with the equipment from third parties. This might have been an advantage but apparently was not.

95. Dr Frisch concluded that not only was it not profitable to use any of the transaction based methods it was also not possible to use the profit based transactional net margin method. In place of these conventional methods he simply inquired "whether Roche Australia['s]… management made decisions and took actions that were consistent with what arm's length parties might have done in the same circumstances". Dr Frisch concluded "that Roche Australia['s]… experiences and transfer prices during income years 1994 to 1999 may well have been consistent with the experiences that an arm's-length party would have had in the same circumstances".

Dr Frisch's conclusions

96. Dr Frisch concluded that all of Roche Australia's relevant activities were arm's length.

Dr Brian Becker

97. Dr Brian Becker's first report was prepared in August 2007. It commented on the reports of both Dr Wright and Dr Frisch.

Prescription Division

98. Dr Becker adopted the same approach as Dr Frisch except that he confined his analysis to the Alphapharm transactions and he excluded the sales of isotretinoin 20mg. These sales were excluded because they were relatively low. Alphapharm generally had a market share of less than five per cent for this product. It generally had the lowest market share of all of the preparations it purchased from Roche. Nevertheless, the absolute quantities of isotretinoin 20mg which Alphapharm purchased were much larger than the quantities of isotretinoin 10mg. The quantities were at least 10 times greater and sometimes much greater than that. In 1999, for example, Alphapharm purchased 7 kilograms of isotretinoin 10mg and 433 kilograms of isotretinoin 20mg. Yet Dr Becker excluded the 20mg purchase and included the 10mg purchase.

99. With this exclusion, Dr Becker found that "Roche Australia did purchase the presentations at prices consistent with those paid by the unrelated parties". However, the sales matching the comparables covered only eight of the years under consideration and the comparables matched only some of Roche Australia's products. Dr Becker did not take the step of finding that a like conclusion could be drawn with respect to the other products and years. He undertook a gross margin comparison.

100. Dr Becker found that Roche Australia earned a combined gross margin of 40.5 percent on the comparable sales in the eight years in which there were comparable sales. He excluded the sales of isotretinoin 20mg. He found that the gross margin on non-comparable products averaged 35.5 percent. The difference of 5.0 percent, when applied to the value of sales which were not used in the comparables calculation, amounted to $65,530,000. Dr Becker then excluded moclobemide from his comparables. This gave a gross margin of 46.1 percent. Applying the different gross margin of 10.5 percent and making the same calculation, the price adjustment increased to $138,075,000. After applying validating tests Dr Becker concluded that the appropriate amount was the adjustment of $65,530,000.

101. Dr Becker could not find any comparable transaction to assist him with assessing the clinical research activities of Roche Australia. He turned to comparisons of profitability. He chose eleven companies. The middle 50 percent of these, or the interquartile range, "earned an average mark-up on their costs (from 1992-2002) of approximately 5.0 percent to 11.8 percent with a median of 6.4 percent". Dr Becker chose the median. Applying that to the actual mark-up by Roche Australia he concluded that the clinical trial profit should be increased by $2,205,000.

Consumer Division

102. Dr Becker again employed a profitability approach because of the absence of comparable transaction prices. He chose eleven companies. These were not the same eleven as he had previously used. They reported operating profit margins with an interquartile range of 1.2 percent to 2.7 percent with a median of 1.7 percent. The figure for Roche Australia was -9.3 percent. Applying the differential of about 11.0 percent yielded an adjustment of $11,957,000.

Diagnostics Division

103. Dr Becker used the same approach again, this time identifying five comparable companies. The interquartile range for the comparable companies was 1.2 percent to 4.7 percent with a median of 4.6 percent. Comparing this with the Roche Australia figures he arrived at an arm's length adjustment of $10,138,000.

Dr Becker's conclusions

104. Dr Becker summarised his adjustments (somewhat rounded) to $89.8 million which would represent an overall arm's length operating profit margin for Roche Australia of 6.5 percent.

Mr Murray Hammond

105. Mr Murray Hammond also provided an expert report. However, Mr Hammond was retained by Roche Australia to assist it in the preparation of its case. Accordingly, the report was not the report of an independent expert. Nor was it the report of an economist experienced in transfer pricing issues. It was the report of a financial analyst. Nevertheless, the content of the report was little criticised and cross examination on it was not extensive. The report does contain a collection of primary material relating to the financial activities of Roche Australia. It also contains a comparative analysis of the Alphapharm transactions. It compares product and strength with like product of the same strength.

106. In line with the analysis of Dr Becker the analysis of Mr Hammond shows that the Alphapharm purchases were largely for more than the purchases by Roche Australia. It shows a period when isotretinoin 20mg was sold to Roche Australia for more than the price to Alphapharm. It also shows a slight excess in the price to Roche Australia for moclobemide 50mg. Similar analysis of sales to Arrow, Hexal and Biochemie yields similar results.

Inhibace and Quinodis

107. One of the Roche Group's products is the drug Inhibace (cilazapril). It also markets a drug called Quinodis.

108. In 1995 Inhibace was a new drug. Roche Basel was keen for it to be launched. According to Mr Nadjarian, Roche Basel considered it to be a profit driver. It thought it would be successful. The Roche Group annual report for 1993 said it "showed strong sales growth". This statement was replaced in the 1994 report with a reference to "strong volume growth". However, Mr Nadjarian did not wish to market the drug in Australia. He did not think it would make money.

109. In March 1996 Roche Basel entered into an agreement with Bayer AG for Inhibace to be sold in Australia.

110. In April 2006 René Maier, the head of international pricing at Roche Basel, sent an email to Mr Hammond commenting on the Inhibace agreement. He said that the tablets were invoiced to Bayer AG in Swiss Francs "… at 40 percent of the Australian wholesale price…". This was in response to an email from Mr Hammond suggesting that they were sold "at a rate that was 35 percent of the price to the wholesaler giving them a 65 percent Gross Margin". A spreadsheet prepared by Mr Hammond supports a 65 percent gross margin. The agreement was a gross margin contract under which it was expected that the margins would remain the same. The wholesale price upon which it was based was the price paid by the Pharmaceutical Benefits Scheme.

111. Mr Hammond did not make any calculations with respect to the Inhibace agreement. He said "… we were not to use it…". Dr Frisch was aware of the Inhibace agreement, but he did not use it. He gave the following reasons for not using it:

"[Roche] told me about the arm's length transactions with Alphapharm, etcetera, and they said, By the way, there are these - there's Inhibace, and they told me about another one, Quinodis I believe, and I said, Well, are those significant transactions or are they immaterial transactions and idiosyncratic one-off transactions? And it was clear to me that Inhibace was not - in no way could be thought of as, you know, a major drug… Fred didn't even - Fred - Mr Nadjarian didn't even want to bother with it. So I made the judgment that this was not going to be an important - no way was this expected to be an important drug for Australia or for Roche or for anyone, and instead it's one of these isolated one-off transactions that, you know, I - I didn't think was even worth pursuing as being representative of the important drugs that Roche Australia was carrying."

While I do not doubt that this reflects what Dr Frisch was told, and appears to accord with the views expressed by Mr Nadjarian in his evidence, it does not reflect Roche Basel's opinion of Inhibace.

112. In other parts of his evidence Dr Frisch said that it was his practice "to use everything… It's all arm's length information, let's use it all." He also agreed that "it is possible that Inhibace was comparable to some of the drugs that Roche Australia carried".

QUINODIS

113. There is evidence that Quinodis was a drug which the Roche Group proposed to market in Australia. In a letter in 1993 Mr Nadjarian said that Quinodis would "hopefully be launched in 1994 but would not contribute a great deal to growth until the year after". The evidence suggests that there was a proposal that the drug should be marketed by the Bayer Group, like Inhibace. However, no agreement was produced; nor were any other records relating to the marketing of the drug. The drug may never have been marketed in Australia.

The correct approach

The Prescription Division

114. I can understand that, at a time when the Commissioner was conducting an audit of Roche Australia, without financial data for what might be comparable sales, he would seek expert advice based on general financial information from an expert such as Dr Wright. However, that information will always be second best to information relating to actual sales. After all Division 13 is concerned with costs of acquisition of property. Even the double tax treaties, in a case such as the present, will operate on sales although any adjustment will relate to profit. This is because the costs of sales affects profit directly. It was not argued on the part of the Commissioner that the double tax treaties justified some wider approach.

115. Apart from admitting to an intuitive belief that the way to evaluate transfer prices is to look at comparable prices at arm's length, rather than comparing different aspects of the subject's business to a range of other businesses, there seem to me to be more concrete reasons why the evidence of Dr Wright relating to the Prescription Division is no longer of particular assistance, given the presence of other evidence based on actual transactions:

116. During the hearing I was specifically informed on behalf of the Commissioner that reliance continued to be placed on the first report of Dr Wright. This position conflicts somewhat, at least so far as the Prescription Division is concerned, with the following statement in the Commissioner's final submissions:

"There now is direct evidence before the Tribunal which would enable it to decide whether the prices paid by the applicant to Roche Basel for products acquired by its [Prescription] Division were consistent with the correct arm's length consideration. In respect of the [Prescription] Division it is submitted that the expert evidence now has only indirect relevance; it is but a 'sanity check' for any conclusion that is drawn from the direct evidence."

The Commissioner's position, at the time of submissions, was to rely upon the Inhibace agreement as the only comparable product. This is the direct evidence referred to. Dr Wright did not deal with Inhibace because she did not deal with any direct comparable. Neither did Dr Frisch, although he was aware of the Inhibace agreement, nor Dr Becker.

117. I have come to the conclusion that Dr Wright's method is not of any real assistance in dealing with the Prescription Division. In addition to the problems I have found with her method, I also found problems with Dr Wright's application of the method. Part of this follows from the problems I find in the method. For example, she was driven to use the profitability of advertising agents to determine a level of profit for the marketing aspect of sales and marketing. She explained this on the basis that she was trying "to value the marketing functioning… And you do that by reference to companies that employ such individuals and render creative marketing as their core business, and that is an ad agency". I do not think that this explanation justifies using the profitability of international advertising agencies, as comparable to internal marketing deliberations of a pharmaceutical company. The advertising agency may be dealing, for example, with a media advertising campaign for coffee. That does not seem to me to be comparable to a pharmaceutical company deliberating internally on how it might best train and equip a group of sales representatives to make the best out of detailing its pharmaceutical products.

118. In addition, Dr Wright separated sales from marketing and arrived at cumulative profitability figures for both activities. I have difficulty in accepting that such an approach may not overstate appropriate profitability when the subject carried out both activities in-house and without any apparent internal division.

119. There was one aspect of Dr Wright's evidence which particularly troubled me. In addition to treating sales and marketing both separately and cumulatively, Dr Wright applied the top of the interquartile range for selling expenses. She accordingly chose 10.3 percent by contrast with the median of 6.6 percent. She chose the median in other cases. The explanation she gave for this in her initial report was "because I believe that to be the appropriate mark-up".

120. In her evidence in chief Dr Wright was asked to explain further. She began by saying that the figures her researchers had picked up were "very volatile". Further research showed that the companies were "entering new business". This part of her explanation only highlights the inherent problems in the method to which I have already referred. She went on:

"This is where my experience in the pharmaceutical industry came to bear. In the last 25 years I have been working for pharmaceutical companies. This is the only case I have had that has worked against the industry. There has not been a single time in that 25 years that I haven't had at least one pharmaceutical company client and the analysis that you see in this report is the same analysis that I have done for those companies. Because I had access to the company resources I knew and know what kinds of contracts exist. I know how pharma companies deal with contract sales forces. I know what the results are likely to be. If you say with Quintiles [a company providing services to the pharmaceutical industry], you see that in the period '96 to '99 that the operating margin is in the 9 to 10 per cent range in each one of those years. I know that that is about right for the third party contracts that I have seen. So what I chose to do was to let that knowledge override the allocation of corporate overhead expenses that was creating the volatile operating margins."

121. It is to be observed that Dr Wright was using companies which contracted with the pharmaceutical industry as a means of determining the selling cost components which would enable her to arrive at operating profit. Her method depends upon finding proper comparables. Yet, here, she seems to have abandoned the exercise in favour of anecdotal knowledge from dealing with pharmaceutical companies. She gave no further details. This approach does not seem to me to be appropriate for an expert such as Dr Wright. This is particularly so when it contradicts the figures which have been derived from a complex process which should, by definition, be reliable. Yet the figures were readily departed from without any proper, justified basis. Dr Wright made no effort, for example, to collect or present the figures on which her ultimate conclusion was based.

122. Dr Wright arrived at a median mark-up of 5.4 percent for Marketing and Selling expenses. This is by far the largest component in the increased taxable incomes leading to the amended assessments. The total amount flowing from Dr Wright's calculations for marketing and selling based on the median figures (subject to the median for selling being shown as 10.3 percent) is $80,037,151 and the figure applied was 5.4 percent. The proportion related to selling was 1.9 percent. This is derived by taking 10.3 percent of the percentage of selling expenses to sales. Had Dr Wright chosen the actual median of 6.6 percent the figure would have been 1.21 percent and the amount of $80,037,151 would have been reduced by approximately $10,000,000. This is a substantial sum to depend upon anecdotal knowledge.

123. In the result I do not gain any direct assistance in determining the ultimate matters before me relating to the Prescription Division from either the method employed by Dr Wright or her implementation of it.

124. That leaves me with the possibility of using comparable sales which, in any event, all the experts, including Dr Wright, considered to be the preferred method.

125. The problem is identifying the comparable sales and making appropriate adjustments. In the present case there do not seem to me to be any large issues relating to adjustments. There was one issue as to whether the terms of payment available to Roche Australia warranted an adjustment, but no other substantial issues arise. The main issue is what are the appropriate comparables. The dispute surfaced at two levels. First, were the generic sales of isotretinoin, moclobemide and calcitriol comparable sales and was the Inhibace sale a comparable sale? The parties adopted opposing positions. To Roche Australia, only the generic sales were comparable. To the Commissioner, only the Inhibace sale was comparable. Secondly, within the generic sales should any sales, such as isotretinoin 20mg, be excluded? Before looking in detail at specific sales it seems sensible to look at the evidence more generally.

126. There is a significant amount of evidence relating to what was called "the 30 per cent rule". As described by Mr Nadjarian this was "an unwritten… rule that everyone knows that when you try and list a product on the PBS if you have a margin that is greater than 30 per cent, that can be one of the reasons… to reject approval of the product at the price you requested". It seems, however, that the Pharmaceutical Benefits Scheme had to act on the applicant's figures.

127. The evidence of both Mr Nadjarian and Mr Maier was that the prices which Roche Basel required Roche Australia to pay were calculated from a gross profit margin of 30 percent.

128. The agreements for the sales to the generic companies were negotiated to realise a 40.0 percent gross margin. For example, Mr Nadjarian agreed that "what happened in practice was that a contract with Alphapharm is negotiated, and if not negotiated it has certainly been implemented after negotiations based upon a gross profit margin of 40.0 percent".

129. There was ultimately no dispute over the figures presented to me. The dispute concerned the way in which they should be used. A slight confusion does arise because Roche Australia prepares its accounts on a calendar year and not a financial year basis. Although Dr Frisch's figures are nearly the same as Dr Becker's, Dr Frisch refers to an income year and Dr Becker to the calendar year. Accordingly, for example, the identical figures appear under 1995 for Dr Frisch and 1994 for Dr Becker. In addition, Dr Frisch did not include figures for income year 1993. This is because no detailed income statement was available for the 1992 calendar year. Dr Becker took the actual sales, assumed that costs and margins were the same as in 1993 and made his calculations accordingly. By definition, he arrived at the same margin for 1992 as he calculated for 1993. This leads to a slight difference in margin expressed as a percentage between Dr Frisch and Dr Becker for the total period under consideration.

130. A further confusion arose in relation to the inclusion of period costs in the calculation of gross margins. Period costs are defined by Mr Hammond in his statement as "balancing figures that make up the difference between the cost of goods sold in Roche Australia's product contribution report with the cost of goods sold in the individual accounts." Dr Becker observed that Dr Frisch, in calculating the gross margins in his first report, did not consistently account for period costs. Dr Becker pointed out that because period costs could not be assigned by product for each year, Dr Frisch excluded these costs from his resale price analysis for the years 1998 to 2002. Dr Frisch did not, however, exclude these costs for the years 1993 to 1997. This meant that Dr Frisch compared gross margins on comparable products (without period costs) to gross margins on non-comparable products (with period costs). Dr Frisch, in his second report, acknowledged this error, and ultimately agreed with Dr Becker's figures for the total sales and cost of goods sold by Roche Australia.

131. The average gross margin achieved by Roche Australia in its Prescription Division was 37.54 percent (Dr Frisch) and 36.1 percent (Dr Becker). This margin does not reflect clinical trials. The sole cause of the difference is the inclusion of figures for calendar year 1992 by Dr Becker. The gross margin he arrived at for that year was 36.6 percent which reproduces the agreed gross margin for calendar year 1993.

132. The gross margin figures for each year, subject to the above comments, are as follows:

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
36.6% 36.6% 38.0% 31.0% 31.1% 33.8% 40.7% 36.7% 38.7% 34.5% 37.5%

133. The method employed by both Dr Frisch and Dr Becker was to compare the profit margin earned with what they found to be a comparable arm's length profit margin. They took an average profit margin for all years as that comparator, although asking themselves the question whether this was a fair approach. While the approach may be reasonable for an expert considering whether there is evidence of transfer pricing at other than arm's length prices, I do not think that the Commissioner or this Tribunal can ultimately act on that basis. This is because the task of both the Commissioner and the Tribunal is to consider taxation assessments for separate years. The focus must be on the separate prices in each of the years under consideration. It accordingly seems to me to be necessary to look at each year separately and to the gross profit margin in each year.

134. The controversial question is what products should be used as comparables. The Commissioner primarily asserts that Inhibace should be the sole comparable. If this approach were to be accepted the conclusion is simple. If Roche Australia should have paid a price for its prescription drugs in each year which would return a gross margin of 60 percent or 65 percent then in each year under consideration the present assessments would be based on an understatement of Roche Australia's income. The assessments would plainly not be excessive.

135. The position becomes more complicated when the experts' opinions are addressed. It is to be remembered that none of the experts considered Inhibace. Both Dr Frisch and Dr Becker considered third party sales of isotretinoin, moclobemide and calcitriol.

136. The third parties who acquired the three drugs were Alphapharm, Arrow, Hexal and Biochemie. Dr Frisch considered all strengths of the drugs sold to all of these companies. He made adjustments to compensate for packaging and other differences to enable comparison with the prices paid by Roche Australia. Dr Becker undertook the same exercise except that he relied only on the sales to Alphapharm and only on some of them. He made adjustments "which were essentially the same as the adjustments performed in the FRISCH REPORT".

137. Both Dr Frisch and Dr Becker prepared adjusted prices for the generic sales. However, the adjusted results are little different although they slightly favour the case of Roche Australia. The unadjusted prices appear in Attachment 1. There prices are used for ease of comparison although I have also examined and compared the adjusted prices. Where the figures for Dr Frisch show a range there was a price change during the year.

138. The parties made no effort to resolve the differences between the experts' figures. This was no doubt because the differences were not substantial and the time and effort involved would not have been productive. The figures broadly lead to the same conclusions.

139. The prices paid for the three drugs by Roche Australia were, on average, less than the prices paid by the generics and they were less in most, but not all, instances. A particular exception is isotretinoin 20mg in 2000 and 2001.

140. The technique chosen by both Dr Frisch and Dr Becker was to isolate comparables, adjust them and then compare the comparables with the prices paid by Roche Australia for the same drugs. Dr Frisch used all the drugs as comparables. Dr Becker primarily relied on all the Alphapharm purchases other than isotretinoin 20mg.

141. Dr Frisch concluded that Roche's purchases of the comparable drugs were at arm's length prices because the overall purchase prices for the comparable drugs exceeded the prices Roche Australia paid overall, though not for every drug in every year. In other words, the average prices they paid were less than the average prices paid by the generics.

142. That left the prices paid for drugs for which there were no comparable drugs to be assessed. Dr Frisch originally calculated that the average percentage gross margin earned by Roche Australia on the drugs with comparable prices was 37.1 percent. The average percentage gross margin earned on the other drugs was 37.0 percent. Dr Frisch concluded that the difference was immaterial and it was "valid to conclude that, since the transfer prices for the [comparable] products were arm's length, the transfer prices for the non-[comparable] products were arm's length as well".

143. Dr Becker raised issues relating to Dr Frisch's calculations of 37.1 percent and 37.0 percent, including the period costs. Dr Becker concluded that the actual margins were 37.2 percent and 35.8 percent. Dr Frisch accepted the corrections and the reasons for them, but said that they did not affect his conclusion.

144. Dr Becker undertook a similar exercise to Dr Frisch with the Alphapharm drugs other than isotretinoin 20mg. This yielded comparisons of 40.5 percent and 35.5 percent for the gross margins for the comparable drugs and the non-comparable drugs. The difference of 5 percent, applied to sales of the non-comparable drugs of $1,309,246, lead to an uplift in profit of $65,530,000.

145. The comparative figures are set out in Attachment 2.

146. The approach of Dr Frisch was to include all generic sales of comparable products. That seems to me to be a principled approach. It removes one area of subjective decision-making. All arm's length transactions are included. It is accepted that Roche Australia had a wide range of products and market sizes. There will be anomalous or atypical sales but the generic sales under consideration do not seem to me to fall into that category.

147. I was troubled by Dr Becker's excluding isotretinoin 20mg. He did this because of the low market percentage which Alphapharm had with the product. He equated this with "low volumes", making it less appropriate to use the purchase as a "pricing benchmark". The reality is, however, that the absolute sales of isotretinoin 20mg were quite high. It is plain that the market for the 20mg presentation was much greater than the market for the 10mg presentation. In 1999 Roche Australia sold 5,520kgs of the former and 97kgs of the latter. Alphapharm purchased 433kgs of the 20mg presentation and 7kgs of the 10mg presentation. It struck me as surprising that acquisitions of 433kgs in one year would be excluded on the ground that the volume is low while acquisitions of 7kgs would be included. I find it hard to believe that one quantity should be included on the grounds of low volume although a quantity 60 times as great was excluded. The argument that the market percentage is what matters does not seem to me to be persuasive.

148. The issue of whether isotretinoin 20mg should be excluded is quite important because it is one of the few drugs which Alphapharm was able, for a time, to acquire more cheaply than Roche Australia. Excluding isotretinoin 20mg from the bases changes the average profit margin of Roche Australia on the comparables from 37.2 percent to 40.5 percent.

149. I propose to explore the experts reasoning further without excluding isotretinoin 20mg from the comparables to be considered. I do not think that a better approach would be to exclude both isotretinoin presentations. I accept Dr Frisch's opinion that as many comparables should be included as possible. This is particularly so in a case, such as this, in which comparables are hard to find. I accept that atypical transactions should be excluded, but I do not find either isotretinoin presentation to be in that category. Nor do I think that moclobemide should be excluded, although Dr Becker has raised the possibility. In any event, excluding both isotretinoin and moclobemide would unnecessarily reduce the base from which the comparisons are to be made.

150. I accordingly prefer Dr Frisch's method of dealing with the sales of isotretinoin, moclobemide and calcitriol to that of Dr Becker, and accept that Roche Australia earned a gross margin from those sales of 37.2 percent and a gross margin of 35.8 percent from the other sales. If there were nothing more, like Dr Frisch, I would find that Roche Australia's purchase of the comparable products were arm's length purchases and that, at most, the balance of sales required an uplift of 1.4 percent from 35.8 percent to 37.2 percent.

151. However, my task is to arrive at a decision as to what were arm's length prices for the acquisitions in question. I will be guided by the words of the legislation extracted above. The experts' opinions will assist me, but they are not determinative. I must arrive at my own decision. That may require me to look at other matters, provided that they are relevant and probative.

152. The Commissioner relies particularly on two matters. The first is the Inhibace sales. The second is the low level of profitability of Roche Australia generally. The Commissioner says that the Inhibace profit margin should be taken as the sole guide. I do not agree with that, but I do not think it can be dismissed. The general profit level of Roche Australia is of less significance but I consider that I can refer to it. It also seems to me that I can have regard to the evidence relating to the structuring of the generic sales to Alphapharm, namely that they were generally structured to yield a profit margin of 40 percent.

153. Dealing first with profitability, the evidence is that the gross profit margin for the Roche Group's Prescription Division was 75 percent. This is contrasted with Roche Australia's 36.1 percent. However, it is necessary to note what it is that the parent division is marketing by comparison with the subsidiary. Roche Basel owned the intellectual property associated with the drugs. It is the intellectual property which is really the product, not the pill or capsule by which it is dispensed. The intellectual property included patent rights. The intellectual property came from very substantial expenditure on research and development, much of which would have produced no results. The profits from the exploitation of the intellectual property rights was something to which Roche Basel had a special claim even though the profit would be collected for Australian sales by the Australian subsidiary. The distinction between the two types of profits is discussed in the patent extension cases such as
Bayer AG v Minister for Health (1988) 96 FLR 50 where they are referred to as "the profits of the patentee as such" (s 93(b) of the Patents Act 1952 (Cth)) and manufacturing or distributing profits (see p 69). In the absence of a substantial market for arm's length sales of prescription pharmaceuticals it is not easy to determine how a subsidiary should be rewarded for its part in the marketing of the intellectual property component of prescription pharmaceuticals.

154. The retail market for pharmaceuticals is quite unique. There are two important factors which contribute to this. First, many drugs sold in the market are patented. Accordingly, some participants are entitled to use monopolist conduct in their marketing. However, not all drugs have this patent protection. Secondly, drugs can only be sold to persons for whom they have been prescribed by a medical practitioner. General media advertising is largely prohibited.

155. New patented drugs need to be launched and promoted. Although a pharmaceutical company may have a monopoly, it does not follow that there are not other drugs, also protected by patent, which are used to treat the same conditions. Some drugs are destined to be successful because they involve a considerable advance in heath care. Other drugs will need more promotion. Further clinical trials are among the methods used to assist in establishing a reputation. Promoting drugs through detailing them to medical practitioners is another practice.

156. The marketing of older drugs at the end of their patent life is different. No doubt the patentee continues promotion through detailing. Medical practitioners are likely to be encouraged to continue prescribing the established brand: Roaccutane rather than isotretinoin; Aurorix rather than moclobemide; Rocaltrol rather than calcitriol. However, independent generic companies, like Alphapharm, will be entering the market either because the patent protection has expired or because the patentee is prepared to wholesale the drug to give it some control over the wider marketing of the drug near the end of its patent life in anticipation of other companies entering the market.

157. Because the period covered by the assessments under consideration is eleven years, the activities of Roche Australia cover the whole spectrum. Indeed, a drug could go through the whole range during a period that long.

158. The comparable drugs selected by Dr Frisch and Dr Becker are within the latter category. Indeed, one reason why there is data relating to comparable sales is that Roche Australia was prepared to sell the drugs to the generics. That is unusual for a drug with potential in its early life.

159. Inhibace is much more representative of the first category. It was covered by patent. That suggests it might provide a balance to the comparable sales to the generics. However, Roche Australia says Inhibace is atypical and should be disregarded. Setting aside the fact that it yielded a much higher gross margin than the comparable sales which Roche Australia wishes to rely on, what reasons do they give?

160. Mr Nadjarian says he did not want to market the drug in Australia. He thought the market was oversupplied with competitors. He did not think it would be profitable. He says that subsequent events bore this out because Bayer Australia ceased to sell it. Nevertheless, Roche Basel thought it was a "profit driver" which would be successful. An assessment of whether the price for which it was sold was an arm's length price will depend more upon whether the patentee thought it would be successful than upon whether the patentee turned out to be correct. In any event, the sale stands as a de facto arm's length sale of a drug that was marketed in Australia. Dr Frisch said that there may have been similar drugs in Roche Australia's portfolio. His approach of including everything might well have led him to including it in his analysis had not Mr Nadjarian said that it should be disregarded.

161. It seems to me that Inhibace is a drug to which reference should be made. However, I need to be very careful in its use because the evidence about it is limited and because the experts did not deal with it. Nevertheless, there is a lot of evidence in this case about the pharmaceutical market and the way it operates and there is clear evidence of the gross margin basis upon which Inhibace was sold to Bayer.

162. In 1998 Roche Australia achieved a profit margin of 40.7 percent. In two other years it achieved 38.0 percent or more. The time during which the gross margins for the Prescription Division were at their highest was in 1998 and after. The year 1994 is an exception. There is evidence that after the audit commenced, Roche Basel reduced its transfer prices to Roche Australia. The Taxation Department of Roche Basel was involved. The evidence is that the prices for the sales to Alphapharm and the other generic companies were negotiated around a predicted 40 percent gross margin. The sales of Inhibace were made at a predicted gross margin of 60 percent to 65 percent. All of this needs to be balanced with the average margins of 36.1 percent earned by Roche Australia from its prescription sales overall and its margin of 37.2 percent in its sale of comparable products and 35.8 percent in its sale of non-comparable products.

163. It was one thing for Dr Frisch to say that 37.1 percent and 37.0 percent were not sufficiently distinguishable for a conclusion to be drawn that the sales that yielded one were, and the sales that yielded the other were not, arm's length. However, it is another thing to come to the same conclusion about the difference of 1.4 percent between 35.8 percent and 37.2 percent. Small differences in gross margin can reflect significant differences in price. In the present case an uplift of 1.4 percent would increase profit by more than $16,000,000 over the eleven year period.

164. I do not think that there is a rational basis for distinguishing between profit margins earned by Roche Australia for comparable and non-comparable drugs. I prefer not to calculate a gross margin for the drugs that have comparables and a gross margin for those that do not, determine if the former is arm's length and then evaluate the latter. There is nothing in the evidence which suggests a different process was adopted in pricing the drugs. It is unlikely that one group is arm's length and the other not. It also seems to me appropriate to have regard to the whole of the evidence and not simply the material coming from the work of Dr Frisch and Dr Becker. That necessarily precludes an approach which is limited to comparing Roche prices paid for comparable drugs with prices paid for non-comparable drugs. Whether there were comparable or non-comparable drugs is largely a matter of accident. Using an arm's length price for one group of drugs to determine a profit margin for all other drugs when there is other useful evidence does not seem to me to be appropriate. I do accept, however, that the evidence does not permit determination of prices for individual drugs and that it is necessary to take some overall view.

165. When I take into account all these matters I conclude that an arm's length price for prescription pharmaceuticals would have yielded Roche Australia a gross profit margin of at least 40.0 percent throughout its range. I base this conclusion on a finding that 40.0 percent is the gross margin that arm's length parties would generally negotiate about. They might negotiate about a higher price. However, a margin of 40.0 percent would still be arm's length. This is what happened with the generics. The Inhibace sale which, on one view, might be the closest comparable was based on a gross profit margin of at least 60.0 percent. That might justify a finding that the proper gross profit margin is of the order of 50.0 percent or higher. However, I recognise that there must be a range. If all the evidence pointed to 40.0 percent it might be said that 38.0 percent was within the range. Adopting a very cautious approach to Inhibace I find that it has the effect at least of putting 40.0 percent at the bottom of any range. I also note that the arm's length price might not yield the anticipated margin. Again, the Inhibace agreement leads me to conclude that 40.0 percent remains the bottom of the appropriate range.

166. The 40.0 percent margin should be applied to each year. The resulting uplift of profits would be as follows (figures in millions):

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Sales 57,318 67,658 80,349 90,649 128,704 140,722 135,569 156,411 194,595 208,027 213,094
COGS (36,347) (43,364) (51,603) (63,535) (82,291) (82,654) (84,198) (94,516) (114,600) (138,934) (133,291)
Gross Profit 20,971 24,294 28,746 27,114 46,413 58,068 51,371 61,895 79,995 69,093 79,803
Gross Margin % 36.6% 35.9% 35.8% 29.9% 36.1% 41.3% 37.9% 39.6% 41.1% 33.2% 37.4%
Uplift 1,956 2,769 3,393 9,145 5,069 - 2,857 669 - 14,118 5,434
Gross Profit 22,927 27,063 32,139 36,259 51,482 - 54,228 62,564 - 83,211 85,237
Gross Margin % 40.0% 40.0% 40.0% 40.0% 40.0% - 40.0% 40.0% - 40.0% 40.0%
  Total uplift 45,410

There should not be any adjustment for 1997 or 2000 because that would be the same as proceeding from the averages. In any event, a generous arm's length price secured in one year should determine taxable income for that year and not a higher arm's length price attributed to other years.

Clinical trials

167. Although the profits in this part of the Prescription Division were raised at an earlier point in time and have been dealt with by the experts, the Commissioner no longer relies on any aspect of these activities to support the assessment. Accordingly, although a brief examination of this aspect of Roche Australia's activities was relevant to the context and background in which the matters before me arose, I do not need to deal with them further.

The Consumer Division

168. The experts accept that no comparable sales are available. They are driven to profits based methodology.

169. The substantial issue between the parties is whether it is legitimate to isolate a group of products for separate and individual treatment. Roche Australia says that it is not legitimate to use the operating profits of chosen comparable organisations to test the operating margins on particular products of the subject. It is argued that individual products such as the Category 1 products cannot be separated out in this way.

170. The Consumer Division revenue returned an EBIT to sales percentage of 8.1 percent during the income years 1994 to 2002, which are the years in which the Division was trading. Aspro was successful (4.2 percent) and Interdens was very successful (40.7 percent) but Rennies (-56.6 percent) and Elevit (-55.2 percent) and particularly Aleve (-317.9 percent) were unsuccessful.

171. Mr Nadjarian has given explanations for this situation. Interdens was very profitable. Aspro was profitable but not as profitable. I need not trouble with what Mr Nadjarian said about them. Much hope was apparently held for Rennies, but it was not justified. It returned a gross profit of 44.6 percent on sales. A table prepared by Dr Frisch, which I am not aware of having been challenged, shows the figures. Admittedly, it covers the period as a whole. It does not isolate individual tax years. However, it shows (the gross profit margins are mine):

Sales COGS Gross Profit Gross Profit Margin
Aspro 92,041,000 47,952,919 44,088,681 47.9%
Interdens 1,242,200 325,933 916,267 73.8%
Rennies 12,390,800 6,870,692 5,520,108 44.6%
Aleve 740,900 699,931 40,969 5.5%
Elevit 1, 796,000 1,023,905 772,095 43.0%

172. One starts with the proposition, perhaps irrelevant, that the gross profit margin for each drug (other than Aleve) is substantial. I say "perhaps irrelevant" because neither the parties nor the experts appear to deal with the gross margins in this aspect of the case. In a case in which the primary matter for concern is the cost of acquisition of goods and not profits (subject to considerations relating to the different tests in the double tax treaties and Division 13) this is surprising. I wonder whether there was enough "standing back and looking at the canvas in this case" or whether the case was too influenced by the ideas of US economists steeped in their traditions of transfer pricing issues rather than the application of Australian legislation. In a case involving very substantial sums with teams of two senior counsel and two junior barristers on each side together with a hearing room containing up to 20 persons instructing each side, with an average of not much less, on each of the 12 hearing days, it is surprising that some of these fundamental matters were not addressed.

173. The basic reason why the experts did not address comparable purchases was that none could be found. A second best method was chosen. However, I find it difficult to see that addressing gross profit margins was not at least relevant, particularly as not much time was spent dealing with operating margins in connection with the Prescription Division.

174. The problem is compounded by the fact that a comparison of gross profit margins and EBIT to sales for each of the products shows that it was the operating expenses that caused potentially profitable operations to result in losses. This is particularly so with Aleve and Elevit. Yet operating expenses were hardly considered when Dr Frisch and Dr Becker were examining the Prescription Division.

175. The problems of the approach are highlighted in the submissions on behalf of Roche Australia. Dr Becker ignored the overall operating margin of Roche Australia's Consumer Division of 8.1 percent and concentrated on the five Category 1 products. His analysis of other companies led to the conclusion that an operating margin of 1.7 percent would be arm's length. He gleaned this from undertaking a process not unlike that undertaken by Dr Wright to isolate comparable company profits. As Roche Australia points out they were overall profits, not profits on particular products. When the arm's length profit of 1.7 percent is applied to the Category 1 products an uplift of $11,957,000 is required, bringing an EBIT of -$10,086,000 up to $1,872,000. This is to be contrasted with an uplift on the part of Dr Wright to $6,087,904 representing an operating profit margin on EBIT to sales of 5.6 percent if the median is taken.

176. The uplift is more than the total cost of goods sold for all of the Category 1 products except Aspro. Accepting that Aspro and Interdens both earned operating profits which must be accepted as arm's length the unavoidable conclusion of the uplift proposed by Dr Wright is that the arm's length market price for Rennies, Aleve and Elevit should have been nothing, or, even worse, a negative price. This cannot be right.

177. The only proper conclusion, accepting that the overall operating profit of the Consumer Division is well in the arm's length range, is that the acquisition prices for the Category 1 products were arms length as well. There can be no doubt of that for Interdens. The same must be true of Aspro. It showed a gross margin of 47.9 percent and EBIT to sales ratio of 4.2 percent, well above Dr Becker's figure, although below Dr Wright's.

178. Elevit also showed a healthy gross margin of 43.0 percent. It is apparent that the problem with this drug was not the price paid for it but the lack of a market for it notwithstanding every attempt to find one. That leaves Aleve. It did not even return a reasonable gross margin. However, the problem with Aleve was so overwhelmingly associated with its marketing failure at the cost of more than $2,000,000 against purchase costs of $699,931 that I do not think it can be said that it was bought at an overvalue. The disaster which this drug represented is associated with what must have been grossly disappointed expectations. I cannot find that it was acquired at an overvalue. It would still have been a disaster if it had been given to Roche Australia. The loss on this product had very different causes than acquisition at an overvalue. As to the problems with the Category 1 drugs which were not successful I accept Mr Nadjarian's explanations. I find that the acquisition of property in the Consumer Division were arm's length.

179. It follows that the decisions under review will be set aside insofar as the Consumer Division is concerned and assessments will be issued in accordance with the returns.

Diagnostics Division

180. The Diagnostic Division was unprofitable in every year from 1992 to 1998 with the exception of 1996. Its gross profit margin varied between 15.7 percent and 50.0 percent. The figures are as follows:

Calendar Year
($ 000) 1992 1993 1994 1995 1996 1997 1998 Total
Sales 5,110 5,643 5,878 6,716 8,685 11,296 6,875 50,202
Cost of Goods Sold (2,533) (2,948) (3,277) (3,611) (4,293) (8,653) (5,797) (31,112)
Gross Profit 2,577 2,695 2,600 3,105 4,392 2,643 1,078 19,090
Gross Margin
Operating Expenses
50.4%
(2,802)
47.8%
(3,090)
44.2%
(3,125)
46.2%
(2,732)
50.6%
(3,759)
23.4%
(3,526)
15.7%
(3,646)
38.0%
(22,680)
Other Operating Income (416) (831) (794) (486) (109) 3 63 (2,571)
Other Addbacks/Adjustments (696) (292) (48) (397) 245 (202) (290) (1,680)
Earning before Interest and Taxes (EBIT)
EBIT/Sales
(1,337)
−26.2%
(1,519)
−26.9%
(1,366)
−23.2%
(511)
−7.6%
769
8.9%
(1,082)
−9.6%
(2,795)
−40.7%
(7,841)
−15.6%

181. Both Dr Wright and Dr Becker were unable to find comparable products. They again adopted a profit based method.

182. Dr Becker calculated that an arm's length operating margin would be 4.6 percent. He used the profits of companies he found to be comparable. The median chosen by Dr Wright was 6.7 percent. The effect of Dr Becker's approach was to turn a total loss of $7,841,000 into a profit of $2,298,000 and to reflect this wholly by a reduction in cost of goods sold of $10,138,000. The figures for Dr Wright were higher.

183. Dr Becker's approach would effectively reduce the total cost of goods sold from $31,112,000 to $20,974,000. That would make the gross profit $29,228,000 and the gross profit margin 58.2 percent.

184. There is evidence before me that one of the reasons the Division was unsuccessful was that it did not tie purchasers of its equipment to use of its products.

185. One of the problems of profit based methodology is that, when applied to transfer pricing, it inevitably attributes any loss to the pricing. Where operating expenses are higher these may place some of the emphasis of the cause of the loss on the wrong area. After all, it is certainly true that there are companies which make losses for reasons other than the prices for which they acquire their stock . The Australian operations of multinational companies are not necessarily excluded from this.

186. Mr Nadjarian says he was advised to close down the Diagnostics Division because of its losses. In his submissions the Commissioner said that he "accepts [Roche Australia's] evidence about the commercial circumstances facing the diagnostics division". However he criticised Roche Australia for not adducing "any evidence of the goods acquired, the prices paid, how they were set, and what the arm's length price was".

187. There is, of course, evidence of the total sales and costs of goods sold for each year. The gross profit margin is known. How they were calculated is ultimately not the issue because the process adopted to arrive at price was not an arm's length process. The real question is what was an arm's length price. That primarily involves looking at outside sales. None of the experts were able to find any. Because of the nature of the industry, namely that it largely involves sales from holding company to subsidiary, rather than third party, there may not be any such sales. The Commissioner invokes
Jones v Dunkel (1959) 101 CLR 298; [1959] ALR 367. However, I do not think that the rule in
Jones v Dunkel enables me to make any positive finding relating to comparable sales in the absence of evidence. Moreover, this Tribunal is exercising administrative power. It is making the correct or preferable decision (taking into account the onus) not resolving a dispute associated with an assertion and a rebuttal. Although some of the considerations lying behind the rule in
Jones v Dunkel will apply to proper administrative decision-making, it does not necessarily apply to administrative decision-making as a rule. Indeed, Tamberlin J in the Federal Court of Australia has recently held that it does not apply (
Green v Minister for Immigration and Citizenship [2008] FCA 125 at [41]).

188. Roche Australia has satisfied me, based on the totality of the evidence before me, that the prices for which it acquired the products sold in its Diagnostic Division were arm's length prices. This is just like the Consumer Division. The bad results flowed from operating expenses not acquisition prices.

General consideration

189. Submissions were put to me on two particular matters I have not dealt with so far. The first was whether the double tax treaties as incorporated into Australian law conferred a power to assess. The second was whether it was open to the Tribunal to exercise the powers of the Commissioner conferred by s 136AD(4).

190. So far as the first is concerned I note that the submissions were limited (particularly those of the Commissioner) and both parties accepted that the result in this case would not be affected if the treaties conferred no power to assess. This is because the issues in this case concerned pricing and, to the extent that the double tax treaties relate to profits, the only ultimate relevance of profit was that it reflected prices. Notwithstanding the different tests of independent pricing and arm's length dealing it was accepted that these are essentially the same tests, a proposition which is supported by the OECD Guidelines.

191. In the result I do not need to decide the issue although I note that there is a lot to be said for the proposition that the treaties, even as enacted as part of the law of Australia, do not go past authorising legislation and do not confer power on the Commissioner to assess. They allocate taxing power between the treaty parties rather than conferring any power to assess on the assessing body. On this basis Division 13 should be seen as the relevant legislative enactment pursuant to the power allocated.

192. It was suggested to me that by the time this matter came to the Tribunal the power in s 136AD(4) was effectively spent. I am not sure that this is the case. I do not read
W R Carpenter Holdings Pty Ltd v Federal Commissioner of Taxation 2007 ATC 4679; (2007) 161 FCR 1 as requiring a contrary conclusion. As I read s 136AD(4) it empowers the Commissioner to issue an assessment notwithstanding that there is not sufficient evidence which would ordinarily enable him to do so. I do not see why, on review, the Tribunal does not have the same power. The power is to use material which might otherwise be less than persuasive, or to reason from information in circumstances where reasoning might not otherwise be fully justified. Nevertheless, the process needs to be a rational one. The discretion needs to be exercised in accordance with reason. The present case comes close to such a case. Very little in the material is satisfying in terms of persuasiveness. Nevertheless, I have been able to come to the conclusion to which I have arrived without resorting to s 136AD(4).

Conclusion

193. The result is that Roche is successful with respect to the Consumer Division and the Diagnostics Division. The taxable income for the Prescription Division will be higher than that contended for by Roche but, in most cases, will not be as high as the income on which the amended assessments are based. The amended assessments must now be addressed.

Impact on assessments

194. Before me are applications for review of objection decisions made by the Commissioner for each of the income years ending 30 June 1993 to 2003 (calendar years ending 31 December 1992 to 2002). The uplift I have arrived at should be applied to the income years ending 1993 to 1996 and 1998 to 2001. Amended assessments will be issued in the appropriate amounts. However, the position is not so simple for 1997, 2002 and 2003.

195. In each of those years the amended assessments raised in accordance with the Commissioner's objection decisions are for less than the amount which would flow from these reasons. The question is whether the Tribunal can arrive at a decision which will have the effect of increasing the assessment for any year. The figures are as follows:

Year of income ended 30/6/97 30/6/02 30/6/03
Calendar Year 31/12/96 31/12/01 31/12/02
Taxable income as returned 4 6,642,576 15,061,618
Taxable income as assessed following audit 13,665,239 21,032,908 16,512,736
Taxable income as assessed after objection 13,665,239 18,440,398 15,275,059
Taxable income as determined in these proceedings 14,556,005 20,760,681 20,496,425

196. Section 170(2) of the Assessment Act empowers the Commissioner to amend assessments. In the absence of fraud or evasion amendment is generally permitted within 4 years. The amendments following the Commissioner's objection decisions were made within that period for the years 2002 and 2003 but outside the period for the year 1997.

197. Section 170(9B) refers to amendment "giving effect to a prescribed provision or a relevant provision". Section 136AD is a prescribed provision (s 170(14)). The subsection provides that "nothing in this section prevents the amendment, at any time, of an assessment for the purpose of giving effect to a prescribed provision…". However, s 170(9C) contains a limitation on the power. It provides that subs (9B) does not authorise the Commissioner to amend an assessment where "the prescribed provision has been previously applied, in relation to that supply or acquisition" (s 170(9C)(a)).

198. Broadly speaking, the Commissioner is authorised by s 170(9B) to amend at any time to raise an assessment under s 136AD but not to vary an assessment previously made under that section.

199. Since the assessments subject to the objection decisions under review for 1997, 2002 and 2003 were all amended assessments which raised income tax under s 136AD, Roche argues that there was no power to further amend to increase the amount of assessable income at the time the Commissioner considered the objections. Roche says that the power to amend given by s 170(9B) is the only power to amend an assessment raised under s 136AD. It submits that the general power under s 170(2) is not available. Accordingly, even though the objection decisions were made less than four years after the assessments for 2002 and 2003 were made, the Commissioner did not have power, when making the objection decisions, to amend under s 170(2).

200. The power of the Commissioner at the time of the making of the objection decisions is important because that may determine what power this Tribunal has. Section 170(7) provides that nothing in the "section shall prevent the amendment of any assessment in order to give effect to the decision upon any appeal or review". The subsection facilitates the making of amendments to assessments where required by properly made decisions of this Tribunal. However, the subsection says nothing about the circumstances in which such decisions may be made. The Tribunal's authority comes from s 43 of the Administrative Appeals Tribunal Act 1975(Cth) and the Act conferring the review power on it. The present review power, conferred by s 14ZZ of the Administration Act, authorises, relevantly, the review of objection decisions.

201. Difficult questions can arise as to the point of time to which a Tribunal review is directed. Aspects of the issues that arise are presently before the High Court of Australia in an appeal from
Shi v Migration Agents Registration Authority (2007) 158 FCR 525; (2007) 240 ALR 23; (2007) 95 ALD 260 which has been argued but not decided. However, these questions do not directly arise here.

202. Section 43(1) gives to the Tribunal "all the powers and discretions that are conferred by any relevant enactment on the person who made the decision…". That power must confer upon the Tribunal all the powers and discretions which the Commissioner had when making the objection decisions. If the powers included the power to amend under s 170(2) then that power must be available to the Tribunal.

203. The initial question is whether, on consideration of an objection decision, the Commissioner, and on review, the Tribunal, has power to exercise powers and discretions not raised by the objection. If the Commissioner retains a power generally to amend, for example, under s 170(2), it would seem that he must have that power, at least incidentally, at the time the objection decision is made. In that event I see no reason why the Tribunal would not have the same power on review. This result seems to me to be required by the words of s 43 even though, at the time the Tribunal hears the matter, the time for amendment has expired.

204. This conclusion accords with the decision of Jenkinson J in
Stevenson v Commissioner of Taxation 91 ATC 4476; (1991) 29 FCR 282 (at 299) and the decision on which Jenkinson J's conclusion was based,
Fletcher v Commissioner of Taxation 88 ATC 4834; (1988) 19 FCR 442 (at 452-454).

205. The question remains, however, whether at the time he made his decision on the objections the Commissioner had power to amend the assessments to increase the taxable income pursuant to s 170(2).

206. It is clear that s 170(7) does not confer power on the Tribunal which was not otherwise conferred on the Commissioner. However, ss 170(2) and 170(9B) and (9C) are capable of standing together. The general power to amend will be available for four years. After that time has expired the Commissioner will be empowered to amend an assessment to include assessable income under s 136AD but not to increase the amount of any liability previously raised. Section 170(7) would appear to authorise a reduction because it extends past amendment consequent upon a decision "upon any appeal or review" to "amendment by way of reduction in any particular in pursuance of an objection…".

207. Subsection 170(9B) is couched in negative terms. Effectively, it states a double negative. It provides that "nothing in this section prevents the amendment, at any time, of an assessment" under s 136AD. Double negatives are not equivalent to positives. The section focuses on what the section does not do, not on what it does do. The words of s 170(2) are clear and, relevantly, unqualified. I do not think the words of s 170(9B) and (9C) can be read as if they contained a provision that no amendment can be made, at any time, to an assessment under s 136AD. The better reading of the two sections is that s 170(9B) and (9C) extend the time for a first assessment without disturbing or extending the power to amend otherwise conferred. I note that this reading of the sections accords with the Explanatory Memorandum accompanying the Income Tax Assessment Amendment Bill 1982 (Cth) (at 79).

208. It follows that the Tribunal has the power, which the Commissioner had at the time he considered the objection decisions, to increase the amount of assessable income.

209. Questions may arise whether the Tribunal has a discretion and, if so, how it should be exercised, with respect to such an increase. However, I have decided that, even assuming the existence of a discretion, it should be exercised in favour of amendment to increase the taxable incomes.

210. I say this notwithstanding the fact that at the time of final addresses the Commissioner accepted that individual assessments could not be amended upwards. It was only after these reasons were published in an earlier form that the Commissioner sought to argue that the assessments could be increased.

211. I conclude that the assessments should be increased because these applications were at all times dealt with on an overall basis. During the hearing the parties hardly addressed individual tax years or the existing assessments. The possibility that results contended for by the Commissioner might require increases in individual assessments effectively went unnoticed. Although I had material before me from which I could have made the calculations, that material was not emphasised. Further, the evidence, as appears from the analysis in the body of these reasons, addressed tax years overall. However this matter was to be decided, whether on Roche's expert evidence, the Commissioner's expert evidence, or otherwise, it could not have been decided other than by reference to broad considerations and analysis of the kind discussed in these reasons. In this regard I note that while Roche's evidence, if accepted in full, would have resulted in the objections being allowed in full, that evidence nevertheless approached the issues from a similar, although not identical, perspective to the Commissioner's evidence. It proceeded on the same broad basis. The case having been conducted in this way, so that when the result was determined it might lead to increases in some assessments, it seems to me that, there being power to do so, the assessments should reflect that result when it did occur.

212. It follows that the assessments for 2002 and 2003 should be increased in accordance with the table above.

213. That leaves 1997. The problem is that when the Commissioner dealt with the objection the time for amendment under s 170(2) had passed. Were there nothing more than I have considered above, the Tribunal would have no power to increase the assessable income. However, the Commissioner has mounted an elaborate argument based on a submission that the Tribunal is not addressing the assessment as such and does not have the power to amend assessments (see, e.g. Stevenson at 294). The argument is that what s 136AD(3)(d) provides is that "the Commissioner determines that this subsection should apply in relation to the taxpayer in relation to the acquisition" with the consequence that "consideration equal to the arm's length consideration… shall be deemed to be the consideration…". Assessment does not figure, so the argument goes. There is a search for the true consideration which continues until proceedings are concluded. When the amount is finally determined an appropriate assessment must issue.

214. Apart from being rather too clever it does not seem to me that the argument is right in principal. The fact that the Tribunal does not have power to amend an assessment should not be allowed to ignore reality. The assessment is a document produced within the Australian Taxation Office. For practical reasons, it is appropriate that it should have control over the process. If the Tribunal could amend an assessment, the Commissioner's records, particularly electronic records, would be capable of misleading. However, that does not mean that the Tribunal does not have power to make decisions which will compel the amendment of an assessment. It would be a triumph of form over substance if the wording of a section such as s 136AD were to have the effect contended for. I do not think it does. What spurred the objection decisions was assessments (deemed or actual). It is the assessments upon which the case is focussed. Unless they are to be varied, in the event of success by one or other party, the proceedings will have no utility.

215. Objections and applications for review can be taken from "an assessment, determination, notice or decision…" (s 14ZL(1) Administration Act), where authorised. The present objections were made under s 175A of the Assessment Act which authorises objections against assessments. The actual objection purports to be an objection against the assessment. The fate of the assessments was always at the heart of these proceedings.

216. I see nothing in s 136AD or Division 13 of Part III which would lead to any different result to the normal result that I have found follows from the provisions of s 170(2), namely that this Tribunal's powers of amendment expire when those of the Commissioner expire.

217. It follows that the assessment for 1997 will stand.

218. The parties agreed to forms of decision for each of the years other than 1997, 2002 and 2003. I have slightly varied those forms. The decisions for 2002 and 2003 will follow the same pattern. For 1997 the objection decision will be affirmed although on the basis appearing from these reasons rather than on the basis that it reflects the correct assessable income for Roche.

Attachment 1

Comparison of prices paid in dollars as calculated by Dr Frisch/Dr Becker (income year for Dr Frisch; calendar year for Dr Becker - Dr Frisch's figures shown under prior year for better comparison)

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Isotretinoin 10mg                      
Roche       53.07/53.07     48.70/48.70 45.94/45.76 45.44/45.26 46.74/48.31  
Alphapharm       65.51/68.96   65.51/ 52.09/55.64 49.15/51.84 48.61/51.59 55.74/56.97  
Arrow                      
Hexal                 47.32/ 54.81/  
Biochemie                      
Isotretinoin 20mg                      
R       101.01/101.01   82.04/82.04 92.46/92.46 87.86-/86.89
86.05
85.89-/85.94
91.37
98.53-/91.73
88.75
75.80/75.80
Al       124.54/127.96   83.72/88.34 99.05/102.52 94.12-/86.98
78.76
76.71-/80.87
80.80
87.13-/89.66
88.36
90.33/92.06
Arrow                      
H                 74.16/ 86.28/ 88.20/
B                     72.05/
Moclobemide 150mg                      
R         18.19/18.19 16.32/16.32 16.62/16.62 15.68/15.61 15.51/ 17.71-/16.48
15.95
13.71/13.71
Al         22.92/24.81 17.03/18.71 14.80/18.29 13.97/15.04   14.19-/16.89
14.39
14.71/17.31
Ar                      
H                 12.58/ 14.36/  
B                      
Moclobemide 300mg                      
R                 29.17/ 38.31-/
30.00
27.42
Al                 31.25/33.54 35.68-/37.98
36.19
36.99/38.97
Ar                      
H                 26.04/   30.82/
B                      
Calcitriol                      
R           22.25/22.25 26.32/26.32 24.83/24.74 24.56/24.47 28.05-/26.11
25.25
25.25/25.25
Al           26.15/28.48 30.94/32.92 29.19/30.66 28.87/30.50 32.97-/34.02
33.44
34.18/34.93
Ar                   32.33-/
32.79
33.52/
H                 28.31/ -32.79 33.53/
B                      

Attachment 2
Calender Year
($ 000) 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Total
Sales 57,318 67,658 80,349 90,649 128,704 140,722 135,569 156,411 194,595 208,027 213,094 1,473,095
Cost of Goods Sold (36,347) (42,904) (49,838) (62,590) (88,740) (93,121) (80,326) (99,040) (119,311) (136,233) 133.132) (941,582)
Gross Profit 20,971 24,754 30,511 28,059 39,964 47,602 55,243 57,370 75,284 71,794 79.962 531,513
Gross Margin 36.6% 36.6% 38.0% 31.0% 31.1% 33.8% 40.7% 36.7% 38.7% 34.5% 37.5% 36.1%
Dr Frisch Comparables                        
Sales - - - 17,607 17,074 47,534 45,543 46,794 51,064 44,771 31,752 302,139
COGS - - - 11,827 11,102 29,669 24,520 29,042 33,134 30,580 19,931 189,805
Gross Profit - - - 5,779 5,972 17,865 21,022 17,752 17,930 14,191 11,821  
Gross Margin - - - 32.8% 35.0% 37.6% 46.2% 37.9% 35.1% 31.7% 37.2% 37.2%
Non-Comparables                        
Sales 57,318 67,658 80,349 73,042 111,630 93,189 90,026 109,617 143,531 163,256 181,341 1,170,956
COGS 36,347 42,904 49,838 50,762 77,638 63,452 55,806 69,999 86,177 105,653 113,201 751,777
Gross Profit 20,971 24,754 30,511 22,280 33,992 29,737 34,220 39,618 57,353 57,603 68,140 419.179
Gross Margin 36.6% 36.6% 38.0% 30.5% 30.5% 31.9% 38.0% 36.1% 40.0% 35.3% 37.6% 35.8%
Dr Becker Comparables                        
Sales - - - 397 17,074 25,268 22,287 21,959 32,740 26,276 17,847 163,849
COGS - - - 265 11,102 15,010 11,359 12,884 19,434 16,563 10,824 97,441
Gross Profit - - - 132 5,972 10,258 10,928 9,075 13,307 9,713 7,023 66,408
Gross Margin - - - 33.3% 35.0% 40.6% 49.0% 41.3% 40.6% 37.0% 39.4% 40.5%
Non-Comparables                        
Sales 57,318 67,658 80,349 90,252 111,630 115,454 113,282 134,451 161,855 181,751 195,247 1,309,246
COGS 36,347 42,904 49,838 62,325 77,638 78,110 68,967 86,156 99,878 119,670 122,308 844,140
Gross Profit 20,971 24,754 30,511 27,926 33,992 37,344 44,315 48,295 61,977 62,081 72,939 465,105
Gross Margin 36.6% 36.6% 38.0% 30.9% 30.5% 32.3% 39.1% 35.9% 38.3% 34.2% 37.4% 35.5%

 

Disclaimer and notice of copyright applicable to materials provided by CCH Australia Limited

CCH Australia Limited ("CCH") believes that all information which it has provided in this site is accurate and reliable, but gives no warranty of accuracy or reliability of such information to the reader or any third party. The information provided by CCH is not legal or professional advice. To the extent permitted by law, no responsibility for damages or loss arising in any way out of or in connection with or incidental to any errors or omissions in any information provided is accepted by CCH or by persons involved in the preparation and provision of the information, whether arising from negligence or otherwise, from the use of or results obtained from information supplied by CCH.

The information provided by CCH includes history notes and other value-added features which are subject to CCH copyright. No CCH material may be copied, reproduced, republished, uploaded, posted, transmitted, or distributed in any way, except that you may download one copy for your personal use only, provided you keep intact all copyright and other proprietary notices. In particular, the reproduction of any part of the information for sale or incorporation in any product intended for sale is prohibited without CCH's prior consent.