COMMISSION DECISION (EU) 2017/502

of 21 October 2015

on State aid SA.38374 (2014/C ex 2014/NN) implemented by the Netherlands to Starbucks

(notified under document C(2015) 7143)

(Only the Dutch text is authentic)

(Text with EEA relevance)

THE EUROPEAN COMMISSION,

Having regard to the Treaty on the Functioning of the European Union, and in particular the first subparagraph of Article 108(2) thereof,

Having regard to the Agreement on the European Economic Area, and in particular Article 62(1)(a) thereof,

Having called on interested parties to submit their comments pursuant to the provisions cited above(1) and having regard to their comments,

Whereas:

1.   

PROCEDURE

(1) By letter dated 30 July 2013, the Commission requested the Dutch authorities to provide information on the tax ruling practice in the Netherlands as well as all rulings related to Starbucks Coffee EMEA BV (hereinafter: ‘Starbucks Coffee BV’) and Starbucks Manufacturing EMEA BV (hereinafter: ‘SMBV’), both companies indirectly controlled by Starbucks Corporation. Starbucks Corporation and all the companies controlled by that corporation are referred to hereinafter collectively as ‘Starbucks’ or the ‘Starbucks group’.

(2) By letter dated 2 October 2013, the Dutch authorities submitted the requested information to the Commission, including the advance pricing agreement (hereinafter ‘APA’)(2) concluded in 2008 between the Dutch tax administration and Starbucks Coffee BV (hereinafter: the ‘Starbucks Coffee BV APA’), the APA concluded in 2008 between the Dutch tax administration and SMBV (hereinafter: the ‘SMBV APA’) and supporting documents. Those documents concern, in particular, a transfer pricing report supporting the request for the two aforementioned APAs (hereinafter: the ‘transfer pricing report’) and other exchanges between the Dutch tax administration and the tax advisor of Starbucks Corporation, [the tax advisor]

 (

*1

)

, (hereinafter: ‘the tax advisor’) on behalf of Starbucks Coffee BV and SMBV(3).

(3) On 9 January 2014, in preparation of a meeting to be held on 15 January 2014, the Commission sent an email to the Dutch authorities listing a number of questions concerning, among others, the transfer pricing arrangement agreed upon in the Starbucks Coffee BV APA and the SMBV APA concluded by the Dutch tax administration.

(4) On 15 January 2014, a meeting was held between the Commission services and representatives of the Dutch tax administration in which the Commission services sought, among others, further clarifications on the adjustments made to the cost base in the transfer pricing report as regards the SMBV APA and the fluctuating royalty payments made by SMBV.

(5) By letter dated 28 January 2014, in response to the questions posed in the meeting of 15 January 2014, the Dutch authorities provided information on the comparability adjustments, the choice of the comparable companies and the fluctuating royalty. Further information on the documents provided is described in recitals 59 to 62 of the Opening Decision as mentioned in recital 9.

(6) By letter dated 7 March 2014(4), the Commission informed the Dutch authorities that it was considering whether the APAs in favour of Starbucks Coffee BV and SMBV could constitute new State aid within the meaning of Article 107(1) of the Treaty and invited the Dutch authorities to comment on the compatibility of such aid. The Commission invited the Dutch authorities to provide any additional information relating to the Starbucks Coffee BV and SMBVAPAs, as well as the tax returns of Starbucks Coffee BV and SMBV and companies related to those two companies in the Netherlands.

(7) By letter dated 21 March 2014, the Dutch authorities responded to the letter of 7 March 2014 and provided the requested tax returns. The Dutch authorities also confirmed that all relevant documents regarding the APAs submitted previously to the Commission had already been provided to the Commission.

(8) On 6 May 2014, a meeting was held between the Commission services and representatives of the Dutch tax administration.

(9) On 11 June 2014, the Commission adopted the decision to open the formal investigation procedure under Article 108(2) of the Treaty on the SMBV APA on the grounds that that APA could constitute State aid within the meaning of Article 107(1) of the Treaty (hereinafter ‘the Opening Decision’)(5).

(10) By letter dated 16 July 2014, the Dutch authorities submitted their comments on the Opening Decision. The submission included, among others, the Roasting Agreement between Alki Limited Partnership (hereinafter: ‘Alki LP’) and SMBV and the Green Coffee Purchase Agreement between SMBV and Starbucks Coffee Trading Company SARL (hereinafter: ‘SCTC’).

(11) By letter dated 25 November 2014, the Commission requested the Dutch authorities to provide the information asked for in the Opening Decision which was only partly submitted by the Dutch authorities on 16 July 2014 and to provide additional information necessary to analyse the SMBV APA.

(12) By letter dated 19 December 2014, the Dutch authorities replied to the letter of 25 November 2014, indicating that part of the requested information is not in the possession of the Dutch authorities.

(13) On 19 December 2014, the Opening Decision was published in the

Official Journal of the European Union

. The Commission invited interested parties to submit their comments on the measure.

(14) By letter dated 16 January 2015, Starbucks submitted its observations on the Opening Decision. Comments on the Opening Decision were also submitted by the Dutch Association of Tax Advisors (De Nederlandse Orde van Belastingadviseurs, hereinafter: ‘NOB’), the Confederation of Netherlands Employers and Industry (Verbond van Nederlandse Ondernemingen & Nederlands Christelijk Werkgeversverbond, hereinafter: ‘VNO-NCW’), ATOZ Tax Advisers Luxembourg, Oxfam International and the Austrian Chamber of Commerce (the Bundesarbeitskammer Österreich, hereinafter: ‘BAK’).

(15) By letter dated 8 January 2015, in response to the Commission's letter of 25 November 2014, the Dutch authorities provided the limited partnership deed constituting Alki LP.

(16) On 12 February 2015, the Commission informed the Netherlands that, in accordance with Article 6a of Council Regulation (EC) No 659/1999(6), it had identified the formal investigation procedure on the SMBV APA as ineffective to date(7). By letter dated 6 February 2015 and in accordance with Article 6a(2)(b) of Regulation (EC) No 659/1999, the Commission requested the agreement of the Dutch authorities to contact Starbucks directly to obtain the missing information.

(17) By letter dated 18 February 2015, the Commission informed the Dutch authorities that it had received observations by a competitor on the value added of the roasting process to green coffee beans and invited the Dutch authorities to comment on this observation. The Dutch authorities submitted their comments on those observations by letter dated 11 March 2015.

(18) By letter dated 12 March 2015, the Netherlands provided its permission to contact Starbucks directly in response to the Commission's letter of 6 February 2015. Following that permission, by letter dated 16 March 2015, the Commission requested Starbucks, based on Article 6(a)(6) of Regulation (EC) No 659/1999, to provide information on the legal structure, the business model with regard to the Starbucks shops, and the raw material used by SMBV, i.e. the green coffee beans (hereinafter: the ‘Starbucks MIT request’).

(19) By letters dated 20 and 26 March 2015, the Dutch authorities submitted their observations on the comments of third parties to the Opening Decision.

(20) On 7 April 2015, following the adoption of the decision of 12 February 2015 and in accordance with Article 6a(6) of Regulation (EC) No 659/1999, the Commission contacted four competitors of Starbucks to provide market information on their business model and their value creating activities so as to enable the Commission to complete its assessment of the case (hereinafter: the ‘competitor MIT request’). Those four competitors included Company Y, Alois Dallmayr Kaffee oHG (hereinafter: ‘Dallmayr’), Nestlé SA (hereinafter: ‘Nestlé’) and Melitta Europa GmbH & Co. KG (hereinafter: ‘Melitta’). The Commission simultaneously informed the Dutch authorities that it had sent requests for information to competitors of Starbucks.

(21) On 13 April 2015, Starbucks submitted the information requested in the Commission's letter of 16 March 2015.

(22) By letters dated 27 April 2015, Dallmayr and Company Y replied to the Commission's request for market information of 7 April 2015.

(23) On 29 April 2015, a meeting was held between the Commission services and Starbucks at which the Commission services provided clarifications on how certain questions in the Starbucks MIT request should be understood in the context of the investigation.

(24) By letter dated 6 May 2015, following the reply of Starbucks of 13 April 2015, the Commission requested Starbucks to provide additional information.

(25) By letter dated 11 May 2015, the Commission requested Company Y to provide further clarifications on the submitted market information. Those clarifications were provided by Company Y by letter dated 21 May 2015.

(26) By letters dated 20 May 2015 and 26 May 2015, Nestlé and Melitta replied to the Commission's competitor MIT request of 7 April 2015.

(27) By letter dated 27 May 2015, the Dutch authorities submitted their comments on the information provided by Company Y and Dallmayr.

(28) By letter dated 29 May 2015, Starbucks submitted their replies to the Commission's request of 6 May 2015.

(29) By letter dated 19 June 2015, the Dutch authorities provided their comments on the information submitted by Starbucks on 13 April 2015 and 29 May 2015.

(30) By letter dated 26 June 2015, the Dutch authorities submitted their comments on the market information provided by Nestlé, Melitta and the clarifications provided by Company Y.

(31) On 29 June 2015, in addition to its submissions of 13 April 2015 and 29 May 2015, Starbucks provided a further substantiation of the supposed arm's-length nature of the transfer prices applied by SCTC for the supply of green coffee beans.

(32) By letter dated 24 July 2015, Starbucks spontaneously submitted additional information on the functions performed by SCTC, SMBV and Starbucks US(8) and provided new figures relating to Alki LP.

(33) By letter dated 5 August 2015, the Commission requested Starbucks to provide clarifications and other documents with regard to its submission of 24 July 2015 to be able to fully analyse the new information.

(34) By letter dated 24 August 2015 and email sent on 26 August 2015, Starbucks partially submitted the information requested by the Commission in its letter of 5 August 2015. The Commission forwarded this information to the Dutch authorities on 28 August 2015.

(35) By letter dated 10 September 2015 and email sent on 11 September 2015, Starbucks submitted the remaining information to the Commission. By letter dated 23 September 2015, Starbucks complemented the information submitted to the Commission on 10 and 11 September 2015.

(36) By letters dated 25 September 2015 and 7 October 2015, the Dutch authorities provided their comments on the information submitted by Starbucks on 10, 11 and 23 September 2015.

2.   

DESCRIPTION OF THE CONTESTED MEASURE

2.1.   

Description of the beneficiary

(37) The beneficiary of the measure is SMBV. SMBV is a subsidiary incorporated in the Netherlands of the Starbucks group. The Starbucks group is composed of the Starbucks Corporation and all the companies controlled by that corporation. The Starbucks Corporation is headquartered in Seattle, United States of America (hereinafter ‘US’). The corporate structure of the Starbucks group is explained in more detail in recital 27 and Figure 1 of the Opening Decision.

(38) Starbucks is a roaster, marketer and retailer of specialty coffee, operating in 65 countries. It purchases and roasts coffees that are sold, along with handcrafted coffee, tea and other beverages and fresh food items, through company-operated stores. It also sells a variety of coffee and tea products and licenses its trademark through other channels, such as licensed stores, grocery and national foodservices accounts(9). In 2014, the Starbucks group had worldwide net revenues of USD 16 448 million and post-tax earnings of USD 2 067 million(10).

(39) SMBV is the only wholly controlled Starbucks group entity outside of the US which roasts coffee.

2.2.   

The contested measure

(40) This Decision concerns the SMBV APA, an advance pricing agreement concluded by the Dutch tax administration with SMBV on 28 April 2008 (

Vaststellingsovereenkomst APA

). The SMBV APA is binding for 10 years, from 1 October 2007 to 31 December 2017(11).

(41) An APA is an agreement between a tax administration and a taxpayer on the application of tax law regarding (future) transactions, i.e. it determines the amount of profit that the taxpayer generates from its activities that are taken into account in that tax jurisdiction. An APA determines, in advance of intra-group transactions, an appropriate set of criteria (e.g. method, comparables and appropriate adjustments thereto, critical assumptions as to future events) for the determination of an arm's-length pricing for those transactions over a fixed period of time. An APA is formally initiated by a taxpayer.

2.2.1.   

The SMBV APA

(42) By concluding the SMBV APA, the Dutch tax administration accepted that the remuneration determined by Starbucks' tax advisor in the transfer pricing report for the functions performed by SMBV in the Netherlands (including risk assumed and assets used) constitutes an arm's-length remuneration(12).

(43) That remuneration consists of a mark-up of [9-12] % of the relevant cost base. The relevant cost base used to calculate that remuneration includes all personnel costs engaged in both manufacturing and supply chain activities, the cost of production equipment (i.e. depreciation) and plant overheads. It does not include the costs of the Starbucks cups, paper napkins, etc., the costs of green coffee beans (cost of raw materials), the logistics and distribution cost for services provided by third parties, the remuneration for activities provided by third parties under so-called ‘consignment manufacturing contracts’ and the royalty payments to Alki LP.

(44) In the SMBV APA, the Dutch tax administration further accepted that the level of the royalty payment from SMBV to Alki LP would be determined at the end of each year as the difference between the realised operating profit before royalty expenses and the aforementioned [9-12] % mark-up on operating expenses. The SMBV APA further provides that ‘

this royalty payment is deductible for corporate income tax purposes and is not subject to Dutch withholding tax

’(13).

(45) The SMBV APA thus endorses a profit allocation to SMBV within the Starbucks group that enables it to determine its corporate income tax liability to the Netherlands on a yearly basis for 10 years. Since the APA entered into force on 1 October 2007, this Decision analyses the SMBV APA under the State aid rules as from that date.

2.2.2.   

The transfer pricing report

(46) The remuneration accepted by the Dutch tax administration in the SMBV APA is based on the transfer pricing analysis prepared by Starbucks' tax advisor in the transfer pricing report, which forms an integral part of that APA.

(47) The objective of the transfer pricing report is to support the proposed profit allocation to SMBV within the Starbucks group as being based on an arm's-length pricing of intra-group transactions. The transfer pricing report presents a company overview, a functional analysis and a selection of transfer pricing methods. The report presents the following relevant information about Starbucks Coffee BV and SMBV(14):

2.2.2.1.   

Starbucks Coffee BV

(48) Starbucks Coffee BV functions as the Starbucks group's head office for the Europe Middle-East and Africa (‘EMEA’) region, supporting the group's EMEA business operations. Starbucks Coffee BV assists with identifying developers to develop and operate Starbucks retail stores in the EMEA territories. In its capacity as head office, Starbucks Coffee BV licenses certain Starbucks trademarks, technology and know-how(15) from its shareholder, Alki LP, in return for a royalty payment. Starbucks Coffee BV enters into what is referred to as an ‘Area Development and Operation Agreement’ (hereinafter: ‘ADOA’) with related and unrelated operators of Starbucks shops in the EMEA region. Those operators are called Developers, also referred to hereinafter as ‘Shops’. Starbucks Coffee BV sublicenses intellectual property (hereinafter: ‘IP’) rights to the Developers to enable them to develop and operate their shops as Starbucks Shops. For the use of the IP, the Developers pay a royalty and other fees to Starbucks Coffee BV based on a percentage of turnover. According to the submission of the Netherlands of 2 October 2013, both related and unrelated Developers pay the same percentage of turnover as a royalty to Starbucks Coffee BV. Therefore, Starbucks' tax advisor considers that a comparable uncontrolled price method(16) is applied to determine the arm's-length price of intra-group royalty payments to the EMEA head office Starbucks Coffee BV(17).

2.2.2.2.   

SMBV

(49) Starbucks' tax advisor, when presenting the most important transactions and inter-company flows for SMBV in the transfer pricing report(18), limits itself to describing that SMBV primarily processes green coffee and sells roasted coffee to affiliated and non-affiliated parties. It also explains that the green coffee beans sourced by SMBV are sourced from an affiliated Starbucks' subsidiary in Switzerland (SCTC). Furthermore, it explains that SMBV operates as an intermediary distribution entity for a variety of non-coffee items. In addition to the supply chain function for its own manufacturing activities, the tax advisor further explains that certain markets also receive some supply chain support from SMBV. The tax advisor does not describe the roasting IP licensing arrangement under which SMBV pays royalties to Alki LP for licensed IP among the most important transactions and inter-company flows, but only shows it in a graph and describes it in a separate section in the transfer pricing report describing the EMEA market and Netherlands operations(19).

(50) According to the functional analysis provided in the transfer pricing report(20), the focus of SMBV's activities is its Amsterdam-based roasting facility. The main raw material component of that roasting process is green coffee beans. The actual roasting process for a particular coffee blend depends on the particular type of green coffee bean used in the recipe and the desired flavour profile. SMBV is responsible for executing roasting forecasts provided by Starbucks US and ensuring that the resulting product meets the quality standards of Starbucks US. SMBV buys the green coffee beans from SCTC, a designated supplier(21). The beans for the EMEA market are roasted and packaged in the Netherlands by SMBV.

(51) SMBV licenses a subset of IP from Alki LP, which is not mentioned in the functional analysis, but described in a separate section on the EMEA market and Netherlands operations as ‘necessary to utilise the coffee roasting manufacturing process and the right to supply coffee to [D]evelopers. In return [SMBV] remits a royalty to Alki LP for the licensed IP’(22). That coffee roasting related IP consists, in particular, of roasting curves, which according to the transfer pricing report dictates the temperature and the length of time required to complete the roasting process.

(52) According to the transfer pricing report, SMBV employed [40-60] people, of which [20-30] perform supply chain operations including procurement, planning, logistics and distribution planning(23). SMBV has a distribution centre in the United Kingdom (hereinafter: ‘UK’) operated under contract by a third party(24). SMBV has a supply and logistics agreement with a third party in the Netherlands, whereby the third party purchases inventory for resale to developers at arranged prices(25). SMBV also engaged another third party to open a distribution centre in Germany during 2006.

(53) SMBV also has a relationship with a consignment manufacturer, [unaffiliated manufacturing company 1]. [Unaffiliated manufacturing company 1] mainly produces [a trademark registered coffee product] powder for the Starbucks [trademark registered coffee] product. SMBV is responsible(26) for managing this toll manufacturing relationship(27) and sells the majority of the product produced by [unaffiliated manufacturing company 1] to Starbucks' US resident companies.

(54) Figure 1 is based on the description of the transfer pricing report to illustrate the part of the structure of Starbucks which is relevant for this Decision(28).

Figure 1

Structure of Starbucks based on the description in the transfer pricing report

[Bild bitte in Originalquelle ansehen]

(55) According to the transfer pricing report, the transactional net margin method (hereinafter: ‘TNMM’(29)) was selected by the tax advisor to estimate an arm's-length remuneration due to SMBV over other transfer pricing methods, because, ‘

in the specific set of Starbucks' circumstances; the net margin is less affected by transactional differences and functional differences, as in the case with measures employed in standard traditional methods

’(30). For a more detailed description of the choice of the method and of the peer analysis in the transfer pricing report, see recitals 40 to 58 of the Opening Decision.

(56) In applying the TNMM to SMBV's roasting activities, the tax advisor considered the relevant base(31) for the net profit indicator to be the costs of the services rendered by SMBV, in line with the cost plus method, which is considered an appropriate methodology for supply chain and manufacturing services. However, the mark-up is only applied to those costs where, according to the tax advisor, SMBV adds value. The transfer pricing report lists those costs as main operating costs, such as cost of personnel and depreciation, excluding COGS.

(57) To determine the appropriate arm's-length range of profitability for the activities performed by SMBV, the tax advisor conducted a search to identify companies operating in Europe with similar functions and risks. A comparable companies search in the Amadeus database(32) using the primary NACE Rev 1.1 code 1586 — Processing of tea and coffee to identify companies engaged in the trade of coffee (the buying and selling of coffee not processed by the company itself were not deemed comparable) followed by elimination through the use of additional financial selection criteria and a manual screening by the tax advisor(33), ultimately resulted in 20 companies as potentially comparable(34).

(58) The net profit indicator calculated for each company was a mark-up on total costs, which is defined as operating profit divided by total operating costs. The median of the unadjusted mark-up on total costs for those companies from 2001 to 2005 was estimated at 7,8 %.

(59) However, according to the tax advisor, this set of comparable companies includes full-fledged manufacturers that typically perform more functions and incur risk relating to their raw materials. To correct for this difference and to allegedly increase the reliability of the comparison, the tax advisor performed a first adjustment to account for the fact that the proposed application of the mark-up to SMBV's cost-base does not include a cost component for green coffee beans.

(60) Additionally, the comparable companies' returns would, according to the tax advisor, reflect a return on a cost base that includes such raw materials. Therefore, a raw materials cost adjustment was applied by the tax advisor to modify the total cost mark-up(35) obtained from the comparable companies set. The tax advisor indicated that this adjustment is associated with what the tax advisor refers to as ‘taking title to the raw materials’. Concretely, the tax advisor subtracted from the profit of each company in the set their estimated price of raw material multiplied by 12-month EURIBOR plus 50 basis points.

(61) The combination of those two adjustments lowered the estimated remuneration from a median of 7,8 % on total costs to an estimated median of 9,9 % on operating costs(36). A (rounded) mark-up of [9-12] % of operating costs was considered on that basis to reflect an arm's-length mark-up for the provision of roasting services and associated supply chain activities by SMBV for its intra-group transactions.

2.3.   

Description of OECD guidance on transfer pricing

(62) The Organisation for Economic Cooperation and Development (hereinafter the ‘OECD’) OECD provides guidance on taxation for its member countries. The OECD's guidance on transfer pricing can be found in its Transfer Pricing Guidelines (hereinafter the ‘OECD TP Guidelines’), which is a non-binding legal instrument providing guidance on transfer prices.

(63) Transfer prices refer to prices charged for commercial transactions between various parts of the same corporate group. Multinational companies have a financial incentive to allocate as little profit as possible to jurisdictions where those profits are subject to higher taxation. This could lead to exaggerated transfer prices which should not be accepted as a basis for calculating taxable income. To avoid this problem, tax administrations should only accept transfer prices between intra-group companies that are remunerated as if they were agreed to by independent companies negotiating under comparable circumstances at arm's length(37). This is known as the ‘arm's-length principle’.

(64) The authoritative statement of the arm's-length principle is found in paragraph 1 of Article 9 of the OECD Model Tax Convention, which forms the basis of bilateral tax treaties involving OECD member countries and an increasing number of non-member countries. Article 9 provides: ‘[Where] conditions are made or imposed between the two [associated] enterprises in their commercial or financial relations which differ from those which would be made between independent enterprises, then any profits which would, but for those conditions, have accrued to one of the enterprises, but, by reason of those conditions, have not so accrued, may be included in the profits of that enterprise and taxed accordingly’.

(65) On 27 June 1995, the OECD's Committee on Fiscal Affairs adopted transfer pricing guidelines (hereinafter: the ‘1995 OECD TP Guidelines’(38)) which were the object of a Recommendation of the OECD Council(39). The latest version of the OECD TP Guidelines was adopted in July 2010 (hereinafter the ‘2010 OECD TP Guidelines’)(40).

(66) Given their non-binding nature, the tax administrations of the OECD member countries are simply encouraged to follow the Guidelines. However, in general, the OECD TP Guidelines serve as a focal point and exert a clear influence on the tax practices of OECD member (and even non-member) countries. Moreover, in numerous OECD member countries those Guidelines have been given the force of law or serve as a reference for the purpose of interpreting domestic tax law. To the extent the Commission cites the OECD TP Guidelines in the present Decision, it does so because those guidelines are an existing manual in the area of transfer pricing that are the result of expert discussions in the context of the OECD and elaborate on techniques aimed to address common challenges of the application of the arm's-length principle to concrete situations. The OECD TP Guidelines therefore provide useful guidance to tax administrations and multinational enterprises on the application of the arm's-length principle. They also capture the international consensus on transfer pricing.

2.3.1.   

Transfer pricing methods

(67) Both the 1995 and the 2010 OECD TP Guidelines describe five methods to approximate an arm's-length pricing of transactions and profit allocation between companies of the same corporate group: (i) the comparable uncontrolled price method (hereinafter ‘CUP’); (ii) the cost plus method; (iii) the resale minus method; (iv) the TNMM and (v) the transactional profit split method. Both Guidelines also draw a distinction between traditional transaction methods (the first three methods) and transactional profit methods (the last two methods). Both Guidelines further explain that multinational corporations retain the freedom to apply methods not described in those Guidelines to establish transfer prices, provided those prices satisfy the arm's-length principle.

(68) The 1995 OECD TP Guidelines declare an express preference for traditional transaction methods, such as the CUP, over transactional methods, such as the TNMM, as a means to establish whether transfer pricing is at arm's length. Paragraph 3.49 of the 1995 OECD TP Guidelines provides: ‘Traditional transaction methods are to be preferred over transactional profit methods as a means of establishing whether a transfer price is at arm's length, i.e. whether there is a special condition affecting the level of profits between associated enterprises. To date, practical experience has shown that in the majority of cases, it is possible to apply traditional transaction methods.’

(69) In this regard, paragraph 2.3 of the 2010 OECD TP Guidelines provides: ‘As a result, where, taking account of the criteria described at paragraph 2.2, a traditional transaction method and a transactional profit method can be applied in an equally reliable manner, the traditional transaction method is preferable to the transactional profit method.’

(70) The CUP and the TNMM are relevant for the present Decision and are therefore described in more detail in recitals 71 to 75.

(71) The CUP method compares the price charged for the transfer of property or services in a controlled transaction (i.e. a transaction between two enterprises that are associated enterprises with respect to each other) to the price charged for the transfer of property or services in a comparable uncontrolled transaction (i.e. a transaction between enterprises that are independent enterprises with respect to each other), conducted under comparable circumstances.

(72) The TNMM is one of the ‘indirect methods’ to approximate an arm's-length pricing of transactions and profit allocation between companies of the same corporate group. It approximates what would be an arm's-length profit for an entire activity, rather than for identified transactions. It does not seek to establish the price of goods sold, but estimates the profits independent companies could be expected to make on an activity, such as the activity of selling goods. It does this by taking an appropriate base (‘a profit level indicator’), such as costs, turnover or fixed investment, and applying a profit ratio reflecting that observed in comparable uncontrolled transactions to that base.

(73) Because the TNMM does not set a price for individual transactions, the taxable profit of an entity estimated using the TNMM might not have a direct effect on the taxable profit of another entity of the same corporate group. The method is therefore different to using, for example, the CUP method, where transfer pricing establishes the price of a specific good or service which is then recorded in the taxable profit for the same amount by the group company selling and the group company buying the particular good or service.

(74) The use of the TNMM is often associated with paragraph 3.18 of the 2010 OECD TP Guidelines, according to which the ‘tested party’, i.e. the party to the transaction for which a financial indicator (mark-up on costs, gross margin, or net profit indicator) is tested, should, in principle, be the company which has the less complex function. In detail, paragraph 3.18 of the 2010 OECD TP Guidelines provides for the following recommendation: ‘When applying a cost plus, resale price or transactional net margin method as described in Chapter II, it is necessary to choose the party to the transaction for which a financial indicator (mark-up on costs, gross margin, or net profit indicator) is tested. The choice of the tested party should be consistent with the functional analysis of the transaction. As a general rule, the tested party is the one to which a transfer pricing method can be applied in the most reliable manner and for which the most reliable comparables can be found, i.e. it will most often be the one that has the less complex functional analysis’(41).

(75) That paragraph of the 2010 OECD TP Guidelines is often interpreted by tax advisors in situations of transactions between two related companies as allowing to estimate the arm's-length profitability of only one of them, i.e. the less complex one, and attributing any other profit observed in the accounts to the second company, which is considered more complex. This is done regardless of whether the company considered as more complex earns an arm's-length remuneration as a result of the transactions between it and the less complex company, and regardless of the fact that the 2010 OECD TP Guidelines themselves do not seem to lift the requirement for resulting transactions to be priced at arm's length when paragraph 3.18 of the 2010 OECD TP Guidelines is relied upon.

2.3.2.   

Arm's-length range

(76) The OECD TP Guidelines describe as an acceptable arm's-length outcome from a comparison analysis a range of outcomes rather than one specific outcome(42). In practice, what tax advisors refer to as a ‘range’ is the interquartile range. Quartiles in a series of data are three points which divide the figures in the set ranked from smallest to largest into three equally populated sets, that is 25 % of the data is in the 25th percentile (also called lower quartile), 50 % of the data is below or equal to the second quartile, which is the median of the set, and 75 % of the data is below or equal to the 75th percentile (also called upper quartile). Such an interquartile range is mentioned in paragraph 3.57 of the 2010 OECD TP Guidelines for situations where comparability defects cannot be fully addressed. According to the 2010 OECD TP Guidelines, in such cases, if the range includes a sizeable number of observations, statistical tools that take account of central tendency to narrow the range (e.g. the interquartile range or other percentiles) might help to enhance the reliability of the analysis. In practice, tax advisors often consider that any outcome within an interquartile range is acceptable and equally correct.

(77) Furthermore, when making adjustments to pricing by the taxpayer that is outside a correct range, the tax administration in presence of comparability defects should according to paragraph 3.62 of the 2010 OECD TP Guidelines(43) rely on measures of central tendency (for instance the median, the mean or weighted averages, etc.) to determine the appropriate point in the range.

2.4.   

Description of main accounting and financial terms used in the transfer pricing analysis

(78) The transfer pricing analysis in the transfer pricing report included in the SMBV APA request contains a number of accounting concepts and financial profit indicators. A brief overview of financial indicators and accounting concepts frequently used in transfer pricing assessment and relevant for the present case are explained below.

2.4.1.   

Main profit and loss accounting items (non-financial company)

(79) A typical profit and loss account first records the income that a company receives from its normal business activities, usually from the sale of goods and services to customers. This accounting item is referred to as ‘Sales’ or ‘Turnover’ or ‘Revenue’.

(80) Cost of goods sold (hereinafter: ‘COGS’) represents mainly the value of material used for the production of goods (raw materials) or the purchase price of goods that have been resold if the company does not process the goods sold. COGS is deducted from sales to calculate gross profit.

(81) Operating expenses covers principally salary expenses(44), energy expenses and other administrative and sales expenses. In the case of SMBV, the royalty paid to Alki LP is classified as an operating expense, but it is excluded from the operating expenses used to calculate the taxable income according to the SMBV APA(45).

(82) Table 1 provides a simplified overview of a profit and loss account(46).

Table 1

Simplified profit and loss account

Sales (or Turnover or Revenue)

— Cost of goods sold (COGS)

Gross Profit

— Operating Expense (OpEx)

Operating profit (EBITDA)

— Interest and depreciation

Taxable income (EBT)

— Tax

Net profit

2.4.2.   

Profit margins

(83) Performance and profitability is often measured using ratios presented as ‘margins’ or ‘mark-ups’. Margins are also used in peer comparisons in transfer pricing.

(84) Some margins are conventionally defined. This is the case for gross margins, which are in principle defined as gross profits divided by sales, and for net margins defined in principle as the net profits divided by sales. Net profit margins used in transfer pricing analyses will often use as a starting point (nominator) the taxable income, rather than the net profits, in particular when using the TNMM, which serves to approximate the taxable income of a tested party.

2.5.   

Description of the Dutch rules on arm's-length pricing

2.5.1.   

Article 8b of the corporate income tax 1969

(85) The SMBV APA was concluded on the basis of Article 8b(1) of the Dutch Corporation Tax Act 1969

(Wet op de Vennootschapsbelasting 1969)

(hereinafter ‘CIT’).

(86) Article 8b(1) CIT, which was inserted in the CIT in 2002, lays down the arm's-length principle in the domestic tax law of the Netherlands and reads as follow: ‘Where an entity participates, directly or indirectly, in the management, control or capital of another entity, and conditions are made or imposed between these entities in their commercial and financial relations (transfer prices) which differ from conditions which would be made between independent parties, the profit of these entities will be determined as if the last mentioned conditions were made’.

2.5.2.   

Dutch Transfer Pricing Decree 2001

(87) Guidance as to how the Dutch tax administration interprets the arm's-length principle laid down in Article 8b(1) CIT is provided in the Dutch Transfer Pricing Decree of 30 March 2001 (hereinafter ‘the Decree’)(47). The preamble to the Decree states the following: ‘[…] The policy of the Netherlands on the arm's-length principle in the field of international tax law is that this principle forms an integral part of the Netherlands' system of tax law as a result of its incorporation in the broad definition of income recorded in Section 3.8 of the Income Tax Act 2001’(48). In principle, this means that the OECD Guidelines apply directly to the Netherlands under Section 3.8 of the Income Tax Act 2001. There are a number of areas in which the OECD Guidelines provide scope for individual interpretation by the member countries. In a number of other areas, practical experience has shown that the OECD Guidelines are in need of clarification. This decree explains the Netherlands' position in relation to these particular points and seeks, where possible, to remove any confusion'.

(88) With regard to the use of a range, the Decree determines under Chapter 1.2: ‘In some cases it will be possible to apply the arm's-length principle and arrive at one single figure that is the most reliable to determine the arm's-length character of the transfer prices. Because, however, transfer pricing is not an exact science, a particular transfer pricing method will often generate a range of figures all of which are equally reliable’.

(89) With regard to the transfer pricing methods, under Chapter 2, the Decree states the following: ‘Chapter II of the OECD Guidelines discusses the three traditional transaction methods introduced in Paragraphs 1.68 to 1.70 (i.e. the comparable uncontrolled price method, the resale price method and the cost-plus method), whilst Chapter III examines the methods known as the transactional profit methods (i.e. the profit-split method and the transactional net margin method or TNMM). Depending on the circumstances, a choice of one of these five accepted methods has to be made. The methods can supplement each other. The OECD Guidelines are based on a certain hierarchy of the methods where a preference exists for the traditional transaction methods. On the one hand, transactional profit methods are considered more or less as methods of last resort. On the other hand, the OECD Guidelines state that the tax authorities need to start a transfer pricing audit from the perspective of the method chosen by the taxpayer (see Paragraph 4.9 of the OECD Guidelines).

In accordance with Paragraph 4.9 of the OECD Guidelines, whenever the Netherlands' tax administration undertakes a transfer pricing audit, it should start from the perspective of the method adopted by the taxpayer at the time of the transaction. This complies with Paragraph 1.68 of the OECD Guidelines. The implication is that taxpayers are in principle free to choose a transfer pricing method, provided that the method adopted leads to an arm's-length outcome for the transaction in question. In certain situations, however, some methods will generate better results than others. Although taxpayers may be expected to base their choice of a transfer pricing method on the reliability of the method for the particular situation, taxpayers are definitely not expected to weigh up the advantages and disadvantages of all of the various methods and then explain why the method that was ultimately adopted generates the best results in the prevailing conditions (i.e. the best method rule). Certain situations are also suited for a combination of methods. At the same time, taxpayers are not obliged to use more than one method. The only obligation resting on the taxpayer is to explain why the decision was taken to adopt the particular method that was adopted.’

(90) Under Chapter 2.1 of the Decree, the CUP method is described. With regard to the preference for the use of this method, the Decree states: ‘(…) If a comparable price is available, the comparable uncontrolled price method (commonly known as the CUP method) will, in general, be the most direct and the most reliable method in determining the transfer price, so that this method is to be preferred over other methods’.

2.6.   

Additional information received after the opening of the formal investigation procedure from Starbucks and the Netherlands

(91) The Netherlands and Starbucks provided complementary information and figures regarding SMBV, Alki LP, the Starbucks Shops and SCTC.

2.6.1.   

Information on the activities and risks of SMBV

(92) In the transfer pricing report, SMBV is presented as a low risk manufacturer. The Commission requested information to verify whether the main activity of SMBV was indeed low-risk manufacturing, which in this case would be the roasting of coffee. The Commission also requested information about the structure of SMBV's costs to verify whether the underlying activities generating those costs confirm that the tasks performed by SMBV are indeed routine execution tasks and do not present any degree of complexity, in line with the assumptions in the transfer pricing report. Information on the degree of business risk assumed by SMBV is presented in different sections, where contracts are described.

(93) In order to analyse what products are the main drivers of SMBV's sales and to verify whether coffee roasting is indeed the main economic activity of SMBV, the Commission requested the breakdown of SMBV's sales by product. Starbucks provided the amounts received from the Shops for the coffee beans (highlighted in Table 2) and separately the amounts of any other payments from the Shops for other activities such as sale of cups, presented in Table 2.

Table 2

Revenues of SMBV by category of products sold

(94) Starbucks indicated that the revenues under the description ‘REV PACKAGED COFFEE’ in Table 2 relate to SMBV's roasting and packaging function. The remainder of the revenues relate, according to Starbucks, to SMBV's administrative and logistics support function. A small portion of the coffee roasted in the Netherlands is further processed by third-party manufacturers. It concerns: ‘REV SINGLE SERVE PODS-COFFEE’, ‘REV SINGLE SERVE PODS-VERISMO’, ‘REV SOLUBLE COFFEE’, ‘REV FRAP COFFEE BASE’ and ‘REV EXTRACT’. Those revenues should also be classified as administrative and logistics support revenue, as the underlying revenue invoiced to Developers represents the value created by third parties rather than SMBV. In any case, according to Starbucks, those revenues represent only a small portion of the total roasting output of SMBV.

(95) Regarding the pricing of products, Starbucks provided […] price list examples for different quarters in 2013 and 2014 that provide prices for coffee and non-coffee products sold by Starbucks. According to Starbucks, […].

(96) Regarding the pricing of coffee products, the prices are calculated on the basis of formulas referring to costs and they vary per product. Starbucks provided, as an illustration, the cost build up formula for the item ‘[certification programme coffee product e]’ which is the highest volume item sold by SMBV. The coffee prices charged to Shops by SMBV are calculated based on the costs that SCTC pays for green coffee beans, to which cost […] of coffee acquisition and coffee […] are added. To this amount, ‘[…]’ costs are added for […], as well as a […] summing up to the roasted coffee price. Finally a licensing fee paid to a third party is paid relating to the [certification programme] designation(49).

(97) Regarding the prices of non-coffee products sold to Shops by SMBV, Starbucks indicated that the prices of the non-coffee products delivered by SMBV to Shops are determined by adding a so-called [cost recovery margin] to the purchase price of the non-coffee products. That [cost recovery margin] is charged by SMBV to recover the […] expenses. However, it covers not only the […] expenses of SMBV, but also all the relevant […]costs of the […]. In fact, as the final prices for non-coffee products charged by SMBV to Shops are calculated by applying the [cost recovery margin] to the product costs, the profit on non-coffee products that SMBV records corresponds to this [cost recovery margin].

(98) The most important expense item of SBMV is salary expense, which amounted to EUR [1-10 million] in 2014 out of a total operating expense of EUR 16 124 000 that year. Other large operating expense items of SMBV in 2014 were wage related expenses, payroll tax, property rents and lease payments, equipment depreciation and intangibles amortisation. Although prior to 2013 SBMV did not have intangibles amortisation expenses, it was one of the largest operating expense items in 2014, amounting to EUR 628 000, which was a larger expense in that year than equipment repair and maintenance, electricity, bad debt, office and computer supplies and others (for example software licence expenses, which are recorded as a separate item). The intangible amortisation expense seems to relate to an IP right first recorded in the financial accounts in 2012, which would have been acquired for more than EUR 4 million. Additionally, in 2014, a large operating expense item was represented by ‘Market research’ amounting to EUR [100-200 thousand] that year.

2.6.2.   

Royalty payments

(99) The taxable profits of SMBV in the Netherlands, which are determined based on the SMBV APA in reference to the operating expense of the company, are reduced by a royalty paid to Alki LP. The Commission requested further information from the Netherlands and Starbucks on the amount of the royalty payment and the exact calculation of the tax base.

(100) SMBV's profit and loss account as presented in its financial statements are reproduced in Table 3.

Table 3

SMBV's profit and loss accounts 2001-2014

(101) Under the notes to the financial statements, the position ‘Other expenses’ in Table 3 is defined as follows: ‘Other expenses relate to a royalty agreement held with the affiliated company [CV 1], which was assigned to Alki LP on December 13, 2006 and is based on a tax ruling with the Dutch tax authorities’. The APA to which this footnote relates is the SMBV APA and thus indicates that SMBV's auditor interpreted the SMBV APA to determine the royalty payments by SMBV to Alki LP.

(102) That royalty is calculated as a residual in the profit and loss account. When constructing the profit and loss account of SMBV, all the input figures other than the royalty are either observed or assumed to be priced at arm's length. Based on the pricing agreed in the SMBV APA, a taxable profit (the position ‘Result before taxation’ in Figure 2) is calculated at approximately [9-12] % of SMBV's operating expenses (the position ‘General and administrative expenses’ in Figure 2). However, as the position ‘Sales’ in Figure 2 minus all the accounting costs before the royalty payment(50) do not sum up to this taxable profit calculated based on the SMBV APA, the excess profit beyond that [9-12] % mark-up is paid by SMBV in the form of a tax deductible royalty to Alki LP for the coffee roasting related IP (the position ‘Other expenses’ in Figure 2). Based on SMBV's accounts, the royalty payment thus takes place as determined by the SMBV APA issued by the Dutch tax administration.

Figure 2

[Bild bitte in Originalquelle ansehen]

(103) As the tax base in SMBV's accounts for the period of application of the SMBV APA is equal to but not exactly [9-12] % of each year's ‘General and administrative expense’, the Commission requested Starbucks to provide the actual costs for each year used to calculate the tax base. Starbucks provided information on the amount of operating expenses used to calculate the taxable income each year on the basis of the SMBV APA and also for the previous accounting periods (since 2001), when a previous APA was in force. For each period of the application of the SMBV APA, the costs taken into account for the calculation of the tax base are lower but close to the amounts reported as ‘General and administrative expense’. For example, for the period 2012/2013 those costs are EUR 15 694 137 and for the period 2007/2008 those costs are EUR 15 055 253. In the periods preceding the application of the SMBV APA, the costs taken into account for the calculation of the tax base are much higher, as they would, according to Starbucks, also include the costs charged by [unaffiliated manufacturing company 1]. For example, for the period 2006/2007, the costs used to calculate the tax base were EUR [30-40 million]. This explains why the corporate tax liability decreased by more than half when the SMBV APA entered into force(51), i.e. from EUR 844 309 in 2006/2007 to EUR 383 909 in 2007/2008, as shown in Table 3.

2.6.3.   

Information on Alki LP

(104) In response to the questions of the Opening Decision, the Netherlands provided a description of the licence rights held by Alki LP to use and sublicense the IP right for which SMBV pays a royalty to Alki LP. In this regard, the Netherlands explained that the IP right includes three categories: (i) trade mark rights(52); (ii) Starbucks system rights(53); and (iii) coffee-related rights(54). The Netherlands further explained that the knowledge and information relating to coffee includes the use and applications of confidential information, technology and specific knowledge with regard to the recipes for mixing green coffee beans, the process of roasting green coffee beans and the production of other derived coffee products(55).

(105) The Netherlands further explained that under the Roasting Agreement, further described at recital 142, Alki LP acts as the principal that bears all the company risks and performs the associated activities(56). Therefore, according to the Netherlands, the royalty does not solely represent remuneration for the IP right, but also for Alki LP's role as principal.

(106) Regarding Alki LP, Starbucks specified that Alki LP has no employees on its payroll to perform its role as principal under the Roasting Agreement, nor do its partners have employees on the payroll who perform this role. According to Starbucks, Alki LP performs that role through the guidance, expertise and required functions provided by Starbucks entities in the US pursuant to the [Cost Sharing Agreement] (hereinafter: ‘CSA’)(57). According to Starbucks, and as evidenced by the limited partnership deed constituting Alki LP, SCI UK I, Inc.(58) and Starbucks Coffee International, Inc.(59) are the partners of Alki LP.

(107) Starbucks clarified that Alki LP was added to the group structure in December 2006 as part of the expansion of the CSA to cover the entire EMEA region by including the UK market. By adding Alki LP to the structure, Starbucks avoided that the royalty income received by Alki LP would be directly integrated in the income of the partners of Alki LP under US tax law.

2.6.4.   

Payments by Alki LP to Starbucks Corporation under the Cost Sharing Agreement

(108) As neither Alki LP, nor its partners, have any employees(60), and Starbucks claims that Alki LP is able to perform its role as principal through the CSA, the Commission requested information about the payments under the CSA and the services those payments relate to, as well as the way in which the payments are financed by Alki LP. Starbucks provided the following overview of payments by Alki LP to Starbucks US broken into different types of payment.

(109) Under the CSA, Alki LP makes royalty payments to Starbucks US for the three categories of IP that Starbucks US licenses to Alki LP. Those royalty payments consist of (i) a buy-in royalty allocated to the brand/trademark rights; (ii) a buy-in royalty allocated to the business format rights; and (iii) a permanent royalty allocated to the coffee knowledge.

(110) The amount of the royalty payment is calculated on the basis of a formula that takes as a starting point the residual profit pool. The residual profit pool is the sum of the royalties paid by SMBV and Starbucks Coffee BV to Alki LP(61) and income that Alki LP receives from Starbucks Corporation(62). From that amount, an arm's-length return for routine functions performed by Alki LP is deducted(63). […]. In this respect, a difference is made between the IP rights for the brand/trademark and the business format, on the one hand, and the coffee knowledge, on the other. That difference is explained by the fact that Alki LP holds the beneficial ownership of the business format related intangibles and the trademark/brand for the EMEA region, whilst the legal and beneficial ownership of the coffee knowledge rights are held by Starbucks US.

(111) The beneficial ownership of the business format related intangibles and the trademark/brand for the EMEA region was transferred from Starbucks Corporation in 2002 to [CV 1] and assigned to Alki LP since 2006. For the beneficial ownership, [CV 1] and Alki LP made buy-in royalty payments, where the percentage of the US share of the cost which Alki LP needed to pay to Starbucks US digressively reduced from [65-70] % in 2002 to [0,5-1] % in 2010 and onwards for the trademark, and [70-75] % in 2002 to [0-0,5] % in 2010 and onwards for the business format IP. With those buy-in payments, the beneficial ownership rights for these two categories of IP for the EMEA region are deemed to be paid off by Alki LP. With regard to the coffee knowledge IP, Alki LP pays a permanent royalty of [70-75] % of the residual profit allocated to this IP category to Starbucks US.

(112) Starbucks submitted an overview of the permanent royalty and the buy-in payments by [CV 1] and Alki LP for the period 2005-2014. With regard to the buy-in payments, [CV 1] paid a total of EUR [1-10] million as buy-in payments for the trademark and business format IP to Starbucks Corporation(64) (65). In December 2006, the CSA was assigned by [CV 1] to Alki LP. From 2007 to 2014, Alki LP paid EUR [1-10] million for the trademark and EUR [20-30] million for the business format IP(66). This means that, in total, EUR [10-20] million was paid as buy-in payments for the trademark and EUR [20-30] million for the business format IP. In 2014, Alki LP transferred the beneficial ownership of the IP rights to Starbucks EMEA Ltd for which it recorded a gain of EUR [1-1,5] billion as other operating income in its profit and loss account. With regard of the permanent royalty, Alki LP paid on average EUR [1-10] million a year (EUR [70-80] million in total) over the period 2005–2014 to Starbucks Corporation for the coffee knowledge IP(67).

(113) In addition to the buy-in and permanent royalty, Alki LP pays service fees for services provided by Starbucks Corporation and cost sharing payments for the development of certain intangible assets(68). Over the period 2008–2014, Alki LP paid on average EUR [1-10] million per year on cost sharing payments and EUR [1-10] million per year on service fees.

2.6.5.   

Information on SCTC and on the prices of green coffee beans paid by SMBV

(114) Given that green coffee beans are procured by SMBV through a group transaction, more precisely they are purchased from SCTC, the Commission requested information about how the prices for this intra-group transaction are set and about the price set for each year throughout the application of the SMBV APA.

(115) Starbucks submitted the balance sheet and profit and loss accounts for SCTC since 2005. Table 4 reproduces information contained in those accounts. The calculation of the mark-up on the costs of green coffee beans (COGS) applied by SCTC was added by the Commission for the purpose of this Decision. Based on the accounting information available, which starts at 2005, the average mark-up on the costs of green coffee beans stood at [around 3 %] for the period 2005 to 2010, compared to an average mark-up of [around 18 %] over the period 2011 to 2014.

(116) Starbucks submitted that from 2011 onwards, Starbucks adjusted the transfer pricing policy applicable to the sale of green coffee beans to increase the mark-up on product costs given the growing importance of SCTC's operations, especially its increasing expertise in coffee procurement and, more importantly, its ownership and operation of the evolving C.A.F.E. Practices Programme.

(117) The C.A.F.E. Practices Programme was launched in 2004 in conjunction with the opening of SCTC's farmer support centre in Costa Rica to provide an evolving programme where farmers, exporters, and buyers could have an ongoing conversation to (i) improve coffee quality; (ii) provide economic accountability for payments along all levels of the supply chain; (iii) encourage social responsibility in working conditions, protection of workers' rights, and adequate living conditions; and (iv) inspire environmental leadership for growing and processing green coffee beans.

(118) In order to verify whether the growing operations, which Starbucks invokes to explain the increased mark-up, are confirmed in the accounts, Table 4 also presents the assets of SCTC. Although the total assets have indeed grown significantly since 2010, this is mainly due to increased cash holding. Operating assets net of cash have increased by less than a third over the four years from 2010 to 2014.

(119) The annual reports of SCTC submitted by Starbucks also do not describe in the accompanying notes any difference in risks, functions or assets from 2010 to 2011. The activities and risks are described in identical terms over this period, i.e. the principal activity of SCTC is according to both annual reports the purchase and sale of high-quality green coffee beans and related activities of quality assurance and operations (logistics, warehousing, blending, decaffeinating, etc.).

Table 4

Financial data of SCTC

(120) As Starbucks invoked the C.A.F.E. Practices Programme to explain the increase in the margin charged by SCTC, the Commission requested Starbucks to provide the costs of that programme and the Farmer Support Centres costs. Starbucks provided estimations of the yearly costs of (i) the C.A.F.E. Practices Programme; and of (ii) the Farmer Support Centres, presented in Table 5.

Table 5

Direct and indirect costs of C.A.F.E. Practices Programme and Farmer Support Centres

(121) Over the period 2011-2014, the combined costs of the C.A.F.E. Practices Programme and the Farmer Support Centres present at most [0,5-1] % of the value of the green coffee beans purchased by SCTC(69).

(122) In response, Starbucks indicated that the C.A.F.E. Practices Programme should be considered IP of which the value is unrelated to underlying costs.

(123) The Commission requested Starbucks to provide the transfer pricing report covering the pricing of green coffee beans sold by SCTC to SMBV. According to Starbucks, historically, no transfer pricing documentation has been prepared that covers the relationship between SCTC and SMBV. For the purpose of responding to the Commission's request, Starbucks provided an ad hoc report on the SCTC transfer prices to the Commission on 13 April 2015.

(124) According to the ad-hoc transfer pricing report provided, SCTC determines the prices to its affiliates by applying a mark-up to the product costs associated with the green coffee beans sourced by it.

(125) In addition, to determine a current arm's-length mark-up on product costs for SCTC for green coffee beans procurement, three separate components were identified:

— Intellectual Property — C.A.F.E. Practices Programme: SCTC manages the C.A.F.E. Practices Programme and uses valuable know-how that, when incorporated into Starbucks' business operations, ensures consistent supply and supports the Starbucks brand for sustainability. Starbucks analysed this transaction using comparable licensing agreements relating to food and agricultural technologies.

— Procurement: SCTC provides procurement functions for green coffee beans. Starbucks analysed this transaction using comparable sourcing agreements between third parties.

— Financing: SCTC should generate a return for financing costs it incurs when holding unsold inventory and net receivables for green coffee beans. Starbucks analysed an appropriate return for financing that should be returned to SCTC.

(126) Combining the results for each separate component analysed by SCTC would yield a combined result for green coffee beans overall from 2005 through 2014 as demonstrated in Table 6:

Table 6

Mark-up on product costs

(127) The combined arm's-length range for the total mark-up on product costs charged by SCTC for the coffee sold to SMBV is according to the ad hoc transfer pricing report between a lower quartile of [around 9 %] and an upper quartile of [around 21 %], with a median of [around 12 %], for the overall period 2005 through 2014.

(128) According to Starbucks, SCTC's actual realised gross margin(70) falls squarely within the arm's-length range. This would be evidenced by the fact that the gross margin over Sales of SCTC stood at [around 12 %] for the years 2005 through 2014, based on a weighted average.

(129) The Commission requested Starbucks to provide the data and the components of the mark-up used to establish the quartile and median figures for the period 2005 through 2014, submitted to the Commission. Starbucks indicated that this data is not available.

(130) Starbucks submitted information on 29 June 2015 to substantiate the pricing of green coffee beans in the ad hoc transfer pricing report provided and in particular the figures presented in Table 6 for which the Commission requested the underlying data.

(131) Contrary to what was indicated in the table provided by Starbucks on 13 April 2015, the values in the table regarding the C.A.F.E. Practices Programme 1,4 % and 9,9 % are not the lower quartile (25th quartile) and the upper quartile (75th quartile). Rather, they would be the two extreme values, i.e. lowest value and highest value, of a comparison of licence agreements. In its submission of 29 June 2015, Starbucks identified eleven licence agreements relating to technology licensing, which they consider comparable to the C.A.F.E. Practices Programme. Those agreements relate to food and beverage products. The range of percentages that Starbucks arrives to is presented in Table 7. Starbucks explained that ‘Percentage Mark-up on Product Cost’ in Table 7 ‘is calculated by the average of each observation divided by one minus the average of the observation’, although there is no explanation as to why that step was taken to calculate a mark-up on product costs in complete absence of information or consideration on the costs of the companies to the respective agreements.

Table 7

Explanation by Starbucks of previously provided data on C.A.F.E. Practices Programme pricing interquartile range of 1,4 % to 9,9 %

(132) In its submission of 29 June 2015, Starbucks also provided a pricing of the procurement function of SCTC estimated by using comparable transactions identified by Starbucks using the PowerK database and LIVEDGAR database. Thirteen agreements whereby buying agency services are provided by a third-party buying agent are considered consistent with the procurement services that SCTC provided to SMBV.

(133) The results of the analysis are presented as percentages of ‘Free on Board’ product costs, which is the costs charged to the client for delivered products. The median of the observations presented is 6 %. Starbucks presents the result in terms of an interquartile range of product costs from 4,7 % to 8,0 %.

(134) In detail, the observations used refer to agreements with the following principals and the commissions as percentage of Free on Board rates specified in the brackets: Aeropostale, Inc. (5 %); Ateca Production International, Inc. (7 %); BELL SPORTS CORP. (6 %), COLUMBIA SPORTSWEAR CO. (1,5 %); Designs Apparel Inc. (5 %); F.I.S. Inc. (10 %); He-Ro Industries Incorporated (10 %); JONES APPAREL GROUP INC/NINE WEST GROUP INC. (10 %); Mannesmann Pipe&Steel Corporation (2 %), Skin Shoes LLC (8 %); Skin Inc. (6 %); TI Sportswear Inc. (4,7 %) and WW Mexicana (1 %). The interquartile range on these observations is 4 % to 8,5 %. As in the case of the C.A.F.E. Practices Programme pricing, Starbucks indicated that ‘percent of sales is calculated as the percent of FOB divided by one plus the percent of FOB’(71). This modification of the figures from the comparative analysis is not explained and does not seem to correspond to any customary financial adjustment.

(135) Finally, Starbucks calculates the pricing of what is presented as a financing function and which SCTC should earn as a return according to Starbucks for ‘holding green coffee and financing of net receivables for its own risk and account’. This return is calculated by Starbucks by applying a weighted average cost of capital (‘WACC’) to the account receivables and inventory stock of SCTC for 2014 and is on this basis set at [around 3 %].

(136) Finally, to calculate the profits that SMBV generated on the roasting of the green coffee beans purchased from SCTC, the Commission requested Starbucks for the price paid by SMBV to SCTC. Starbucks provided an overview of the average prices (per pound) charged by SCTC to SMBV and the amounts paid by SMBV to SCTC, which are presented in Table 8. No other payments have been made by SMBV to SCTC.

Table 8

Amounts paid by SMBV to SCTC for the purchase of green coffee beans

Purchases from SCTC

2.6.6.   

Information and figures on Starbucks Shops

(137) The products distributed by SMBV, either produced by SMBV or bought by SMBV from other suppliers, are sold to Shops. Some of the Shops are independent and some are owned by Starbucks. The Commission requested information to verify whether there is a difference in the commercial conditions applied to Shops owned by the group and independent Shops.

(138) With regard to the Starbucks Shops, Starbucks provided information on the different licensee programmes and eligibility criteria to develop Starbucks Shops in the EMEA region, an overview of the stores in the EMEA region, and figures regarding the turnover of the Shops and the licence fees paid in 2012 by those Shops to Starbucks Coffee BV. According to this information, the licence fees percentage over turnover paid by the three different types of licence programmes (independent geographic licensees, company-owned geographic licensees and channel licensees –the three different types of licence programmes) varies across countries with percentage ranges from [5 to 10] %.

2.6.7.   

Information on the profitability of other roasting facilities operated by Starbucks

(139) The Commission requested from Starbucks financial information on the profitability of any other roasting facility operated by the group and, in particular, Starbucks Manufacturing Corporation (hereinafter ‘SMC’), which was identified from publicly disclosed information as belonging to the Starbucks group. The Commission requested this information because the 2010 OECD TP Guidelines recommend the use of internal data to test the arm's-length nature of the division of profits(72). Starbucks indicated that SMC was the only roasting facility operated by the group other than SMBV. As SMC does not have audited accounts, Starbucks provided pro-forma accounts based on the internal reporting of the company.

(140) The data provided by Starbucks is presented in Table 9. The ratios of profits over sales and profits over operating expense(73) were added by the Commission for the purpose of this Decision, considering a ratio of [9-12] % of profits over operating expense is agreed in the SMBV APA as an arm's-length remuneration of SMBV. For SMC, this ratio stood at around 500 % over the past four accounting periods.

Table 9

Financial information on US manufacturing company Starbucks Manufacturing Corporation

2.7.   

Description of the various contracts setting out the roles and responsibilities of SMBV

(141) When SMBV was set up in the Netherlands, it entered into various agreements that set out the contractual divisions of the roles and responsibilities of SMBV and its counterparties. The three agreements relevant for this Decision are the Roasting Agreement between SMBV and Alki LP, the Green Coffee Purchase Agreement between SMBV and SCTC, and the Supply Agreement between SMBV and the Shops.

(142) The Roasting Agreement, which entered into force on 1 January 2002, governs the manufacturing process. Under the Roasting Agreement, Alki LP acts as the principal and SMBV as the owner of the roasting facility. SMBV pays a licence to Alki LP in exchange for which Alki LP grants SMBV access to the IP rights and takes over the entrepreneurial risk in the business of SMBV such as cost overruns and the sale of the products. SMBV will roast the green coffee beans and make the roasted coffee and other coffee related products available to the Shops (according to the Supply Agreement discussed below and the ADOA(74)). As owner of the facility, SMBV has to ensure that the correct equipment is used, the correct processes are carried out and the goods are produced in line with the specifications provided by Alki LP. SMBV will be the owner of the purchased products, which it has to buy from suppliers designated and/or approved by Alki LP. SMBV has to ensure that the products at the time of delivery comply with Alki LP's product specifications and SMBV is obliged to promptly replace any quantity of defective or non-performing products at its own expense. Alki LP however bears the risks relating to losses which may be incurred as a result of the production, sale, transport, storage, treatment or other use of the products and has to indemnify SMBV against any compensation claims.

(143) The Green Coffee Purchase Agreement entered into force on 22 April 2002 and stipulates the conditions for the sale of green whole-bean coffee from SCTC to SMBV. SCTC will sell the beans to SMBV at prices set forth on price lists issued by SCTC. SMBV will issues a purchase order for coffee on the basis of purchasing forecasts. SCTC will deliver the beans […].

(144) The Supply Agreement sets out the responsibilities in the supply of essential Starbucks goods between SMBV and the Shops. The Shops commit to purchase all branded roasted coffee and other essential coffee-related products as defined in that agreement exclusively from SMBV or approved suppliers at prices which are either determined under a formula attached to that agreement (for coffee and coffee-related products) or pursuant to a […] price list (for other goods). The Shops must provide SMBV with a [periodic] purchase forecast(75) whereas SMBV guarantees to the Shops that at the time when the coffee and other goods leave SMBV's warehouse, they are free from defects and, where this is not the case, SMBV replaces free of charge any defective product or grants a credit to the Shops(76).

2.8.   

Cost Sharing Agreement

(145) The CSA between Alki LP and Starbucks Corporation of 1 August 2006 determines the direct rights of Alki LP to license and sublicense to third parties the right to: (i) […] operate Starbucks stores […], (ii) […] operate ‘[…]’ and (iii) manufacture and/or distribute products using the Trademarks and Technology and Know-How.

(146) Under the CSA, Alki LP is to pay Starbucks Corporation a royalty, service fees for services provided by Starbucks Corporation, and cost sharing payments for the development of certain intangible assets. The royalty payments consist of: (i) a buy-in royalty allocated to the trademark and business format IP and a (ii) permanent royalty allocated to the coffee knowledge IP. Payments under this agreement are detailed in recitals 112 and 113.

2.9.   

Contracts between Starbucks and third parties relating to the manufacturing and the sale of coffee

(147) In order to compare the commercial conditions fixed between SMBV and the Starbucks group and in particular the royalty payment for the roasting IP, on the one hand, with commercial conditions fixed between the Starbucks group and third parties, on the other, the Commission requested from Starbucks all contracts whereby Starbucks licenced IP and all contracts where Starbucks outsourced the roasting of coffee. The Commission also requested information about the commercial conditions applied between Starbucks Corporation and its group roasting company located in the US.

(148) In response to the Commission's request, Starbucks provided contracts with:

— [Unaffiliated manufacturing company 2], (‘[unaffiliated manufacturing company 2]’), […].Starbucks entered into two types of agreements with [unaffiliated manufacturing company 2] to subcontract the roasting of coffee, which have been amended at several instances. In a technology licence agreement of [before 2008], an affiliate of Starbucks, […], grants a non-exclusive licence to [unaffiliated manufacturing company 2] to use, amongst others, the technology and know-how of Starbucks to produce and sell roasted coffee to selected third parties with which Starbucks has entered into supply agreements, such as [unaffiliated manufacturing company 5] […]. [Unaffiliated manufacturing company 2] has to perform the services so that the Roasted Coffee is of high quality, for which [unaffiliated manufacturing company 2], among others, has to comply with certain quality assurance standards established by Starbucks. The technology licence agreement stipulates that [unaffiliated manufacturing company 2] does not have to pay any fees for the licence(77). A green coffee supply agreement stipulates that [unaffiliated manufacturing company 2] has the obligation to buy green coffee beans exclusively from Starbucks for a fixed fee [per a certain quantity].

— [Unaffiliated manufacturing company 3], (‘[unaffiliated manufacturing company 3]’), […]. Under a roasting licence agreement of [after 2008], [unaffiliated manufacturing company 3] provides coffee roasting services which it sells to Starbucks and a joint-venture between [unaffiliated manufacturing company 3] and Starbucks called [unaffiliated manufacturing company 3-Starbucks joint-venture] that operates the Starbucks coffee stores in [a certain country]. [Unaffiliated manufacturing company 3] pays a roasting fee to Starbucks […] of USD […] [per a certain quantity] of green coffee produced and sold to the [unaffiliated manufacturing company 3-Starbucks joint-venture].

— [Unaffiliated manufacturing company 4], (‘[unaffiliated manufacturing company 4]’), […]. In order to subcontract the roasting of coffee [unaffiliated manufacturing company 4] entered into three types of agreements, of which only the […] purchase agreement of [after 2008] was submitted. Accordingly Starbucks appoints [unaffiliated manufacturing company 4] to manufacture, package, produce and supply coffee product. The agreements do not provide for any fee or royalty to be paid to Starbucks. The manufacturing has to take place under certain product specifications and standards provided by Starbucks. The agreement stipulates that [unaffiliated manufacturing company 4] shall sell the products to Starbucks and its affiliates for a price set at the level of the green coffee beans (according to Starbucks, the green coffee beans are therefore just a pass-through cost for [unaffiliated manufacturing company 4]) and supply cost to which a conversion fee is added. This conversion fee includes fees for roasting […]. The pricing formula is then translated into a sales price [per a certain quantity] of roasted coffee. Starbucks claims that this results in a margin of [5-10] % on total conversion costs, without further indications on how this alleged mark-up was arrived to.

— [Unaffiliated manufacturing company 5] (‘[unaffiliated manufacturing company 5]’), […]. Starbucks entered into three types of agreements with [unaffiliated manufacturing company 5] on [before 2008]. In a supply agreement, Starbucks takes the obligation to supply to [unaffiliated manufacturing company 5] roasted coffee beans, concentrate and other coffee ingredients for a price based on a formula used to develop coffee ingredient pricing for [Starbucks'] retail coffee shop business, […]. The […]agreement, which is a manufacturing and distribution agreement, grants [unaffiliated manufacturing company 5] the right to exclusively distribute certain pre-packed, ready-to-drink beverages under the Starbucks Trademark in a specific territory. The manufacturing has to take place under certain product specifications and standards provided by Starbucks. The […]agreement does not provide for any fee or royalty to be paid by [unaffiliated manufacturing company 5] to Starbucks. The trademark and technology licence agreement allows [unaffiliated manufacturing company 5] to use the trademark of Starbucks as well as its technology and know-how for the purpose of manufacturing the agreed products. [Unaffiliated manufacturing company 5] pays a licence fee of [10-15] % of net sales subject to an adjustment mechanism, which according to Starbucks results in a fee of approximately [10-15] % on net sales of ready-to-make drinks(78).

— [Unaffiliated manufacturing company 6] (‘[unaffiliated manufacturing company 6]’), […]. Starbucks entered into three types of agreements with [unaffiliated manufacturing company 6] on [before 2008], which have subsequently been amended. In a supply agreement, Starbucks takes the obligation to supply to [unaffiliated manufacturing company 6] roasted coffee beans, concentrate and other coffee ingredients for a price based on a formula used to develop coffee ingredient pricing for Starbucks' […] retail coffee shop business, […]. The manufacturing and distribution agreement grants [unaffiliated manufacturing company 6] the right to exclusively distribute certain pre-packed, ready-to-drink beverages under the Starbucks Trademark in [a certain country]. The manufacturing has to take place under certain product specifications and standards provided by Starbucks. The manufacturing and distribution agreement does not provide for any fee or royalty to be paid by [unaffiliated manufacturing company 6] to Starbucks. The trademark and technology licence agreement allows [unaffiliated manufacturing company 6] to use the trademark of Starbucks as well as its technology and know-how for the purpose of manufacturing the agreed products. [Unaffiliated manufacturing company 6] pays a licence fee depending on the product. According to Starbucks, this fee amounts to between [5 and 10] % of net sales of ready-to-make drinks(79).

— [Unaffiliated manufacturing company 7] (‘[unaffiliated manufacturing company 7]’), […]. Starbucks entered into three types of agreements with [unaffiliated manufacturing company 7] on [after 2008], which have subsequently been amended. In a supply agreement, Starbucks takes the obligation to supply to [unaffiliated manufacturing company 7] coffee ingredients, […], for a price set [per a certain quantity]. The manufacturing and distribution agreement grants [unaffiliated manufacturing company 7] the right to exclusively distribute certain pre-packed, ready-to-drink beverages under the Starbucks Trademark in a specific territory. The manufacturing has to take place under certain product specifications and standards provided by Starbucks. The manufacturing and distribution agreement does not provide for any fee or royalty to be paid by [unaffiliated manufacturing company 7] to Starbucks. The trademark and technology licence agreement allows [unaffiliated manufacturing company 7] to use the trademark of Starbucks as well as its technology and know-how for the purpose of manufacturing the agreed products. [Unaffiliated manufacturing company 7] pays a licence fee depending on the territory of between [10 and 20] % of net sales.

(149) The contracts that Starbucks has with [unaffiliated manufacturing company 5], [unaffiliated manufacturing company 6] and [unaffiliated manufacturing company 7] listed above were classified in three categories: (i) supply agreements, similar to the Supply Agreement between SMBV and Developers described at recital 144; (ii) manufacturing and distribution agreements, similar to the agreement between SMBV and Alki LP; and (iii) trademark and technology licence agreements, similar to the ADOA between Starbucks Coffee BV and the Shops described at recital 48. Of the three types of agreement between Starbucks and [unaffiliated manufacturing company 5], [unaffiliated manufacturing company 6] and [unaffiliated manufacturing company 7], only the last category of agreements requires Starbucks' counterparties to pay a royalty.

(150) In addition, four other roasting agreements with [unaffiliated manufacturing company 8], [unaffiliated manufacturing company 1], [unaffiliated manufacturing company 9], and [unaffiliated manufacturing company 10] were provided by Starbucks to the Commission.

— The [unaffiliated manufacturing company 8] manufacturing and supply agreement dates from [after 2008] and stipulates that [unaffiliated manufacturing company 8] is to produce flavoured coffee for Starbucks for a specified price(80). […] stipulates that [unaffiliated manufacturing company 8] must strictly comply with a technical manual regarding, among others, the manufacturing of the products. […] stipulates that Starbucks grants [unaffiliated manufacturing company 8] a royalty-free licence for the use of the know-how in connection with the manufacturing process.

— The [unaffiliated manufacturing company 1] manufacturing and supply agreement is dated [after 2008] and provides that [unaffiliated manufacturing company 1] shall produce certain coffee products for Starbucks for a specified price(81). […] stipulates that [unaffiliated manufacturing company 1] must strictly comply with a technical manual regarding, among others, the manufacturing of the products. The agreement does not provide for any fee or royalty to be paid by [unaffiliated manufacturing company 1] to Starbucks.

— The [unaffiliated manufacturing company 9] manufacturing and supply agreement is dated [after 2008] and provides that [unaffiliated manufacturing company 9] shall produce flavoured coffee for Starbucks for a specified price(82). […] stipulates that [unaffiliated manufacturing company 9] must strictly comply with a technical manual regarding, among others, the manufacturing of the products. The [manufacturing and supply agreement] does not provide for any fee or royalty to be paid by [unaffiliated manufacturing company 9] to Starbucks. The agreement also refers to a technology and trademark licence agreement, which was however not submitted.

— The [unaffiliated manufacturing company 10] manufacturing agreement is dated [after 2008] and concerns the manufacturing and roasting of green coffee beans for Starbucks by a […] roaster who sources its coffee beans directly. The agreement does not provide for any fee or royalty to be paid by [unaffiliated manufacturing company 10] to Starbucks. […] of the agreement clarifies that Starbuck owns the IP rights in the coffee blend, the time temperature curve of the roast progression and the roaster end point, etc. Starbucks pays a fee to [unaffiliated manufacturing company 10] for the coffee [per a certain quantity], following a specific formula which charges less for roasting than for packaging.

2.9.1.   

Arguments presented by Starbucks regarding the agreements with third parties and with Starbucks Manufacturing Corp.

(151) In addition to the contracts listed in recital 148, Starbucks supplied an analysis of the comparability of those contracts with the royalty payments made by SMBV to Alki LP.

(152) Among the agreements whereby Starbucks licensed out intangibles (trademarks and certain specific coffee related know-how) for exploitation by a third party, Starbucks distinguishes between (i) third-parties that utilised those intangibles to manufacture products containing Starbucks coffee ingredients in roasting agreements ([unaffiliated manufacturing company 3] and [unaffiliated manufacturing company 2]), and (ii) third parties which distributed such manufactured products directly to customers ([unaffiliated manufacturing company 7], [unaffiliated manufacturing company 6] and [unaffiliated manufacturing company 5]). [Unaffiliated manufacturing company 7], [unaffiliated manufacturing company 6] and [unaffiliated manufacturing company 5] use the intangibles to produce ready-to-drink coffee, which they sell to retailers (mainly supermarkets). Starbucks considers the royalties paid by [unaffiliated manufacturing company 7], [unaffiliated manufacturing company 6] and [unaffiliated manufacturing company 5] to be comparable to the royalties paid by SMBV, which it estimates to represent [5-10] % of sales(83).

(153) More generally, Starbucks made a distinction among all agreements concluded with third parties since 2002. To the best of Starbucks' knowledge, the main categories would be the following:

(1) Agreements whereby third parties exploit Starbucks intangibles (coffee related intangibles and trademarks) on the market:

— The roasting licence agreements with [unaffiliated manufacturing company 2] and [unaffiliated manufacturing company 3].

— The trademark and technology licence agreements with [unaffiliated manufacturing company 7], [unaffiliated manufacturing company 6] and [unaffiliated manufacturing company 5] for the exploitation of the Starbucks trademark and certain specific coffee related know-how.

(2) Various agreements which grant third parties access to Starbucks intangibles (technology, know-how and trademarks) under a royalty-free licence due to the specific nature of these agreements:

— Various co-manufacturing and co-packaging agreements where the IP licence merely serves to enable the third party to produce in accordance with Starbucks' specifications and to protect Starbucks' IP rights. Starbucks' co-manufacturing or co-packaging partners typically supply the products back to Starbucks and do not exploit the intangibles on the market.

— Various distribution agreements where the IP licence merely serves to allow the distributor to resell Starbucks branded products and to protect Starbucks' IP rights. The distributor does not exploit the Starbucks intangibles on the market.

(3) Licence agreements with Developers to enable them to exploit the Starbucks system and operate Starbucks coffee stores.

2.9.2.   

Commercial arrangement between Starbucks US and the Starbucks Manufacturing Corp.

(154) Starbucks indicated that SMC(84), the only other company ensuring roasting activities within the Starbucks group, does not pay any royalty to Starbucks for the use of the IP relating to the roasting technology and know-how or any other IP. According to Starbucks, this would be the case because SMC does not have separate audited accounts and because for US tax purposes SMC is consolidated with other Starbucks US entities.

3.   

GROUNDS FOR INITIATING THE PROCEDURE

(155) The Commission decided to initiate the formal investigation procedure because it took the preliminary view that the SMBV APA, which accepts the remuneration proposed by Starbucks' tax advisor for the functions performed by SMBV in the Netherlands, concluded by the Dutch tax administration in 2008 appeared to constitute State aid within the meaning of Article 107(1) of the Treaty that is incompatible with the internal market.

(156) In particular, the Commission expressed doubts that the remuneration agreed for the functions performed by SMBV complied with the arm's-length principle.

(157) More specifically, in the Opening Decision, the Commission raised the following three doubts as regards compliance of the SMBV APA with the arm's-length principle:

(1) Whether the Dutch tax administration correctly accepted SMBV's classification as a low-risk toll manufacturer when it concluded the SMBV APA;

(2) Whether the Dutch tax administration was right to accept adjustments made by Starbucks' tax advisor when it concluded the SMBV APA; and

(3) Whether the Dutch tax administration was right to accept SMBV's interpretation as regards the calculation of royalties in its profit and loss accounts, insofar as the level of those royalties is not linked to the value of the IP in question.

(158) In more detail, under first doubt the Commission questioned the tax advisor's assumption in the transfer pricing report that SMBV does not bear any risk and that it should therefore be classified as a toll or contract manufacturer. In particular, the Commission referred to the evidence of inventory risk recorded in the financial accounts of SMBV that would call this assumption into question.

(159) Under the second doubt, the Commission questioned two consecutive adjustments by the tax advisor, which seem both to have the same purpose of addressing one comparability concern. The first adjustment consists in reducing the cost base retained to calculate SMBV's taxable base to operating expense. The tax advisor considered this appropriate because SMBV would be a toll or contract manufacturer. A second adjustment, designated a ‘Conversion Mark-up Adjustment’ in the transfer pricing report, deducts a multiple of COGS from the profit of companies used as comparables for transfer pricing purposes. That second adjustment, presented by the Netherlands as a ‘working capital adjustment’, reduces SMBV's taxable base in the Netherlands, but neither the adjustment nor the methodology used by the tax advisor seemed justified.

(160) Finally, under its third doubt, the Commission questioned the arm's-length nature of the royalty paid by SMBV to Alki LP, since the amount of the royalty did not seem related to the value of the IP it is meant to remunerate. Due to the use of the TNMM in the transfer pricing analysis, the royalty corresponds in reality to the residual profit of SMBV, i.e. any profit recorded by SMBV above [9-12] % of operating expense is transformed into a tax deductible royalty (see Figure 2).

4.   

COMMENTS FROM THE NETHERLANDS

(161) The Netherlands submitted its comments to the Opening Decision on 16 July 2014. They focus, first, on why the remuneration agreed upon in the SMBV APA is at arm's length and why the method chosen by the tax advisor is the appropriate method to determine that remuneration. Second, they focus on why the SMBV APA does not confer a selective advantage to SMBV.

4.1.   

Comments on transfer pricing

(162) According to the Netherlands, the remuneration agreed upon in the SMBV APA is at arm's length and the TNMM is the appropriate method to reach an arm's-length outcome in this case. The Netherlands argues that transfer pricing is not an exact science and that there is therefore a range of figures within which the transfer price can lie.

(163) The Netherlands submits that, according to the OECD TP Guidelines, compensation for transactions between two independent companies will usually reflect the functions that each company performs and that therefore the functional analysis should be at the centre of the assessment of the arm's-length nature of the remuneration of SMBV. Pursuant to the OECD TP Guidelines, the contractual conditions are the starting point when it comes to determining whether the arm's-length principle has been applied correctly.

(164) According to the Netherlands, the IP assigned by Starbucks to Alki LP in respect of the knowledge and information relating to coffee includes the use of technology and knowhow regarding the recipe for blending coffee beans, the process of roasting coffee and the production of other derived coffee products.

(165) The Netherlands notes that the Roasting Agreement between Alki LP and SMBV cannot be considered a simple IP licence agreement. That agreement, a fifty-year manufacturing agreement, relates to the roasting of green coffee beans by SMBV. The Roasting Agreement is a contract between a client and a contractor, which also governs the provision of an IP right.

(166) Alki LP is not based in the Netherlands and, in accordance with the international apportionment of taxation powers, is not liable to pay tax in the Netherlands. The activities carried out by the employees of SMBV are, and have always been:

(1) Manufacture/coffee roaster activities: these activities are performed by production staff, coffee roasting technicians, maintenance staff, quality control staff and warehousing staff.

(2) Logistics and administrative support activities.

(167) The Netherlands states that Starbucks intention has always been to set up an operating, low-risk coffee roasting plant and that the facts and circumstances have not changed significantly over the years. They argue that as SMBV was not involved in any business restructuring, the Commission cannot cite passages from Chapter 9 of the 2010 OECD TP Guidelines, as the 2010 OECD TP Guidelines were not yet available when the SMBV APA was agreed upon. The use of hindsight should be avoided under the OECD TP Guidelines.

(168) According to the Netherlands, SCTC is responsible for purchasing the green coffee beans. It further argues that SMBV is not involved in the sourcing of raw materials as, according to the Roasting Agreement, SMBV can only source raw materials from parties which have been designated by Alki LP. SMBV only performs an administrative role in the management of stocks and does not bear the ultimate stock risk.

(169) More specifically with regard to the administrative role, the Netherlands states that although under the Green Coffee Purchase Agreement it is SMBV that has to provide the information required with regard to the green coffee beans to be bought, this information is obtained by SMBV from Alki LP and, in a manner prescribed by Alki LP, from the Developers. The specifications are therefore only passed on by SMBV once it has obtained the information required for this purpose. According to the Netherlands, this combination results in a situation in which the role of the raw materials is comparable with that of providing goods on a consignment basis.

(170) With regard to the ultimate stock risk, the Netherlands adds that, although SMBV retains the legal title to all products and materials used for the production activities, and it has put in place provisions for losses in the value of stock, the costs for which those provisions have been put in place are ultimately not born by SMBV. Given the way the royalty payment is determined under the Roasting Agreement, those costs are ultimately borne by Alki LP.

(171) According to the Netherlands, no employees of SMBV are involved in business negotiations with Starbucks's developers/buyers. Regarding the price-setting, the Netherlands argues that Starbucks […] determines the global pricing formulae. The accounting team at Starbucks […] provides SMBV with the price list (‘[…] Price List’) by way of Alki LP and those prices are calculated using the globally adopted pricing formulae on which the Supply Agreements between SMBV and the Developers are based. The purchase price for green coffee beans paid by SMBV is derived from a pricing formula which includes aspects such as […]. If the underlying costs rise or fall, the cost base applied to set the prices between SMBV and its Developers is also amended. Moreover, Developers are obliged to purchase coffee and essential goods under the ADOA concluded between Starbucks Coffee BV and Developers.

(172) According to the Netherlands, SMBV does not operate under a toll manufacturer contract. The contractual relationship results in a situation where the green coffee beans are being purchased legally, but without a functional contribution, from SCTC and invoiced to the buyers. Despite the lack of functionality as regards the purchase and sale, the stocks need to appear on SMBV's balance sheet in line with accounting standards.

(173) The contractual relationships between SMBV, Alki LP, SCTC and the Developers described above lead, according to the Netherlands, to the conclusion that SMBV is a manufacturer with a low-risk profile and that conclusion is further supported by the functional analysis. SMBV is therefore to be regarded by the Netherlands as the ‘least complex entity’ (tested party) whose arm's-length remuneration must be determined using a benchmarking study.

(174) According to the Netherlands, the databases used to carry out a benchmarking study do not provide any details such as the transaction prices or terms and conditions. The information available is limited to a comparison of operational results for the entity as a whole. The TNMM is the most commonly used method internationally. According to the Netherlands, this is confirmed in the 2010 OECD TP Guidelines as well as the 1995 OECD TP Guidelines, which applied when the SMBV APA was concluded.

(175) In view of the production functionality of SMBV, a cost-oriented profit level indicator is used by SMBV (TNMM method with a profit mark-up on the costs). In this respect, the Netherlands stresses that, in accordance with paragraph 4.9 of the OECD TP Guidelines(85), the Dutch tax administration always begins its investigation into transfer prices from the perspective of the method used by the taxpayer. Chapter 2 of the Decree also states that the Netherlands does not apply a ‘best method-rule’. This starting point, which is, according to the Netherlands, in line with the OECD TP Guidelines, is also used in the SMBV APA and has resulted in the acceptance of the TNMM with the profit level indicator proposed by SMBV.

(176) In view of SMBV's role as a toll manufacturer, the relevant costs to determine the cost basis used to calculate the profit mark-up are the costs which have resulted in added value. In the case of SMBV, the operational costs are, according to the Netherlands, the relevant costs with added value to which a profit mark-up is applied. Thus, based on the benchmark, the profit mark-up is [9-12] % of the operational costs. SMBV has a low risk profile and no added value with regard to the raw materials and plays a supporting role for the non-coffee-related products. The operational costs associated with these limited support activities are included in the cost base. As a result, these activities are also paid with a [9-12] % profit mark-up.

(177) With regard to the comparability adjustments, the Netherlands argues that at the time that the transfer pricing report was drawn up on which the SMBV APA is based, there was no indication on how to deal with manufacturers with a low risk profile. It acknowledges that the 2010 OECD TP Guidelines do contain more instructions on comparability adjustments, including working capital adjustments, but argue that those examples are only guidelines and that this means that other positions can also result in an arm's-length result. Moreover, the Netherlands argues that the assessment of whether the SMBV APA is in line with the arm's-length principle must be based on the knowledge and existing OECD TP Guidelines that were available at the time, i.e. the 1995 OECD TP Guidelines. Therefore, also the Annex to Chapter III of the 2010 OECD TP Guidelines, which shows a working example of a comparability adjustment, could not be applied.

(178) The Netherlands has also performed a sensitivity analysis on the arm's-length range, calculated according to the methodology in the transfer pricing report, if some of the parameters are modified. The Netherlands modified the time period of the comparison and recalculated ranges considered as arm's length for periods 2008-2012 and 2003-2012, compared to the range of 2001-2005 in the transfer pricing report. Additionally the Netherlands recalculated the range if the percentage of COGS deducted from the remuneration under the second adjustments would not be EURIBOR plus 50 basis points, but rather EURIBOR minus 50 basis points and finally if it would be what the Netherlands designate as ‘current account rate’. The Netherlands also calculated the margin achieved by SMBV as a percentage of operating cost and raw material costs (excluding the cost associated with tea and other COGS with an intermediary nature) for the periods 2008-2012 and 2003-2012 and compared this with the mark-up on the total cost for the comparable companies on both EBT and EBIT level for the same time periods. On the basis of those simulations, the Netherlands' conclusion is that if any of the doubts raised by the Commission were to be accepted, this would still result in a remuneration within the arm's-length range. In accordance with the OECD TP Guidelines, no corrections are permitted to be made when the remuneration falls within the arm's-length range.

(179) Finally, on a more general note, the Netherlands disputes the allegation that it failed to request the underlying contracts from which it could be shown that the level of risk carried by SMBV was limited, arguing that this is not necessary as every APA always includes a critical assumption that the facts and circumstances presented appear to be correct and that, in case the facts were not presented correctly, the APA can be terminated. Moreover, the Netherlands argues that, as it had at its disposal the historical background of how the roasting plant was set up, it did not need to request all the contracts(86).

4.2.   

Comments on the application of Article 107(1) of the Treaty

(180) The Netherlands further submits that no selective advantage is being conferred on SMBV and that no State aid is involved within the meaning of Article 107(1) of the Treaty.

(181) The Netherlands explains that the arm's-length principle has been incorporated into Article 8b(1) CIT and elaborated further in the Decree, which is fully in line with Article 9 of the OECD Model Tax Convention. The Netherlands reiterates what is also stated in the Decree itself, that the Decree is specifically aimed at aspects which the OECD TP Guidelines leave open to interpretation or where there is a lack of clarity.

(182) In particular, the Netherlands argues that, in situations relating to transfer pricing, a selective advantage can only be involved if it is demonstrated that the OECD TP Guidelines and Decree are expressly deviated from and that obvious errors of judgment are made in the application of the arm's-length principle or if an established national policy is being deviated from. The tax authorities' discretion in assessing and approving methods and results in individual cases does however not imply any selectivity in any way or arbitrary treatment. Given that the Netherlands considers the result of the SMBV APA as an acceptable approximation of a market price, they do not consider it to confer an advantage to SMBV.

(183) The Netherlands further argues that the reference system applied by the Commission, which is the ordinary tax system based on the difference between profits and losses of an undertaking carrying on its activities under normal market conditions, is not correctly identified. According to the Netherlands, the correct reference system should be the corporate income tax law which includes the arm's-length principle under Article 8b(1) CIT and the Decree that provides further guidance on the application of the arm's-length principle. The Netherlands argues that as long as the SMBV APA does not deviate from Article 8b(1) CIT and the Decree, there cannot be a selective advantage.

(184) With regard to the TNMM method used, the Netherlands argues that on the basis of the Decree every taxpayer is in principle free to choose a transfer pricing method, provided the method chosen leads to an arm's-length outcome for the specific transaction. Therefore, the doubts about the use of the TNMM which were raised by the Commission go beyond the doubts that the Dutch tax administration could have expressed under the Decree with regard to this transfer pricing method. Further, the Netherlands insists that the Decree requires only that a transfer pricing analysis results in a range of arm's-length results and not in a precise arm's-length price.

(185) In addition, the Netherlands argues that the decision does not appear to take into account the fact that affiliated and unaffiliated companies are not always in a similar legal and factual situation. There are always differences between affiliated and unaffiliated companies which is something, according to the Netherlands, also acknowledged in the Commission's

Groepsrentebox

decision(87).

(186) Finally, the Netherlands observes that if the Commission were to impose its own interpretation of tax principles of the Member States, it would encroach on the sovereignty of the Netherlands.

5.   

COMMENTS FROM INTERESTED PARTIES

5.1.   

Comments by Starbucks to the Opening Decision and the MIT request

(187) Starbucks sent its comments to the Opening Decision on 16 January 2015. In addition, Starbucks sent market information to the Commission by letters dated 13 April 2015, 29 May 2015, 10 September 2015 and 23 September 2015. Starbucks's comments largely resemble those of the Netherlands, in as much as both argue that the remuneration in the SMBV APA was at arm's length and did not confer a selective advantage to SMBV.

(188) First, Starbucks argues that SMBV only performs limited, low-risk functions in support of the worldwide Starbucks organisation serving the EMEA region. In support of that argument, […]. According to Starbucks, the primary responsibility of SMBV consists in the roasting and packaging of coffee, as well as logistic and administrative activities to ensure a smooth and efficient delivery of the various goods to the Developers.

(189) With regard to the structure chosen, Starbucks argues that a roasting arrangement on a consignment basis was considered, but that this would have been unpractical and would have resulted in administrative complexities and too many inter-company transactions. To align the administrative and legal structure with the physical flow of the goods and for efficiency reasons, Starbucks decided to use SMBV as the contracting and invoicing entity. A consequence of that structure is that the inventories (for green coffee beans, non-coffee products and non-strategic goods) appear on the balance sheet of SMBV, since accounting standards and practice follow the legal product flow. However, Starbucks argues that SMBV's role as contracting and invoicing entity is just of an administrative nature and does not result in any transfer of risks or commercial responsibilities to it as under the Roasting Agreement it is rather Alki LP, supported by Starbucks US, that bears all the economic risk of SMBV, including the inventory risk. Moreover, even though SMBV is the contracting party with the Developers, this is done on the basis of standardised agreements and underlying terms and conditions that are determined by Starbucks US. Finally, as SMBV acquires the legal ownership, the inventory would also have to appear in its balance sheet, which according to Starbucks also explains why SMBV took provisions for inventory obsolesce, which it is compensated for under the SMBV APA.

(190) As regards the transfer pricing method chosen, although not present in the transfer pricing report, Starbucks sets out the role of Alki LP against the role of SMBV. Starbucks argues that since SMBV is only engaged in routine execution activities in the areas of roasting, packaging and supporting logistic and administrative services, while Alki LP licenses the valuable intellectual property and bears the entrepreneurial risk, SMBV is the least complex entity. Therefore, the TNMM constitutes the most appropriate transfer pricing method. Starbucks argues that, because there are no comparable transactions similar to the arrangement between Alki LP and SMBV, the CUP method would not have been suitable for transfer pricing purposes. In any event, the Dutch tax administration has to start the transfer pricing examination on the basis of the methodology selected by the taxpayer.

(191) Starbucks recalls that the application of any transfer pricing method typically produces a range of figures, which could be equally defensible, since transfer pricing is not an exact science and that any transfer pricing analysis will inherently result in a range of arm's-length outcomes and a conclusion on

an

arm's-length price and not

the

arm's-length price.

(192) As regards the adjustments, the purpose of the adjustments is to account for important differences in the functional profiles of SMBV and the comparable companies included in the sample to arrive at an appropriate arm's-length remuneration. According to Starbucks, those adjustments were appropriate, in certain respects conservative, and certainly did not understate SMBV's remuneration for the functions performed. To further substantiate the reasonableness of the [9-12] % mark-up, Starbucks asked [the tax advisor] to make a comparison between the actual results realised by SMBV with the actual results realised by the comparable companies in the period 2008–2012. This backward looking analysis demonstrates that the [9-12] % mark-up has remained comfortably within the appropriate ranges. That further confirms the arm's-length nature of the applied transfer pricing methodology for SMBV as agreed upon in the APA concluded with the Dutch tax authorities. Starbucks also requested [a law firm] to provide a second opinion on whether the 2007 transfer pricing report properly applied the arm's-length principle. [The law firm] did not conducted an own factual investigation but reviewed the transfer pricing report and the documents available to the tax advisor. It concluded that the arm's-length principle had been reasonably applied to SMBV's intragroup transactions.

(193) Similarly to the Netherlands, Starbucks further argues that the Commission did not correctly identify the reference framework. Starbucks argues that the reference system should be the Dutch system of corporate taxation and more specifically Article 8b(1) CIT, the Decree and the administrative practice of the Dutch tax authorities. According to both parties, there could only be a State aid concern if the SMBV APA deviates from the normal interpretation and application of the arm's-length principle in the Netherlands. Moreover, Starbucks argues that an identification of a benchmark group of taxpayers is missing and argues that related and unrelated companies are not always in a comparable legal and factual situation(88). Finally, Starbucks criticises the Commission's approach for being at odds with the Dutch tax administration's ‘substance over form approach’.

(194) In its submissions in response to the Commission's Starbucks MIT request, Starbucks expressed views presented in recitals 151 to 154. Starbucks further indicated that the Commission cannot use information which post-dates the SMBV APA for its assessment.

(195) On 24 July 2015, Starbucks submitted a report by [a management consulting firm], produced in response to its request to analyse remuneration models for manufacturing and logistic/administrative services. Starbucks claims that that report supports Starbucks' position regarding the remuneration model and profit level indicator chosen(89).

5.2.   

Nederlandse Orde van Belastingadviseurs

(196) The NOB argues that the determination whether a particular tax treatment of a taxpayer under an APA constitutes State aid should be based on Dutch legislation, administrative practice and application of the arm's-length principle at the time that that APA was entered into. It also argues that under the Decree, the Dutch tax administration does not apply ‘a best-method rule’ (that is, it does not require that the best method is used for tax base calculation) and that the application of the arm's-length principle usually results in an arm's-length range instead of a single arm's-length price. The NOB notes that the reference to the prudent independent market operator in the Opening Decision seems to introduce a new sort of EU standard above and beyond the OECD TP Guidelines for assessing the arm's-length nature of the underlying arrangement. It asks for a confirmation that the Commission will use the domestic legal system as a reference framework and no other standard. It further argues that taxpayers should have legitimate expectations that APAs that are concluded on the basis of a national interpretation of the domestic laws do not constitute State aid.

5.3.   

VNO-NCW

(197) VNO-NCW expresses its worries on the application of the prudent independent market operator test and urges for the use of the nationally applied transfer pricing rules as the benchmark for assessing selectivity. It argues that the application of the arm's-length principle usually results in an arm's-length range instead of a single arm's-length price.

5.4.   

ATOZ

(198) ATOZ's main argument relates to the legal basis for the Opening Decision. According to ATOZ, the decision does not distinguish whether the pricing agreement in the SMBV APA diverges from Dutch administrative practice, from any other tax authorities' habitual practice or from OECD standards, but merely concludes that it does not respect the arm's-length principle. ATOZ argues that it seems that the Commission takes the view that there is an objective arms-length standard, based on OECD principles and somehow enshrined in EU law, which transcends Member States' national law and practice. However, according to ATOZ, the Commission should consider whether the SMBV APA is consistent with Dutch law. ATOZ argues that the Commission's approach creates, amongst others, legal uncertainty among multinationals.

5.5.   

Oxfam

(199) Oxfam in its comments expressed support for the Commission's investigation, encouraging the Commission to increase its investigation capacity also in view of the fact that it may be better placed than national bodies to structurally assess the tax ruling practices of the Member States. It calls on the Commission to ensure that adequate sanctions are adopted in cases where selective advantages are confirmed and that harmful tax practices are phased out quickly.

5.6.   

BAK

(200) The BAK supports the Commissions arguments from the Opening Decision and argues that, in general, those sorts of agreements and legal structures lower the worldwide taxes paid, which has negative consequences for consumers and employees.

6.   

OBSERVATIONS BY COMPANY X

(201) Company X, which does not want its identity to be disclosed, presented observations to the Commission in response to the Opening Decision. According to that company, based on its evaluation as a competitor of Starbucks, the value added by the roasting process (roasting and packing, not considering the cost for packaging materials) to green coffee would be on average equal to 13-17 % of the green coffee cost, in case of roast and ground coffee or coffee in beans. According to that company, those levels would be applicable to all distribution channels.

7.   

INFORMATION SUPPLIED BY STARBUCKS' COMPETITORS IN RESPONSE TO THE MIT REQUEST

7.1.   

Companies contacted by the Commission in the context of the MIT request

(202) As explained in recital 20, the Commission contacted four competitors of Starbucks to provide market information on their business model and their value creating activities so as to enable the Commission to complete its assessment of the SMBV APA. The four competitors concerned are Company Y, Dallmayr, Nestlé and Melitta. The choice of the four companies was based on the consideration that all companies are active in the coffee roasting segment and, being group companies, could provide insight into the organisation of coffee roasting activities within an integrated company.

7.2.   

Dallmayr

(203) By letter of 27 April 2015, Dallmayr replied to the request for market information by the Commission.

(204) Dallmayr informed the Commission that coffee roasting is either performed as a stand-alone business or vertically integrated within a company. Larger companies usually perform roasting in-house. The sourcing function is typically integrated with the roasting function. Dallmayr does not outsource the roasting function.

(205) Dallmayr considers the payment of a royalty by a third party that provides the roasting services rather unusual. In fact, Dallmayr would expect the customer to pay the roaster, not the other way around.

7.3.   

Nestlé

(206) By letter of 20 May 2015, Nestlé replied to the Commission's request for market information. Nestlé indicated that the three elements important in the value creation for coffee are the quality (the type of bean), the darkness (achieved by roasting), and the grind size. They also indicated that they do not and would not outsource the roasting function due to the importance of roasting to the flavour development of the product.

7.4.   

Melitta

(207) By letter of 26 May 2015, Melitta replied to the Commission's request for market information. Melitta indicated situations in which they could outsource the roasting of coffee beans. These situations are: (i) when machinery is not available in its factory to produce specific types of products, such as specific packaging formats or soft pads; (ii) when machinery is not available in its factory to produce soluble coffee; and (iii) when sales exceed the available roasting and packaging capacities at its factory. In that last case, when roasting and packaging is outsourced due to capacity constraints, green coffee beans from Melitta are sent to the supplier who roasts and packs the coffee to a finished product. That finished product is then delivered to Melitta. Such a contract was in place with the same supplier for a number of years.

(208) To assure the quality and taste of the finished product, Melitta either provides roasting curve prescriptions or defined taste profiles to the third party to whom the roasting is outsourced.

(209) The contractual arrangement does not foresee any royalties paid to Melitta when outsourcing the roasting activity.

7.5.   

Company Y

(210) By letter of 27 April 2015, Company Y replied to the Commission's request for market information. The Commission requested a clarification by a letter of 11 May 2015, to which Company Y replied on 21 May 2015.

(211) Company Y does not outsource the coffee roasting function to third parties. The roasting is ensured by a group company designated as a toll manufacturer by Company Y. This coffee roasting company does not pay any royalty for the use of the intellectual property or the know-how used in the roasting process.

(212) The company pays licence fees for the use of IT systems. For the remuneration of the group company, classified by Company Y as a toll-manufacturer, the cost base is the production costs, excluding raw material costs. The production costs are particularly energy, depreciations on machines (as for example roasting and packaging line), personnel costs, IT costs and maintenance of equipment.

8.   

COMMENTS FROM THE NETHERLANDS ON THIRD PARTIES' COMMENTS, ON COMMENTS BY COMPANY X AND THE REPLIES TO THE MIT REQUEST

8.1.   

The Netherlands' comments on third parties' comments

(213) By letters dated 20 and 26 April 2015, the Netherlands expressed their complete agreement to the observations of Starbucks, the NOB, VNO-NCW and ATOZ. With regard to the comments of Oxfam, the Netherlands indicated that this concerns an observation on detrimental tax competition in general and does not examine the SMBV case in particular. The Netherlands considers the claims made by BAK incorrect and refrains therefore from commenting on them.

8.2.   

The Netherlands' comments on Company X's comments

(214) By letter dated 11 March 2015, the Netherlands stated that they cannot provide any substantive reply to the comments by company X, as the Netherlands were not provided with the functional analysis and benchmark of the anonymous competitor.

8.3.   

The Netherlands' comments on Dallmayr's and Company Y's comments

(215) By letter dated 27 May 2015, the Netherlands provided its comments to the market information provided by Dallmayr and Company Y. As a general comment, the Netherlands states that the functional analysis and the contractual arrangements are missing and that making a comparison is therefore very difficult as both parties do not outsource the roasting function to an independent third party.

(216) With regard to Dallmayr, the Netherlands states that Dallmayr's definition of roasting includes more than just coffee roasting, as the sourcing function is integrated with the roasting function. Furthermore, the Netherlands argues that SMBV does not perform sales activities with regard to coffee and non-coffee items but that it seems that Dallmayr has been differently organised on this point. Furthermore, under Dallmayr's client — contractor relationship, the Netherlands argue that the remuneration takes place on the weight and price of green coffee beans which puts the occupation degree risk on the contractor where this is not the case with SMBV.

(217) With regard to Company Y, the Netherlands states that coffee roasting is considered a routine function and that the roasting facility is remunerated on a cost plus margin, where the green coffee bean cost does not form part of the cost base. According to the Netherlands, this approach is fully in line with the SMBV APA.

8.4.   

The Netherlands' comments on Nestlé's comments

(218) With regard to Nestlé, the Netherlands states that the three elements that create value for coffee according to Nestlé are all performed by other foreign companies of the Starbucks group, not by SMBV (sourcing and quality control of the beans is performed by SCTC, the roasting curves are provided by Alki LP, and the Starbucks coffee Shops perform the grinding).

8.5.   

The Netherlands' comments on Melitta's comments

(219) With regard to Melitta, the Netherlands states that the three situations described by Melitta where the coffee roasting function is outsourced is in some ways different to the situation of SMBV. In situation 1 and 3, similarity lies in the fact that the sourcing function is also not performed by the producer, but according to the Netherlands the difference is the length of the roasting contract (annual contract compared to the 50 year valid roasting agreement) and the occupation degree risk (which would be at the level of Alki LP according to the Netherlands compared to at the level of Melitta). The second situation the Netherlands considers too different to be able to make a comparison, as it not only concerns the roasting of coffee beans but also the production of all-inclusive products.

8.6.   

The Netherlands' comments on Starbucks' letters of 13 April, 29 May 2015, 10 and 11 September 2015 and 23 September 2015

(220) By letter dated 19 June 2015, the Netherlands provided its comments to the market information provided by Starbucks. In those comments, the Netherlands repeated its statement that they consider that the correct reference framework should be the Dutch national tax system and, in particular, Article 8b of the CIT 1969 and the Decree. It argues that Article 8b and the Decree always apply for intra-group transactions, whether a company asked for an APA or not. Moreover, the Netherlands states that the TNMM is internationally the most commonly used method and that the Netherlands does not apply a best method rule. It also argues that an advantage of the TNMM is that a country only has to consider the transfer price unilaterally and that a possible higher or lower transfer price for the green coffee beans would not affect the tax base of SMBV, as the cost of the beans are excluded from the cost base to which the margin is applied.

(221) In their letter of 25 September 2015, the Netherlands repeat their statements that based on the transfer pricing report, the contractual relationships between SMBV and its counterparties, as well as the actual behaviour of SMBV, SMBV should be considered as a manufacturer with a low risk profile. The Netherlands further argue that it is common practice in the Netherlands that unique functions such as a combination of holding intangible assets and taking on entrepreneurial risks cannot be benchmarked and that it is therefore logic that the royalty payments is calculated as a residual. According to the Netherlands, their statement that the SMBV's APA is at arm's length and fully in line with the OECD TP Guidelines is supported by the second opinion performed by [a law firm] (previously submitted, see recital 192), the submission of Company Y(90), the sensitivity analysis performed by the Netherlands (previously submitted see recital 178) and by various articles by tax experts. Moreover, although according to the Netherlands a CUP could not be applied to the situation of SMBV, a CUP type of analysis of the contracts between Starbucks and third parties on coffee roasting or co-manufacturing of coffee products, see (recital 152) show according to the Netherlands that the royalty payments from SMBV to Alki LP were not too high(91).

(222) In its letter of 7 October 2015, the Netherlands repeats its argument that the CUP method is not applicable to the royalty payment by SMBV to Alki LP. In addition, the Netherlands indicates that the price paid for green coffee beans by SMBV to SCTC would be at arm's length, based on the fact that [unaffiliated manufacturing company 2] was paying a higher price for green coffee beans than SMBV under its contract with Starbucks. Finally, the Netherlands provided financial information on 11 companies from different Union Member States, all registered under the NACE code ‘Processing of tea and coffee’, indicating that on balance all these companies have a similar or lower profitability than SMBV and that some of those companies even have been loss-making over several financial periods.

9.   

ASSESSMENT OF THE CONTESTED MEASURE

9.1.   

Existence of aid

(223) According to Article 107(1) of the Treaty, any aid granted by a Member State or through State resources in any form whatsoever which distorts or threatens to distort competition by favouring certain undertakings or the provision of certain goods shall be incompatible with the internal market, in so far as it affects trade between Member States.

(224) According to settled case-law, for a measure to be categorised as aid within the meaning of Article 107(1) of the Treaty, all the conditions set out in that provision must be fulfilled(92). It is thus well established that, for a measure to be categorised as State aid within the meaning of that provision, there must, first, be an intervention by the State or through State resources; second, the intervention must be liable to affect trade between Member States; third, it must confer a selective advantage on an undertaking and, fourth, it must distort or threaten to distort competition(93).

(225) As regards the first condition for a finding of aid, the SMBV APA was concluded by the Dutch tax administration (

Belastingdienst

), which is part of the public administration of the Netherlands. That APA entails an acceptance by the Dutch tax administration of a profit allocation proposed by Starbucks on the basis of which SMBV determines its corporate income tax liability to the Netherlands on a yearly basis. The SMBV APA is therefore imputable to the Netherlands.

(226) As regards the measure's financing through State resources, the Court of Justice has consistently held that a measure by which the public authorities grant to certain undertakings a tax exemption which, although not involving a positive transfer of State resources, places the persons to whom it applies in a more favourable financial situation than other taxpayers constitutes State aid(94). Below, the Commission will demonstrate that the SMBV APA results in a lowering of SMBV's tax liability in the Netherlands by deviating from the tax that SMBV would otherwise have been obliged to pay under the general Dutch corporate tax system. Consequently, the SMBV APA should be considered to give rise to a loss of State resources, since any reduction of tax for SMBV results in a loss of tax revenue that would otherwise have been available to the Netherlands.

(227) As regards the second condition for a finding of aid, SMBV is part of the Starbucks group, a globally active entity operating in all Member States of the Union, so that any aid in its favour is liable to affect intra-Union trade. Similarly, a measure granted by the State is considered to distort or threaten to distort competition when it is liable to improve the competitive position of the recipient compared to other undertakings with which it competes(95). To the extent the SMBV APA relieves SMBV of a tax liability it would otherwise have been obliged to pay under the general Dutch corporate income tax system, that APA distorts or threatens to distort competition by strengthening its financial position, so that the fourth condition for a finding of aid is also fulfilled in the present case.

(228) As regards the third condition for a finding of aid, the Commission will demonstrate from recital 252 onwards why it considers the SMBV APA to confer a selective advantage upon Starbucks, in so far as it results in a lowering of SMBV's tax liability in the Netherlands by deviating from the tax SMBV would be due under the general Dutch corporate income tax system, therefore fulfilling all the conditions for a finding of aid under Article 107(1) of the Treaty.

9.2.   

Existence of a selective advantage

(229) According to settled case-law, ‘Article 107, paragraph 1 of the Treaty requires it to be determined whether, under a particular statutory scheme, a State measure is such as to favour 'certain undertakings or the production of certain goods’ in comparison with others which, in the light of the objective pursued by the scheme in question, are in a comparable legal and factual situation. If it is, the measure concerned fulfils the condition of selectivity'(96).

(230) In fiscal cases, the Court of Justice has devised a three-step analysis to determine whether a particular tax measure is selective(97). First, the common or normal tax regime applicable in the Member State is identified: the ‘reference system’. Second, it is determined whether the tax measure in question constitutes a derogation from that system, in so far as it differentiates between economic operators who, in light of the objectives intrinsic to the system, are in a comparable factual and legal situation. If the measure constitutes a derogation from the reference system, it is then established, in the third step of the analysis, whether that measure is justified by the nature or the general scheme of the reference system. A tax measure which constitutes a derogation to the application of the reference system may be justified if the Member State concerned can show that that measure results directly from the basic or guiding principles of that tax system(98). If that is the case, the tax measure is not selective. The burden of proof in that third step lies with the Member State.

9.2.1.   

Determination of the reference system

9.2.1.1.   

Reference system composed of the general Dutch corporate tax system

(231) As a general rule, for the purposes of the selectivity analysis a reference system is composed of a consistent set of rules that apply on the basis of objective criteria to all undertakings falling within its scope as defined by its objective.

(232) In the present case, the Commission considers the reference system to be the general Dutch corporate income tax system which has as its objective the taxation of profits of all companies subject to tax in the Netherlands(99). Companies established in the Netherlands are resident taxpayers(100); they are subject to corporate income tax on their worldwide income(101). Companies that are not established in the Netherlands (non-resident companies) are subject to tax with regard to income from Dutch sources(102). Whether a company is deemed to be established in the Netherlands for tax purposes is assessed on the basis of factual circumstances. Important factors for this assessment include the place of actual management and the head office location. Under the CIT, all companies incorporated under Dutch law are deemed to be residents of the Netherlands(103).

(233) According to Article 3.25 of the Income Tax Act 2001 (

Wet inkomstenbelasting 2001

), which through article 8 CIT 1969 also applies to corporate taxpayers, the taxable yearly profits must be determined on the principles of sound business practice and in a consistent manner independently of the likely outcome(104). According to sound business practice, for example, allowances may be made for unrealised losses, while profits not yet realised may be disregarded.

(234) In general, the taxable profits correspond to the accounting profits as reflected in the company's profit and loss accounts. However, adjustments can be made based on specific tax provisions, such as applicable tax incentives, the participation exemption, corrections to the tax result from transactions not executed at arm's length and the application of different depreciation rules under tax and accounting rules(105).

(235) While the determination of taxable profits in the case of non-integrated/domestic standalone companies that transact on the market is rather straightforward, as it is based on the difference between income and costs in a competitive market, the determination of taxable profits in the case of integrated group companies like Starbucks requires the use of proxies. Standalone, non-integrated companies can take their accounting profits as a starting point for determining the tax base to which the Dutch corporate income tax applies, since those profits are dependent on prices dictated by the market for the inputs acquired and the products and services sold by the company. By contrast, an integrated company that transacts with companies of the same corporate group will first have to estimate the prices applied to those intra-group transactions for determining their taxable profits, that estimate being determined by the same company controlling the group instead of being dictated by the market.

(236) However, this difference in determining the taxable profits of non-integrated companies, i.e. those not belonging to a corporate group and thus ‘standalone’, and integrated companies, i.e. those belonging to a corporate group, has no bearing on the objective of the Dutch corporate income tax system which aims to tax profits of all companies subject to tax in the Netherlands, whether non-integrated or integrated. Since under the general Dutch corporate income tax system, the profits of all companies resident in the Netherlands are taxed in the same manner without any distinction as to group and uncontrolled companies, both types of companies should be considered to be in a similar factual and legal situation in light of the intrinsic objective of that system(106). Indeed, since the aim of the SMBV APA is to determine the tax base of SMBV for the purpose of levying corporate income tax under that system, it is the general Dutch corporate income tax system that constitutes the reference system against which that APA should be examined to determine whether Starbucks has benefitted from a selective advantage. Accordingly, the different manner in which the taxable profit is necessarily arrived at for integrated and non-integrated companies has no relevance for determining the reference system for the selectivity analysis in the present case.

(237) This is also confirmed by Article 8b(1) CIT and the Decree that implements the OECD's arm's-length principle into Dutch tax law. In its introduction, the Decree states: ‘The policy of the Netherlands on the arm's-length principle in the field of international tax law is that this principle forms part of the Netherlands' system of tax law as a result of its incorporation in the broad definition of income recorded in Section 3.8 of the Income Tax Act 2001.’ Section 3.8 of the Income Tax Act 2001 reads: ‘[t]he profit from a business enterprise is the amount of the aggregate benefits that, under whatever name and in whichever forms, are derived from a business enterprise’. Thus, the Decree, through its reference to Section 3.8 of the Income Tax Act 2001 and the concepts of income and profit ‘under whatever name and in whichever form derived from a business enterprise’, does not make any distinction between income and profits derived by a group company or a standalone company.

(238) According to the Netherlands and Starbucks, the Commission has previously decided, in its decision on

Groepsrentebox

(107), that groups and uncontrolled companies do not belong to the same reference system(108). They thereby seem to imply that the reference system to assess selectivity can only include companies being subject to transfer pricing rules, i.e. group companies.

(239) At the outset, the Commission recalls that it is not bound by its decisional-practice. Each potential aid measure must be assessed on the basis of its own merits under the objective criteria of Article 107(1) of the Treaty, so that even if a contrary decisional practice were shown to exist, that could not affect the findings of the present decision(109).

(240) In any event, contrary to what the Netherlands and Starbucks claim, the

Groepsrentebox

decision does not confirm that where a tax measure is granted in favour of an integrated company, the reference system must necessarily be limited to those types of companies. Moreover, the objective of the tax measure at the basis of the

Groepsrentebox

decision is not comparable to the present case and therefore, the conclusions the Netherlands and Starbucks draw from that decision are not applicable to the present case.

(241) The

Groepsrentebox

scheme was set up at the time by the Dutch authorities to reduce the difference in tax treatment between the provision of equity capital and loan capital in a group context and, as such, to reduce arbitrage between these two forms of intra-group financing(110). The Commission observed in its final decision that given the objective of the measure, which was to reduce the difference in tax treatment between the provision of equity capital and loan capital in a group context and, as such, to reduce arbitrage between these two form of intra-group financing, it is ‘only group companies, [and not stand-alone companies,] that are confronted with arbitrage between equity capital and loan capital within their group’(111). It was in the light of that observation, as well as of the objective of the scheme which was ‘to reduce incentives for arbitrage between financing through a capital injection and a loan, and ensuring tax neutrality in this regard’(112), that the Commission considered the reference system in that case to include only companies subject to corporation tax and engaged in intra-group financing transactions(113).

(242) By contrast, the objective of the SMBV APA is to determine SMBV's tax base to calculate the tax due for the purposes of levying the Dutch corporate income tax on that amount. First, while it could be argued that the objective underlying the

Groepsrentebox

decision is only valid in a group context (such as the fact that stand-alone companies are not faced with the issue of arbitrage between different forms of financing), the determination of the tax base for the computation of the annual corporate income tax liability is equally relevant and applicable to entities that are part of a group as well as stand-alone companies.

(243) Second, while it is true that SMBV provides services to other companies of the group and thus operates in a group context, the transaction it carries out could also be carried out outside a group context. SMBV is a coffee roaster and distribution company. As the contracts submitted by Starbucks demonstrate, other companies of the Starbucks group outsource this function to third parties(114). Moreover, the answers to the MIT request by the competitors further demonstrate that roasting is outsourced to non-group companies(115). Thus, the activities of SMBV can be carried out by independent companies and not only in a pure intra-group setting(116).

(244) The Commission therefore concludes that the reference system against which the SMBV APA should be examined is the general Dutch corporate tax system in the form of the Dutch corporate income tax rules (CIT). In particular, that reference system is composed of a consistent set of rules that apply on the basis of objective criteria for the taxation of profits of stand-alone companies, where the determination of the taxable profit usually coincides with the accounting profit (subject to certain adjustments based on tax law) and of group companies, which resort to transfer prices to allocate profits, alike. In light of the intrinsic objective of that system, both types of companies — non-integrated and integrated companies — should be considered to be in a similar factual and legal situation.

9.2.1.2.   

Article 8b(1) CIT and the Decree do not constitute the appropriate reference system

(245) The Netherlands considers that the reference system should be the Decree and consequently that SMBV should be considered in a similar factual and legal situation only to group companies that fall under the Decree(117). Accordingly, the existence of selectivity would require evidence that SMBV has received a different treatment as compared to other group entities tax resident in the Netherlands falling under the Decree.

(246) The Commission does not accept this line of reasoning.

(247) As explained in recital 236, the objective of the Dutch corporate income tax system is to tax the profits of all companies that fall under its tax jurisdiction, irrespective of whether those companies are integrated or non-integrated companies. As explained in recital 232, Dutch corporate income tax is levied on the worldwide profits of companies resident in the Netherlands (unless a tax treaty applies), while non-resident companies, including Dutch branches of foreign companies, are only taxed on specific Dutch-sourced income.

(248) By considering, as the Netherlands does, that the reference system only includes group companies, since only they need to revert to the arm's-length principle as required by Article 8b CIT and the Decree when allocating profit, an artificial distinction is introduced between companies based on their company structure for the purpose of determining their taxable profits that the general Dutch corporate income tax system does not recognise when taxing profits of companies falling within its tax jurisdiction. Indeed, the Decree is meant precisely to ensure that group and stand-alone companies are treated in a similar manner under the general Dutch corporate income tax system and taxed on profits that derive from their activities, whether those activities are carried out in an intra-group context or not.

(249) The Commission considers that the purpose of the Decree is and cannot be to establish special rules for related companies, but to clarify the application of the arm's-length principle under Article 8b(1) CIT in light of the OECD TP Guidelines given that the purpose of the Decree is, as acknowledged by the Netherlands, ‘specifically aimed at aspects which the OECD TP Guidelines leaves open to interpretation or where there is a lack of clarity’. The rules laid down in the Decree are therefore meant to align the tax treatment of related companies with the treatment of unrelated companies to the extent that transactions between related parties should be priced at arm's length and therefore mirror the situation of unrelated parties for the purposes of levying corporate income tax on their profits.

(250) In any event, the Commission observes that if the reasoning of the Netherlands and Starbucks were accepted that the Decree does establish special rules for integrated companies, the existence of those special rules could, in itself, lead to a finding of selectivity. In fiscal cases, selectivity exists when a Member State exempts a (certain category of) undertaking(s) from a general rule that applies to all undertakings in a comparable factual and legal situation. It also exists when a special regime is set up that deviates from that general rule for the benefit of certain but not all undertakings in a comparable factual and legal situation. Thus, considering the Commission concludes that integrated and non-integrated companies are in a comparable factual and legal situation as regards the imposition of Dutch corporate income tax to profits, the creation of a special regime that applies only to integrated companies, which deviates from the general Dutch corporate income tax rules, is in itself selective in nature, so that any benefit granted on the basis of that regime is selective in nature.

(251) The Commission therefore concludes that, in the present case, the reference system against which the SMBV APA should be examined is the general Dutch corporate income tax system, irrespective of whether corporate income tax under that system is imposed on group or stand-alone companies.

9.2.2.   

Selective advantage due to a derogation from the general Dutch corporate income tax system

(252) Having determined that the general Dutch corporate income tax system constitutes the reference system against which the SMBV APA should be assessed, it is necessary to establish whether that APA constitutes a derogation from that reference system, leading to unequal treatment between companies that are factually and legally in a similar situation.

(253) In relation to that second step of the selectivity analysis, whether a tax measure constitutes a derogation from the reference system will generally coincide with the identification of the advantage granted to the beneficiary under that measure. Indeed, where a tax measure results in an unjustified reduction of the tax liability of a beneficiary who would otherwise be subject to a higher level of tax under the reference system, that reduction constitutes both the advantage granted by the tax measure and the derogation from the system of reference.

(254) According to the Court, in the case of an individual aid measure, as opposed to a scheme, ‘the identification of the economic advantage is, in principle, sufficient to support the presumption that it is selective’(118). In the present case, the individual aid measure from which SMBV benefits is the SMBV APA, which endorses a methodology for determining its taxable profit in the Netherlands for the functions it performs within the Starbucks group, which are subsequently taxed under the general Dutch corporate income tax system.

9.2.3.   

Selective advantage resulting from a deviation from the arm's-length principle

(255) In principle, the function of an APA is to establish in advance the application of the ordinary tax system to a particular case, given a set of facts and circumstances specific to that case, for a certain period of time and provided that there is no material change over the application of the APA in that specific set of facts and circumstances. Where an APA is based on a method of assessment that deviates from what would result from a normal application of the ordinary tax system without justification, that APA will be considered to confer a selective advantage upon its beneficiary in so far as that selective treatment results in the lowering of that beneficiary's tax liability in the Member State concerned as compared to companies in a similar legal and factual situation.

(256) An advantage pursuant to Article 107(1) of the Treaty is any economic benefit that an undertaking would not have obtained under normal market conditions, i.e. in the absence of the State intervention(119). Thus, whenever the financial situation of an undertaking is improved as a result of a State intervention, an advantage is present. Such improvement is shown by comparing the financial situation of the undertaking as a result of the contested measure with the financial situation of that undertaking had the measure not been granted(120). An advantage can consist both in the granting of positive economic advantages as well as in the mitigation of charges normally included in the budget of an undertaking(121).

(257) As explained in recital 42 et seq., by concluding the SMBV APA, the Netherlands accepted a methodology for determining SMBV's taxable profit in the Netherlands, as proposed by the Starbucks' tax advisor in the transfer pricing report, which allows SMBV to determine its corporate income tax liability in the Netherlands on a yearly basis for the duration of which that APA is valid. More specifically, the transfer pricing report endorsed by the SMBV APA determines, in the absence of transactions dictated by the market as would exist for a non-integrated independent company, the profit to be allocated to that company of the Starbucks group resulting from the transactions it concludes with the other group companies of the Starbucks group.

(258) The Court of Justice has already held that a reduction in the taxable base that results from a tax measure that enables a taxpayer to employ transfer prices in intra-group transactions that do not resemble prices which would be charged in conditions of free competition between independent undertakings negotiating under comparable circumstances at arm's length confers a selective advantage on that taxpayer, by virtue of the fact that its tax liability under the ordinary tax system is reduced as compared to independent companies which rely on their accounting profits as a basis to determine their taxable base(122).

(259) In its judgment on the Belgian tax regime for coordination centres(123), the Court of Justice assessed a challenge to a Commission decision which concluded, inter alia, that the method for determining taxable income under that regime conferred a selective advantage on those centres(124). Under that regime, taxable profits were set at a flat-rate amount which represented a percentage of the full amount of operating costs and expenses, from which staff costs and financial charges were excluded. According to the Court, ‘in order to decide whether a method of assessment of taxable income such as that laid down under the regime for coordination centres confers an advantage on them, it is necessary, […], to compare that regime with the ordinary tax system, based on the difference between profits and outgoings of an undertaking carrying on its activities in conditions of free competition.’ The Court then held that ‘the effect of the exclusion of [staff costs and the financial costs] from the expenditure which serves to determine the taxable income of the centres is that the transfer prices do not resemble those which would be charged in conditions of free competition’, which the Court found to ‘[confer] an advantage on the coordination centres’(125).

(260) The Court has thus accepted that a tax measure which results in a group company charging transfer prices that do not reflect those which would be charged in conditions of free competition, that is prices negotiated by independent undertakings negotiating under comparable circumstances at arm's length, confers an advantage on that group company in so far as it results in a reduction of its taxable base and thus its tax liability under the ordinary corporate income tax system.

(261) The principle that transactions between intra-group companies should be remunerated as if they were agreed to by independent companies negotiating under comparable circumstances at arm's length is generally referred to as the ‘arm's-length principle’. In the Belgian coordination centres judgment, the Court of Justice endorsed the arm's-length principle as the benchmark for establishing whether a group company receives an advantage for the purposes of Article 107(1) of the Treaty as a result of a tax measure that determines its transfer pricing and thus its taxable base.

(262) The purpose of the arm's-length principle is to ensure that transactions between group companies are treated for tax purposes by reference to the amount of profit that would have arisen if the same transactions had been executed by independent companies. Otherwise, group companies would benefit from a favourable treatment under the ordinary corporate income tax system when it comes to the determination of their taxable profits that is not available to independent companies, leading to unequal treatment between companies that are factually and legally in a similar situation in light of the objective of such a system, which is to tax the profits of all companies falling under its tax jurisdiction.

(263) The Commission's assessment of whether the Netherlands granted a selective advantage to SMBV must therefore consist in verifying whether the methodology accepted by the Dutch tax administration by concluding the APA for the determination of SMBV's taxable profits in the Netherlands departs from a methodology that results in a reliable approximation of a market-based outcome and thus from the arm's-length principle. In so far as the methodology the Netherlands accepted by the SMBV APA results in a lowering of SMBV's tax liability under the general Dutch corporate income tax system as compared to non-integrated companies whose taxable profit under that system is determined by the market, that APA will be deemed to confer a selective advantage to SMBV for the purposes of Article 107(1) of the Treaty.

(264) The arm's-length principle therefore necessarily forms part of the Commission's assessment under Article 107(1) of the Treaty of tax measures granted to group companies independently of whether a Member State has incorporated this principle into its national legal system. It is used to establish whether the taxable profits of a group company for corporate income tax purposes has been determined on the basis of a methodology that approximates market conditions, so that that company is not treated favourably under the general corporate income tax system as compared to non-integrated companies whose taxable profit is determined by the market. Thus, for any avoidance of doubt, the arm's-length principle that the Commission applies in its State aid assessment is not that derived from Article 9 of the OECD Model Tax Convention, which is a non-binding instrument, but is a general principle of equal treatment in taxation falling within the application of Article 107(1) of the Treaty, which binds the Member States and from whose scope the national tax rules are not excluded(126).

(265) Consequently, in response to the Netherlands' argument that the Commission, in undertaking such an assessment, replaces the national tax administration in its interpretation of their national law(127), the Commission recalls that is not examining whether the SMBV APA complies with the arm's-length principle as laid down in Article 8b(1) of the CIT or the Decree, but whether the Dutch tax administration conferred a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty by concluding an APA that endorses a profit allocation that departs from the amount of profit that would have been taxed under the general Dutch corporate income tax system if the same transactions had been executed by independent companies negotiating under comparable circumstances at arm's length.

(266) In response to the argument invoked by the Netherlands and Starbucks that because transfer pricing is not an exact science, the assessment by the Commission of the transfer pricing arrangement agreed in the SMBV APA should necessarily be limited(128), the Commission recalls that the approximation component of transfer pricing has to be viewed in the light of its objective. While the 2010 OECD TP Guidelines do indeed acknowledge that transfer pricing is not an exact science in paragraph 1.13 thereof, that same paragraph first explains that ‘[i]t is important not to lose sight of the objective to find a reasonable estimate of an arm's-length outcome based on reliable information’. The objective of the OECD TP Guidelines is to develop, for the benefit of tax administrations and multinational enterprises, the most appropriate methods for estimating arm's-length prices of cross-border transactions between associated enterprises for taxation purposes. The pursuit of that objective would be impossible if the approximative nature of the transfer pricing exercise could be used to disregard the consensus on appropriate transfer pricing methodologies which those guidelines represent. The approximative nature of the arm's-length principle can therefore not be invoked to justify a transfer pricing analysis that is either methodologically inconsistent or based on an inadequate comparables selection.

(267) In conclusion, if it can be shown that the methodology accepted by the Dutch tax administration, by concluding the SMBV APA, for the determination of SMBV's taxable profits in the Netherlands departs from a methodology that leads to a reliable approximation of a market-based outcome and thus from the arm's-length principle, that APA will be found to confer a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty in so far as it leads to a lowering of SMBV's tax liability under the general Dutch corporate income tax system as compared to non-integrated companies whose tax base is determined by the profits they generate under market conditions.

9.2.3.1.   

Methodological choices, parameters and adjustments underlying the SMBV APA

(268) The SMBV APA accepts a methodology for determining a profit allocation to SMBV within the Starbucks group, which is based on a transfer pricing report prepared by Starbucks' tax advisor that calculates a remuneration for a function performed by SMBV (roasting/manufacturing).

(269) In calculating the remuneration due to SMBV, the tax advisor proceeds to successive methodological choices in the transfer pricing report:

(i) the choice to use the TNMM to estimate a taxable profit;

(ii) the choice of operating expenses as profit level indicator for use in the application of the TNMM(129); and

(iii) the application of a working capital adjustment to address differences between SMBV and the comparables used to estimate an arm's-length mark-up(130).

(270) The transfer pricing report concludes on a remuneration for the roasting function SMBV performs equal to a mark-up of [9-12] % of its operating expense, which is accepted by the Dutch tax administration as constituting an arm's-length remuneration in the SMBV APA. The Dutch tax administration further accepts in the SMBV APA that any profit generated by SMBV in excess of that level of remuneration will be paid out as a royalty to Alki LP.

(271) In the following sections, the Commission will explain why it considers that several of the methodological choices underlying the transfer pricing report should not have been accepted by the Dutch tax administration in the SMBV APA, because their acceptance results in a taxable profit for SMBV that cannot be regarded to constitute a reliable approximation of a market-based outcome and results in a reduction of SMBV's tax liability as compared to non-integrated companies whose taxable profits is determined by the market.

(272) First and foremost, the transfer pricing report fails to identify or analyse SMBV's controlled and uncontrolled transactions(131), which is a necessary first step in assessing the arm's-length nature of commercial conditions applicable between related parties for transfer pricing purposes. More specifically, the transfer pricing report fails to examine whether the royalty payment made by SMBV to Alki LP for the licensing of the roasting IP, which is the intra-group transaction for which the SMBV APA was effectively requested and granted(132), is at arm's length. As the Commission will demonstrate, a transfer pricing analysis of the arm's-length value of that royalty based on comparable uncontrolled transactions leads to the conclusion that no royalty should be due for the roasting IP licensed by Alki LP to SMBV(133). The transfer pricing report also fails to examine whether the price charged for the green coffee beans by SCTC to SMBV is at arm's length. As the Commission will demonstrate, no market-based justification exists for the substantial increase in that price from 2011 onwards, which resulted in a reduction of SMBV's accounting profits from that year onwards(134).

(273) In addition, and without prejudice to the preceding recital, the transfer pricing report fails to analyse the complexity of the functions of all group companies involved in controlled transactions, in particular Alki LP, when accepting that SMBV is to be regarded as the ‘least complex function’ and, thus, the ‘tested party’ for the purpose of applying the TNMM(135).

(274) Finally, and without prejudice to the preceding two recitals, the transfer pricing report misidentifies SMBV's main functions to be remunerated and inappropriately seeks to estimate that remuneration on the basis of operating expenses(136).

9.2.3.2.   

The transfer pricing report fails to examine the intra-group transaction for which the SMBV APA was effectively requested and granted

(275) The SMBV APA agrees to a taxable remuneration for SMBV for its roasting function. It also agrees that any profits generated by SMBV in excess of that level of remuneration will be paid out as a royalty to Alki LP which is not taxed in the Netherlands.

(276) In other words, by concluding the SMBV APA, the Dutch tax administration expressly accepts that the tax advisor's methodology for calculating the taxable remuneration due to SMBV for its roasting function directly determines the level of the royalty paid by SMBV to Alki LP for the roasting IP licensing arrangement between them. It also expressly accepts that the actual level of profits generated by SMBV in the Netherlands is to be reduced for tax purposes by that royalty payment, resulting in a lower taxable profit than that actually recorded. Indeed, if SMBV's accounting profits are higher than the level of remuneration agreed to in the SMBV APA, the royalty payment to Alki LP will be increased by the difference between that level of remuneration and SMBV's accounting profits.

(277) Thus, the royalty is an adjustment variable determined by combining SMBV's accounting profits and the remuneration agreed in the SMBV APA. As such, the roasting IP licensing arrangement between Alki LP and SMBV is the transaction for which the SMBV APA was effectively requested and the methodology for determining the level of that royalty as an adjustment variable is the transaction effectively being priced by the SMBV APA.

(278) Nevertheless, the transfer pricing report upon which that APA is based only proposes a remuneration for SMBV by analysing a function performed by it (roasting/manufacturing) through the application of the TNMM; it does not identify the royalty payment as the adjustment variable in constructing that proposed remuneration. Consequently, that report fails to identify or analyse the roasting IP licensing arrangement for which that royalty is paid as the transaction effectively being priced and therefore fails to establish a methodology for ensuring that that royalty payment is in line with the arm's-length principle.

(279) The purpose of a transfer pricing exercise, however, is to establish whether the conditions of controlled transactions are consistent with the arm's-length principle.

(280) That focus on transactions for the purposes of transfer pricing is clearly expressed in paragraph 1.6 of the 2010 OECD TP Guidelines, which clarifies from the outset that ‘because the separate entity approach treats the members of an MNE group as if they were independent entities, attention is focused on the nature of the transactions between those members and on whether the conditions thereof differ from the conditions that would be obtained in comparable uncontrolled transactions. Such an analysis of the controlled and uncontrolled transactions, which is referred to as a “comparability analysis”, is at the heart of the application of the arm's-length principle’(137).

(281) In other words, commercial transactions between related and unrelated parties must first be clearly identified before being assessed for comparability. That focus on transactions is equally supported under Dutch tax law by the requirement not to allow for deductions of expenses which are not in line with the arm's-length principle(138).

(282) It is only if it is impossible to compare a particular intra-group transaction for which an APA is being sought (taking into account the functions performed) to similar uncontrolled transactions, that resorting to a comparison of functions performed is justified. In this vein, paragraph 1.41 of the 2010 OECD TP Guidelines explains that ‘before broadening the search to include a larger number of potentially comparable uncontrolled transactions based on similar functions being undertaken, thought should be given to whether such transactions are likely to offer reliable comparables for the controlled transaction’(139). This also explains the preference for the CUP method over all other transfer pricing methods expressed both in paragraph 2.14 of the 2010 OECD TP Guidelines and in paragraph 2.7 of the 1995 OECD TP Guidelines, both of which state: ‘Where it is possible to locate comparable uncontrolled transactions, the CUP method is the most direct and reliable way to apply the arm's-length principle. Consequently, in such cases the CUP method is preferable over all other methods’.

(283) Similarly, both the 1995 and 2010 OECD TP Guidelines favour traditional transactional methods over transactional profit methods as a means of establishing whether a transfer price is at arm's length(140). The 1995 OECD TP Guidelines even exclude automatically resorting to transactional profit methods such as the TNMM for transfer pricing purposes, stating in paragraph 3.50 thereof: ‘There are, however, cases where traditional transaction methods cannot be reliably applied alone or exceptionally cannot be applied at all. These would be considered cases of last resort. (…) However, even in a case of last resort, it would be inappropriate to automatically apply a transactional profit methods without first considering the reliability of that method’.

(284) The Netherlands' observation that its tax administration is not bound by a best method rule does not relieve that administration of the obligation to confirm that the transfer pricing method the taxpayer selects results in a reliable approximation of an arm's-length price before accepting an APA request based on that method. That obligation is even reflected in the provision of the Decree to which the Netherlands refers in this regard. Indeed, paragraph 3.1 of the Decree provides ‘[t]he Dutch Tax and Customs Administration should always begin its investigation into the transfer prices from the perspective of the method used by the taxpayer at the time of the transaction. This means that the taxpayer is, in principle, free to choose a transfer pricing method, provided the selected method results in an arm's-length result for the specific transaction. The taxpayer must demonstrate its choice’. In other words, the method proposed by the taxpayer should be the starting point of the tax administration's examination of the APA request. However, whatever method chosen must ensure an arm's-length result for the specific transaction being priced, so that the tax administration remains free to question the appropriateness of the method chosen by the taxpayer. Finally, given that that provision requires the appropriateness of the transfer pricing method chosen to be reasoned by the taxpayer and given the Decree's own express preference for the CUP method where comparable transactions are available(141), the absence of a best method rule does not relieve the tax administration from ensuring, before it agrees to the APA request, that the transfer pricing method selected by the taxpayer is able to provide for a reliable approximation of a market-based outcome in line with the arm's-length principle.

(285) As will be shown in the following section, since the transfer pricing report failed to provide any information on uncontrolled transactions similar to the roasting IP licensing arrangement between SMBV and Alki LP and, thus, since it failed to examine the only intra-group transaction that was effectively being priced by the transfer pricing analysis, the transfer pricing methodology proposed by Starbucks' tax advisor and accepted by the SMBV APA cannot be considered to result in a reliable approximation of a market-based outcome in line with the arm's-length principle. Indeed, since the transfer pricing report engages in an analysis of an arm's-length remuneration for SMBV based on an incorrect point of departure (its roasting function), it results in a remuneration which is improperly estimated on the basis of the TNMM. Rather, recourse should have been had to more reliable comparisons to available information on similar transactions between unrelated parties for transfer pricing purposes, which was in the possession of Starbucks at the time that the APA request was made and should have been requested by the Dutch tax administration to ensure the royalty payment being priced by the SMBV APA was at arm's length.

9.2.3.3.   

The royalty payment to Alki LP resulting from the SMBV APA is not priced at arm's length

(286) The IP for which the royalty payment is made covers the roasting know-how and the roasting curves licensed by Alki LP to SMBV. That royalty payment does not relate to the value of the Starbucks' brand, since the right to use that brand is paid for by the Shops to Starbucks Coffee BV.

(287) In the Opening Decision, the Commission expressed doubts on the arm's-length nature of the royalty payment(142). In particular, the Commission explained that the fact that the level of the royalties due by SMBV to Alki LP is dependent on the difference between the remuneration established in the SMBV APA and the accounting pre-tax profit before the payment of the royalty(143) leads to a situation in which that royalty payment is calculated as profit in excess of the SMBV APA and does not reflect the arm's-length value of that IP(144). The Commission referred, in that respect, to paragraph 6.16 of the 2010 OECD TP Guidelines, according to which, ‘a royalty would ordinarily be a recurrent payment based on the user's output, sales, or in some rare circumstances, profits’(145). In the SMBV APA, the royalty payment to Alki LP is not related to the output, sales, or profits of SMBV. The Commission further indicated, in recital 120 of its Opening Decision, its doubts on the arm's-length level of the royalty payment, considering the royalty is disconnected through the pricing method from the economic value of any underlying IP, the value to SMBV of the roasting intangibles.

(288) To compare the value of the royalty payment to royalty levels observed on the market set in terms of percentage of turnover, the Commission calculated the amount of the royalty paid by SMBV to Alki LP as a percentage of the sales of roasted coffee by SMBV to the Shops each year. According to those calculations, the royalty payment varies between [1-10] % to [30-40] % of SMBV's revenues from sales of coffee over the life of the SMBV APA, confirming the Commission's doubts about the fluctuation of the royalty(146). Moreover, for three of those years the resulting levels were above [30-40] %, as presented in Table 10:

Table 10

Fluctuation of the royalty payment over the life of the SMBV APA

(289) In this specific context, the variable nature of the royalty payment gives a first indication that the level of that payment bears no relation to the value of the IP for which it is being paid.

(290) For the reasons explained in recitals 291 to 338, the Commission considers that a comparison to comparable uncontrolled transactions using the CUP method, in particular the IP licensing arrangements in several roasting and manufacturing and distribution agreements Starbucks has concluded with third parties, demonstrates that the arm's-length value of the royalty paid by SMBV to Alki LP for the roasting IP should be zero. In other words, no royalty should be due for that IP in that specific relationship, since SMBV does not derive any benefit from the use of the roasting IP licensed from Alki LP.

(a)   Comparison to roasting agreements concluded by Starbucks with third parties and comparison to similar arrangements on the market

(291) Certain roasting and manufacturing contracts concluded by Starbucks group companies with third parties, listed in recitals 148 to 150, were provided by Starbucks to the Commission during the investigation. For the reasons explained in recitals 292 to 298, the Commission considers those transactions to constitute a direct comparable for determining the level of the royalty payment due by SMBV to Alki LP under the roasting IP licensing arrangement.

(292) Paragraph 1.38 of the 2010 OECD TP Guidelines provides the following guidance on the comparability examination ‘the examination of the […] comparability factors is by nature two-fold i.e. it includes an examination of the factors affecting the taxpayer's controlled transactions and an examination of the factors affecting uncontrolled transactions. Both the nature of the controlled transaction and the transfer pricing method adopted […] should be taken into account when evaluating the relative importance of any missing piece of information on possible comparables […]’.

(293) Paragraph 1.36 of the 2010 OECD TP Guidelines lists five comparability factors(147) which ‘include the characteristics of the property or services transferred, the functions performed by the parties (taking into account assets used and risks assumed), the contractual terms, the economic circumstances of the parties, and the business strategies pursued by the parties.’

(294) The Commission observes, first, that the characteristics of the property transferred under the roasting IP arrangement between Alki LP and SMBV is identical to the property transferred in the transactions between Starbucks and the third parties in the roasting agreements listed in recitals 148 to 150. Both sets of transactions consist of roasting technology, coffee blends and roasting curves.

(295) Second, while not all of those third parties roast coffee (some are engaged in the production of ready-to-drink beverages or other products and ingredients for drink preparation), in those transactions where the third parties do roast coffee, the function of the third party relates to the exact same product as SMBV's roasting function in its contractual relationship with Alki LP. In particular, Starbucks contractual arrangements with [unaffiliated manufacturing company 2], [unaffiliated manufacturing company 3], [unaffiliated manufacturing company 4] and [unaffiliated manufacturing company 10] all pertain to the roasting of green coffee.

(296) Third, in none of those transactions was the role of Starbucks more limited than the role of Alki LP. In some of those transactions, Starbucks assumed more functions vis-à-vis the third parties than Alki LP vis-à-vis SMBV. In particular, most contractual arrangements were associated with the sourcing of the coffee by Starbucks(148). Moreover, in some of those arrangements, Starbucks also buys the roasted coffee from the third party. Therefore, the remuneration of Starbucks in those arrangements presents a maximum arm's-length remuneration for the licensing arrangement, if any remuneration were in fact due to Alki LP.

(297) Fourth, there is no indication that the economic circumstances of the third parties affect their arrangement with Starbucks. In particular, as Starbucks provided many roasting IP licensing arrangements of which none could be identified as containing a royalty paid to Starbucks for the roasting IP, this cannot be related to the specific economic circumstances of an individual third party. The arrangements cover different geographic regions, including the Union and Switzerland, which was considered as a relevant geographic delimitation for the comparables search in the transfer pricing report.

(298) Fifth, the business strategies of the third parties are discussed below. The arrangements differ principally depending on whether the third party exploits the IP directly on the market by selling products to end customers or not.

(299) Accordingly, with those uncontrolled transactions, the level of an arm's-length royalty payment between SMBV and Alki LP can be determined using the CUP method, that is, comparing the payment due in a controlled transaction (from SMBV to Alki LP) to the payment due in comparable uncontrolled transactions (from third parties to other Starbucks group companies), conducted under comparable circumstances.

(300) In this regard, the Commission notes that under similar agreements concluded by Starbucks with [unaffiliated manufacturing company 2], [unaffiliated manufacturing company 3], [unaffiliated manufacturing company 4], [unaffiliated manufacturing company 9], [unaffiliated manufacturing company 8], [unaffiliated manufacturing company 1] and [unaffiliated manufacturing company 10], third parties do not pay a royalty under their licensing arrangements with Starbucks if they do not exploit the roasting IP directly on the market.

(301) Indeed, [unaffiliated manufacturing company 3] only pays a royalty to Starbucks when it sells its production to the [unaffiliated manufacturing company 3- Starbucks joint-venture]. In that case, [unaffiliated manufacturing company 3] directly exploits the roasting IP on the market through a related party, so that the royalty payment appears to cover the distribution of Starbucks' branded products to third parties by the joint venture. This conclusion is confirmed by the fact that when [unaffiliated manufacturing company 3] resells the roasted coffee to the Starbucks group, rather than to the joint venture, and the distribution and exploitation on the market of the brand is ensured by the Starbucks group, no royalty is paid by [unaffiliated manufacturing company 3] to Starbucks for the roasting IP.

(302) As regards [unaffiliated manufacturing company 2], while Starbucks claims that the higher mark-up on the green coffee beans purchased for Starbucks in the contract with [unaffiliated manufacturing company 2] represents a remuneration for roasting IP, this mark-up appears to be passed on to [unaffiliated manufacturing company 5]. Indeed, the price at which [unaffiliated manufacturing company 5] buys coffee from [unaffiliated manufacturing company 2] is also defined as a mark-up to the cost of the acquired green coffee beans. In its relationship with [unaffiliated manufacturing company 2], which would apparently be remunerated through a higher mark-up, Starbucks acts as supplier, which is a different function than the function assumed by Alki LP in its relationship with SMBV. In response to Starbucks' argument that because [unaffiliated manufacturing company 2] is ready to pay a top-up on the price charged for beans by SCTC and that, therefore, SCTC's prices would be arm's length, the Commission considers that the purchase price for green coffee beans cannot be analysed in isolation of [unaffiliated manufacturing company 2]'s obligation under its contracts with Starbucks to sell its production to [unaffiliated manufacturing company 5] and that, therefore, the pricing arrangements between [unaffiliated manufacturing company 5] and [unaffiliated manufacturing company 2] must also be taken into consideration. There are no indications that any mark-up to a purchase price would not be passed on directly to [unaffiliated manufacturing company 5] or otherwise affect the commercial conditions between [unaffiliated manufacturing company 5] and [unaffiliated manufacturing company 2], as this contractual arrangement was not concluded independently of the contractual arrangement between Starbucks and [unaffiliated manufacturing company 5].

(303) In addition, the Commission observes that in its relationship with [unaffiliated manufacturing company 5], [unaffiliated manufacturing company 6] and [unaffiliated manufacturing company 7], only the trademark and technology licence agreements concluded by Starbucks with those third parties contain a royalty payment. That royalty payment, however, is comparable to the royalty paid by the Shops to Starbucks Coffee BV for the exploitation of the Starbucks' brand IP on the market, since all three companies sell products to final consumers. The master and manufacturing and distribution agreements between the three companies and Starbucks, which concern the manufacturing process, do not provide for a royalty for the Starbucks IP.

(304) Furthermore, the Commission notes that SMBV outsources the production of [a trademark registered coffee product] and soluble coffee to third parties, in particular, [unaffiliated manufacturing company 1]. Although Starbucks' [trademark registered coffee product] product is a brand-protected product, [unaffiliated manufacturing company 1] does not pay any royalties to SMBV or any other Starbucks company for the blend and other production information. Similarly, although the production of the technology used in the production of the VIA soluble coffee products is advertised as innovative by Starbucks, its production is outsourced to a third party and SMBV does not perceive any royalties from the third party producing VIA. Indeed, the accounts of SMBV evidence that no royalty is perceived by the company for any of its outsourced production(149).

(305) Finally, for the purposes of assessing whether SMBV pays an arm's-length remuneration to Alki LP for the roasting IP, the arrangement between Alki LP with SMBV can also be compared to arrangements between Starbucks' competitors with third party roasters.

(306) For example, in response to the MIT request regarding arm's-length commercial conditions between unrelated parties in which a company is entrusted with the function of roasting green coffee beans, Melitta explained that when outsourcing the roasting of coffee to a third party it does not perceive royalties, although its puts its roasting curves at the disposal of the third party(150).

(307) According to the submission of Company Y, which performs a roasting activity through a group company designated as a toll manufacturer, its in-house roasting company also does not pay any royalty to the group either for the IP or the know-how used in the roasting process(151).

(308) The same is also true of Dallmayr, which indicated that it considers the payment of a royalty by a company providing roasting as unusual, since it would rather expect the customers to pay the roaster, not the other way around(152).

(309) It follows that, since in the manufacturing agreements Starbucks concluded with third parties, several of which existed at the time the SMBV APA request was being considered by the Dutch tax administration, no royalty is required for the use of the roasting IP, the Commission considers that a transfer pricing analysis of the arm's-length value of the royalty paid by SMBV to Alki LP for roasting IP leads to the conclusion that no royalty should be due for that IP in that specific relationship. This conclusion also follows from a comparison with arrangements between Starbucks' competitors with third party roasters.

(b)   SMBV does not capture the value of the roasting IP in its relationship with Alki LP

(310) Whereas roasting know-how and curves can have a value, in the specific relationship between Alki LP and SMBV that value is not captured by the roaster. That is because the importance of the roasting know-how and curves lies, in the case of the Starbucks structure, in ensuring a consistent taste associated with the brand and individual products. Accordingly, the value of Starbucks' roasting know-how and curves is only exploited when Starbucks products are sold under the Starbucks brand by the Shops. On their own, the roasting know-how and curves do not generate value for the roaster on an on-going basis if they cannot be exploited on the market.

(311) That conclusion is supported by the submission of Starbucks, according to which licence agreements in which the IP licence merely serves to enable third parties to produce in accordance with Starbucks' specification and to protect Starbucks' IP rights, but where third parties do not exploit the intangibles on the market, are not associated with any royalty payments by those third parties to Starbucks(153).

(312) In addition, in the case of SMBV, the roasting know-how and curves appear to constitute a technical specification according to which the roasting should proceed due to a preference or a choice of the purchasing company. Roasting curves are described by Starbucks in the transfer pricing report as being dictated to SMBV. They allow SMBV and the third parties, with whom roasting agreements and manufacturing and supply agreements have been concluded, to meet the requirements of Starbucks. Roasting preferences are imposed on manufacturers, for example through the roasting standards requirements and quality assurance standards. Such specifications are part of each roasting or manufacturing and supply agreement.

(313) In this regard, the Roasting Agreement indicates that SMBV must adopt roasting processes provided by Alki LP and employ proper equipment, machinery and production methods to ensure that the products meet the product specifications prescribed by Alki LP(154). SMBV does not seem to derive any benefit from the use of the product specification in that relationship. The fact that the specifications laid down by Alki LP regarding the roasting process and, in particular, the roasting curves allow SMBV to roast coffee that is sold under the Starbucks brand does not bring any benefit to SMBV in terms of increased sales or sales price, considering that SMBV does not, in principle, sell its production to final customers which value the Starbucks brand. SMBV sells virtually all its production to Starbucks franchised Shops, which all pay a royalty to the group for exploiting the Starbucks IP on the market, representing a value for their distribution business. SMBV does not exploit the roasting IP directly on the market; the Shops are the interface with final customers who value the consistent taste associated with the Starbucks brand.

(314) Finally, based on the financial information provided in Table 2 and Table 8, Figure 3 presents SMBV's profit margin on its coffee roasting activities, which is obtained by subtracting the price paid by SMBV to SCTC for green coffee beans from the revenues of roasted coffee recorded under the description ‘REV PACKAGED COFFEE’ for each year(155).

Figure 3

Losses generated from SMBV's roasting activities since 2010

(revenues and purchase price of coffee in EUR)

[Bild bitte in Originalquelle ansehen]

(315) Figure 3 demonstrates that, since 2010, SMBV is loss-making on its roasting activities, when the margin on green coffee beans required by SCTC increased. Taking that fact into account, along with the data in Table 10 that show the percentage of royalty paid by SMBV to Alki LP over the revenue from sales of coffee, the roasting know-how and curves do not appear to create positive value for SMBV. For example, in 2013, although SMBV seems to have recorded a gross loss (before operating expense is deducted) of around EUR [1-10] million on its roasting activities, it paid a royalty of EUR 22,8 million to Alki LP for the roasting IP.

(316) In other words, the royalty payment from SMBV to Alki LP is financed through the other activities of SMBV(156). While it is conceivable that an economic operator would bear losses to finance a royalty that could lead to future increased profits, the loss-bearing nature of the activity is related to the increase in the price of green coffee beans by SCTC and does not seem to offer any prospects of future profits(157). Indeed, SMBV's business strategy appears stable over the lifetime of the APA and there is no concrete actions that SMBV could undertake to return to profitability, in absence of any decisive influence on the sales. Since operating costs represent a small portion of SMBV's total costs, cost cutting would not be sufficient to return to profitability because it would not make up for the financial impact of the increased mark-up on the green coffee beans. The royalty paid by SMBV to Alki LP for the roasting IP in an intra-group context appears to serve structurally the sole purpose of shifting profits derived from SMBV's reselling function to Alki LP.

(317) The fact that the roasting activity does not generate sufficient profit to allow for royalty payments therefore further confirms, given the specific intra-group relationship between SMBV and Alki LP, that the methodology for determining the level of that royalty as an adjustment variable as accepted by the SMBV APA is not in line with the arm's-length principle.

(318) Accordingly, in light of the comparison with Starbucks' roasting arrangements with third parties, the Commission considers that a transfer pricing analysis of the arm's-length value of the royalty paid by SMBV to Alki LP for the roasting IP leads to the conclusion that no royalty should be due for that IP, since SMBV does not seem to derive any benefit from the use of the roasting IP in that relationship. None of the arguments advanced by the Netherlands or Starbucks during the investigation invalidate that conclusion.

(c)   The royalty payment does not reflect a remuneration for taking over entrepreneurial risks

(319) In response to the Commission's doubts on the arm's-length nature of the royalty payment, the Netherlands and Starbucks claimed that that payment does not only reflect a remuneration for the use of the roasting IP, but also a payment for the taking over of entrepreneurial risk by Alki LP(158). The Netherlands invokes the fact that SMBV would not carry the economic risk of loss of inventory, since, given the way the royalty payment is determined under the Roasting Agreement, those costs are ultimately borne by Alki LP(159).

(320) The Commission notes, first, that this claim is not supported by the transfer pricing report supporting the SMBV APA request. Rather, that report clearly contradicts that claim, stating that ‘[SMBV] licenses a sub-set of IP from Alki LP that is necessary to utilise the coffee roasting manufacturing process and the right to supply coffee to [D]evelopers. In return [SMBV] remits a royalty to Alki LP for the licensed IP’(160). More important, nowhere in the transfer pricing report is any mention made of the fact that the royalty payment would partly constitute a payment by SMBV to Alki LP for the taking over of entrepreneurial risks. In fact, SMBV retains more risks than presented in the transfer pricing, in particular, inventory risks and certain risks related to the supply of beans and the manufacturing capacity utilisation.

(321) The Commission observes, second, that the Netherlands appears to consider that the way in which the contractual arrangement under the Roasting Agreement is structured(161) gives legal substance to the actual allocation of risk and responsibilities between Alki LP and SMBV. However, according to SMBV's accounts, the royalty payment is generated by the SMBV APA and not by the contractual arrangement between Alki LP and SMBV.

(322) Third, and most important, the Commission considers that if the Netherlands' claim were accepted, the business risk of any group company could be eliminated through intra-group reallocation of risks by simple means of contract. For example, a company undertaking all strategic decisions for a group, in particular, investment and R & D decisions, which would in principle be considered as a complex function concentrating entrepreneurial risk, could be considered as ‘low risk’ and taxed accordingly, so long as an intragroup contract would be put in place setting its remuneration at any random level of operating expense and passing its residual profits to any other group company. Accepting that claim would render the application of the arm's-length principle for the pricing of intragroup transactions meaningless, since contractual arrangements would be considered to trump economic reality.

(323) The Commission recalls(162), in that respect, paragraphs 9.44 to 9.46 of the 2010 OECD TP Guidelines(163) which clarify in the context of a business restructuring that ‘it is the low (or high) risk nature of a business that will dictate the selection of the most appropriate transfer pricing method, and not the contrary’(164). Contrary to the views expressed by the Netherlands(165), this idea that the application of the transfer pricing rules should follow risks and not contractual arrangements when those arrangements do not reflect the underlying business risk, is also valid outside the context of a business restructuring. While it is true that agreeing on a transfer pricing method (i.e. such as the TNMM method based on operating costs) and constructing contractual arrangements such as to adjust the pre-tax profit each year to that arrangement results in limiting the fluctuation of the tax liability of the company over fiscal years, such a structure constructed through contractual arrangements does not, however, necessarily correspond to the economic reality underlying the transactions and the risks associated with the activity of the company. The 1995 OECD TP Guidelines present this consideration in paragraph 1.26, according to which, ‘in relation to contractual terms, it may be considered whether a purported allocation of risk is consistent with the economic substance of the transaction. In this regard, the parties’ conduct should generally be taken as the best evidence concerning the true allocation of risk'.

(324) Fourth, as regards the Netherlands' and Starbucks' further claim that it is the low risk nature of SMBV's business that in fact dictates the transfer pricing method accepted by the SMBV APA, the Commission observes that the information submitted during the investigation does not support that claim. In recital 113 of the Opening Decision, the Commission expressed doubts on the capacity of Alki LP to bear and control any business risk. When a company assumes a risk, it should be able, on the one hand, to control the risks(166) and, on the other hand, to financially assume such a risk(167). The capacity to assume a risk should therefore be analysed before the proposed structure is considered in line with normal competitive conditions. The 2010 OECD TP Guidelines clarify that control should be understood, in this context, as the capacity to make decisions to take on the risk and to manage it. This would require the company to have people who perform those control functions(168).

(325) However, according to the information submitted by Starbucks(169), Alki LP has no employees of its own and while, as a partnership, it cannot in principle be excluded that the partners could contribute to Alki LP's activities while not being employees, according to the information submitted(170), Alki LP's partners are corporations and none of them has any employees. Consequently, the operating capacity of Alki LP to assume risks appears limited.

(326) In addition, Alki LP's capacity to bear financial risk is limited to its financial resources and to the financial resources of its partners. According to the information submitted by Starbucks to the Commission(171), the latter do not declare separate accounts, but are themselves limited liability companies. Therefore, the financial capacity of Alki LP is inferior to and cannot be equated to the overall financial capacity of the group.

(327) In response, Starbucks argues that Alki LP's operating capacity is provided through employees of Starbucks Inc., occasionally contracted to support Alki LP in its tasks(172). The Commission finds this argument unconvincing. There is no reason why the hiring of a management support employee should proceed through Alki LP: SMBV could have hired an employee to support the management of the company directly. Since the accountability of outsourced management is weaker than the accountability of direct managers, outsourced management presents increased agency costs from a corporate governance perspective. Given SMBV's activities that include, among others, operating a roasting facility, the management of SMBV requires day-to-day monitoring and cannot be effectively ensured by a part-time employee.

(328) Therefore, Alki LP cannot ensure the management of SMBV's business risks as effectively as direct employees or managers of SMBV could, and therefore that risk could, at best, be partially mitigated through its contractual transfer to Alki LP.

(329) That conclusion supports the doubt expressed in recital 89 of the Opening Decision on the economic rationality of the structure, as Alki LP seems superfluous in this structure. Paragraphs 1.64 to 1.66 of the 2010 OECD TP Guidelines refer(173), in that respect, to situations where structures are not determined by normal commercial conditions and may have been structured by the taxpayer to avoid or minimise tax. In such cases, the tax authorities should analyse the taxable basis based on a corrected structure that would have been determined by normal commercial conditions. The information submitted by Starbucks, according to which Alki LP was added to the structure for US tax reasons(174), confirms the Commission's doubts on the economic rationality of the structure.

(330) Fifth, while the Commission accepts the claim of the Netherlands and Starbucks that certain sales risks of SMBV are mitigated through the arrangements with the Shops, this fact has no bearing on SMBV's relationship with Alki LP. In any event, regarding the limited scope of SMBV in the negotiations with its counterparties of commercial conditions and of prices, the Commission accepts that the forecast system by the Shops seems to reduce business uncertainty for SMBV(175). Under the contractual arrangements between the Shops and SMBV, the Shops cannot significantly depart from the forecast provided to SMBV and they also are under the obligation to purchase products from SMBV. The off taking of the production of SMBV is ensured through the requirements for the Shops to purchase production from SMBV. However, key risks of a manufacturing business relate not only to inventory risks but mainly to the manufacturing capacity and neither the Netherlands nor Starbucks have established that the underlying risks of demand variations on the manufacturing capacity utilisation have been completely annulled through those contractual arrangements. Specifically, the contracts with third party manufacturers, [unaffiliated manufacturing company 1] in particular, are not fully aligned with any possible demand variation.

(331) Sixth, the contractual arrangement invoked by the Netherlands by which the SCTC guarantees the quality of the green coffee beans supplied equally has no bearing on the relationship between SMBV and Alki LP. In any event, it is a standard business arrangement that does not reduce the risk of SMBV compared to normal market conditions.

(332) In conclusion, the Commission rejects the claims that any effective risk transfer takes place from SMBV to Alki LP through contractual arrangements. Therefore, any component of the royalty meant to compensate for an entrepreneurial risk transfer cannot be justified.

(d)   The level of the royalty payment is not justified by the amounts Alki LP pays for technology to Starbucks US under the Cost Sharing Agreement

(333) Starbucks further seems to invoke that the royalty paid by SMBV to Alki LP is justified because Alki LP pays amounts of the same magnitude to Starbucks US for the coffee roasting technology.

(334) The Commission notes, at the outset, that the SMBV APA is not concerned with the relationship between Alki LP and Starbucks US, but with the relationship between SMBV and Alki LP. Accordingly, all that matters for an assessment of whether the transfer pricing analysis of the transactions between those two entities has been determined in accordance with a methodology that gives a reliable approximation of a market-based outcome in line with the arm's-length principle is the value of those transactions as determined by Starbucks' tax advisor, and not the value of any payments outside that relationship.

(335) In any event, the Commission does not contest that important sums, designated as a ‘permanent royalty’, are paid by Alki LP to Starbucks US(176). However, the designation of this payment under the CSA as remunerating the roasting technology cannot be confirmed based on the way in which those payments are structured under the CSA.

(336) First, the payments by Alki LP to Starbucks US are financed to a larger extent by the royalty income it receives from the brand royalty paid to it by Starbucks Coffee BV than from the royalty received from SMBV(177).

(337) Second, the payments from Alki LP to Starbuck US under the CSA do not appear to be set at arm's length. In particular, regarding the acquisition of the IP relating to business format and brand, Alki LP (including its predecessor [CV 1]) paid in aggregate EUR [30-40] million for the acquisition of this IP, as of 2005, while that IP was sold by Alki LP for EUR [1-1,5] billion in 2014. That difference in value seems to indicate that that arrangement did not reflect an arm's-length remuneration for the IP; the payments under the permanent royalty could therefore be seen as compensating for the sale of the brand in the same transaction for a value that was too low.

(338) In sum, Starbucks' argument that the payments under the CSA justify the royalty payments between SMBV and Alki LP as being arm' length must be rejected.

(e)   Conclusion on the arm's-length nature of the royalty payment

(339) In light of the above, the Commission considers that a comparison to comparable uncontrolled transactions, in particular the royalty arrangements in several roasting and manufacturing and distribution agreements Starbucks has concluded with third parties, demonstrates that the arm's-length value of the royalty paid by SMBV to Alki LP for roasting IP should be zero, thus, that no royalty should be due for that IP in that specific relationship, since SMBV does not derive any benefit from the use of the roasting IP licensed from Alki LP.

(340) Given that conclusion, this royalty does not need to be estimated. Rather, SMBV's accounting profits, with no deduction of the royalty payment from SMBV to Alki LP from those profits for the licensing of the roasting IP, should be the starting point from which SMBV's tax liability in the Netherlands is determined to ensure that SMBV taxable profits correspond to a level obtained by non-integrated companies whose tax liability is determined by the profits they generate under market conditions. In other words, the un-taxed profits paid as a royalty by SMBV to Alki LP for the roasting IP should have been fully taxable in the Netherlands.

(341) Consequently, the methodology accepted by the SMBV APA for determining the level of that payment, according to which all profits generated by SMBV in excess of [9-12] % of operating expense is transferred to Alki LP(178), departs from a methodology that leads to a reliable approximation of a market-based outcome in line with the arm's-length principle. Since the application of that methodology leads to a lowering of SMBV's tax liability under the general Dutch corporate income tax system as compared to non-integrated companies whose taxable profits is determined by the market, the SMBV APA, by accepting that methodology, should be considered to confer a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty.

(f)   The arm's-length nature of the prices SCTC charged to SMBV for green coffee beans

(342) As explained in recital 272, the identification and analysis of SMBV's controlled and uncontrolled transactions is a necessary first step in assessing the arm's-length nature of commercial conditions applicable between related parties for transfer pricing purposes.

(343) In recital 116 of the Opening Decision, the Commission explained that if the royalty payment was estimated using a direct transfer pricing method, such as the CUP method, the prices charged for the green coffee beans would be the outstanding related controlled transaction that needed to be assessed for transfer pricing purposes and the price charged for the green coffee beans by SCTC to SMBV would have to be assessed to establish whether the level of that price, reflected in SMBV's profit and loss account, was not exaggerated, leading to a reduction of SMBV's taxable profits.

(344) The transfer pricing report lists the green coffee bean sourcing agreement between SMBV and SCTC among the most important transactions and inter-company flows, but fails to examine or analyse whether the price charged for the green coffee beans by SCTC to SMBV is at arm's length.

(345) The pricing of green coffee beans by SCTC is described in recitals 114 to 119. Table 6 presents the breakdown by Starbucks of the mark-up of green coffee beans purchased by SCTC charged to SMBV. Based on SCTC's financial data provided in Table 4, Table 11 presents the actual mark-up recorded by SCTC and the actual gross margin charged to its customers each year.

Table 11

Mark-up on green coffee bean sourcing recorded and gross margin applied by SCTC

(346) The average mark-up on the costs of green coffee beans supplied by SCTC for the period 2005 to 2010 is [around 3 %], compared to an average mark-up of [around 18 %] over the period 2011 to 2014. The corresponding gross margin on COGS for the period 2005 to 2010 is [around 6 %], compared to an average gross margin on COGS of [around 18 %] over the period 2011 to 2014.

(347) According to Starbucks, the mark-up of [around 3 %] applicable on average for the period 2005 to 2010, corresponds to an arm's-length mark-up. When deducted from the mark-ups applied since 2011, that mark-up seems to result in a remuneration recorded by SMBV on the roasting activities in line with the estimated range presented by Company X of 13 to 17 % on COGS for roasting activities(179). The [around 3 %] mark-up is also within the range for supply function remuneration put forward by Starbucks in the comparable analysis provided on 29 June 2015(180). The Commission can therefore accept that the [around 3 %] mark-up on the costs of green coffee beans during the period 2005 to 2010 was at arm's length.

(348) However, since Starbucks did not provide any valid justification for the increase in the average mark-up to [around 18 %] from 2011 onwards, the Commission considers, as explained in recitals 349 to 357, that no corresponding deduction to SMBV's accounting profits as a result of that increase should be accepted from that period onwards. In fact, had the transfer pricing report properly analysed the price charged by SCTC to SMBV for green coffee beans as a controlled transaction, that price would have been within the scope of the SMBV APA, which is to determine SMBV's tax base in the Netherlands(181). Indeed, since the price paid for the green coffee beans is deducted from SMBV's accounting profits, the SMBV APA should have prescribed an arm's-length price in 2008 from which no deviation would have been possible in 2011, including an increase in the mark-up, unless that APA were replaced or amended.

(349) Starbucks claims that the increase of that mark-up in 2011 was due to the growing importance of SCTC's operations, particularly the increased expertise in coffee procurement and, more important, its ownership and operation of the evolving C.A.F.E. Practices Programme. That justification does not, however, seem to correspond to the information provided during the investigation. In particular, the C.A.F.E. Practices Programme has been in place since 2004. Furthermore, as SCTC's remuneration is proportionate to the green coffee beans sold, any increase in capacity should have been remunerated accordingly. Indeed, so long as the remuneration constitutes a stable percentage of COGS, the increase of turnover resulted in a proportionate increase of profits.

(350) Starbucks further claims that the mark-up ‘throughout the period’ 2005 to 2014 was arm's length in nature(182). In its submission of 29 June 2015, Starbucks clarified that the breakdown of Table 5, provided in its submission of 13 April 2015, was constructed based on a comparables analysis. The Commission does not accept that claim for the following reasons.

(351) First, as regards the pricing of the C.A.F.E. Practices Programme in the comparables analysis provided by Starbucks on 29 June 2015(183), the Commission observes that the figures submitted by Starbucks present consistency issues. Information presented on 13 April 2015 and reproduced in Table 6 contradicts the information provided on 29 June 2015 presented in Table 7. More specifically, in relation to the C.A.F.E Practices Programme, what was initially presented as the upper quartile, was later presented by Starbucks as the highest observed value. Inconsistencies also concern the use of a transformation factor to express a royalty agreed as a percentage of sales, where Starbucks performs an uncommon adjustment to arrive at a royalty expressed as a percentage of costs(184).

(352) In addition, the identified agreements(185) relate to licensing of technology. It could be compared to a situation where Starbucks through Alki LP licences technology to SMBV, although agreements where coffee technology is licenced out do not seem to be existing standalone from a licence agreement on coffee brands. SCTC does not licence any technology to SMBV for which SMBV should receive a remuneration through a higher mark-up on green coffee beans purchased.

(353) Second, the Commission observes that the costs of both the C.A.F.E. Practices Programme and the Farmers Support Centres was on average not more than [0,5-1] % of the value of green coffee beans purchased by SCTC(186). Those costs are of a comparable magnitude to the cost of the [certification programme] designation of [1,5-2] % in proportion of the green coffee beans price used for the ‘[certification programme coffee product e]’, which would be SMBV's most-sold product(187). In fact, the C.A.F.E. Practices Programme seems to consist of a certification programme more comparable to the [certification programme] designation than to an intellectual property or technology licence as presented by Starbucks.

(354) Furthermore, the prices of coffee products sold by SMBV to the Shops is determined on a cost basis as explained in recital 96. Therefore, the Commission considers that the direct and indirect costs of C.A.F.E. Practices Programme is, contrary to what Starbucks claims, a more appropriate way to approach the arm's-length pricing of the programme on the price of green coffee beans charged to SMBV.

(355) Regarding the financing mark-up presented in Table 6, the Commission notes that the method adopted does not calculate the working capital cost because client payables, which contribute in a normal business to finance account receivable and inventories, are not deducted from the estimated amount to finance(188). Starbucks possibly considers that financing costs are to be added because the agreements regarding the estimated supply costs are based on buying agents, who might not take title to the goods sold. This is not, however, assessed or evidenced in the analysis provided by Starbucks. Therefore, a financing mark-up comparable to the one put forward by Starbucks does not seem justified.

(356) Third, Figure 4 shows the impact on SCTC's profit in Swiss Francs (CHF) from the increase of the mark-up in 2011. Figure 4 also presents the operating expense of SCTC, which did not demonstrate any such increase that could have been expected from an increased importance of SCTC as claimed by Starbucks. The operating expense in percentage of total costs remained stable and the operating costs would therefore have been covered through a remuneration linked to COGS. The increase in the mark-up from [around 3 %] to [around 18 %] on average for 2005 to 2010 and from 2011 to 2014, respectively(189), resulted in a quadrupling of SCTC profits, as illustrated in Figure 4.

Figure 4

Profitability of SCTC

[Bild bitte in Originalquelle ansehen]

(357) For the sake of completeness, the Commission observes that SMBV's losses on its roasting activities since 2010 can be placed in relation to the increased mark-up on the costs paid to SCTC for the green coffee beans(190), as shown in Figure 5.

Figure 5

Profitability of the roasting activity of SMBV

[Bild bitte in Originalquelle ansehen]

(358) In light of those observations, the Commission considers that to arrive at a reliable approximation of an arm's-length mark-up for the period from 2011 onwards, the [around 3 %] average mark-up for the period 2005 to 2010 should be increased by the costs of C.A.F.E. Practices Programme and up to the amount of costs of the [certification programme] designation. The cost of the [certification programme] designation represented [1-1,5] % of the costs of green coffee beans purchased by SCTC at the end of 2014 and translated into [0,5-1] % of the price charged to SMBV(191). An arm's-length mark-up recorded by SCTC for the period 2011 onwards would therefore be up to [around 6 %] of the costs of green coffee beans purchased by SCTC, corresponding to a gross margin of up to [around 9 %] on SCTC COGS, charged by SCTC to SMBV. Consequently, the [around 18 %] average mark-up on the costs of green coffee beans supplied by SCTC to SMBV effectively applied from 2011 to 2014 does not reflect a reliable approximation of a market-based outcome in line with the arm's-length principle.

(359) In sum, since the transfer pricing report fails to examine or analyse whether the price charged for the green coffee beans by SCTC to SMBV is at arm's length, the methodology proposed in that report to determine SMBV's taxable profits departs from a methodology that leads to a reliable approximation of a market-based outcome in line with the arm's-length principle. Since the application of that methodology leads to a lowering of SMBV's tax liability under the general Dutch corporate income tax system as compared to non-integrated companies whose taxable profits is determined by the market, the SMBV APA, by accepting that methodology, should be considered to confer a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty.

(g)   Conclusion on the selective advantage granted by the SMBV APA

(360) As concluded in recitals 339 to 341, the Commission considers that a comparison to comparable uncontrolled transactions, in particular the roasting IP licensing arrangements in several roasting and manufacturing and distribution agreements Starbucks concluded with third parties, demonstrates that the arm's-length value of the royalty paid by SMBV to Alki LP for the roasting IP should be zero. Consequently, the SMBV APA, by accepting a methodology for determining the level of the royalty according to which all profits generated by SMBV in excess of [9-12] % of operating expense is transferred to Alki LP(192), confers a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty.

(361) In addition, the transfer pricing report's failure to examine or analyse whether the price charged for the green coffee beans by SCTC to SMBV is at arm's length means that the methodology proposed in that report, accepted by the SMBV APA, for determining SMBV's taxable profits in the Netherlands, confers a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty.

9.2.3.4.   

SMBV was incorrectly qualified as the less complex function in the transfer pricing report

(362) As explained in recital 282, it is only if it is impossible to compare a particular intra-group transaction for which an APA is being sought, taking into account the functions performed, to similar uncontrolled transactions, that resorting to a comparison of functions performed is justified. As explained in Section 9.2.3.3, the Commission considers that the royalty arrangements in several roasting and manufacturing and distribution agreements Starbucks concluded with third parties constitute comparable uncontrolled transactions to SMBV's royalty arrangement with Alki LP as follows from the SMBV APA.

(363) Without prejudice to this conclusion that no functional comparison was warranted in the present case, the Commission further submits that the transfer pricing report's analysis of the functions performed in the application of the TNMM does not result in a reliable approximation of a market-based outcome in line with the arm's-length principle.

(364) To appropriately estimate the arm's-length remuneration of functions, the transfer pricing report should have engaged in a comparison of the functions performed by each party to the related transactions.

(365) In one-sided transfer pricing methods such as the TNMM, only the remuneration of the ‘tested party’ is analysed for transfer pricing purposes, regardless of the resulting remuneration of the other parties to the transaction. On the assumption that SMBV is the ‘least complex function’ in the relationship between SMBV and Alki LP, Starbucks' tax advisor took SMBV as the ‘tested party’ and did not consider whether the residual profit allocated to Alki LP is in proportion to Alki LP's functions, risks and assets. Starbucks justifies that choice on the grounds that SMBV does not own valuable IP and does not incur meaningful business risks in performing routine activities; SMBV would therefore be the least complex entity in that relationship(193).

(366) However, Starbucks' line of reasoning demonstrates a confusion between the complexity of functions and risks assumed. Paragraph 3.18 of the 2010 OECD TP Guidelines(194) explains that the choice of the ‘tested party’ in the application of the TNMM should be consistent with the functional analysis of the transaction(195). As a general rule, the ‘tested party’ is the one to which a transfer pricing method can be applied in the most reliable manner and for which the most reliable comparables can be found, i.e. it will most often be the one that has the less complex functional analysis.

(367) Complexity and risk are distinct, although considerations of risk flow into the functional analysis. Complexity is to be assessed in relative terms, that is, by comparison to the other parties involved in the transactions. Paragraph 3.18 of the 2010 OECD TP Guidelines refers for the choice of the tested party in this regard to a ‘less’ complex function, rather than in absolute terms to a function which would not be complex(196). The requirement to assess the complexity of functions in comparison to the other parties to the transactions is also found in paragraphs 1.21, 1.22 and 1.23 of the 1995 OECD TP Guidelines. Accordingly, where the TNMM is relied upon for transfer pricing purposes, the functions of the other party to the transaction, in this case Alki LP, also have to be analysed. Information on the functions of Alki LP was certainly available to the Dutch tax administration at the time of the SMBV APA request.

(368) However, if direct observables can be identified in respect of the related transactions, such observables should serve to determine the remuneration of the company engaging in comparable transactions(197). The existence of direct comparables to determine the arm's-length remuneration is what the Commission effectively argues in Section 9.2.3.3.

(369) Nevertheless, in the following section, the Commission analyses the relative complexity of the group entities engaging in transactions with SMBV.

(370) First of all, SMBV performs a number of functions besides its roasting function. This fact, as such, presents a difficulty in finding appropriate comparables for transfer pricing purposes(198). In addition, the roasting function performed by SMBV is critical for coffee producers. SMBV also undertakes or buys market research, holds IP and engages in contracts with toll manufacturers. Finally, the price setting responsibilities are not clearly defined and SMBV seems to have a degree of control over prices received for goods sold.

(371) The routine nature of the roasting function performed by SMBV claimed by Starbucks' tax advisor in the transfer pricing report is contradicted by the submissions of competitors, as three of the four competitors having received a MIT request by the Commission do not subcontract roasting. In fact, two of those competitors indicated that they consider the roasting function to be critical and therefore, in principle, they do not consider that it could be subcontracted(199). However, that the roasting function is an important function does not necessarily mean that it is also very complex.

(372) Yet, in the specific case of SMBV, an analysis of its operating expense shows a significant expense for market research(200). Moreover, one of SMBV's most important operating expenses is the amortisation of intangible assets. SMBV conducts market research, holds significant IP, and acquired additional IP in 2012 for a value of EUR 4 million. A routine manufacturer does not engage in such activities.

(373) The Netherlands claims that those expenses, which relate to software and an IT system licence payment for standard IT systems, does not as such evidence that SMBV would be using valuable IP. For example, in the case of Company Y a company designated as toll manufacturer pays IT system licences(201). However, in the case of SMBV, the IP amortisation payments do not relate to such a software fee, considering software fees are recorded under a distinct item in accounts of SMBV(202).

(373) Such considerations are not sufficient to conclude on their own that SMBV is not the less complex function. The complexity of the functions ensured by Alki LP must also be assessed for this purpose.

(375) The activities of Alki LP are assessed in recitals 324 to 329. It follows from that assessment that Alki LP's operating capacity is extremely limited to non-existent, given that it has no employees and none of its partners have any employees.

(376) The Commission therefore concludes that Starbucks' tax advisor was unjustified in designating SMBV as the less complex function compared to Alki LP for the transfer pricing analysis.

(377) Consequently, since the methodology for determining SMBV's tax base in the transfer pricing report was premised on the flawed assumption that SMBV should be the ‘tested party’ for the application of the TNMM, that methodology does not result in a reliable approximation of a market-based outcome in line with the arm's-length principle. Since the SMBV APA's endorsement of that methodology leads to a lowering of SMBV's tax liability under the general Dutch corporate income tax system as compared to non-integrated companies whose taxable profit under that system is determined by the market, that APA should be considered to confers a selective advantage to SMBV for the purposes of Article 107(1) of the Treaty.

9.2.3.5.   

The transfer pricing report misapplies the TNMM

(378) In the preceding sections the Commission demonstrated that the transfer pricing report fails to examine whether the royalty payment made by SMBV to Alki LP for the licensing of the roasting IP, which is the intra-group transaction for which the SMBV APA was effectively requested and granted(203), is at arm's length. It subsequently demonstrated that the transfer pricing report incorrectly analyses the complexity of all group companies for the application of the TNMM. It is without prejudice to these conclusions, which in and of themselves allow the Commission to further conclude that the SMBV APA confers a selective advantage on SMBV in the form of the lowering of its tax liability in the Netherlands, that the Commission also submits, for the sake of completeness, that Starbucks' tax advisor improperly applied the TNMM in the transfer pricing report.

(a)   Starbucks tax advisor inappropriately identified SMBV's main functions in the transfer pricing report

(379) For transfer pricing purposes, the choice of transfer pricing method and the selection of comparables are determined on the basis of the functional analysis of the company for which an APA is being requested. According to the functional analysis prepared by Starbucks' tax advisor, SMBV's primary functional contribution to the Starbucks group relates to roasting/manufacturing activities(204). On the basis of that assumption, the tax advisor considers SMBV as a low-risk coffee manufacturer performing routine functions and therefore the ‘least complex entity’ for the purposes of applying the TNMM.

(380) In the Opening Decision, the Commission expressed the doubt that the Netherlands incorrectly accepted SMBV's classification as a low-risk toll manufacturer in the transfer pricing report(205).

(381) Based on the information submitted by the Netherlands and Starbucks during the investigation, the Commission considers that the tax advisor incorrectly designated SMBV as a low-risk manufacturer. More specifically, information provided on SMBV's revenues demonstrates that roasting is not the main source of income of SMBV. As illustrated in Table 2, income from the sales of roasted coffee as recorded under the item ‘REV PACKAGED COFFEE’(206) represents only [15-20] % of the revenues of SMBV in 2013 and 2014. In 2007, that is, at the time that the SMBV APA was concluded, that proportion was substantively at the same level at [15-20] %.

(382) Instead, SMBV derives most of its profits recorded in the Netherlands from an activity different from roasting. Indeed, SMBV's reselling function, referred to as providing logistic and administrative services by the Netherlands and Starbucks(207), appears more important than its roasting activity and accounted for [80-85] % of SMBV's revenues in 2013 and 2014. In 2007, based on data which was available at the time the SMBV APA was requested, that proportion was [75-80] %. Moreover, at that moment, [10-30] out of [40-60] SMBV's employees were active in that activity. SMBV was also managing three contractual relationships related to distribution and logistics.

(383) Contrary to the claim by the Netherlands that the costs of non-coffee products are pass-through costs for SMBV(208), SMBV in fact records a margin on the resale of non-coffee products. Moreover, SMBV's reselling function not only represents the main source of SMBV's income, but also represents the only source of SMBV's profit since 2010, considering SMBV's roasting activity has been loss making since that year(209), without prejudice to the question whether those losses on the roasting activities were caused by incorrect pricing of the green coffee beans(210).

(384) The Netherlands further argues that it did not need to request more information from Starbucks than was provided to its tax administration at the moment of the APA request, because it was familiar with SMBV's business, as the APA was the renewal of a previous ruling(211). However, the fact that the Netherlands ignored that most of the SMBV's income derives from the resale of non-coffee products clearly contradicts its claim that its tax administration was familiar with SMBV's business(212).

(385) Although the transfer pricing report does acknowledge that SMBV resells production bought from third parties to the Shops, it ignores this function when choosing comparables and a profit level indicator for the application of the TNMM. Accordingly, Starbucks' tax advisor failed to perform a critical assessment to identify SMBV's principal functions for the transfer pricing analysis(213).

(386) SMBV's principal function was thus insufficiently identified in the transfer pricing report by Starbucks' tax advisor and wrongly accepted by the Dutch tax administration as the basis for calculating the remuneration accepted in the SMBV APA.

(b)   Starbucks' tax advisor inappropriately used operating expenses as profit level indicator in the application of the TNMM

(387) Paragraph 2.87 of the 2010 OECD TP Guidelines indicates that typically and subject to the facts of the case, sales or distribution operating expenses might be an appropriate base for distribution activities when using the TNMM(214).

(388) According to the Netherlands, even if the activities of SMBV were not correctly determined in the transfer pricing report, operating expenses are an appropriate indicator for profitability in the application of the TNMM. However, since SMBV's profits are generated and recorded through a margin on products distributed, the Commission considers sales as a more adequate indicator of SMBV's profit generating reselling function. Moreover, between 2008 and 2014(215), SMBV's total sales have increased almost three-fold, while the gross margin has more than doubled over that same period(216). By contrast, SMBV's operating expenses increased by only 6 %, so that operating expenses cannot be considered an adequate indicator of SMBV's profit generating resale activity.

(389) Indeed, SMBV's profits arising from its reselling activities are recorded in the Netherlands and are not attributed to any other member of the group in a position to generate such profits. In fact, the payment of royalties as a residual profit effectively shifts profits SMBV derives from the resale of non-coffee products to Alki LP. However, none of the alleged components of the royalty payment has any relation to profits from SMBV's reselling function, since Alki LP is not in a position to generate active profits from the resale of non-coffee products, as it has no operating capacity either directly or through its partners(217). Therefore, attributing those profits to Alki LP through the payment of royalty based on residual profits is not in line with the arm's-length principle.

(390) Moreover, recorded profits have to be attributed, because they are an economic reality, which cannot be superseded by the use of an economic or transfer pricing model that has as its objective approximating economic reality in absence of (direct) observables. In the case of SMBV, profits from the resale of non-coffee products are not attributed through a remuneration in any form to any other group entity nor to any third party that would be in a position to generate active profits from the resale of non-coffee products and, therefore, they should be attributed to SMBV.

(391) The tax advisor thus inappropriately used operating expenses instead of sales as profit level indicator in the application of the TNMM.

(c)   A functional analysis based on SMBV's reselling function and a remuneration based on margin of sales would have led to a higher level of remuneration

(392) To illustrate the impact of Starbucks' tax advisor's incorrect identification of SMBV's main functions and its inappropriate selection of operating expenses as profit level indicator, the Commission replicated the tax advisor's analysis with a corrected peer group of companies based on SMBV's reselling function and calculated a mark-up on sales for that corrected peer group.

(393) Given that SMBV's main business activity was incorrectly identified in the transfer pricing report, the peer group of comparables used by the tax advisor to determine a remuneration for the functions performed by SMBV — NACE code ‘Processing of tea and coffee’(218) — was equally inappropriate in the application of the TNMM. To correct for the peer group, the Commission conducted a similar analysis, based on the Amadeus database as used in the transfer pricing report, using the NACE code ‘Wholesale of coffee, tea, cocoa and spices’. The other selection criteria employed by the Commission to select the peer group were identical to those used by Starbucks' tax advisor in the transfer pricing report(219).

(394) For comparison purposes, the Commission then removed companies from the corrected peer group that mainly distribute products other than coffee and tea (such as spices, sugar or companies selling equipment only), while the retaining companies engaged in roasting, resulting in twelve comparable companies.

(395) Since half of those twelve companies did not report data on operating expenses(220), and since that data is sensitive to accounting differences between tax jurisdictions, the Commission calculated a mark-up on sales for the companies in the corrected peer group. The result of those calculations is represented in Table 12.

Table 12

Comparables analysis based on wholesale distribution

(396) A peer group analysis of companies active principally in wholesale distribution of coffee results in a median return on sales of 3,1 %, with an interquartile range of 1,5 % to 5,5 %. Although the interquartile range is only presented in one point of the 2010 OECD TP Guidelines among other possible ranges to address comparability concerns(221), it is frequently used by tax advisors. However, to avoid granting an advantage to companies resorting to transfer pricing, the point in the range closest to the most likely market outcome should be used for the purposes of pricing controlled transactions.

(397) In the present case, SMBV tax base as determined by the SMBV APA is outside the interquartile range calculated on the basis of the corrected peer group with a remuneration based on a mark-up on sales. As presented in Table 13, in each year since 2008 SMBV's taxable profit calculated on the basis of the SMBV APA was lower than its taxable profit in the lower point of that range, i.e. 1,5 % of sales:

Table 13

(398) The interquartile range of 1,5 % to 5,5 % is based on financial data from the period 2005 to 2007, which is the most recent period pre-dating the SMBV APA. The finding that the tax base agreed to in that APA falls outside of the range is equally valid for different observation periods(222).

(399) That Starbucks' tax advisor was incorrect to consider SMBV as a manufacturing company with a remuneration based on operating expenses for the purposes of estimating an arm's-length remuneration for SMBV is further confirmed by an internal data comparison with SMC. SMC is the only other company of the Starbucks group ensuring roasting activities internally(223). Since that company is consolidated for tax purposes with Starbucks US, there are no incentives for Starbucks to shift the profits of that company by means of a royalty paid to Starbucks Corporation. Therefore, although SMC's accounts are not audited, they are established for genuine business purposes in the absence of tax considerations. Based on those accounts, SMC is more than [40-50] times more profitable than the remuneration agreed for SMBV in the SMBV APA when considering a return on operating expenses. As explained in recital 140 and shown in Table 9, while the SMBV APA agrees on a [9-12] % mark-up of profits over operating expense, that ratio stood at around 500 % for SMC over the past four accounting periods.

(400) The purpose of the exercise undertaken by the Commission in recitals 392 to 398 is not to calculate an arm's-length remuneration for the functions performed by SMBV within the Starbucks group. The Commission acknowledges that the range presented above is not backed by a sufficient comparability analysis and that the exercise is simply meant to replicate and duplicate the tax advisors analysis if the functions would have been correctly identified. Rather, the purpose of the exercise undertaken by the Commission is to show that even if the conclusions reached by the Commission in Sections 9.2.3.3 and 9.2.3.4 were incorrect, the tax advisor's misidentification of SMBV's main functions and its inappropriate selection of operating expenses as profit level indicator in the application of the TNMM confirm that the methodology proposed by it in the transfer pricing report and accepted by the SMBV APA for determining SMBV's tax base in the Netherlands does not result in a reliable approximation of a market-based outcome in line with the arm's-length principle. Since that methodology results in a lowering of SMBV's tax liability under the general Dutch corporate income tax system as compared to non-integrated companies whose taxable profit under that system is determined by the market, the SMBV APA, by accepting that methodology, should be considered to confer a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty.

(d)   The inappropriateness of the working capital adjustment

(401) In the transfer pricing report, Starbucks' tax advisor proposes a ‘Conversion Mark-up Adjustment’, which is presented by the Netherlands as a working capital adjustment, but the methodology used does not factor in the level of the working capital of either the comparables or of SMBV. Working capital is the sum of the inventories and trade receivable to be financed net of trade payables. There is no constant relation between the COGS used in the adjustment and working capital needs. In particular, a company with a high amount of raw material cost might have low working capital needs if it processes its stock efficiently. Working capital adjustments are aimed at capturing possible differences in the stock and trade receivable and payables processing, which is not captured by the amount of raw material used by the company.

(402) The Commission therefore considers the tax advisor's ‘working capital adjustment’ ill-fitted for the declared purpose and adjusting for differences in working capital use. There is no justification for that adjustment in the set of facts presented in the transfer pricing report. The Netherlands' argument that the method presented in the Annex to Chapter III of the 2010 OECD Guidelines, although available at the time, was not part of the 1995 OECD Guidelines, which were applicable at the moment the APA was entered into(224), is therefore irrelevant

(403) In response to the doubts expressed by the Commission on the working capital adjustment in the Opening Decision(225), the Netherlands provided a simulation of a mark-up calculated using the methodology of the adjustment for different time periods, using a different interest rate and finally using the same interest rate as a reference, but deducting rather than adding 50 basis points. Those simulations do not, however, address those doubts. The Commission did not express doubts on the time period used by Starbucks' tax advisor in the transfer pricing report. Rather, in recital 101 of the Opening Decision, the Commission notes that the hypothetical raw material remuneration was estimated by reference to the EURIBOR to which a spread of 50 basis points was added, the level of which is not explained. The Netherlands does not explain that level by presenting simulations with different interest rates and a different spread of minus 50 basis points, which remains unexplained.

(404) Regarding the simulation provided by the Netherlands, where SMBV's pre-tax profits divided by full costs, including COGS, are compared to the ratio of pre-tax profit to total costs of competitors(226), the Commission notes that the Netherlands take a range that is based on information dating from after the SMBV APA. If an inter-quartile range of pre-tax profit over total costs is established on the basis of the financial data of peers for the period 2001-2005, as used in the transfer pricing report, this range would be 4,9 % to 13,1 % and the profits of SMBV over that period would be outside that range(227). For example, in 2008 SMBV's pre-tax profits over total costs were at 1,2 % and decreased to 0,5 % in 2014.

(405) The Netherlands further refers to an article of the 2012 International Transfer Pricing Journal that would argue that for both fully-fledged and toll or contract manufacturers the total costs, including COGS, are the most suitable profit level indicator. The article would also present a comparison between margins on total costs of fully-fledged manufacturers and toll or contract manufacturers. This analysis would show, on the basis of the sample chosen, that toll and contract manufactures would have higher margins on total costs than fully-fledged manufacturers. However, notwithstanding the general validity of the outcome of the empirical study presented in the article, the findings of the article contradict rather than support the Netherlands' argument that the profitability of low-risk manufacturers should be adjusted downwards compared to fully fledged manufacturers. This is because the Netherlands accepted a reduction of the margin on costs to factor in the fact that SMBV would not be a fully-fledged manufacturer, whereas the empirical findings of the article seem to indicate that a higher margin would have been appropriate.

(406) Finally, the transfer pricing report also accepts a considerable reduction in the cost base used to calculate the tax base in 2008 compared to the previous arrangement by excluding the costs of [unaffiliated manufacturing company 1], although the activities of SMBV did not change and the commercial relationship with [unaffiliated manufacturing company 1] also did not change. Combined with the misclassification of the actual activities of SMBV, that adjustment does not seem sufficiently reasoned.

(407) In sum, even if the comparables analysis had not been based on an incorrect classification of SMBV's activities as coffee roasting and Starbucks' tax advisor had not improperly used operating expense instead of sales as profit level indicator in the application of the TNMM, the use of the working capital adjustment and the exclusion of [unaffiliated manufacturing company 1]'s costs from SMBV's tax base mean that the methodology proposed by the transfer pricing report and accepted by the SMBV APA does not result in a reliable approximation of a market-based outcome in line with the arm's-length principle.

(408) By accepting that methodology, which leads to a lowering of SMBV's tax liability under the general Dutch corporate income tax system as compared to non-integrated companies whose taxable profit under that system is determined by the market, the SMBV APA confers a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty.

9.2.4.   

Subsidiary line of reasoning: Selective advantage due to a derogation from the Decree

(409) The Netherlands and Starbucks argued that the Decree constitutes the appropriate reference system against which the existence of a selective advantage resulting from SMBV APA must be determined.

(410) As noted at recitals 245 to 251, the Commission does not agree with the arguments of the Netherlands and Starbucks concerning the applicable reference system. However, in a subsidiary line of reasoning, the Commission concludes that the SMBV APA also grants SMBV a selective advantage in the context of the more limited reference system composed of group companies applying transfer pricing to which Article 8b(1) CIT and the Decree apply.

(411) Article 8b(1) CIT and the Decree are considered to establish the ‘arm's-length principle’ under Dutch tax law, according to which transactions between intra-group companies should be remunerated as if they were agreed to by independent companies negotiating under comparable circumstances at arm's length. The preamble to the Decree, in particular, explains that the arm's-length principle as set out in the OECD TP Guidelines has been transposed into domestic law and that those guidelines apply directly to the Netherlands(228).

(412) Considering that the Commission has already demonstrated in Section 9.2.3.1 that the SMBV APA endorses certain methodological choices made by Starbucks' tax advisor for transfer pricing purposes that cannot be considered to result in a reliable approximation of a market-based outcome resulting in a reduction of SMBV's tax liability in the Netherlands, the Commission can similarly conclude that that APA also gives rise to a selective advantage under the more limited reference framework of Article 8b(1) CIT and the Decree.

9.2.5.   

Justification

(413) Neither the Netherlands nor Starbucks have advanced any possible justification for the selective treatment of SMBV as a result of the SMBV APA. The Commission recalls, in this respect, that the burden of establishing such a justification lies with the Member State.

(414) In any event, the Commission has not been able to identify any possible ground for justifying the preferential treatment from which SMBV benefits as a result of the SMBV APA that could be said to derive directly from the intrinsic, basic or guiding principles of the reference system or that is the result of inherent mechanisms necessary for the functioning and effectiveness of the system(229), whether that reference system is the general Dutch corporate income tax system, as established by the Commission, or the Decree, as advocated by the Netherlands and Starbucks.

9.2.6.   

Conclusion on the existence of a selective advantage

(415) The Commission concludes that the SMBV APA, by endorsing a method for arriving at a profit allocation to SMBV within the Starbucks group that cannot be considered to result in a reliable approximation of a market-based outcome in line with the arm's-length principle and that results in a lowering of SMBV's tax liability under the general Dutch corporate income tax system as compared to non-integrated companies whose taxable profit under that system is determined by the market, confers a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty.

(416) By a subsidiary line of reasoning, the Commission concludes that the SMBV APA, by endorsing under the Decree, based on Article 8b(1) CIT, a method for arriving at a profit allocation to SMBV that cannot be considered to result in a reliable approximation of a market-based outcome in line with the arm's-length principle and that results in a lowering of SMBV's tax liability as compared to other group companies taxable in the Netherlands, confers a selective advantage on SMBV for the purposes of Article 107(1) of the Treaty.

9.3.   

Beneficiary of the contested measure

(417) The Commission considers the SMBV APA to grant a selective advantage to SMBV within the meaning of Article 107(1) of the Treaty, since it leads to a lowering of that entity's taxable profit in the Netherlands as compared to non-integrated companies whose taxable profits are determined by transactions concluded on market terms. However, the Commission notes that SMBV forms part of a multi-national corporate group, i.e. the Starbucks group, the remuneration of SMBV's role within that group being the subject-matter of the SMBV APA.

(418) Separate legal entities may be considered to form one economic unit for the purpose of the application of State aid rules. That economic unit is then considered to be the relevant undertaking benefitting from the aid measure. As the Court of Justice has previously held, ‘[i]n competition law, the term “undertaking” must be understood as designating an economic unit […] even if in law that economic unit consists of several persons, natural or legal’(230). To determine whether several entities form an economic unit, the Court of Justice looks at the existence of a controlling share or functional, economic or organic links(231). In the present case, SMBV is fully controlled by Alki LP, which in turn is controlled by companies of the Starbucks group(232).

(419) Moreover, it is the Starbucks group which took the decision to establish SMBV in the Netherlands and thus the Starbucks group which benefits from the SMBV APA as that APA, as indicated in recital 45, establishes the profit that should be allocated to SMBV within that corporate group for the functions it provides to the companies of that group. The SMBV APA is, after all, a ruling that accepts a transfer pricing methodology for transactions within the Starbucks group, so that any favourable tax treatment afforded to SMBV by the Dutch tax administration, benefits the Starbucks group as a whole by providing additional resources not only to SMBV, but to the entire the group. In other words, as discussed in recital 257, where transfer pricing is required to set prices for products and services within various legal entities of one and the same group, the effects of setting a transfer price affects by its very nature more than one group company (a price increase in one company reduces the profit of the other).

(420) Accordingly, notwithstanding the fact that the group is organised in different legal personalities, in the context of a transfer pricing arrangement those companies must be considered as a single group benefitting from the contested aid measure(233).

(421) Finally, in the present case the determination of SMBV' tax base in the Netherland influences the royalty payments to Alki LP as the royalty corresponds to any profit recorded by SMBV above [9-12] % of operating expense as agreed by the SMBV APA(234). The reduction of SMBV's tax liability in the Netherlands therefore not only benefits SMBV, but also Alki LP and therefore the Starbucks group.

9.4.   

Conclusion on the existence of aid

(422) In light of the foregoing, the Commission concludes that the SMBV APA concluded by the Dutch tax administration grants SMBV and the Starbucks group a selective advantage which is imputable to the Netherlands, financed through State resources and which distorts or threatens to distort competition and is liable to affect intra-EU trade. The SMBV APA therefore constitutes State aid within the meaning of Article 107(1) of the Treaty.

(423) Since the SMBV APA gives rise to a reduction of charges that should normally be borne by SMBV in the course of its business operations, that APA should be considered as granting operating aid to SMBV and the Starbucks group.

9.5.   

Evidence relied upon by the Commission for a finding of aid

(424) One of the arguments advanced by the Netherlands during the administrative procedure is that some of the information relied upon by the Commission in its Opening Decision, as well as data relied upon during the formal investigation procedure was not available to the Dutch tax administration on the date on which it entered into the SMBV. Thus, the Netherlands accuses the Commission of enjoying ‘the benefit of hindsight’ when examining the SMBV.

(425) For instance, while the assessment of SMBV's classification as a toll manufacturer refers to the functions performed by SMBV at the time of the APA, it also relates to other activities, such as market research or the amortisation of intangible assets, which only occurred later and therefore could not have been taken into account by the Dutch tax administration when it entered into the SMBV APA in 2008. Equally, the fact that the costs for green coffee beans increased significantly after 2010, which implies that the royalty payments were financed through other activities than the roasting activity in contravention of the arm's-length principle, only became evident after the APA was concluded.

(426) The Commission notes, in this regard, that a considerable number of arguments supporting the Commission's conclusion that the SMBV APA does not comply with the arm's-length principle rely on information and data available to the Dutch tax administration at the time that APA was concluded. This relates, in particular, to several of the roasting and manufacturing agreements between Starbucks and third parties listed in recitals 148 to 150, information on the complexity of the functions performed by SMBV and Alki LP, information questioning the classification of SMBV as a low-risk toll manufacturer and the appropriateness of using operating expenses as profit level indicator instead of sales and the increase by 50 basis points of the raw material mark-up(235).

(427) As regards information on the costs for green coffee beans, the Commission observes that the transfer pricing report lists the green coffee bean sourcing agreement between SMBV and SCTC among the most important transactions and inter-company flows, but fails to examine or analyse whether the price charged for the green coffee beans by SCTC to SMBV is at arm's length, since the tax advisor instead relied on the TNMM to calculate a remuneration based on the roasting function of SMBV.

(428) Had the transfer pricing report properly analysed that agreement, an arm's-length remuneration for the green coffee beans would have had to have been estimated for the purposes of the SMBV APA request(236). Had the SMBV APA consequently covered the green coffee beans transaction, an arm's-length remuneration agreed to in 2008 would have left no room for the unjustified increases in the price of green coffee beans from 2011 and onwards.

(429) In any event, as explained in recital 348 et seq., Starbucks did not provide any valid justification for the increase in the price of green coffee beans after 2010, so the argument is irrelevant.

(430) Moreover, information on the mark-up applied to that transaction was available to the Dutch tax administration at the time that the SMBV APA was requested and, had that administration requested that information, it would have resulted in an arm's-length mark-up of [around 3 %] of COGS for SCTC(237), which is in any event lower than the [around 6 %] mark-up and the corresponding gross margin of [around 9 %] on GOCS which the Commission considers at arm's length in recital 358. If that mark-up had subsequently been accepted in an APA, SCTC would have been prevented from increasing that mark-up to an average of [around 18 %] during the period 2011 onwards, absent a request to modify the APA. In this regard, the Commission recalls that the SMBV APA declares itself to be valid for 10 years provided the critical assumptions presented by SMBV are correct and, in case they are not correct, the APA should be regarded as terminated.

9.6.   

Compatibility of the aid

(431) State aid shall be deemed compatible with the internal market if it falls within any of the categories listed in Article 107(2) of the Treaty(238) and it may be deemed compatible with the international market if it found by the Commission to fall within any of the categories listed in Article 107(3) of the Treaty. However, it is the Member State granting the aid which bears the burden of proving that State aid granted by it is compatible with the internal market pursuant to Articles 107(2) or 107(3) of the Treaty.

(432) The Netherlands has not invoked any of the grounds for a finding of compatibility in either of those provisions for the State aid it has granted by concluding the SMBV APA.

(433) Moreover, as explained in recital 423, the SMBV APA should be considered as granting operating aid to SMBV and the Starbucks group. As a general rule, such aid can normally not be considered compatible with the internal market under Article 107(3)(c) of the Treaty in that it does not facilitate the development of certain activities or of certain economic areas, nor are the tax incentives in question limited in time, digressive or proportionate to what is necessary to remedy to a specific economic handicap of the areas concerned.

(434) Consequently, the State aid granted to SMBV and the Starbucks group by the Netherlands by concluding the SMBV APA is incompatible with the internal market.

9.7.   

Unlawfulness of the aid

(435) According to Article 108(3) of the Treaty, Member States are obliged to inform the Commission of any plan to grant aid (notification obligation) and they may not put into effect any proposed aid measures until the Commission has taken a final position decision on the aid in question (standstill obligation).

(436) The Commission notes that the Netherlands did not notify the Commission of any plan to grant the contested aid measure, nor did it respect the standstill obligation laid down in Article 108(3) of the Treaty. Therefore, in accordance with Article 1(f) of Regulation (EU) 2015/1589(239), the SMBV APA constitutes unlawful aid, put into effect in contravention of Article 108(3) of the Treaty.

10.   

RECOVERY

(437) Article 16(1) of Regulation (EU) 2015/1589 establishes an obligation on the Commission to order recovery of unlawful and incompatible aid. That provision also provides that the Member State concerned shall take all necessary measures to recover unlawful aid that is found to be incompatible. Article 16(2) of Regulation (EU) 2015/1589 establishes that the aid is to be recovered, including interest from the date on which the unlawful aid was at the disposal of the beneficiary until the date of its effective recovery. Commission Regulation (EC) No 794/2004(240) elaborates the methods to be used for the calculation of recovery interest. Finally, Article 16(3) of Regulation (EU) 2015/1589 states, that ‘recovery shall be effected without delay and in accordance with the procedures under the national law of the Member State concerned, provided that they allow for the immediate an effective execution of the Commission decision’.

10.1.   

Legitimate expectations

(438) Article 16(1) of Regulation (EU) 2015/1589 provides that the Commission shall not require recovery of the aid if this would be contrary to a general principle of law.

(439) NOB is the only interested party that raises the issue of legitimate expectations. From the outset, it should be noted that the principle of legitimate expectations can only be invoked by those liable to repay the aid(241), namely Starbucks, and that Starbucks did not submit any argument to that effect.

(440) In any event, for a claim of legitimate expectations to succeed, the expectation must arise from prior Commission action in the form of precise assurances(242). This means that the legitimate expectation must arise from a previous behaviour of the Commission that, for instance, had already approved the same or a similar aid scheme. NOB did not refer to any such acts of the Commission.

(441) The argument of legitimate expectations by NOB is therefore without merit for the purposes of recovery of the aid unlawfully granted by the Netherlands to Starbucks by way of the contested tax ruling in favour of SMBV.

10.2.   

Methodology for recovery

(442) In accordance with the Treaty and the Court of Justice's established case-law, the Commission is competent to decide that the Member State concerned must abolish or alter aid when it has found that it is incompatible with the internal market. The Court has also consistently held that the obligation on a State to abolish aid regarded by the Commission as being incompatible with the internal market is designed to re-establish the previously existing situation. In that context, the Court has stated that that objective is attained once the recipient has repaid the amounts granted by way of unlawful aid, thus forfeiting the advantage which it had enjoyed over its competitors on the market, and the situation prior to the payment of the aid is restored.

(443) No provision of Union law requires the Commission, when ordering the recovery of aid declared incompatible with the internal market, to quantify the exact amount of the aid to be recovered. Rather, it is sufficient for the Commission's decision to include information enabling the addressee of the decision to work out that amount itself without overmuch difficulty(243).

(444) In relation to unlawful State aid in the form of tax measures, the Notice on business taxation provides in point 35 thereof that the amount to be recovered should be calculated on the basis of a comparison between the tax actually paid and the amount which should have been paid if the generally applicable rule had been applied.

(445) As concluded in recitals 339 to 341, the Commission considers that a comparison to comparable uncontrolled transactions using the CUP method, in particular the roasting IP licensing arrangements in several roasting and manufacturing and distribution agreements Starbucks concluded with third parties, demonstrates that the arm's-length value of the royalty paid by SMBV to Alki LP for the roasting IP should be zero. In other words, no royalty should be due for that IP in that specific relationship, since SMBV does not seem to derive any benefit from the use of the roasting IP licensed from Alki LP.

(446) Given that conclusion, the Netherlands should take SMBV's accounting profits, with no deduction of the royalty payment from SMBV to Alki LP from those profits for the licensing of the roasting IP, as the starting point from which SMBV's tax liability in the Netherlands is determined to properly ensure that the aid granted by the SMBV APA is eliminated through recovery.

(447) In addition, given the Commission's conclusion in recital 358 that an average mark-up of up to [around 6 %] of the costs of green coffee beans sold by SCTC to SMBV for the period from fiscal years 2011 onwards and the prices of green coffee beans sold to SMBV corresponding to a gross margin of [around 9 %] for SCTC constitutes a reliable approximation of an arm's-length price, SMBV's accounting profits for 2011 fiscal years and onwards should be increased by the difference in the gross margin on green coffee beans effectively applied during that period and a gross margin on COGS of SCTC of [around 9 %].

(448) It is the difference between the corresponding amount of accounting profits arrived at following the two steps detailed in recitals 446 and 447 fully taxed under the rules of the general Dutch corporate income tax system and the corporate income taxes effectively paid by SMBV to the Netherlands since 1 October 2007 that constitutes the amount of aid that must be recovered from SMBV and the Starbucks group to eliminate the advantage SMBV and the Starbucks group received from the Netherlands as a result of the SMBV APA.

10.3.   

Entity from which the aid is to be recovered

(449) In light of the observations in recitals 417 to 421, the Commission considers that the Netherlands should, in the first place, recover the unlawful and incompatible aid granted by the SMBV APA from SMBV. Should SMBV not be in a position to repay the full amount of the aid received as a result of the SMBV APA, the Netherlands should recover the remaining amount of that aid from Starbucks Corporation, since it is the entity which controls the Starbucks group, so as to ensure that the advantage granted is eliminated and the previously existing situation on the market is restored through recovery.

11.   

CONCLUSION

(450) In conclusion, the Commission finds that the Netherlands has unlawfully granted State aid to SMBV and the Starbucks group by concluding the SMBV APA, in breach of Article 108(3) of the Treaty, which the Netherlands is required to recovery by virtue of Article 16 of Regulation (EU) 2015/1589 from SMBV and, if the latter fails to repay the full amount of the aid, from Starbucks Corporation for the amount of aid outstanding,

HAS ADOPTED THIS DECISION:

Article 1

The advanced pricing arrangement entered into by the Netherlands on 28 April 2008 with Starbucks Manufacturing EMEA B.V., which enables the latter to determine its corporate income tax liability in the Netherlands on a yearly basis for a period of 10 years, constitutes aid within the meaning of Article 107(1) of the Treaty on the Functioning of the European Union that is incompatible with the internal market and that was unlawfully put into effect by the Netherlands in breach of Article 108(3) of the Treaty.

Article 2

(1)   The Netherlands shall recover the incompatible and unlawful aid referred to in Article 1 from Starbucks Manufacturing EMEA B.V.

(2)   Any sums that remain unrecoverable from Starbucks Manufacturing EMEA B.V., following the recovery described in the preceding paragraph, shall be recovered from Starbucks Corporation.

(3)   The sums to be recovered shall bear interest from the date on which they were put at the disposal of the beneficiaries until their actual recovery.

(4)   The interest shall be calculated on a compound basis in accordance with Chapter V of Regulation (EC) No 794/2004.

Article 3

(1)   Recovery of the aid granted referred to in Article 1 shall be immediate and effective.

(2)   The Netherlands shall ensure that this Decision is implemented within four months following the date of notification of this Decision.

Article 4

(1)   Within two months following notification of this decision, the Netherlands shall submit information regarding the methodology used to calculate the exact amount of aid.

(2)   The Netherlands shall keep the Commission informed of the progress of the national measures taken to implement this Decision until recovery of the aid granted referred to in Article 1 has been completed. It shall immediately submit, on simple request by the Commission, information on the measures already taken and planned to comply with this Decision.

Article 5

This Decision is addressed to The Netherlands.

Done at Brussels, 21 October 2015.

For the Commission

Margrethe VESTAGER

Member of the Commission

(1)  

OJ C 460, 19.12.2014, p. 11

.

(2)  Throughout the decision, the terms ‘tax ruling’ and ‘APA’ are used synonymously.

(

*1

)

  Parts of this text have been hidden so as not to divulge confidential information; those parts are enclosed in square brackets. Profit margins and mark-ups of SCTC concerning the green coffee beans are rounded to the closest multiple of 3 %.

(3)  A previous tax ruling concluded between the Dutch tax administration and Starbucks Coffee BV and SMBV in 2001 was also submitted as part of the supporting documents.

(4)  That letter was sent to the Netherlands in the English language on 7 March 2014, followed by a version in Dutch of this same letter sent on 14 March 2014.

(5)  See footnote 1.

(6)  Council Regulation (EC) No 659/1999 of 22 March 1999 laying down detailed rules for the application of Article 108 of the treaty on the functioning of the European Union (

OJ L 83, 27.3.1999, p. 1

). Regulation (EC) No 659/1999 is repealed by Council Regulation (EU) 2015/1589 of 13 July 2015 laying down detailed rules for the application of Article 108 of the Treaty on the Functioning of the European Union (

OJ L 248, 24.9.2015, p. 9

), with effect from 14 October 2015. All the procedural steps taken during the course of the proceedings were adopted under Regulation (EC) No 659/1999. Any reference to Regulation (EC) No 659/1999 may be construed as a reference to Regulation (EU) 2015/1589 and should be read in accordance with the correlation table in Annex II to the latter regulation.

(7)  C (2015) 862, 12.2.2015.

(8)  Starbucks US refers to all companies that are members of the Starbucks group that are resident in the United States of America.

(9)  This description is based on Starbucks' 2014 Annual Report, p. 2.

(10)  See recital 20 of the Opening Decision for more detailed information on the Starbucks Group.

(11)  The SMBV APA states that after the sixth accounting year, i.e. end 2013, a half-term check will be made to see whether all the facts and circumstances have remained the same.

(12)  From the original Dutch: ‘[

SMBV] wordt geacht een arm's length vergoeding te ontvangen voor haar activiteiten zoals beschreven [in het transfer pricing report] indien de operationele marge [9-12] % van de relevante kostgrondslag bedraagt

.’

(13)  From the original Dutch: ‘

De jaarlijks door SMBV aan Alki LP te betalen royalty wordt aan het einde van het jaar vastgesteld op het verschil tussen de gerealiseerde operationele winst met betrekking tot de productie en distributie functie als genoemd [in het transfer pricing report], vóór royalty uitgaven (“gerealiseerde operationele winst voor royalty uitgaven”) en de hierboven omschreven beloning van [9-12] % cost-plus. Deze royalty betaling is aftrekbaar voor de vennootschapsbelasting en is niet onderworpen aan Nederlandse dividendbelasting

.’ For an explanation of why the royalty payment is not subject to tax in the Netherlands or in the UK, see recital 28 of the Opening Decision.

(14)  Recitals 27 and 28 of the Opening Decision provide a full description of the legal structure of Starbucks as presented in the transfer pricing report.

(15)  These trademarks, technology and know-how include mainly the Starbucks Shop format and the Starbucks corporate identity.

(16)  See recital 71 for a description of the CUP method.

(17)  This definition is based on paragraphs 2.13 until 2.20 of the 2010 OECD TP Guidelines. A (external) Comparable Uncontrolled Price is applied if independent third parties under the same circumstances pay the same price for the same product or service as related parties. In this case, the intra-group royalty payments concern the payments made by the EMEA Starbucks-owned Developers to Starbucks Coffee BV.

(18)  Section II.G of the transfer pricing report.

(19)  Section II.E of the transfer pricing report.

(20)  Section III.B of the transfer pricing report.

(21)  SCTC buys those beans for the benefit of the entire Starbucks group worldwide and its independent licensees.

(22)  See footnote 19.

(23)  In 2011, SMBV employed approximately [70-80] people. According to the observations of Starbucks to the Opening Decision [40-50] persons were active in coffee roasting and [30-40] persons in logistic and administrative services.

(24)  According to information submitted by Starbucks to the Netherlands at the moment of the rulings request, SMBV and Starbucks Corporation entered into an agreement with [unaffiliated logistics company A] in 2003 under which [unaffiliated logistics company A] will provide inventory management services (supporting forecast provided by Starbucks), import coordination and export documentation services, vendor contracts administration and warehousing operations.

(25)  According to information submitted by Starbucks to the Netherlands at the moment of the rulings request, SMBV entered into a supply chain and coffee operations warehousing services agreement with [unaffiliated logistics company B] in 2004 under which [unaffiliated logistics company B] will perform warehouse services with respect to various Starbucks products including administrative and handling services, vehicle loading and unloading, inspections and storage.

(26)  According to the transfer pricing report, SMBV is responsible for the contractual relationship with [unaffiliated manufacturing company 1]. The contracts between [unaffiliated manufacturing company 1] and SMBV are however negotiated by Starbucks […] according to Starbucks.

(27)  Toll manufacturing is usually understood to mean an arrangement in which a company processes raw materials or semi-finished goods for another company.

(28)  This graph in not contained in the report, it was produced by the Commission in order to facilitate the reading of this Decision.

(29)  See recital 72 for a description of the TNMM.

(30)  Transactional differences refers to differences between the transactions concluded by the company for which the taxable basis is approximated through a transfer pricing method and the transactions concluded between independent companies used to determine the arm's length pricing. In the TNMM method this refers to the transactions concluded by comparable companies used to approximate an arm's length margin. Functional differences refers to differences between the functions performed by the company of which the taxable basis is approximated by a transfer pricing method and the functions performed by comparable companies used to approximate an arm's length margin.

(31)  The report refers in this context to paragraph 3.26 of the 1995 OECD TP Guidelines.

(32)  The Amadeus Database is a database of comparable financial information for public and private companies across Europe. It is maintained by Bureau van Dijk, a publisher of company information and business intelligence.

(33)  Automated searches in Amadeus resulted in a selection of 240 companies; additional financial selection criteria applied by the tax advisor eliminated 88 companies, reducing the sample to 152 potentially comparable companies.

(34)  The main elimination criteria were that the companies were engaged in unrelated functions (i.e. distribution, repair, etc.), producing unrelated products (i.e. candies, other food products, etc.) or that the companies belonged to a group. The 20 companies accepted for comparison purposes in the transfer pricing report are presented in appendices to that report and reproduced in Table 5 of the Opening Decision.

(35)  The tax advisor uses the term mark-up although the calculation of the mark-up is based on the operating profit of the comparable companies divided by a cost basis and not on gross profit.

(36)  In the case of SMBV, the operating costs represent a small fraction of the total costs. Therefore, although the percentage of mark-up applied is higher, the resulting estimate of tax base is lower.

(37)  Tax administrations and legislators are aware of this problem and tax legislation generally allows the tax administration to correct tax declarations of integrated companies that incorrectly apply transfer prices, by substituting prices which correspond to a reliable approximation of those agreed to by independent companies negotiating under comparable circumstances at arm's length.

(38)  Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations, OECD, 13 July 1995.

(39)  Recommendation C(95)126/FINAL of 13 July 1995.

(40)  Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations, OECD, 22 July 2010.

(41)  For the 1995 OECD TP Guidelines, see paragraph 3.43, which contains a comparable description.

(42)  Paragraphs 1.45 to 1.48 of the 1995 OECD TP Guidelines and paragraphs 3.55 to 3.62 of the 2010 OECD TP Guidelines.

(43)  Paragraph 1.48 of the 1995 OECD TP Guidelines.

(44)  For completeness it is noted that a portion of the labour costs can be included in COGS, when it is directly associated with the production.

(45)  See recital 56 of the Opening Decision.

(46)  In Table 1, EBITDA stands for the conventional acronym of ‘earnings before interest, taxes, depreciation and amortisation’ and EBT stands for the conventional acronym of ‘earnings before tax’.

(47)  Transfer Pricing Decree 2001 (

Besluit verrekenprijzen

) of 30 March 2001, IFZ2001/295M. The Decree was replaced in 2013 (together with the Decree of 21 August 2004, IFZ 2004/680M which Decree supplemented the 2001 Decree and amended it with respect to certain points. However, the 2004 Decree is not relevant in this case) by the Decree of 14 November 2013, IFZ 2013/184M, International Tax Law. Transfer pricing method, application of the arm's length principle and the Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations. As the ruling dates from 2007, the Decision will refer to the 2001 Transfer Pricing Decree.

(48)  As in 2001 article 8b was not yet codified in the CIT, transfer pricing adjustments, both for personal as for corporate income tax purposes, were made based on Article 3.8 of the Income Tax Act 2001 (which through article 8 CIT also applied to corporate taxpayers).

(49)  For example, at the end of […] this fee represented [1,5-2] % of the price SCTC expected to pay for the purchase of green coffee beans and [0,5-1] % of the price charged to the […] by SMBV. The [certification programme] designation was used and paid for by Starbucks since […].

(

*2

)

  Additionally provided, see Table 8.

(50)  In detail, the pre-tax profit before the royalty payment is equal to ‘Sales’ minus ‘Direct Costs of Sales’ (which represent the costs of raw material consumed in the production process), minus ‘General and Administrative expenses’, minus ‘Foreign currency exchange’, plus ‘Interest income’, minus ‘Interest expense’ in Figure 2. For example, for the year 2010/2011, the pre-tax profit before payment of royalty would be equal to EUR 13 783 458. In order to lower the pre-tax to the level agreed in the SMBV APA of around [9-12] % of agreed costs, a tax-deductible royalty of EUR 12 352 838 is paid out to Alki LP, as recorded in the position ‘Other expenses’.

(51)  Under the previous ruling, SMBV's tax base was calculated as 8 % of the costs taken into account.

(52)  These include the Starbucks figurative mark and the Starbucks business plan.

(53)  These include the furnishings and fittings, the exterior and the perception of the Shops, so-called systems relating to the exterior/the ‘front’ of the Shops and so-called systems for the set-up/‘rear’ of the Shops.

(54)  These include the ‘brand curves’ and the formula for the coffee mixes.

(55)  Comments of the Netherlands to the Opening Decision, p. 25.

(56)  Comments of the Netherlands to the Opening Decision, p. 27.

(57)  See recitals 145 and 146.

(58)  A company incorporated under the laws of the State of Washington, US (General Partner).

(59)  A company incorporated under the laws of the State of Washington, US (Limited Partner).

(60)  See recital 106.

(61)  For the period 2008-2014, Alki LP received in total EUR [400-500] million in licence fees from Starbucks Coffee BV & SMBV. Out of this EUR [400-500] million, EUR [300-400] million ([80-85] %) was received from Starbucks Coffee BV and EUR [60-70] million ([15-20] %) from SMBV.

(62)  According to Starbucks, this payment from Starbucks US to Alki LP concerns a global cost adjustment intended to compensate Alki LP for the lower licence fees received from SMBV as a result of Starbucks' pricing policy for roasted coffee on a global basis.

(63)  This ‘arm's-length return’ is calculated under US transfer pricing rules and is determined for all the activities of Starbucks Coffee BV and SMBV.

(64)  Starbucks could not provide a breakdown of the buy-in payments for the brand or for the business format for the years 2005 and 2006. There are also some discrepancies between the licence fee and buy-in payments by [CV 1] as provided in the overview by Starbucks and the profit & loss statements of [CV 1]. Starbucks has not been able to establish the exact cause of these discrepancies.

(65)  No buy-in payments were made before 2005 to Starbucks Corporation as [CV 1] was loss-making until 2005.

(66)  However, this amount would on average be […] lower if the income received from Starbucks Corporation would not be taken into account in the calculation of the residual profit pool.

(67)  However, this amount would on average be […] lower if the income received from Starbucks Corporation would not be taken into account in the calculation of the residual profit pool.

(68)  See recital 146.

(

*3

)

  Including an extraordinary post tax profit of [1-10 million] CHF and a loss of [800-900 thousand] CHF in 2006 and 2007 respectively.

(69)  Value in 2014, calculated by comparison of the total estimated costs of the C.A.F.E. Practices Programme and Farmer Support Centres Programme with the COGS of SCTC, adjusted by the USD:CHF exchange rate.

(70)  Gross margin is gross profit (i.e. Sales minus COGS) divided by Sales, see recital 84.

(71)  FOB refers to Free on Board.

(72)  See paragraph 2.141 of the 2010 OECD TP Guidelines: ‘Where comparable uncontrolled transactions of sufficient reliability are lacking to support the division of the combined profits, consideration should be given to internal data, which may provide a reliable means of establishing or testing the arm's length nature of the division of profits. The types of such internal data that are relevant will depend on the facts and circumstances of the case and should satisfy the conditions outlined in this section and in particular at points 2.116-2.117 and 2.132. They will frequently be extracted from the taxpayers’ cost accounting or financial accounting'.

(73)  The amount reported under ‘Deductions’ is taken as presenting operating expenses, as it is mainly constituted of labour costs and the cost of depreciation and repair.

(74)  Under this agreement, Starbucks Coffee BV has the right to grant third parties a licence to develop, own and operate Starbucks stores operating under the Trademarks and using the Technology and Know-How.

(75)  According to the Supply Agreement, […].

(76)  Starbucks specified that the Supply Agreements are contracted by Starbucks […] and that these are standardised agreements where the underlying terms and conditions are determined by Starbucks […], cf. observations of Starbucks, recital 6.14.

(77)  […].

(78)  […].

(79)  

Ibid

, […].

(80)  […].

(81)  […].

(82)  […].

(83)  This estimation is based on the sum of royalties paid to Alki LP over the period 2008-2014, divided by EUR [900-1 000] of revenues from all products, which are recorded in Table 5 under the categories ‘REV READY TO DRINK’, ‘REV PACKAGED COFFEE’, ‘REV SINGLE SERVE PODS-COFFEE’, ‘REV SINGLE SERVE PODS-VERISMO’, ‘REV SOLUBLE COFFEE’, ‘REV PACKAGED TEA’, ‘REV BLENDED BEVG MIX’, ‘REV FRAP COFFEE BASE’ and ‘REV EXTRACT’. Although Starbucks has indicated in their submissions that ‘REV SINGLE SERVE PODS-COFFEE’, ‘REV SINGLE SERVE PODS-VERISMO’, ‘REV SOLUBLE COFFEE’, ‘REV FRAP COFFEE BASE’ and ‘REV EXTRACT’ should also be classified as administrative and logistics support revenue as the underlying revenue invoiced to Developers represents the value created by third parties rather than SMBV and that in any case, this only represents according to Starbucks a small portion of the total roasting output of SMBV. Additionally categories ‘REV PACKAGED TEA’, ‘REV BLENDED BEVG MIX’ have not been indicated to contain a coffee by Starbucks.

(84)  See recital 140 for the financial data of the company.

(85)  However, paragraph 4.9 of the 1995 OECD TP Guidelines (and also paragraph 4.9 of the 2010 OECD TP Guidelines) refer to situations where ‘because of the complexity of the facts to be evaluated, even the best-intentioned taxpayer can make an honest mistake’.

(86)  They refer in this context also to a non-binding Commission Communication on the work of the EU Joint Transfer Pricing Forum (cf. COM(2007)71) where it is states that ‘tax administrations should make every effort to keep the burden of the evaluation to a minimum by requiring only pertinent information’.

(87)  Commission Decision 2009/809/EC of 8 July 2009 on the groepsrentebox scheme which the Netherlands is planning to implement (C4/07(ex N 465/06)) (

OJ L 288, 4.11.2009, p. 26

).

(88)  Relying again on Decision 2009/809/EC.

(89)  According to the report, a margin on COGS is only warranted in cases where the service provider is responsible for areas such as product development/R & D, strategic supply chain planning, strategic procurement and revenue generating activities (i.e. marketing, pricing and demand generation) and also incurs and manages the associated risks. According to Starbucks, SMBV does not perform these functions. The report further states that functions such as day-to-day planning, non-strategic procurement, conversion, order processing, warehousing/logistics and invoicing without commercial risks in relation to the products involved only warrant a margin on the service provider's own operating costs.

(90)  In the case of Company Y, the mark-up is also only applied on operational expenses.

(91)  According to the Netherlands, although most of the third parties perform other specific activities apart from the roasting activity, carry entrepreneurial risk or own self developed IP, the fees that those third parties pay to Starbucks (either in the form of a higher purchase price for coffee beans or in the form of a royalty on revenue) are at similar levels as those of SMBV. However, according to the Netherlands, lower fees or prices could have been expected because of the more complex profile of some of the third parties.

(92)  Case C-399/08 P

Commission

v

Deutsche Post

ECLI:EU:C:2010:481, paragraph 38 and the case-law cited.

(93)  Case C-399/08 P

Commission

v

Deutsche Post

ECLI:EU:C:2010:481, paragraph 39 and the case-law cited.

(94)  See Joined Cases C-106/09 P and C-107/09 P

Commission

v

Government of Gibraltar and United Kingdom

, ECLI:EU:C:2011:732, paragraph 72 and the case-law cited.

(95)  Case 730/79

Phillip Morris

ECLI:EU:C:1980:209, paragraph 11. Joined Cases T-298/97, T-312/97, etc.

Alzetta

ECLI:EU:T:2000:151, paragraph 80.

(96)  C-172/03

Heiser

ECLI:EU:C:2005:130, paragraph 40.

(97)  Joined Cases C-78/08 to C-80/08

Paint Graphos

, ECLI:EU:C:2009:417.

(98)  Joined Cases C-78/08 to C-80/08 Paint Graphos ECLI:EU:C:2009:417, paragraph 65.

(99)  See also C-78/08 to C-80/08

Paint Graphos

ECLI:EU:C:2009:417, paragraph 50.

(100)  Article 2 CIT 1969.

(101)  The standard CIT rate is 25 %. There are two taxable income brackets. A lower rate of 20 % applies to the first income bracket, for taxable income up to EUR 200 000.

(102)  Article 3 CIT 1969 and Chapter III of the CIT 1969, according to which non-resident companies are taxed on: (1) business income derived from a Netherlands permanent establishment or permanent representative, (2) income and capital gains derived from immovable property located in the Netherlands, (3) income and capital gains from rights related to the exploration for or exploitation of natural resources situated in the Netherlands or the Netherlands part of the continental shelf, (4) all remuneration derived from a directorship of a resident entity, (5) income from rights to the profits of an enterprise (bonds and shares excluded) the management of which is situated in the Netherlands and (6) income and capital gains, from debt claims related to a substantial shareholding.

(103)  Article 2, paragraph 4, CIT 1969.

(104)  From the original Dutch: ‘

De in een kalenderjaar genoten winst wordt bepaald volgens goed koopmansgebruik, met inachtneming van een bestendige gedragslijn die onafhankelijk is van de vermoedelijke uitkomst. De bestendige gedragslijn kan alleen worden gewijzigd indien goed koopmansgebruik dit rechtvaardigt

.’

(105)  Chapter II and III of the CIT 1969.

(106)  In general, all undertakings having an income are considered to be in a similar legal and factual situation from the perspective of direct company taxation.

(107)  Decision 2009/809/EC.

(108)  Observations of SMBV to the Opening Decision, Section 2.20.

(109)  C-138/09

Todaro Nunziatina & C.

ECLI:EU:C:2010:291, paragraph 21.

(110)  The scheme provided that the positive balance between interest received on group loans and interest paid in the context of intra-group financing transactions was not taxed at the standard corporate tax rate of 25,5 % at the time but taxed in a ‘group interest box’ at the rate of 5 %.

(111)  

Groepsrentebox

decision, recital 85.

(112)  

Groepsrentebox

decision, recital 101.

(113)  

Groepsrentebox

decision, recital 107.

(114)  See recitals 148 to 150.

(115)  Cf. the answer of Melitta at recitals 207 to 209.

(116)  It should be noted, however, that even if SMBV were engaged in financing transactions, the Commission would not consider the

Groepsrentebox

decision as applicable given the different objective pursued by the group interest scheme on the one hand and the determination of SMBV's taxable base on the other hand.

(117)  The Netherlands' observations under Section 3.2 to the Opening Decision.

(118)  Case C-15/14 P

Commission

v

MOL

ECLI:EU:C:2015:362, paragraph 60; See also, Case T-385/12

Orange v Commission

ECLI:EU:T:2015:117.

(119)  Case C-39/94

SFEI and Others

ECLI:EU:C:1996:285, paragraph 60; Case C-342/96

Spain

v

Commission

ECLI:EU:C:1999:210, paragraph 41.

(120)  Case 173/73

Italy

v

Commission

ECLI:EU:C:1974:71, paragraph 13.

(121)  See, for instance, Case C-387/92

Banco Exterior de Espana

, ECLI:EU:C:1994:100.

(122)  Joined Cases C-182/03 and C-217/03

Belgium and Forum 187 ASBL

v

Commission

, ECLI:EU:C:2006:416.

(123)  Joined Cases C-182/03 and C-217/03

Belgium and Forum 187 ASBL

v

Commission

, ECLI:EU:C:2006:416.

(124)  Commission Decision 2003/757/EC of 17 February 2003 on the aid scheme implemented by Belgium for coordination centres established in Belgium (

OJ L 282, 30.10.2003, p. 25

).

(125)  Joined Cases C-182/03 and C-217/03

Belgium and Forum 187 ASBL

v

Commission

, ECLI:EU:C:2006:416, paragraphs 96 and 97.

(126)  Joined Cases C-182/03 and C-217/03

Belgium and Forum 187 ASBL

v

Commission

, ECLI:EU:C:2006:416, paragraph 81.

(127)  See recital 186.

(128)  See recital 162 and 191.

(129)  Referred to as the ‘first adjustment’ in recitals 49, 97 and 100 of the Opening Decision.

(130)  Referred to as the ‘second adjustment’ in recital 52 of the Opening Decision. Whereas the first two choices are discussed in the OECD Guidelines, the subsequent choice of applying a working capital adjustment is not covered by the OECD Guidelines in the manner applied by the tax advisor in this case.

(131)  A controlled transaction is a transaction between two enterprises that are associated enterprises with respect to each other, while an uncontrolled transaction is a transaction between enterprises that are independent enterprises with respect to each other.

(132)  Section 9.2.3.2.

(133)  Section 9.2.3.3 (a) to (e).

(134)  Section 9.2.3.3 (f).

(135)  Section 9.2.3.4.

(136)  Section 9.2.3.5.

(137)  See, also, Paragraph 1.6 of the 1995 OECD TP Guidelines.

(138)  See recital 234.

(139)  The focus on transactions is again stated in paragraph 1.33 of the 2010 OECD TP Guidelines clarifying that ‘

application of the arm's length principle is generally based on a comparison of the conditions in a controlled transaction with the conditions in transactions between independent enterprises

.’

(140)  See recitals 68 and 69.

(141)  See recital 90.

(142)  See, in particular, recitals 120 and 122 of the Opening Decision.

(143)  As explained in recital 102.

(144)  See recital 115 of the Opening Decision.

(145)  See paragraph 6.16 of the 1995 OECD TP Guidelines which presents the same consideration.

(146)  For illustration, an analysis using RoyaltyStat, at 2Q 2015, shows that out of the 168 agreements available through the database across sector whereby only technology was licenced, the median value of the royalty was 5 % of sales (based on 143 of these agreements where the licence fee was determined as a percentage of the value of sales rather than amount paid per unit sold). Among all the contracts available through the RoyaltyStat database, no contract was identified whereby remuneration was paid for coffee roasting technology licenced on the market. Such technology was only licenced out in certain instances in combination with trademarks.

(147)  See also paragraph 1.17 of the 1995 OECD TP Guidelines.

(148)  This is, however, not the case for [unaffiliated manufacturing company 10].

(149)  See Table 3.

(150)  See recital 209.

(151)  See recitals 211 to 212.

(152)  See recitals 203 to 205.

(153)  See recitals 148 to 150 and 300 to 304.

(154)  See recital 142.

(155)  Starbucks indicated that the revenues under the description ‘REV PACKAGED COFFEE’ relate to SMBV's roasting and packaging function and that revenues from other items which could cover products processed partially on the basis of coffee roasted by SMBV represent a small portion of the total roasting output of SMBV, see recital 94.

(156)  In fact, according to recital 97, SMBV keeps a margin on non-coffee products covering all the […] costs of […]. This margin is recorded on the reselling function.

(157)  See also paragraph 3.64 of the 2010 OECD TP Guidelines, according to which: ‘An independent enterprise would not continue loss-generating activities unless it had reasonable expectations of future profits.’ and paragraph 1.52 of the 1995 OECD TP Guidelines.

(158)  See recital 105.

(159)  See recital 170.

(160)  See recital 51.

(161)  In which the transfer pricing arrangement for SMBV is reflected in the calculation of the royalty that needs to be paid to Alki LP, see recital 170.

(162)  See also recital 87 of the Opening Decision.

(163)  These paragraphs are part of chapter 9 of the 2010 OECD TP Guidelines dealing with business restructuring but are nonetheless relevant because of the underlying principle stated in these paragraphs.

(164)  Paragraph 9.46 of the 2010 OECD TP Guidelines.

(165)  See recital 167.

(166)  Paragraph 9.23 and 9.26 of the 2010 OECD TP Guidelines.

(167)  Paragraph 9.29 of the 2010 OECD TP Guidelines.

(168)  Paragraph 9.23 of the 2010 OECD TP Guidelines.

(169)  See recital 106.

(170)  

Ibid

.

(171)  

Ibid

.

(172)  See recital 146.

(173)  Similar text can be found in paragraphs 1.36 to 1.41 of the 1995 OECD TP Guidelines.

(174)  See recital 107.

(175)  See recital 144.

(176)  See recital 112.

(177)  Starbucks Coffee BV payments to Alki LP are attributed to the same pool as payments by SMBV, the latter of which present a smaller part of revenues of Alki LP; see recital 110.

(178)  See recital 102.

(179)  See recital 201.

(180)  This interquartile range estimated using Starbucks observations is 4 % to 8,5 %, see recital 134.

(181)  See paragraph 3.42 of the 1995 OECD TP Guidelines, according to which ‘when profit margins of an independent enterprise are used, the profits attributable to the transactions of the independent enterprise must not be distorted by controlled transactions of that enterprise.’

(182)  As the gross margin from which it results would have been arm's length, see recital 128.

(183)  See recital 130.

(184)  See recital 131.

(185)  See recital 131, the 11 identified agreements are technology licensing agreements in relation to food and beverage products.

(186)  See recital 121.

(187)  See recital 96.

(188)  Additionally, SCTC seems to hold considerable amounts of excess cash based on figures in Table 4.

(189)  See recital 346.

(190)  See recital 315.

(191)  See recital 96.

(192)  See recital 102.

(193)  See recital 190.

(194)  The choice of the tested party is only necessary when using the cost plus, resale price or TNMM, see paragraph 3.18 of the 2010 OECD TP Guidelines.

(195)  This requirement is also to be found in paragraph 3.43 of the 1995 OECD TP Guidelines.

(196)  This is also the case of paragraph 2.59 and 9.79 of the 2010 OECD TP Guidelines.

(197)  See paragraph 2.3 of the 2010 OECD TP Guidelines and paragraph 3.49 of the 1995 OECD TP Guidelines.

(198)  According to paragraph 3.42 of the 1995 OECD TP Guidelines ‘it would be inappropriate to apply the transactional net margin method on a company-wide basis if the company engages in a variety of different controlled transactions that cannot be appropriately compared on an aggregate basis with those of an independent enterprise.’

(199)  See recitals 204 to 212.

(200)  See recital 98.

(201)  See recital 212.

(202)  See recital 98.

(203)  Section 9.2.3.2.

(204)  See recital 49.

(205)  See recitals 79 to 96 of the Opening Decision.

(206)  Starbucks indicated that the revenues under the description ‘REV PACKAGED COFFEE’ relate to SMBV's roasting and packaging function. Revenues from other items which could cover products processed partially on the basis of coffee roasted by SMBV represents according to Starbucks a small portion of the total roasting output of SMBV, see recital 94.

(207)  See recitals 166 and 190.

(208)  See recital 97.

(209)  See recital 315.

(210)  See recitals 342 et seq.

(211)  See recital 179.

(212)  See also paragraph 1.43 of the 2010 OECD TP Guidelines and paragraph 1.21 of the 1995 OECD TP Guidelines.

(213)  

Ibid

.

(214)  This is also in line with the example presented regarding distribution function in paragraph 3.48 of the 1995 OECD TP Guidelines.

(215)  From EUR 128 million in 2008 to EUR 350 million in 2014.

(216)  From EUR 20 million in 2008 to EUR 44 million in 2014.

(217)  See recital 325.

(218)  See recital 57.

(219)  Active companies or companies with unknown status were retained, in EU-15 countries, Iceland, Norway and Switzerland, which were not owned for 25 % or more by another company to ensure the independence criterion; these choices are identical to the tax advisor. However, a more recent period of observation was retained, as the tax advisor used data for 2001-2005 and the APA was concluded in 2008, three years after the last data period.

(220)  In the transfer pricing report the operating expense is arrived at by calculating the difference between total costs and what Bureau van Dijk operating Amadeus designated as ‘material costs’ which are meant to reflect COGS. ‘Material costs’ data was not available for many companies of the comparison.

(221)  See paragraph 3.57 of the 2010 OECD TP Guidelines.

(222)  If a five year period of 2003 to 2007 is chosen instead the resulting interquartile range would be identical of 1,5 % to 5,8 % with a median of 2,9 %. If the same five year period of 2001 to 2005, as in the transfer pricing report of Starbucks is chosen, the resulting interquartile range would be 1,8 % to 5,7 % with a median of 2,7 %. Finally, if a five year period is chosen ending 2006 instead of 2007, that is a period of 2002-2006, the interquartile range would be 1,7 % to 5,2 % with a median of 2,6 %.

(223)  See recital 139.

(224)  See recital 177.

(225)  See recitals 101 to 113 of the Opening Decision.

(226)  See recital 178.

(227)  This range results from the figures in Table 5 of the Opening Decision.

(228)  See recital 87.

(229)  Joined Cases C-78/08 to C-80/08 Paint Graphos and others ECLI:EU:C:2009:417, paragraph 69.

(230)  Case C-170/83

Hydrotherm

ECLI:EU:C:1984:271, paragraph 11. See also Case T-137/02

Pollmeier Malchow

v

Commission

ECLI:EU:T:2004:304, paragraph 50.

(231)  Case C-480/09 P

Acea Electrabel Produzione SpA

v

Commission

ECLI:EU:C:2010:787 paragraphs 47 to 55; Case C-222/04

Cassa di Risparmio di Firenze SpA and Others

ECLI:EU:C:2006:8, paragraph 112.

(232)  The corporate structure of the Starbucks group is explained in more detail in recital 27 and Figure 1 of the Opening Decision.

(233)  See, by analogy, Case 323/82

Intermills

ECLI:EU:C:1984:345: paragraph 11 ‘It is clear from the information supplied by the applicants themselves that following the restructuring both SA Intermills and the three manufacturing companies are controlled by the Walloon regional executive and that, following the transfer of the plant to the three newly constituted companies, SA Intermills continues to have an interest in those companies. It must therefore be accepted that, in spite of the fact that the three manufacturing companies each has a legal personality separate from the former SA Intermills, all those undertakings together form a single group, at least as far as the aid granted by the Belgian authorities is concerned […]’.

(234)  See recital 102.

(235)  Moreover, the SMBV APA's non-compliance with the arm's length principle has been further confirmed and exacerbated in the years following its conclusion. For instance, whereas in 2007, income from roasted beans presented only [15-20] % of the total revenues of SMBV, this further decreased in later years to […] % in 2013 and 2014. This is aligned with the fact that SMBV engaged in other activities, such as market research, implying significant expenses and other IP related activities, given the inclusion of intangible amortisation expenses as one of the largest operating expense items in 2014.

(236)  See paragraph 3.42 of the 1995 OECD TP Guidelines, according to which ‘when profit margins of an independent enterprise are used, the profits attributable to the transactions of the independent enterprise must not be distorted by controlled transactions of that enterprise.’

(237)  See Table 11.

(238)  The exceptions provided for in Article 107(2) of the Treaty concern aid of a social character granted to individual consumers, aid to make good the damage caused by natural disasters or exceptional occurrences and aid granted to certain areas of the Federal Republic of Germany, none of which apply in the present case.

(239)  Reference is made to Regulation (EU) 2015/1589 with respect to the unlawfulness of the aid and the recovery obligation.

(240)  Commission Regulation (EC) No 794/2004 of 21 April 2004 implementing Council Regulation (EU) 2015/1589 laying down detailed rules for the application of Article 108 of the Treaty on the Functioning of the European Union (

OJ L 140, 30.4.2004, p. 1

).

(241)  Joined Cases T-116/01 and T-118/01

P&O European Ferries (Vizcaya) SA and Diputacion Floral de Vizcaya

v

Commission

ECLI:EU:T:2003:217, paragraph 115.

(242)  Case T-290/97

Mehibas Dordtselaan

v

Commission

ECLI:EU:T:2000:8, paragraph 59 and judgement in cases C-182/03 and C-217/03

Belgium and Forum 187 ASBL

v

Commission

ECLI:EU:C:2006:416, paragraph 147.

(243)  See Case C-441/06

Commission

v

France

ECLI:EU:C:2007:616, paragraph 29 and the case-law cited.