BNZ INVESTMENTS LIMITED & ORS V THE COMMISSIONER OF INLAND REVENUE HC WN CIV 2004-485-1059

BNZ INVESTMENTS LIMITED & ORS V THE COMMISSIONER OF INLAND REVENUE HC WN CIV 2004-485-1059 15 July 2009IN THE HIGH COURT OF NEW ZEALAND WELLINGTON REGISTRY CIV 2004-485-1059 CIV 2005-485-1045 CIV 2006-485-1028 CIV 2006-485-2084 CIV 2008-485-1056IN THE MATTER OF The Tax Administration Act 1994 and the Income Tax Act 1994 BETWEEN BNZ INVESTMENTS LIMITED & ORS Plaintiffs AND THE COMMISSIONER OF INLAND REVENUE Defendant Hearing: 16-20, 23-27, 30-31 March; 1-3, 6-9, 14-17, 20-22, 27-30 April; 1, 4- 7, 11-13, 18-22, 25-27 May; 4-5, 11-12 June 2009 Counsel: A R Galbraith QC, A S Butler, A R Barker, R J Hollyman and T R Stewart for the Plaintiffs B W F Brown QC, M T Scholtens QC, E J Norris, P J Cassidy and D K Lemmon for the Defendant Judgment: 15 July 2009JUDGMENT OF WILD JCONTENTS Introduction........................................................................................................................................[1] The plaintiffs ....................................................................................................................................[10] Background to these transactions.........................................................................................................Structured finance transactions generally ................................................................................... [11] Tax driven transactions................................................................................................................ [23] Repos............................................................................................................................................ [27] The time span of the BNZ transactions ........................................................................................ [36] The ruled transactions ................................................................................................................. [38]The six transactions in issue..................................................................................................................How they came about................................................................................................................... [46] The detail of the transactions....................................................................................................... [56]Detailed analysis of the CSFB transaction...........................................................................................Negotiation of the CSFB transaction ........................................................................................... [58] Structure of the transaction.......................................................................................................... [69] Incorporation and funding of BNZIS2 ......................................................................................... [70] Establishment of the Trust............................................................................................................ [71] Initial unit holders........................................................................................................................ [73] Investment of CSFB Trust property.............................................................................................. [74] Repurchase of Class A Unit ......................................................................................................... [75] Interest rate swap......................................................................................................................... [76] Guarantees, indemnities and credit support ................................................................................ [78] Letter of credit.............................................................................................................................. [79] Other documentation ................................................................................................................... [80] Still further documentation within the Credit Suisse Group ........................................................ [81] Quarterly cash flows .................................................................................................................... [82] BNZ's returns from the transaction ............................................................................................. [84] CSFB's returns from the transaction ........................................................................................... [87]Differences between CSFB and the other five transactions .........................................................[89] The US tax treatment of the three US transactions ......................................................................[91] The UK treatment of the three UK transactions...........................................................................[99] Foreign tax credit transactions .....................................................................................................[101] The law....................................................................................................................................................The deduction provisions ........................................................................................................... [107] Subvention payments and loss transfers..................................................................................... [108] Foreign tax credit ...................................................................................................................... [109] The conduit regime .................................................................................................................... [110] The general anti-avoidance rules (GAARs) ............................................................................... [111] Ben Nevis and Glenharrow ........................................................................................................ [114] Step 1 – The taxpayer's use of the specific provision(s) ............................................................ [118] Step 2 – The taxpayer's use of the specific provision(s) in the overall arrangement................. [124]Step 1 of Ben Nevis.........................................................................................................................[139] The deductibility of the GAF/GPF (Issue 3)................................................................................[143]BNZ's primary argument ........................................................................................................... [145] Alternative argument.................................................................................................................. [162]Step 2 of Ben Nevis................................................................................................................................. The legislative policies of the specific provisions.................................................................................Introduction............................................................................................................................... [173]The deductibility of expenditure in the transactions......................................................................... Introduction ............................................................................................................................... [176] The legislation............................................................................................................................ [178] Opening contentions .................................................................................................................. [184] Summary .................................................................................................................................... [189]The FTC regime (utilised by the Gen Re 1 transaction) .......................................................... [191]The conduit regime ............................................................................................................................ Background ................................................................................................................................ [206] Legislative policies underlying the relevant provisions of the conduit regime .......................... [217] Summary .................................................................................................................................... [243]Redeemable preference shares (RPS) ..........................................................................................[244]Factual analysis of the transactions..............................................................................................[253]"Off the shelf" nature of the transactions................................................................................ [254]How common were transactions such as these? ...................................................................... [260]Level of disclosure..................................................................................................................... [271]Complexity................................................................................................................................. [281]The benefit split ......................................................................................................................... [285]The distribution rate.................................................................................................................. [292]The GPF ............................................................................................................................................. Introduction ............................................................................................................................... [299] Was the GPF a contrivance? ..................................................................................................... [306] Was the pricing of the GPF within market parameters?............................................................ [331]The interest rate swap ............................................................................................................... [360]The formula............................................................................................................................... [405]The sub-LIBOR funding for the counterparty ......................................................................... [420]Transactions profitable at 'market' pricing ............................................................................. [428]Early termination ...................................................................................................................... [431]Use of tax capacity and ETR..................................................................................................... [437] Tax capacity ............................................................................................................................... [438] ETR ............................................................................................................................................ [448]Transactions pre-tax negative................................................................................................... [456]The transactions "viewed in a commercially and economically realistic way" ...................... [464]Conclusions.............................................................................................................................................Deductibility of the transaction expenses ................................................................................. [478]The foreign tax credit claimed in Gen Re 1 ............................................................................. [484]Conduit relief............................................................................................................................. [488]The interest rate swap ............................................................................................................... [498]The GAFs/GPFs........................................................................................................................ [504]Commerciality ........................................................................................................................... [512]Economic effect......................................................................................................................... [520]Does s BG 1 catch the transactions? (Issue 1) ............................................................................[526] Is the way in which the Commissioner counteracted the tax advantage correct? (Issue 2) ...[527] Result ..............................................................................................................................................[542]Introduction[1] Is each of six similar structured finance transactions entered into by the plaintiffs (the BNZ) a 'tax avoidance arrangement' void under s BG 1 Income Tax Act 1994? [2] That is the primary issue in these five consolidated proceedings brought by the BNZ against the Commissioner, challenging his assessments issued after he voided each of the transactions pursuant to s BG 1. The BNZ claims the transactions are not caught by s BG 1. [3] A second issue, arising only if s BG 1 applies, is the correctness of the way in which the Commissioner has, pursuant to s GB 1, counteracted the tax advantage obtained by the BNZ under the transactions. The Commissioner disallowed the deductions claimed by the BNZ, as its costs of the transactions. The BNZ claims the deductions should be disallowed only to the extent they are excessive or 'over- market'. [4] A third, and perhaps strictly antecedent, issue is whether the guarantee arrangement fee (GAF) or guarantee procurement fee (GPF) charged in each transaction is properly deductible under s BD 2. [5] The transactions are so-called 'repo' deals: the BNZ made an equity investment in an overseas entity on terms requiring the overseas counterparty to repurchase that investment when the transaction terminated. The transactions were structured to enable the BNZ to deduct its expenses of earning the income yielded by its investment, while receiving that income free of tax. In the case of the first transaction, that tax relief resulted from a credit for foreign tax paid. The BNZ's income from the five subsequent transactions was relieved of tax by the conduit regime. That domestic tax 'asymmetry' – tax deductible costs earning tax exempted income - made the transactions highly profitable for the BNZ.[6] The BNZ contends each of these transactions involved real obligations, notably those resulting from the BNZ raising $500 million on the New Zealand money market and advancing that to the counterparty upon a repo obligation. Further, the BNZ's case is that the transactions made legitimate use both of cross- border tax arbitrage and the domestic tax 'asymmetry' just described. Cross-border tax arbitrage refers to the different tax treatment of the transaction in New Zealand and the foreign counterparty's jurisdiction. New Zealand tax law treated the transactions as equity investments, the counterparties' jurisdictions (the United States of America for the first three transactions; the United Kingdom for the later three) as secured loans. That enabled the counterparties to deduct, as interest, the distribution they made which the BNZ received free of tax in New Zealand. The BNZ submits that the evidence establishes that "tax driven structures and tax arbitrage are common and accepted elements of international finance". The Bank says the central or critical question in the case is the appropriate approach in law to the asymmetry between deductible expenditure and tax relieved income around which these transactions or arrangements were structured. [7] The essential bases on which the Commissioner asserts s BG 1 catches the transactions are: a) They substantially altered the incidence of tax for the BNZ. Indeed, that was their only purpose or effect. It certainly was not a merely incidental purpose or effect. The Commissioner adopted the description of one of his witnesses: A prime purpose of the profit-maximising actions of these transactions was to use the tax base to make money.b) They had no commercial purpose or rationale. Absent the tax benefits they generated, the transactions were loss-making, in that the BNZ provided funding to the counterparties at substantially less than its cost of funds. The Commissioner contended "the tax tail wagged the commercial dog".c) They were not within the scheme and purpose of the regimes they utilised to generate tax exempted income, the foreign tax credit (FTC) and conduit relief regimes respectively. In the case of the FTC regime utilised by the first (Gen Re 1) transaction, the transaction also did not comply with the applicable specific provisions. d) The principal deductible expenses claimed by the BNZ (the fixed rate it paid on an interest rate swap and the GAFs or GPFs) were contrived and artificial. e) The transactions were structured on a formulaic basis, which had the artificial consequence that, the higher the transaction costs, the higher the tax benefits they generated. [8] The six transactions in issue span eight income tax years between 1998 and 2005. Three further transactions, two of them earlier in time, also featured in the evidence. While those three further transactions have a similar structure, they have the distinguishing feature of being New Zealand tax positive. The BNZ obtained binding rulings from the Commissioner on each of those three transactions, which I will call 'the ruled transactions'. [9] Approximately $416 million of tax hinges on the outcome of these proceedings. Challenge proceedings brought by the Westpac Banking Corporation began in Auckland on 30 June. Proceedings brought by other New Zealand trading banks have yet to come on for trial. If occasional press reports are accurate, the total amount of tax in issue is over $1.5 billion.The plaintiffs[10] The Bank of New Zealand was deemed by the Bank of New Zealand Act 1988 to be a company limited by shares incorporated and registered under the Companies Act 1955. The BNZ is a wholly owned subsidiary of National Australia Bank Limited, registered in Australia, headquartered in Melbourne and listed on the Australian Stock Exchange. The other plaintiffs are all wholly owned subsidiaries ofthe Bank of New Zealand. I will continue to refer to the plaintiffs as 'the BNZ', unless there is a need for more specificity. The National Australia Bank or its other subsidiaries I will term 'the NAB'.Background to these transactionsStructured finance transactions generally[11] The transactions in issue are structured finance transactions. Such a transaction differs from a common form one, in that the design or structure is specific to the needs of the parties to the transaction. In his evidence for the Commissioner, Professor Evans at para 40 in his statement of evidence quoted a description from Culp, Christopher L., Structured Finance and Insurance: the ART of Managing Capital and risk, Wiley, 2006 which included: Structured financing solutions are more bespoke than standardised and tend to be deliberately designed to package cash flows in a way that tries to equate a highly specific supply of funds and risk transfer with an equally specific demand for those funds and risk transformation scenarios.[12] Professor Evans made the point that structured finance transactions need to be considered as a whole for their commercial rationale and economic effects. The reason, he explained, is that such transactions: can be conceptualised as a mosaic of individual elements – consisting of variables and institutional arrangements – which depends for its characteristics and utility on each element fitting its particular place. this makes it difficult to reasonably assess the commercial and economic effect of individual elements in isolation from the bundle of elements that constitute the transaction. (Primary Brief – 'PB' – para 42)Professor Evans is a well known, widely published and well respected New Zealand economist, whose research interests are in the economics of organisation and markets. He is Professor of Economics at Victoria University here in Wellington. [13] In his evidence for the BNZ, Mr Kyle (PB 2.48) described four types of transaction the BNZ's structured finance team had been involved in during the1990s. They include acquisition finance, project finance and leases. Mr Kyle was former Global Head of Structured Finance at NAB, based in London. Professor Evans said that transactions of the type described by Mr Kyle generally facilitated some real economic activity. Whilst they tended to utilise tax rules and capacity and shared the benefit of doing so amongst the parties, these economic and social costs "may be deemed to be outweighed by the lower cost of capital (the transaction) provides" (PB 45). [14] Professor Evans contrasted structured finance transactions between financial institutions, which can be motivated (and structured) to reduce tax capacity, sometimes by exploiting inter-jurisdictional differences and tax rules e.g. the tax arbitrage. Professor Evans said that this sort of structured finance transaction "will be indicated by unaltered risk profiles of the transacting parties; and by the presence of circular flows of payments between the parties to the transaction" (PB 46). [15] Mr Hooper, until March this year CEO of nabCapital, explained that the structured finance group at nabCapital concentrates on specialised financing transactions, usually with other financial institutions and occasionally large corporates (PB 3.27). He acknowledged that the term "structured finance" is much used in the financial community, and means different things to different people, but explained:At nabCapital, however, what we describe as "structured finance" has tended to focus on forms of financing that rely on the favoured treatment of that structuring for tax purposes. This may result from tax arbitrage opportunities that exist between two different jurisdictions, the use of a concessionary tax regime in particular jurisdictions, or differences in the tax status of the counterparties. (PB 3.29).[16] That emphasis, on structuring for tax advantage, also featured in Professor Rosenbloom's rather more general description:The term "structured financing transaction" is commonly used to denote a transaction developed to take advantage of tax, regulatory, or other provisions. (PB 3.14)Professor Rosenbloom was called by the BNZ. He is Visiting Professor of Taxation and Director of the International Tax Programme (a Master of Laws programme) at New York University School of Law. He has taught international tax law at the Harvard, Stanford, Columbia and Pennsylvania Law Schools, and taught or instructed in tax in a number of countries (New Zealand not included). When away from academia, Professor Rosenbloom is Co-Chairman of the law firm Kaplan & Drysdale. His tax practice since 1968 includes some four years with the US Treasury Department from 1977 to 1981, latterly as the Treasury's International Tax Counsel and Director of its Office of International Tax Affairs. [17] Further evidence for the BNZ about structured finance transactions was given by Mr Gross and Professor Schwarcz. Mr Gross considered all structured financing had the aim of increasing the relative marketability of the financing. The methods of achieving this included taking advantage of differing arbitrage opportunities within the regulatory, tax, financial and/or accounting areas. Mr Gross had 26 years experience in international banking, most of it with the Bank of America or its predecessor banks. [18] Professor Schwarcz identified key motivations for structured finance transactions. From the standpoint of the originator of the transaction, these motivations included achieving funding benefits and allocating risk to third parties. From an investor's viewpoint, the motivation was simpler: achieving a relatively high rate of return with little investment risk (PB 2.2, 2.4). Professor Schwarcz is a Professor of Law and Business at Duke University where, amongst other things, he teaches a course entitled "Structuring Commercial & Financial Transactions". His involvement in structured finance goes back to 1984, when he joined the international law firm Shearman & Sterling. Subsequently, as a partner, he headed the Structured Finance Practice Group in another international law firm, Kaye, Scholer, Fierman, Hays & Handler. His clients there included Citicorp. [19] The Commissioner also offered evidence about structured finance transactions. It was given primarily by Mr Stanton. Mr Stanton is a London based chartered accountant who began his professional career in London in 1975 with Arthur Andersen. In 1982 he joined Hill Samuel Group as Head of Group Tax. Hewas appointed a director in 1987, following which he transferred to the Group's bank to start up a structured finance team specialising in tax-based transactions. In 1993 he joined Robert Fleming & Co., as a Director and Head of Structured Finance. In 1998 he took on the additional role of Chief Operating Officer, Capital Markets, with responsibility for the Bank's equity capital markets and debt capital markets operations. Like Hill Samuel, Robert Fleming was a London based merchant banker and provider of financial services. Robert Fleming was acquired by Chase Manhatten in 2000, and a year later Chase merged with J P Morgan. Mr Stanton left the organisation at that stage and currently holds several non-executive directorships with companies quoted on the LSE. [20] As with Mr Hooper, Mr Stanton's evidence about structured finance began with an acknowledgement that the term has been, and still is, used in the banking industry to describe a number of different banking activities and is used in different contexts (PB 2.14). He said the same thing under cross-examination: "It's a term that's much used" (Transcript 3400). Mr Stanton indicated: I am going to use the term "structured finance" to refer to tax driven transactions which depend on the reduction of a party's tax liabilities, as this accords with how I have seen the term being used most often in practice. (PB 2.19)[21] Mr Stanton listed nine characteristics of structured finance transactions that would not be seen in most banking transactions:(1) Requirement for a tax liability; (2) Substantial financial effect; (3) Performance measured by reference to the size of tax reductions; (4) Involvement of specialist transaction arrangers; (5) Replicable transactions; (6) Limited product life; (7) Inter-bank transactions between structured finance teams; (8) Operation as a niche activity within banks; and (9) Absence of publicity.(PB 2.20)[22] Based on his review of the documents provided to him, Mr Stanton said that all those special characteristics were present in the six transactions in issue here (PB 2.52). He added that he had also reviewed documents provided by the Commissioner relating to a number of transactions undertaken by other New Zealand banks. He said those transactions "adopted a similar structure to the disputed transactions", and also had the nine special characteristics he had identified (PB 2.53-2.54). It is evident from pp 273-283 of Appendix 11 to Mr Stanton's statement of evidence, that the "other bank" transactions he reviewed were entered into by the ASB, ANZ, National (NBNZ) and Westpac. Rabobank also features, though to a minor extent. These are the 16 transactions shown on the diagram at [37]Tax driven transactions[23] Different witnesses defined a "tax driven" transaction in different ways. The description encompasses a transaction significantly motivated by the obtaining of a tax benefit. A transaction that would not have taken place but for the obtaining of a tax benefit is squarely within the description. The transactions in issue are admittedly tax driven transactions, in that the BNZ accepts it would not have entered into them but for the tax benefits they provided. [24] At [260] I address the question: how common were transactions such as the six in issue here? But this is the place to mention evidence given for the Commissioner by Mr Shay. Mr Shay is a tax partner with the law firm Ropes & Gray in Boston, Massachusetts. Before joining that firm in 1987, he was the international tax counsel for the United States Department of the Treasury. The principal focus of his practice is the US federal income taxation of cross border transactions and multi-national taxpayers. Mr Shay is also a lecturer at the Harvard Law School, teaching a course on the international aspects of US income taxation. He holds a JD and an MBA from the Columbia University Schools of Law and Business, respectively.[25] Mr Shay explained that tax driven (or "tax aggressive") transactions such as these were commonplace in the late 1990s and early 2000s when the United States was in a cycle of tax shelters. Four factors had combined to end this cycle. The first was the requirement of the United States Financial Accounting Standards Board that financial statements both disclose and make provision against an uncertain tax position. The second was Sarbanes-Oxley, a US Federal law enacted on 30 July 2002, a reaction to a number of large corporate and accounting scandals, primarily Enron. Broadly, Sarbanes-Oxley requires greater attention to proper corporate governance and financial reporting standards and procedures by the boards and management of US public companies, and public accounting firms acting for them. The third was the introduction in US federal tax legislation of a disclosure regime for reportable transactions. I lack detail about this, but it appears to mirror the United Kingdom disclosure regime referred to in [129] below. Fourth, and most recently, is the current credit crisis and resulting recession. [26] The first three of the transactions in issue were with US counterparties. So were the three "ruled transactions" referred to in [38]. Mr Shay's evidence places at least those six transactions in the context of the "cycle" in which they occurred. They span the period December 1996 (commencement of AIG 1) to August 2006 (termination of AIG 2).Repos[27] In [5] I noted that the contested transactions were 'repo' deals. I had helpful explanations of 'repos' from several witnesses, amongst them Messrs Gross, Nias, Choudhry and Gammie QC. [28] Broadly, a repo is an arrangement under which A, holder of shares or other securities, sells them to B on terms that B will sell them back to A at an agreed time and price. Financial markets regard such a transaction as secured collateralised borrowing. Economically, a repo is similar to a loan, particularly one secured by a pledge of shares.[29] Mr Gross explained the development of classic repo structures in the 1980s between US/UK counterparties. The transactions were based on the different tax treatment of repos in those two countries. The UK regarded the purchase under a repo as transferring ownership of the shares, so that the dividends paid on them belonged to the UK purchaser. The dividends were not taxed provided tax had been paid on the profits from which the dividend was paid. The US treated a repo as a loan by the UK party, secured by the shares. US tax law thus treated the dividends as interest payments, and thus as deductible expenditure. [30] Consistent with this, Mr Kyle said the structure the BNZ used in the six transactions in dispute was a cross-border sale and re-purchase, or "repo" over securities in a special purpose entity. He explained: While this form of financing transaction was novel to BNZ at the time, a repo over fixed rate units or shares is a very common form of structured financing arrangement. Cross-border repos have been entered into in numerous countries around the world. The US-UK market is probably the largest cross-border repo market, reflecting the size of the US and UK financial markets generally. (PB 2.56)[31] Mr Nias explained the versatility of repos in these terms: A repo can be a relatively cheap method of obtaining finance, particularly for non-banks that may not have easy access to the interbank market (which is often the cheapest source of borrowing). As a repo seller, a borrower can obtain better terms than would normally be obtained by borrowing cash on an unsecured basis, as the repoed securities have the economic effect of collateral. Conversely, for a lender, the collateral effect of the repoed securities makes repos a more secure way of lending than, for example, money market deposits – repo buyers are often financial institutions with excess liquidity. The sheer flexibility of repos makes them useful alternatives to other money market instruments and helps explain their popularity. (PB 44)[32] Mr Nias is the Senior Partner and Head of Tax of the law firm McDermott Will & Emery UK LLP. He has 30 years experience in international tax practice. He practises in London. His areas of specialisation include cross-border transactions and structured finance transactions such as those in issue here.[33] Let me give a specific example of a classic (i.e. conventional) equity repo transaction. Investor/borrower A takes a long position in BHP shares, funding its purchase by borrowing cash from B under a repo agreement. A provides its newly acquired BHP shares to the lender of the cash, B, as collateral. B pays any dividends on the collateral securities (the BHP shares) back to A by way of what is known as a "manufactured dividend", and A pays interest to B on the loan. A variation of this is that B retains the dividends against the amount otherwise payable by A to B to repurchase the shares. The dividends then form part of the interest payable under the loan. [34] The transactions here differed from this classic type of repo in a number of respects: a) The counterparties did not use the loans to acquire third party securities that would be used as collateral for the loans. b) The lender (BNZ) retained the distributions received in lieu of receiving interest on the loans. Because the distributions were fixed the repos can be regarded as fixed rate funding transactions. c) The rate of return on the loans was high (relative to the prevailing 5 year swap rates) in spite of collateral being in place. By contrast, the rate of return to the lender of cash in a classic repo is typically lower than the return in a standard unsecured funding transaction, reflecting the lower risk because of the collateral pledged by the borrower. d) The collateral was in the same name as the repo agreement for credit risk purposes i.e. it was Gen Re collateral or Rabobank collateral. Moreover, it was specifically created by Gen Re or Rabobank for the purposes of the transaction. [35] In summary, these were not true classic repos in the conventional market sense. They were essentially straightforward loans rather than repos.The time span of the BNZ transactions[36] The time span of the nine BNZ transactions is best shown diagrammatically: [37] To show the timing of the nine BNZ transactions in relation to the comparable transactions entered into by other New Zealand banks mentioned in [22], I replicate the diagram produced by the Commissioner. The original of that diagram is colour coded. Unfortunately, technology constraints require this to be a colourless judgment. Thus, I have added the bracketed referencing to the various other banks in the right hand margin. The time span and overlap of these transactions are the important points. I hope those, at least, will be legible.The ruled transactions[38] The diagram set out under [36] shows that two of the ruled transactions (AIG 1 and Morgan Stanley) came first in time, and that the third (AIG 2) transaction was the penultimate one in the sequence of nine transactions. [39] I accept the BNZ's submission that the ruled transactions had the same basic structure as the six transactions in issue, in particular they were structured to take advantage of asymmetry in the New Zealand tax legislation i.e. tax deductible costs generating tax exempt income. That emerges from the private binding ruling (Common Bundle 7/4170) on the proposed AIG 1 transaction given on 17 December 1996, which states that interest, GAFs/GPFs and letter of credit fees will be deductible under s DD 1(b), while the dividends received by the BNZ from the AIG counterparty will be exempt from tax.[40] The same is true of the Morgan Stanley transaction (the ruling dated 25 November 1997 is at 12/7576). [41] The first of the Gen Re transactions (referred to in submissions as "Gen Re 0") began as an "almost exact replica" (the BNZ's description at 3.10 in Opening) of the Morgan Stanley transaction, indeed it was negotiated at the same time. Gen Re 0 also received a favourable (draft) ruling (19/13172). The structure of the Gen Re 0 transaction was subsequently altered as a result of Notice 98-5, Foreign Tax Credit Abuse, issued by the Internal Revenue Service of the United States Department of the Treasury (US IRS) on 20 January 1998. Notice 98-5 (16/11519 and Exhibit O) begins:Treasury and the Internal Revenue Service understand that certain U.S. taxpayers (primarily multinational corporations) have entered into or may be considering a variety of abusive tax-motivated transactions with a purpose of acquiring or generating foreign tax credits that can be used to shelter low- taxed foreign-source income from residual U.S. tax. These transactions generally are structured to yield little or no economic profit relative to the expected U.S. tax benefits, and typically involve either: (1) the acquisition of an asset that generates an income stream subject to foreign withholding tax, or (2) effective duplication of tax benefits through the use of certain structures designed to exploit inconsistencies between U.S. and foreign tax laws. This notice announces that Treasury and the Service will address these transactions through the issuance of regulations as well as by application of other principles of existing law, and requests public comment with respect to these and related foreign tax credit issues.The Notice created uncertainty and concern as to whether the United States Revenue would allow Gen Re a foreign tax credit for tax paid in New Zealand. [42] As a result the structure of the Gen Re 0 transaction was altered to relocate the issuer (Gen Re 1 Trust - GRFT) to the United States. So altered, it proceeded as the Gen Re 1 transaction. The BNZ then claimed a foreign tax credit, essentially for the tax paid by GRFT on the income it distributed to BNZIS1. In short, a tax benefit that appeared to have become potentially problematic if claimed in the United States was claimed instead in New Zealand. [43] The BNZ accepts that, in terms of s 91 EA Tax Administration Act 1994, the private binding ruling it obtained for each of AIG 1, Morgan Stanley and AIG 2 binds the Commissioner only in respect of the arrangement ruled upon.[44] The relevance the BNZ places on these rulings is this: a) Except for the tax residence of the issue, the structure of the ruled transactions is indistinguishable from the transactions in issue. b) The different tax residence of the issuer is not a distinction of any principle. Indeed, the Corporates Unit of Inland Revenue viewed the AIG transaction as "on all fours" with the disputed transactions. The Bank referred to a 5 February 2004 Inland Revenue file note observing:Only difference (minor) is that AIG 1 involved a NZ issuer.The Bank also referred to the fact that the Commissioner gave a binding ruling on one of the transactions entered into by the Westpac Banking Corporation (the First Data transaction – US, conduit relief). This was a tax negative transaction, indistinguishable from the latter five transactions in issue in these proceedings. The BNZ did not suggest that it was aware of this ruling before the Commissioner's discovery in these proceedings. c) They were relied upon for ruling favourably on the 2.95% p.a. GPF used in the six transactions in issue. [45] I do not accept that the binding rulings have any relevance to this case, in particular that they in some way support the BNZ's position. First, a binding ruling expressly applies only to the transaction ruled upon. As a matter of law, no reliance can be placed upon the ruling in respect of any other transaction. Second, even if a taxpayer is entitled to place some reliance on a binding ruling as indicating the way in which the Commissioner will treat an identical transaction (and I do not accept that), the six transactions in issue were not identical. They have the distinguishing feature that they were tax negative for New Zealand. Given the size of that negativity, which mounted with each successive transaction, that distinguishing feature was never insignificant, and the BNZ does not suggest that it was. I note thatthe Commissioner, writing to the BNZ on 18 December 1996 about the ruling he had given the previous day on the AIG 1 transaction, stated: We initially had concerns about the structure of the arrangement in relation to the paying of exempt dividends. However, the applicants are entitled to certainty as to how Inland Revenue will treat the arrangement and we were prepared to rule favourably based on the specific facts of the arrangement. In particular, the fact that the arrangement will result in a positive tax gain to New Zealand and that the paying of the exempt dividends is to reduce the risks from potential adverse changes to United States income tax law were important factors we took into account. I interpret this as the Commissioner, although flagging concerns about the structuring of the transaction in terms of its tax effects, ruling favourably because the net effect was to add to the New Zealand tax base.The six transactions in issueHow they came about[46] The genesis of the transaction structure appears to be a proposal in June 1995 put by Clydesdale Bank (one of the NAB's UK subsidiaries) to the Group Credit Bureau of NAB for a preference share transaction between Clydesdale and a subsidiary of the AIG Group (5/2853). [47] Before this proposal proceeded, NAB Group taxation indicated that NAB's UK tax position had become uncertain, and suggested substituting the BNZ. In making this suggestion NAB Group Taxation pointed out: it was considered appropriate that the Bank of New Zealand be substituted for Clydesdale in the transaction. An additional advantage to the NAB Group from the BNZ's involvement [being] the ability of the Group to obtain a binding ruling from the IRD, which is not available from the equivalent UK authority. (5/2917)[48] Mr Birch gave evidence that the "opportunity" was referred to the BNZ by NAB in Melbourne. He explained that an Australian based arranger of financial products, Allco Finance Group Limited, appeared to have originally developed the proposal with AIG Financial Products Corp, a subsidiary of American International Group (PB 3.4). Mr Birch said that the initial proposal (undated, 5/2600) received by the BNZ was for a five year, NZ$250 million financing transaction with AIG Financial Products. The finance was to be provided by way of an equity investment by the BNZ in a subsidiary of AIG Financial Products (PB 3.6). [49] Documentation through the latter part of 1995 and into 1996 records Mr Birch working with Mr Bain of Allco, and with AIG-FP, to develop the transaction structure for use in New Zealand. At the time, Mr Birch was a member of the structured finance team of the BNZ in Wellington. He had joined the team, from the BNZ's tax group, in about 1993. He was appointed head of structured finance at BNZ in 2001, and remained in that position until 2004 when he was seconded to the NAB, ultimately to head the NAB's structured finance team in the UK. Mr Birch holds a degree in accounting and finance. He worked for several years at Deloittes, and then General Finance, before joining the BNZ in 1988. From 1990 to 1993 he worked for KPMP as a tax consultant, before rejoining the BNZ in June 1993. [50] In March 1996 the BNZ structured finance team completed an application for line of credit (ALOC). That ALOC detailed the transaction, and included a summary of the financial returns it would produce. The ALOC was circulated to all the sections within the BNZ and NAB that were required to approve ("sign off on") the transaction. [51] A condition precedent in the AIG 1 transaction agreements was a favourable private binding ruling from the New Zealand IRD. That was applied for on 19 July 1996 by Ernst & Young on behalf of the BNZ (7/3900). As indicated in [39] above, a ruling (two rulings, in fact) were given five months later, both dated 17 December 1996 (7/4157; 7/4170). As it is not in issue, I need not detail the AIG 1 transaction. It suffices to say that, during its five year duration, it was tax positive for New Zealand by some NZ$3 million p.a. (the NZ$21 million tax paid by the AIG NewZealand subsidiary issuer, less the NZ$18 million reduction in the BNZ's tax liabilities resulting from the BNZ deducting its funding costs). [52] The Morgan Stanley transaction followed on 25 November 1997 after a favourable ruling was received for it on 21 November 1997 (12/7533). The financial effect of this was similar to AIG 1 and it also was tax positive for New Zealand, it seems also in the order of NZ$3 million p.a. [53] The transaction that subsequently became Gen Re 1 was to follow with the same structure and New Zealand tax effect. In [41] I explained the intervening issue, in January 1998, of Notice 98-5 by the US IRS, and the subsequent re-structuring of what became Gen Re 1 in the manner I describe in [42]. [54] Notice 98-5 was also the reason for the abrupt termination of the Morgan Stanley transaction. As Mr Kyle noted in an email to Mr Birch on 22 September 2001: the Morgan Stanley deal only lasted a few months before 98-5 was proposed and MS bolted. (11/6817)[55] As I will also explain, the distribution in Gen Re 1 was relieved of tax by the claiming of a foreign tax credit, whereas the distributions in the following five transactions were relieved of tax by the conduit regime.The detail of the transactions[56] Because the six transactions in issue are similar in their form or structure, and in their substance, it is sufficient to detail just one of them, and then to identify significant differences in the other five. I have taken the CSFB transaction. [57] The key dates relating to the CSFB transaction were:• 25 August 1998: transaction closed. Half – NZ$250 million – of the NZ$500 million principal of the CSFB transaction came from the terminated Morgan Stanley transaction.• 25 August 2001: transaction extended for a further three years.• 9 August 2002: transaction terminated early. The proceeds were used by BNZI to fund the Lehman's transaction which closed on 13 August 2002.Detailed analysis of the CSFB transactionNegotiation of the CSFB transaction[58] While the BNZ structured finance team was working with Gen Re to re- structure (what became) the Gen Re 1 transaction, it was approached by Ernst & Young. E & Y had developed a structured finance transaction with Credit Suisse Finance Products, and approached the BNZ as a potential counterparty investor in the transaction. Before E & Y would disclose details of this transaction to the BNZ, Mr Kyle, was required to sign a confidentiality undertaking. He did that by countersigning a letter, dated 12 January 1998 and headed "EXCISE product/structure", from CSFB (23/17704). [59] Details of the proposed transaction followed. They were sent to Mr Kyle by CSFB under cover of a letter dated 22 January 1998 (23/17722). The enclosures comprised an executive summary (23/17724-17727), indicative terms and conditions sheet (23/17728-17730), a supporting tax opinion dated 21 January 1998 from E & Y (Mr Judge, a partner in the Auckland office) (23/17732) and a second tax opinion dated 22 January 1998 from Bell Gully (Mr Smayle of the Wellington office) (23/17739). Since the Bell Gully opinion was directed to the New Zealand tax treatment of the United States partnership in the proposed transaction structure, it attached the written advice dated 22 January 1998 from the United States law firm Skadden Arps about the partnership (23/17745).[60] The executive summary included these statements:• NZ Investor ("NZ Investor") would receive an attractive return on financing by purchasing a US partnership interest from Credit Suisse First Boston, Inc. ("CSFB"), with an agreement to sell such partnership interest back to CSFB in the future.• The structure would give rise to fully tax-credited income distributions from the US partnership. The funding costs and the guarantee procurement fee would be deductible for NZ tax purposes.[61] The sample economics in the executive summary were:AssumptionsNZ Tax Rate 33% 3 yYear NZ Swap Rate 7.95% Partnership Distribution Rate 8.65% Guarantee Procurement Fee 2.95%NZ InvestorPre-Tax Equivalent of Partnership Distribution [8.65% ÷ [(1-33%) 12.91% Less Guarantee Procurement Fee (2.95%) Overall Pre-Tax Equivalent Return 9.96% Advantage over 3 year NZ Swap Rate 201bp [62] As is evident from this executive summary, the proposal relied on the BNZ receiving a foreign tax credit for tax paid on its behalf in the US. The structure was similar to the one the BNZ had been working on with Gen Re, save that it used a US partnership rather than a US trust. [63] By fax on 29 January 1998, Mr Kyle indicated to CSFB:• the pricing of the transaction at 200bps is at the low end of a range that might be acceptable to the BNZ group• we believe BNZ group might be prepared to consider an investment of up to NZ$500m• we have a strong preference to work on this transaction on a club basis with The National Bank of New Zealand. (23/17795)[64] The proposal proceeded from there, and as a "club deal" with the National Bank of New Zealand (NBNZ) through its subsidiary, Harcourt, which was to invest NZ$400 million. [65] By 3 April discussions had reached the point where CSFB was able to send the BNZ and NBNZ a revised term sheet (23/17930). Three days later, on 6 April 1998, the BNZ and NBNZ (Harcourt) sent CSFB a joint letter indicating that the term sheet accurately reflected the business deal agreed, and that their preference was that CSFB instruct its US legal counsel to prepare documentation based on the draft term sheet. [66] By July 1998 BNZ had most of its internal approvals and some of the structured finance team were in New York working with CSFB and advisers in finalising the transaction documentation. While they were there, Harcourt indicated that it may withdraw from the transaction and subsequently did so. Harcourt's withdrawal is noted in a BNZ memorandum dated 16 July 1998. [67] In the meantime, the conduit legislation had been enacted with force from 1 April 1998. In his evidence, Mr Birch said that one of the BNZ's advisers "had been telling us for some time that because BNZ was owned by NAB, the conduit legislation would be well suited to a cross-border repo transaction" (PB 6.41). He stated that the conduit relief regime had advantages over the FTC regime, principally clarity and the reduction of the potential exposure of BNZIS2 to liabilities of the US partnership (PB 6.42). [68] Accordingly, the parties to the CSFB transaction decided to rely on the conduit legislation to relieve the income received by BNZIS2 from tax in New Zealand. In order to obtain conduit treatment, a US unit trust was incorporated into the structure as issuer, in place of the earlier proposed US partnership. That change had no real consequences for CSFB, which could still use the "check the box" US tax procedure, and elect to have the unit trust treated as a corporation for US tax purposes.Structure of the transaction[69] Mr Birch produced a diagram for each of the nine BNZ transactions. I replicate his diagram of the CSFB transaction which assists in an understanding of the entities and cashflows detailed in [70] to [83]. I mention that the Commissioner did not accept that Mr Birch's diagrams showed all the entities and cashflows (particularly those on the counterparty side of the transactions). Accordingly, the Commissioner produced his own set of diagrams, which he claimed were comprehensive. Mr Birch's diagram of the CSFB suffices for the purposes of this judgment:Incorporation and funding of BNZIS2[70] On 2 July 1998 BNZI incorporated a wholly owned subsidiary, BNZIS2, specifically for the transaction i.e. as a special purpose vehicle or 'SPV'. BNZIS2 issued 500 million redeemable preference shares at NZ$1 each to BNZI to provide funding for its proposed NZ$500 million investment. BNZI, in turn, obtained funding to subscribe for the RPS by issuing ordinary shares to its parent, BNZ, for NZ$250m, and by withdrawing NZ$250m from cash it held on deposit with theBNZ, as a result of the termination of the Morgan Stanley transaction on 5 February 1998.Establishment of the Trust[71] On 21 August 1998 two US subsidiaries of Credit Suisse Group ('CS Group') organised the Madison Park Unit Trust "CSFB Trust" as a business trust under the Delaware Business Trust Act. The two subsidiaries were Madison Park LLC (MadPar) and Park Square LLC. A New York division of the Chase Manhattan Bank was appointed Institutional Trustee, and a Delaware offshoot the Delaware Trustee. There were three classes of units in the Trust: i) The Class A Unit: this had an "initial liquidation amount" of NZ$500m. It was an income unit carrying an entitlement to quarterly distributions at the rate of 9.87% per annum on NZ$500m in priority to the income distribution to the holder of the Class B Unit. It carried no voting rights. ii) The Class B Unit: this had an "initial liquidation amount" of US$244,950,000. It was also an income unit, and had the sole voting rights. iii) The Class C Unit: the "initial liquidation amount" was US$1,000. It carried no income entitlement or voting rights. Witnesses analysing the US transactions referred to the CS Group as the "US counterparty" and to the CSFB Trust as the "Trust". [72] Amongst the "Permitted Activities" of the CSFB Trust was a requirement to invest the Trust's property in a "Loan Note" issued by Credit Suisse First Boston Inc. (CSFB Inc), defined as:A note issued by [Credit Suisse First Boston] in favour of the Trust with a principal amount of no less than NZ$1,000,000,000, which provides for the payment of interest in NZ$ at a fixed rate of no less than 7.5923% per annum Witnesses referred to CSFB Inc. as the "US issuer" (i.e. of the Loan Note).Initial unit holders[73] On 25 August 1998: a) MadPar LLC purchased the Class A and B Units for NZ$500m and US$245m (the equivalent of NZ$500m) respectively. Madpar is a subsidiary of the US counterparty, and thus witnesses referred to it as the "US Sub". b) Park Square LLC purchased the Class C Unit for US$1,000.Investment of CSFB Trust property[74] Also on 25 August 1998, the Institutional Trustee used the Trust's property to purchase the Loan Note from CSFB Inc. at its face value of NZ$1,000,002,000. The Loan Note pays interest at 7.5923% p.a. The income from the Loan Note would thus be NZ$75.9 million. After US tax at 35%, this yielded NZ$49.35 million, the amount required to service the dividend on the Class A unit (NZ$500 million @ 9.87% = NZ$49.35 million).Repurchase of Class A Unit[75] On 21 August 1998 MadPar and BNZIS2 entered into a written agreement. This "CSFB Repo Agreement" was terminable unilaterally after the first anniversary of the transaction upon three business days' notice. Under the Agreement: a) MadPar agreed to sell, and BNZIS2 agreed to purchase, the Class A Unit for NZ$500m on 25 August 1998.b) BNZIS2 agreed to re-sell, and MadPar agreed to repurchase, the Class A Unit on or before 25 August 2001 for NZ$500m, plus or minus certain defined adjustments to that repurchase price. c) BNZIS2 agreed to pay MadPar a guarantee procurement fee of 2.95% per annum of NZ$500m, payable quarterly commencing 30 September 1998, in consideration for MadPar procuring its parent Credit Suisse Group to guarantee MadPar's obligations (primarily its re-purchase obligation in b)). d) MadPar gave a range of covenants, including to pay to BNZIS2 the quarterly distributions on the Class A Unit if the CSFB Trust defaulted. e) MadPar agreed, as a condition precedent, to provide additional credit support of at least NZ$200m satisfactory to BNZI.Interest rate swap[76] On 21 August 1998 BNZI entered into an interest rate swap with MadPar. Under this "CSFB swap" the following quarterly payments commencing 30 September 1998 were agreed: a) BNZI to pay MadPar the fixed rate of 9.3507% per annum on a notional NZ$500m. b) MadPar to pay BNZI the 90 day New Zealand bank bill rate (which is a variable or floating rate i.e. one reset in the market) on a notional NZ$500m. [77] The CSFB swap required these fixed and floating rate payments to be netted off, so that it was only the net obligation that was paid by the relevant party on each payment date.Guarantees, indemnities and credit support[78] On 21 August 1998: a) BNZIS2 guaranteed to MadPar BNZI's obligations under the CSFB swap. b) MadPar indemnified (the CSFB indemnity) BNZI against any losses it suffered as a result of BNZIS2 being unable to redeem the RPS it had issued to BNZI or to pay the dividends on those RPS because of default by MadPar. c) Credit Suisse Group guaranteed to BNZI MadPar's obligations under the CSFB Repo Agreement and CSFB indemnity. d) MadPar and BNZI entered into an ISDA (International Swaps and Derivatives Association) credit support annex requiring MadPar to provide highly rated collateral to BNZI for any exposure exceeding NZ$2m BNZI had under the CSFB (interest rate) swap.Letter of credit[79] MadPar satisfied the condition precedent referred to in [75]e) above by obtaining, on 25 August 1998 from the Royal Bank of Canada, an irrevocable standby letter of credit in favour of BNZI for one year. MadPar paid the RBC 30 bps per annum for this.Other documentation[80] Other documentation between the BNZ and CSFB to give effect to the transaction was: a) A Tax Matters Agreement dated 21 August 1998 between the CSFB Trust, MadPar, BNZI and BNZIS2;b) A Liability Assumption Agreement dated 21 August 1998 between MadPar, BNZIS2, BNZI and BNZ; c) An Escrow Agreement dated 21 August 1998 between Credit Suisse Group, MadPar, the CSFB Trust, BNZIS2, BNZI, BNZ International and New Zealand Guardian Trust; d) A Set-off Agreement dated 21 August 1998 between MadPar, the CSFB Trust, Credit Suisse Group, BNZIS2, BNZI, BNZ International and BNZ; e) An Assignment dated 25 August 1998 by MadPar to BNZIS2 of the Class A Unit of the CSFB Trust (giving effect to BNZIS2's purchase of the Class A Unit); f) An RPS Subscription and Guarantee Arrangement Agreement dated 21 August 1998 between BNZIS2 and BNZI.Still further documentation within the Credit Suisse Group[81] Internal to the Credit Suisse Group, there was significant further documentation, for example documenting the loan pursuant to which: a) MadPar lent the NZ$500 million proceeds to CSFB Inc. b) CSFB Inc. agreed to pay MadPar quarterly interest on that amount at the New Zealand 90 day bank bill rate minus 2.43% (243bps).Quarterly cash flows[82] Over the duration of the transaction, quarterly, the following events occur: a) The US issuer (CSFB Inc.) pays to the Trust interest on the NZ$1,000,002,000 Loan Note at 7.5923% p.a.b) The Trust makes a distribution on the NZ$500 million plus a unit to the BNZ Sub (BNZIS2) at 9.87% p.a. c) The BNZ Sub (BNZIS2) pays to the US Sub (MadPar) the guarantee procurement fee at a rate of 2.95% p.a. on NZ$500 million. d) BNZI or MadPar, as applicable, makes a net swap payment equal to the difference between the 9.3507% fixed rate and the BKBM floating rate, on a notional NZ$500 million. e) The BNZ pays interest on the funds it has borrowed on the New Zealand money market. f) MadPar receives interest from CSFB Inc. at BKBM less 2.43% p.a. on NZ$500 million. [83] Mr Shay depicted these quarterly cash flows in this diagram (Figure 1 in PB 21):BNZ's returns from the transaction[84] Mr Stanton set out the BNZ's returns from the CSFB:Floating Payments Percentage Returns % Absolute Amounts (NZ$'m) Trust distribution 9.87 49.35 Less: funding costs charged by BNZ money markets (NZ$ floating) Plus: swap receipt from MadPar NZ$ floating - Less: swap expense payable to MadPar (9.35) 46.75 Less: guarantee procurement fee payable to MadPar (2.95) (14.75) BNZ pre tax loss (2.43) (12.15) New Zealand tax: - Income not taxable - Deduction against other income liable to tax (4.88 plus 1.62 plus 2.95) x NZ corporate tax rate of 33% 4.06 20.30 BNZ post tax benefit 1.63 8.15[85] This table shows:• The BNZ incurred a pre-tax loss of 2.43% or NZ$12.15 million p.a.• The BNZ obtained a reduction in the amount of tax it would have paid but for the transaction of 4.06% or NZ$20.30 million p.a.• A resulting post-tax return to the BNZ of 1.63% or NZ$8.15 million p.a. [86] The pre-tax (or grossed up) equivalent of the 9.87% Trust distribution was 14.73%.CSFB's returns from the transaction[87] Again, Mr Stanton expressed these in tabular form. As CSFB was paying NZD BBR under the interest rate swap, its benefit was expressed as the discount below that rate at which it was being funded by BNZ (through the fixed rate on the swap and GPF):% Trust Distribution (9.87) Fixed leg of the swap 6.92 Guarantee procurement fee 2.95 Discount 2.43[88] As is evident from a comparison of the tables in [84] and [87] the BNZ and CSFB both received the same 2.43% pre-tax equivalent return from the transaction.Differences between CSFB and the other five transactions[89] A convenient start is to replicate table 1 from Professor Evans' evidence, in which he compared the key features of the six transactions:[90] Professor Evans identified eight key differences between the transactions: a) The first three transactions were with US based counterparty entities, the later three with predominantly UK based counterparty entities. b) In the US transactions, the counterparty established the issuer as a trust; in the UK based transactions the issuer was a limited liability company. c) The US based transactions structure the BNZ investment as a repo agreement; the UK based transactions involve separate sale and forward-sale agreements, or so-called "broken repos". d) In the Gen Re 1 and CSFB transactions the investment was made by a BNZI subsidiary (BNZIS1 and BNZIS2 respectively); the other four transactions involved investment directly by BNZI. e) In the Gen Re 2 and Rabo 2 transactions the counterparty issuer was consolidated in the BNZ for accounting purposes. f) In Gen Re 1 it was foreign tax credits that provided the tax exempt distributions; the other five transactions used conduit tax relief. g) The Gen Re 1 and Gen Re 2 transactions featured a negative spread on the fixed and floating legs of the interest rate swap and Rabo 1 and Rabo 2 had a negative spread on the floating leg of the swap for the first year of the transaction. The transactions involved different credit support arrangements:• An LoC in CSFB and Gen Re 1 (in the case of the latter replaced by collateral after one year).• Collateral in Gen Re 2 and Lehman's.• No credit support in Rabo 1.• A risk participation agreement in Rabo 2.The US tax treatment of the three US transactions[91] Mr Hicks and Professor Rosenbloom (called by the BNZ) and Mr Shay (a witness for the Commissioner) were the experts on US taxation. There was no real disagreement between these three witnesses about the US tax treatment of the Gen Re 1, CSFB and Gen Re 2 transactions. It is sufficient, therefore, to base this part of the judgment primarily on the evidence of Mr Shay. [92] The fundamentally different taxation treatment of the three US transactions in New Zealand on the one hand, and the United States on the other, was depicted by Mr Shay in his Figure 2: As is evident from that Figure, New Zealand tax treatment respects the legal form of the transaction; the US tax treatment recognises the economic reality of the transaction. [93] For US tax purposes, the BNZ would be treated as having made a secured loan to the US Sub, and the income units would be treated as collateral held by the BNZ pursuant to that loan. Thus characterised, the distributions on the income units would be deemed to be received by US Sub as owner of the income units. The USSub would be treated as making interest payments to BNZ on the loan, and would receive an interest expense deduction in respect of those deemed interest payments. [94] Mr Shay explained that the analysis by US Courts and the IRS of repos for US tax purposes is based on the particular facts that indicate whether the arrangement truly is a sale (and repurchase) or merely a secured loan. If, during the term of the repo, the "buyer" holds the repo securities merely as collateral for the "seller's" obligation to repurchase those securities, US tax law looks to the underlying economic reality and treats the arrangement as a secured loan for US tax purposes. Based on the factors outlined in Revenue Ruling 74-27 (the 74 indicates this Ruling dates from 1974), Mr Shay considered that the following terms of each of the US transactions indicate that the repo arrangements would be treated as secured loans for US tax purposes: a) BNZ Sub must hold the income units for repurchase by US Sub, and US Sub is entitled to demand and receive the income units when US Sub pays the repurchase price on the repurchase date. b) US Sub is obligated to repurchase the income units from BNZ Sub on the repurchase date. Its failure to do so constitutes an event of default. c) In the event of default, BNZ Sub may sell the income units and seek reimbursement for the difference between the proceeds it receives and the proceeds it would have received had the income units been repurchased by the US Sub. d) The effective interest rate that US Sub pays is specified in advance and is a rate calculated on the amount advanced by BNZ Sub (i.e. NZ$500 million). The US Sub is contractually obligated to pay the distribution rate whether or not such a distribution is actually paid from the Trust.e) The value of the income units, which are non-marketable and not publicly traded, may or may not equal the amount advanced under the repo agreement – NZ$500 million. [95] Consistent with the US tax treatment of the repo as a secured loan, the tax consequences would be as follows (applying them to the CSFB transaction, in particular to the quarterly payments detailed in [82]-[83]: a) Trust (CSFB Trust):• It does not file a consolidated US tax return with any other entity.• It pays tax (at the US corporate rate, having "ticked the box" electing to be taxable as a corporation) on the interest payment it receives from the US issuer (CSFB Inc.).• As the income distribution it makes to BNZ Sub on the Class A unit would be treated for US tax purposes as a deemed dividend from the Trust to US Sub (MadPar) followed by a deemed dividend payment by MadPar to BNZIS2, the deemed dividend payment would not have tax consequences for the Trust.• The net result is that the Trust would pay tax on interest income equal to 7.5923% p.a. on NZ$1,000,002,000. b) US Sub (MadPar):• Does not file a consolidated US tax return with either the CSFB Trust or CSFB Inc.• As explained, because of this the dividend would be treated as taxable income to MadPar. However, because MadPar isdeemed for US tax purposes to own at least 80% of the stock of CSFB Trust, MadPar and the Trust are considered part of an "affiliated group". MadPar would therefore be permitted to take a "dividends received deduction" in the amount of the dividend and thus entirely offset the amount of income inclusion.• Would be entitled to deduct, as interest expense, the amount of the deemed interest payment made to BNZIS2.• The GPF would be treated as ordinary income to MadPar.• The net swap payment would be treated as either net income to, or a net deduction for, MadPar.• The interest on the loan by MadPar to CSFB Inc. would be treated as interest income to MadPar.• The net result to MadPar would be income of .0007% on NZ$500 million. c) The US issuer (CSFB Inc.):• The interest payment on the loan note would be a deductible interest expense.• The interest payment to MadPar on the loan would also be a deductible interest expense.• The net result would be deductions of (1) 7.5923% p.a. on NZ$1,000,002,000 plus (2) BKBM minus 2.43% p.a. on NZ$500 million. d) Combined net US tax consequences to CSFB:• A deduction equal to BKBM minus 2.4307% on NZ$500 million (this assumes the Trust, MadPar and CSFB Inc. would each be subject to a 35% rate of US federal income tax, and that CSFB Inc. has other income against which it could apply the deductions detailed in c).• This result is the same as if the CSFB transaction had been a simple loan by BNZ to CSFB at an interest rate of BKBM minus 2.4307%.• Assuming CSFB receives a return of BKBM on the NZ$500 million lent to MadPar by BNZIS2, and that that return was treated for US tax purposes as ordinary income to CSFB, the net result to CSFB would be taxable income of 2.4307% on NZ$500 million. [96] Mr Shay explained that the US parties to the US transactions had entered into other arrangements that added complexity to the transactions, but preserved their essence as loans for US tax purposes. By way of illustration, he pointed out that the structure of each transaction had been designed so that: a) The trust received taxable income from the US issuer that might result in a US tax liability, but any such liability appeared to be fully offset by the US issuer's corresponding US tax deduction. b) The US Sub was deemed to receive trust distributions without incurring a US tax liability. c) The interest payment by the US Sub to the BNZ was deductible for US tax purposes and was exempt from US withholding tax. [97] Mr Shay's evidence was that various other steps in the transactions did not appear to have significance from a US tax perspective. The interest rate swap was one of those steps. It had the effect of transforming the US Sub's fixed rate interestobligation under the repo agreement into a floating rate interest obligation indexed to BKBM. For US tax purposes, the US Sub would treat any (net) payments it received under the swap as taxable ordinary swap income and any (net) payments it made as a deductible swap expense. As a result, the US Sub could deduct the floating rate payment resulting from the combination of the repo agreement and the interest rate swap as a combination of interest and expense. [98] Mr Shay summarised the net economic result to the US counterparty in the following way:36. After taking U.S. tax consequences into account, the result for the U.S. Counterparty appears to be the same as if each transaction were a simple loan by BNZ to the U.S. Counterparty at an interest rate of BKBM less a fixed-rate discount ("Discount Spread"). The U.S. Counterparty's benefit from the structured loan presumably equals the Discount Spread (that is, the U.S. Counterparty might otherwise need to borrow at a rate equal to BKBM). The Discount Spread, in turn, appears to be a predetermined share of the New Zealand tax benefit generated by the structure of the transactions. 37. Although some tax risk might exist, there appears to be little or no extra U.S. tax cost to the complex structure that was used in the U.S. Transactions compared to a simple loan. The relative complexity of the structure as compared to a simple loan by BNZ to the U.S. Counterparty appears to be designed to generate a fixed amount of New Zealand tax benefits to be shared between the parties at no additional U.S. tax cost to the U.S. Counterparty and no particular detriment to the revenue of the U.S. government.The UK treatment of the three UK transactions[99] As with the three US transactions, there was agreement on the UK tax treatment of the three UK transactions. Broadly, it was similar to the US treatment. The transactions would be taxed as secured lending/borrowing transactions, reflecting their true economics. Only the accounting profit earned by the UK counterparty (i.e. the purchase price differential) would be taxable in the UK. The distribution would be treated as a "manufactured dividend" and deemed "interest". It would be a deductible expense in the UK, while being treated as an exempt receipt in New Zealand (on the tax treatment assumed by the BNZ). [100] Mr Nias summarised the position thus: The net result of the transactions was a deduction for Rabo Holdings (UK) Limited (in the case of the Rabo 1 and Rabo 2 transactions) and LB Investments (UK) Limited (in the case of the Lehman's transaction) for each company's funding costs (in respect of its subscription for the preference shares subject to the repo) and a tax exemption for the receipt of amounts characterised as manufactured dividends in respect of those shares (notwithstanding that they had been disposed of by way of stock loan); i.e. for UK tax purposes the preference shares were treated as having never been sold by each company.(P B 62)Foreign tax credit transactions[101] In [41] I mentioned that the proposed Gen Re 0 transaction was altered after the US IRS issued Notice 98-5. In that Notice the IRS signalled an intention to move against what can conveniently be described as foreign tax credit generator transactions. The replacement Gen Re 1 transaction was restructured so that tax was paid in the US rather than New Zealand, and the transaction became tax negative for New Zealand. To round off this aspect of the case, I note the evidence of Mr Shay that the IRS has subsequently confirmed the applicability of long-standing substance-over-form doctrines to cross-border arbitrage transactions, whether or not a specific anti-arbitrage rule may also be asserted. Mr Shay drew attention to the temporary US Treasury Regulations issued approximately a year ago: T.D.9416, Fed. Reg. 40727, 40728 (July 16, 2008) (Final and temporary regulations under Section 901 of the Internal Revenue Code). Those regulations specifically disallow foreign tax credits attributable to certain foreign tax credit generator transactions. The preamble states that the IRS: will continue to utilize all available tools under current law to challenge the U.S. tax results claimed in connection with these and other similar abusive arrangements, including the substance over form doctrine, the economic substance doctrine (and other broad principles) [102] The targeted arrangements are characterised as those that: exploit differences between U.S. and foreign law in order to permit a person to claim a foreign tax credit for the purported foreign tax payments while allowing the foreign counterparty to claim a duplicative foreign tax benefit.[103] Mr Shay adds that the IRS has in fact challenged foreign tax credit generator transactions using economic substance principles: Principal Life Insurance Company et al v United States (S.D. Iowa) (docketed March 3, 2008); Chief Counsel Advice 200826036 (February 29, 2008); Technical Advice Memorandum 200807015 (November 7, 2007). [104] While not disputing the development outlined by Mr Shay, Mr Hicks made the point that the preamble from which Mr Shay quoted is not a statement of United States tax law, but rather an indication by the IRS of the position the United States may take with regard to foreign tax credits in particular cases. [105] The point is that the US IRS has characterised as "abusive" "foreign tax credit generator" transactions like Gen Re 1, and signalled an intention to move against them. While the BNZ is entitled to protest that this has no relevance to the issue in these proceedings in relation to the Gen Re 1 transaction, I mention it because the BNZ called a significant amount of evidence that such transactions were commonplace and unobjectionable. I refer, for example, to the evidence of Mr Hicks and Professor Rosenbloom mentioned at [262] and [264] respectively. [106] Mr Hicks is an international tax partner in the Washington DC office of the global law firm Skadden Arps. In addition to some 19 years in private practice as a tax lawyer, Mr Hicks spent four years with the US Department of Treasury, the latter two of them as international tax counsel. He has also taught tax since 1992, as an Adjunct Professor at Georgetown University in Washington DC.The lawThe deduction provisions[107] The provisions under which the BNZ deducted its various expenses of the transactions are ss BD2 (the allowable deductions) and DD1(3) (interest). The Commissioner accepts that all the BNZ's expenses fell within the ambit of these provisions with the exception of the guarantee arrangement and procurement fees.To the extent that it is necessary to set out the deduction provisions, I do so at [144], [163] and [165] when ruling on the deductibility of the guarantee arrangement and procurement fees.Subvention payments and loss transfers[108] The Commissioner accepts that the reductions in net income claimed by the BNZ for subvention payments made in relation to net losses resulting from the arrangements, and for transferred net losses arising from expenditure incurred under the arrangements, came within the ambit of the subvention payments and loss transfers provision, s IG2(2). No need, therefore, to set out that provision.Foreign tax credit[109] The BNZ claimed a foreign tax credit in the Gen Re 1 transaction. As the Commissioner accepts that the foreign tax provision, s LC1, applied, it is unnecessary to set out the section.The conduit regime[110] The CIR accepts that the distributions received by the BNZ in the CSFB, Gen Re 2, Rabo 1, Rabo 2 and Lehman's transactions qualified for conduit tax relief under sub-part KH of the Act. It is thus unnecessary to set out the conduit provisions, but I refer to them further in [206] below in considering Parliament's purpose, and what was within Parliament's contemplation, when it enacted the conduit regime.The general anti-avoidance rules (GAARs)[111] The Commissioner avoided the transactions pursuant to s BG 1:BG 1 AVOIDANCE Arrangement void(1) A tax avoidance arrangement is void as against the Commissioner for income tax purposes.Enforcement(2) The Commissioner, in accordance with Part G (Avoidance and Non- Market Transactions), may counteract a tax advantage obtained by a person from or under a tax avoidance arrangement.[112] "Arrangement", "tax avoidance" and "tax avoidance arrangement" are defined in s OB 1. Respectively:"Arrangement" means any contract, agreement, plan, or understanding (whether enforceable or unenforceable), including all steps and transactions by which it is carried into effect. "Tax avoidance", in sections BG 1, EH 1, EH 42, GB 1, and GC 12, includes – (a) Directly or indirectly altering the incidence of any income tax: (b) Directly or indirectly relieving any person from liability to pay income tax: (c) Directly or indirectly avoiding, reducing, or postponing any liability to income tax. "Tax avoidance arrangement" means an arrangement, whether entered into by the person affected by the arrangement or by another person, that directly or indirectly – (a) Has tax avoidance as its purpose or effect; or (b) Has tax avoidance as one of its purposes or effects, whether or not any other purpose or effect is referable to ordinary business or family dealings, if the purpose or effect is not merely incidental.[113] Where a tax avoidance arrangement is void under s BG 1, the Commissioner may counteract the tax advantage:GB 1 AGREEMENTS PURPORTING TO ALTER INCIDENCE OF TAX TO BE VOID GB1(1) Adjustment of income Where an arrangement is void in accordance with section BG 1, the amounts of gross income, allowable deductions and available net losses included in calculating the taxable income of any person affected by that arrangement may be adjusted by the Commissioner in the manner the Commissioner thinks appropriate, so as tocounteract any tax advantage obtained by that person from or under that arrangement Ben Nevis and Glenharrow[114] In applying s BG 1 and its associated definitions, I am guided by the Supreme Court's decisions in Glenharrow Holdings Ltd v CIR [2009] 2 NZLR 359 and Ben Nevis Forestry Ventures Ltd & Ors v CIR [2009] 2 NZLR 289. Ben Nevis is the more relevant decision, since it deals with s BG 1. Glenharrow was concerned with the GAAR (s 76) in the Goods and Services Tax Act 1985. The judgment of Tipping, McGrath and Gault JJ ('the main judgment' – like counsel I baulk at using the 'plurality' epithet), observed that "the current case law has become complex", and set out to "identify a means for determining where permissible use of specific provisions ends and tax avoidance begins" ([13]). In doing that, the main judgment reviewed the case law, rendering it otiose, indeed inappropriate, that I refer to it. That is particularly so since the main judgment sought "to settle the approach which should be applied in New Zealand" in determining the inter-relationship of s BG 1 with applicable specific provisions. [115] Penny v CIR HC CHCH CIV 2007-409-1154 19 March 2009 is, I think, the only case this Court has decided since Ben Nevis. In his judgment, MacKenzie J observed that the essence of the Supreme Court's decision in Ben Nevis "is to endorse a 'scheme and purpose' approach". He then set out [102]-[109] and [113]- [114] of the main judgment in Ben Nevis and then [2]-[3] and [8]-[9] of the separate judgment of the Chief Justice and Anderson J. As my judgment will undoubtedly also be appealed (the Commissioner has appealed the judgment in Penny), I set out the principles and approach as I have extracted them from Ben Nevis andGlenharrow. This should make any error in my understanding more apparent to the appellate eye. [116] I agree with the Commissioner's submission that Ben Nevis:a) Spells an end to the tax mitigation/tax avoidance distinction drawn by the Privy Council in cases such as Challenge Corporation Ltd v CIR[1986] 2 NZLR 513 at 562. ([95]) b) Also stumps the Privy Council's view, in CIR v Auckland Harbour Board [2001] 3 NZLR 289 at [11], that s BG 1 was merely "a longstop for The Revenue" ([100], [103]). c) Ends the "threshold" approach (see fn 113 to [104] in the main judgment in Ben Nevis) favoured by Richardson J in the Court of Appeal in Challenge at 549-550. Because the taxpayer met the threshold of literal compliance with the specific provisions (inChallenge they related to the tax treatment of subvention payments), Richardson J held that such compliance could not consistently be treated as tax avoidance. The main judgment in Ben Nevis observes:[89] The effect was to reconcile conflicting provisions by reading down the scope of (now, s BG 1) so that it did not operate on arrangements that complied with the particular specific provision in the legislation. The scheme and purpose of the legislation required that (s BG 1) be read in the context of the special concession provisions which were dominant.In short, the taxpayer does not avoid the reach of s BG 1 by surmounting step 1 set out in 0 below. ([3], [103], [104], [107]) d) Endorses the approach of treating any artifice or pretence in an arrangement as highly relevant in deciding whether that arrangement has a purpose of tax avoidance: Miller v CIR [2001] 3 NZLR 316 (PC) at [10]; Dandelion Investments Ltd v CIR [2003] 1 NZLR 600 (CA) at [85] and CIR v BNZ Investments Ltd [2002] 1 NZLR 450 (CA) at [40]. ([97], [108]) [117] Ben Nevis stipulates a two step inquiry.Step 1 – The taxpayer's use of the specific provision(s)[118] Has the taxpayer satisfied the Court that its use of the specific provisions is within the intended scope of those provisions? ([107], [3]). This first step involves determining the ordinary meaning of the specific provisions, as established through their text in the light of their specific purpose as per s 5 Interpretation Act 1999 ([103]). [119] The Commissioner's submissions (41-42 of his opening) described step 1 as an inquiry, general in nature, concerned with settling the question of the interrelationship between s BG 1 and specific taxing provisions. The Commissioner submitted:There can be no doubt that the Court considered that this general inquiry should be answered by reference to legislative "scheme and purpose" 29. As the Court's analysis shows, the answer to the inquiry recognises and gives appropriate weight to both the general purpose of the anti avoidance provision, as well as the specific purpose of whatever black letter provisions happen to be at issue. ... ________________________29 Which the Court plainly equates with an orthodox purposive approach to statutory interpretation – see for example the heading before paragraph [84] and the content of paragraph [99].[120] After setting out [103] of Ben Nevis and referring to CIR v Auckland Harbour Board and Challenge, the Commissioner then contended (his 44):Thus the Court determines that s BG 1 can operate whatever specific taxing provision happens to be at issue. The existence of a detailed and specific statutory taxing regime (such as foreign tax credit or conduit in this case, or subvention in Challenge) has little or no bearing on whether or not there is "room" for s BG 1. [121] I am not sure that these submissions correctly describe step 1 i.e. 'the first inquiry' referred to in Ben Nevis at [107]. I accept that the Court, in the early part of [103], makes the point that, as the specific provisions and s BG 1 are meant to work in tandem, they must be construed to give appropriate effect to each. At that point, I think the Court was describing the whole process – the steps combined. In determining Parliament's "overall purpose", there is a strong interrelationship between the specific provisions and s BG 1.[122] I read Ben Nevis as requiring the Court, at step 1, to undertake a discrete inquiry, determining whether the taxpayer has complied with the specific provisions, interpreted as directed by the Court in the latter part of [103]. The Court is engaged in interpreting the specific provisions standing alone, rather than in interpreting them in the context of the whole legislative scheme. If compliance is not conceded, the Court must analyse the transaction and decide whether it complies with the applicable specific provisions. The analysis is to be a 'black letter' one, without the "judicial glosses and elaborations" that the Supreme Court recommended the Court keep to a minimum [104]. [123] The Commissioner's approach to step 1 is apt to shift the focus to s BG 1, when I do not consider step 1 involves recourse to s BG 1. Appellate clarification would assist here, in particular in relation to whether the "in tandem" interpretative approach referred to at [103] applies at step 1 and, if so, how it is to be given practical effect. In making this plea, I am conscious of the rejection of a literal interpretation of the specific provisions at [2] and [3] in the separate judgment of the Chief Justice and Anderson J. But, although they refer in [3] to "the first question", they do not specifically refer to the second question, or say what it is.Step 2 – The taxpayer's use of the specific provision(s) in the overall arrangement[124] If the answer at step 1 is 'yes', the Court next inquires: would the taxpayer's use of the specific provision (to alter the incidence of income tax), viewed in the light of the arrangement as a whole, have been within Parliament's contemplation and purpose when it enacted the specific provision? ([107]) [125] At step 2 the focus shifts to a purposive interpretation of the specific provisions in the context of legislative scheme as a whole. This "scheme and purpose approach" requires the Court to focus on "wider considerations", including s BG 1, and needs the "wider perspective" envisaged by s AA 3(1) of the Income Tax Act and s 5 of the Interpretation Act 1999. [126] Like step 1, step 2 requires the Court to consider the arrangement itself. However unlike step 1, the Court must go beyond simply assessing whether theassessment falls within the literal meaning of the specific provisions. Instead, armed with a thorough grasp of the detail and workings of the arrangement, the Court asks itself: would it have been within Parliament's contemplation that the specific provisions be "deployed" (the word used in [104]) in the manner in which they were deployed by the taxpayer in this particular arrangement? If not, the arrangement will be a tax avoidance arrangement caught by s BG 1, unless the tax avoidance purpose or effect of the arrangement is merely incidental. The Chief Justice and Anderson J explained that s BG 1 tips an arrangement into tax avoidance: if the fiscal effect intended is more than "merely incidental" to the business purpose. The fiscal implications of an arrangement that is "merely incidental" to a business purpose may in some cases be substantial and still within the statutory scheme and purpose. "Merely incidental" may properly be contrasted with the end in view, the "purpose or effect". ([9])[127] Consistent with the inquiry involved at step 2, and whether or not the taxpayer gives evidence of its purpose, the Court ascertains the "purpose or effect" of the arrangement objectively, from its terms: [73]-[74]; Glenharrow [35]-[40]. The effect of a transaction can be different from its purpose – the composite term is not to be collapsed into simply "effect" ([8]). [128] Section BG 1 does not confine the Court as to the considerations which are relevant to the step 2 inquiry. The significance of these will depend on the particular facts, but relevant considerations may include: a) The manner in which the arrangement is carried out. b) The role of all relevant parties and any relationship they may have with the taxpayer. c) The economic and commercial effect of documents and transactions. d) The duration of the arrangement and the nature and extent of the financial consequences it will have for the taxpayer. e) The combination of various elements in the arrangement – "a classic indicator of a use that is outside Parliamentary contemplation is thestructuring of an arrangement so that the taxpayer gains the benefit of the particular provision in an artificial or contrived way. It is not within Parliament's purpose for specific provisions to be used in that manner." ([108]) [129] Counsel for the Commissioner pointed to the similarity between the considerations listed in [128] and those contained in the Australian GAAR, principally s 177D in Part IVA of the Income Tax Act 1936 (Cth). I was informed that the Australian provision had been drawn to the Supreme Court's attention during the argument in Ben Nevis. Although I am conscious of the Supreme Court's caution (in [110]) about reliance on English tax case law, I think it helpful to refer here to one aspect of English tax legislation. In the Finance Act 2004, the English Parliament introduced disclosure regimes. These require disclosure by parties proposing to enter into arrangements fitting descriptions contained in regulations. The descriptions in the Tax Avoidance Schemes (Prescribed Descriptions of Arrangements) Regulations 2006, relating to income tax and corporate tax, include these: a) Confidentiality: arrangements which a promoter or user might wish to keep confidential from the revenue authorities. b) Off market terms: the price of the financial product differs significantly from what might reasonably be expected in the open market. c) Standardised tax products: the arrangement is a standardised tax product, in that it is not tailored, to any material extent, to reflect the circumstances of the client and the promoter makes the arrangement available to more than one other person. d) Loss schemes: an arrangement that has the main benefit of accruing losses to individuals for use to reduce their liability to income tax.I consider these descriptions to be useful indicators of a tax avoidance arrangement caught by s BG 1. The English disclosure regimes were referred to in evidence in a general way by Mr Nias. [130] In applying the considerations listed in [128], the Court considers the use made of the specific provision in the light of the commercial reality and economic effect of that use. The ultimate question is: viewed in a commercially and economically realistic way, does the impugned arrangement make use of the specific provision in a manner consistent with Parliament's purpose? ([109]) In their separate judgment the Chief Justice and Anderson J eschewed, as "allied and dangerous myths" (the description used by Lord Millett, writing extra-judicially), that: a) In tax cases to an extent unknown in other areas of the law, form prevailed over substance; b) The only thing to be regarded was the legal effect of a transaction. They adopted the way in which Ribeiro PJ stated the issue in Collector of Stamp Revenue v Arrowtown Assets Ltd (2003) 6 HKCFAR 517 at [35]:The ultimate question is whether the statutory provisions "construed purposively" were intended to apply to the transaction, viewed realistically.In Glenharrow the Court observed:[49] Transactions which are driven only by commercial imperatives are unlikely to produce tax consequences outside the purpose of the legislation [131] It is at step 2 that the Court must decide "on which side of the line a particular arrangement falls" ([112]). Thus, line-calling remains the Court's function. By contrast, it is not the Court's function "to articulate how the line is to be drawn" ([104]). The certainty which tax advisers desire must continue to elude them. As is stated in the main judgment, also at [112]: Parliament has left the general anti-avoidance provision deliberately general The Courts should not strive to create greater certainty thanParliament has chosen to provide The approach we have outlined gives as much conceptual clarity as can reasonably be achieved [132] My [124] to [131] do not accord with the approach to step 2 contended for by the BNZ. The Bank submitted that the focus at step 2 remained on the legislation – on the specific provisions with, now, the overlay of s BG 1. The question for the Court is essentially whether the asymmetry that was at the core of the transactions is within the scheme and purpose of the legislation, and thus within Parliament's contemplation. Mr Galbraith also posed the question in this way: is there something in the scheme and policy of the Act that prevents the 'packaging up' or 'conjunction' of the specific provisions in the way in which the BNZ packaged them up or conjoined them in the transactions? If the asymmetry at the heart of the transactions was within the purview of the legislation, then the step 2 question posed at [125] is answered 'yes'. [133] Although Mr Galbraith accepted that I can look at the transactions in detail, the approach he contended for seems to render any more detailed analysis of them superfluous. Mr Brown made this point in opening for the Commissioner, submitting that the Court in Ben Nevis (particularly in [107]) makes it plain that step 2 does not involve a further investigation of what Parliament contemplated when enacting the specific provisions. Mr Brown inquired rhetorically: if, as the Court stipulates, there is to be a purposive approach taken to interpreting and applying the specific provisions at the "black letter" (pre-avoidance) stage, why would there be further inquiry as to Parliament's purpose at the BG 1 stage? [134] In [101] in Ben Nevis the Supreme Court again makes the point – it had earlier been made in the submissions of counsel for the Commissioner referred to by Cooke J at 541 in Challenge – that no GAAR can anticipate all the results of taxpayers' ingenuity in crafting arrangements. Thus Parliament could not, and will not, have contemplated the particular arrangement in issue. That arrangement is likely to deploy a number of statutory regimes or provisions. I agree with Mr Brown's submission for the Commissioner that it is unreal to suggest that Parliament, when it enacted the deductibility and subvention provisions and the FTC and conduit regimes, might actually have contemplated transactions structured as are those in issue in these proceedings.[135] It follows that I agree with the Commissioner's submission that the question for the Court at step 2 is necessarily an hypothetical one. Guided by the considerations and the approach set out by the Supreme Court in [108] and [109] inBen Nevis, the Court is essentially asking itself: had Parliament foreseen transactions of this type when enacting the specific provisions deployed in the transactions, would it have viewed them as within the scheme and purpose of those specific provisions? [136] I am unable to reconcile the Bank's step 2 approach with Ben Nevis. Because it involves the same literal interpretation of the specific provisions required at step 1, it seems to me to read down s BG 1. Yet step 2 requires the Court to give s BG 1 full effect in tandem with the specific provisions, by looking to whether Parliament envisaged or intended such a use of specific provisions. The Bank's approach risks being categorised as the "threshold" argument expressly rejected in Ben Nevis i.e. that a taxpayer within the applicable specific provisions is in a safe harbour, immune from attack under s BG 1. And, indeed, that was precisely the way the Commissioner did categorise the Bank's argument – as "no more than a dressed-up "threshold" scheme and purpose argument ". The Commissioner made a similar criticism of the Bank's argument that the arrangement cannot be said to be beyond Parliamentary contemplation because Parliament expressly considered it. Complaining that this argument altogether failed to apply the criteria and indicia set out in Ben Nevis [108] and [109], the Commissioner contended that this was a "subtle attempt to breathe life" into the threshold approach rejected in Ben Nevis. [137] To summarise, I proceed on this basis: a) Step 1 requires me, upon an ordinary interpretation of the applicable specific provisions, to decide whether the arrangements comply with those provisions. b) Step 2 requires me to decide, upon the scheme and purpose of the Act including s BG 1, whether the legislature would have contemplated and intended that the specific provisions be deployed as they were deployed by the taxpayer in the transactions in issue.[138] As the taxpayer has to surmount both steps, the precise scope of each could be said to matter not. But, as a two step approach is stipulated, certainly by the main judgment in Ben Nevis, I have tried correctly to understand what each step entails.Step 1 of Ben Nevis[139] As mentioned, with one exception, the Commissioner concedes that each of the six transactions complied with the 'technical' or 'black letter' requirements of the applicable specific provisions. [140] Both the BNZ and the Commissioner made detailed submissions about the scheme and purpose of those specific provisions. The Bank contended that Ben Nevis confirms the specific statutory provisions "must be front and centre stage in the analysis conducted by the Court, as they provide the framework that defines the scope of any tax avoidance inquiry". [141] I anticipate that the BNZ considers the correct place for me to consider those detailed submissions is at the step 2 stage in Ben Nevis, whereas the Commissioner would favour considering them at step 1. Although, as I have suggested, it may not matter which is correct, I intend considering those submissions at the step 2 stage. [142] I return to the one respect in which the Commissioner disputes compliance with the specific provisions. The Commissioner does not accept that the guarantee arrangement fees and guarantee procurement fees are properly deductible. In [4] I identify that as the third issue, although strictly antecedent, and I deal with it now.The deductibility of the GAF/GPF (Issue 3)[143] As I outlined it at [4], this issue is whether the GAF or GPF charged in each transaction is properly deductible under s BD 2. Some explanation, and consequent refinement, of that issue is needed. [144] The BNZ submits:a) The guarantee fees are expenditure under Part EH (the Financial Arrangement Rules) and therefore deductible regardless of the income to which they relate. b) Alternatively, the guarantee fees are deductible under ordinary deductibility tests.BNZ's primary argument[145] The BNZ submits that a guarantee provided for a fee falls within the definition of "financial arrangement" under the Financial Arrangement Rules in Part EH. As it is a financial arrangement, the net expenditure is deductible under s DD 1(3):DD 1(3) Expenditure on interest by company Subject to section DB 1(1)(e) and despite subsection (1)(b), expenditure on interest is an allowable deduction of a company.[146] As the Gen Re 1 and CSFB transactions were entered into before 20 May 1999, the accrual rules applying for financial arrangements are those in Part EH Division 1, ss EHA1 to EH 18, including this definition of "financial arrangement" in s EH14:"financial arrangement" means (a) (b) any arrangement (whether or not such arrangement includes an arrangement that is a debt or debt instrument, or an excepted financial arrangement) whereby a person obtains money in consideration for a promise by any person to provide money to any person at some future time or times, or upon the occurrence or non-occurrence of some future event or events (including the giving of, or failure to give, notice), and (c) but does not include any excepted financial arrangement that is not part of a financial arrangement.The emphasis is mine because it is the focus of the disagreement as to whether the GAFs meet the definition.[147] The contractual obligation to pay the GAF in the Gen Re 1 transaction is in section 7 of the RPS Subscription and Guarantee Arrangement Agreement dated 1 July 1998 between BNZIS1 and BNZI. In the CSFB transaction it is also in section 7 of the identically entituled agreement, this time dated 21 August 1999 between BNZIS2 and BNZI. In each agreement BNZI paid a 2.95% p.a. fee to BNZIS 1 and BNZIS2 respectively to "arrange" for the repo counterparty to procure the parent guarantee. The agreements describe this internal fee as a "guarantee arrangement fee". It is the deductibility of that GAF which the CIR disallowed. The consequent agreements in the Gen Re 1 and CSFB transactions pursuant to which BNZIS1 and BNZIS2 respectively paid the repo counterparty a GPF are not in issue. [148] The CIR disallowed the GAF for these reasons: a) Upon a plain reading of the RPS Subscription and Guarantee Arrangement Agreement for each of the Gen Re 1 and CSFB transactions, the GAF was not paid in consideration for the parent guarantee or the right to be paid an amount under the parent guarantee in the event of default. The fee was paid in consideration for BNZIS1 or BNZIS2 arranging the parent guarantee. The requirement in the definition for a person to provide money to another person when a future event occurs or does not occur is absent from the act of arranging the parent guarantee; and b) It is not sufficient for the fee to be indirectly connected to the parent guarantee; it needs to be "in consideration for" the guarantee. The BNZ's ex post facto recasting of the arrangement fee as a "guarantee fee" or as a "specific fee payable for the guarantee" acknowledges this requirement, and is symptomatic of the difficulty the BNZ faces in coming within the definition of a financial arrangement. [149] I do not accept these arguments. The definition of "arrangement" in s OB 1 (I have set it out in [112]) encompasses all the agreements that comprised the Gen Re 1 and CSFB transactions respectively.[150] I consider the definition of "financial arrangement" must be given the same wide scope. In its submissions the BNZ referred to Glazebrook and others "The New Zealand Accrual Regime" 2nd edition, 1999 CCH New Zealand Ltd at paragraph 202 as confirming the width of the definition of "financial arrangement": the wording of this part of the "financial arrangement" definition is specifically designed to cover more complex arrangements; for example, it covers the situation where A lends money to B, in consideration for which C makes payments to D. Here, a third party, C, makes payments to a fourth party, D, with respect to a "loan" provided by A to B. In other words, the "financial arrangement" definition cannot be avoided by inserting different parties into a transaction. The wording of this part of the definition was not solely anti-avoidance motivated. It also ensured that complex commercial transactions, such as debt defeasances and assignments of income, are "financial arrangements".[151] For what it is worth, the one counterparty witness, Ms Miller, who was with Gen Re at the relevant time, said Gen Re: did not make much distinction between a guarantee procurement fee and a guarantee fee; it seemed to us that the distinction did not have much of a difference to us. (T 1627-1628)[152] Thus, I accept that each of the Gen Re 1 and CSFB transactions was an "arrangement whereby a person (the repo counterparty) obtained money in consideration for a promise by another person (the parent) to provide money (the guarantee payment) to any person (the BNZ) upon the occurrence of some future event (default by the repo counterparty)", and thus came within the definition of "financial arrangement" in s EH 14. [153] In the case of each of the Gen Re 1 and CSFB transactions, that "financial arrangement" included: a) BNZI's purchase of securities under the repo for $500 million; b) BNZI's sale of the securities back to the repo counterparty for $500 million, subject to any adjustments on settlement;c) The payment of the 2.95% p.a. guarantee arrangement fee by BNZI to BNZIS1 or BNZIS2; d) The payment of the 2.95% p.a. guarantee procurement fee by BNZIS1 or BNZIS2 to the repo counterparty to procure the parent guarantee; and e) The parent guarantor's promise to pay BNZI in the event of default by the repo counterparty of its obligations to BNZI. [154] As the $500 million flows cancelled out, and the guarantee was not called upon, the net result is expenditure in the sum of the GAF. Section DD 1(4) provides that "'interest' includes expenditure incurred under Part EH". Accordingly, that expenditure is deductible under s DD 1(3), as expenditure on interest by BNZI. It follows that I hold that the GAFs paid by BNZI to BNZIS1 and BNZIS2 in the Gen Re 1 and CSFB transactions respectively were deductible under s DD 1(3). [155] As the Gen Re 2, Rabo 1, Rabo 2 and Lehman's transactions were all entered into after 20 May 1999, the accrual rules in Division 2 (ss EH 19 to EH 59) apply. The rules in Division 2 have their own definition of "financial arrangement" in s EH 22(1). This provides:A financial arrangement is (b) an arrangement (that may include a debt or debt instrument or an excepted financial arrangement) under which a person receives money in consideration for a person providing money to any person (ii) when an event occurs in the future or does not occur (whether or not the event occurs because notice is or is not given).[156] Adopting, without repeating, my reasoning in [149]-[153], I consider this somewhat different definition easily encompasses the GPF in the last four transactions, with the result that the GPF is deductible under s DD 1(3).[157] In respect of the Gen Re 2, Rabo 1, Rabo 2 and Lehman's transactions, had the act of procurement been performed on or after the execution of the agreement containing the obligation to pay the GPF, the Commissioner accepted there would have been a short term agreement for the sale and purchase of services, and that the GPF would have been deductible on that basis. That results from a combination of the s OB 1 definition of a "short term agreement for the sale and purchase of property or services" and ss EH 23(2) and EH 24(1)(p). However, the Commissioner pointed out that provision of the parent guarantee was a condition precedent to entering into each of the Gen Re 2, Rabo 1, Rabo 2 and Lehman's repo agreements. It must follow that the act of procurement preceded the transaction documentation containing the obligation to pay the GPF. [158] On this aspect of the case, each party pointed to what it submitted was inconsistency on the part of the other. The CIR pointed to the BNZ's recasting, in these proceedings, of the 2.95% GAFs and GPFs as "guarantee fees". He contended this was inconsistent in two respects. First, it was inconsistent with the deliberate categorisation of those fees, in the transaction documentation, as being paid in consideration for either the arranging or procurement of the guarantee and/or the collateral. Secondly, it was inconsistent with the statements the BNZ made to the Commissioner when applying for private binding rulings referred to in [8]. For example:A procurement fee will be paid to AIG-FP in consideration for AIG-FP procuring the guarantee of AIG. (7/3941 – in he AIG 1 application) Although "guarantee fee" is not defined in the Act, a "guarantee fee" is payment to a guarantor to provide a guarantee BNZI does not pay a fee to a person who provides a guarantee (such as AIG). The only payment made to BNZI is to AIG-FP-sub for procurement services. (43/31319 – in the AIG 2 application)[159] The BNZ countered by suggesting there was an irony about this because, in those binding rulings, the Commissioner had concluded that the GPF payable byBNZI was expenditure under a financial arrangement and deductible accordingly, directly contrary to the Commissioner's submission in these proceedings. [160] Beyond making one point, I do not see these arguments as helpful. My task is to determine the deductibility of the GAFs or GPFs in issue in these proceedings. [161] The one point is that the evidence certainly establishes concern on the BNZ's part, in structuring these transactions, to put "space" between the GAF or GPF, and the provision of the parent guarantee, lest s CN 4 apply requiring tax to be withheld on those fees. That would have adversely affected the economics of the transaction. The BNZ did not withhold tax. Whether it should have is not an issue in these proceedings. But, certainly, the BNZ's position on the GAFs/GPFs when structuring these transactions was the antithesis of its case now.Alternative argument[162] As the BNZ has succeeded on its primary argument, I deal with its alternative argument only briefly. The BNZ relies on the ordinary deductibility tests. It accepts it must bring itself within s BD 2(1) and must also establish that it does not fall within any of the exclusions in s BD 2(2). [163] The BNZ relies on s BD2(1)(b)(i) or (ii) which provide:BD 2 ALLOWABLE DEDUCTIONS BD 2(1) DEFINITION An amount is an allowable deduction of a taxpayer (a) (b) To the extent that it is an expenditure or loss (i) incurred by the taxpayer in deriving the taxpayer's gross income, or (ii) necessarily incurred by the taxpayer in the course of carrying on a business for the purpose of deriving the taxpayer's gross income, or The two clauses are similar, in that the key is expenditure incurred in deriving gross income. [164] The CIR submits, and the BNZ accepts, these provisions require the BNZ to establish a nexus between expenditure and the derivation of gross income. The BNZ's submissions on that nexus are: a) The parent guarantee protected against loss and value of the repo securities held by BNZI, because it secures the repo counterparty's obligation to repurchase those securities; b) The securities are revenue rather than capital assets, and thus their disposal (upon the sale back to the repo counterparty) gives rise to gross income; and c) In the alternative to b), if the disposal of the repo securities did not result in the BNZI deriving gross income (the CIR's submission in relation to Gen Re 1 and CSFB), the guarantee proceeds would have been gross income, providing the necessary nexus with gross income. [165] Next, the BNZ contends that none of the exclusions in s BD 2(2), in particular (b), apply. That sub-section provides:BD 2(2) EXCLUSIONS An amount of expenditure or loss is not an allowable deduction of a taxpayer to the extent that it is (a) (b) incurred in deriving exempt income That is so because the guarantee was not a guarantee of the obligation of the counterparty to pay the tax exempt distribution. It was a guarantee of the counterparty's repo obligation. [166] Although the BNZ accepts that the guarantee fee could have been priced into the tax exempt dividend (i.e. could have been implicit, rather than explicit in theform of the guarantee fee), it was not the parties' choice to transact in that way. The Court is concerned with the legal structures and obligations the parties created, not with any alternative course which they could have adopted but chose not to:Finnigan v CIR (1995) 17 NZTC 12,170 at 12,174 per Richardson J delivering the judgment of the Court of Appeal; Ben Nevis at [47]. [167] This alternative argument cannot succeed. In the Gen Re 1 and CSFB transactions, BNZI did not hold the repo securities; it held securities in BNZIS1 and BNZIS2 respectively. Assuming BNZ's arguments a) and b) are correct in principle, the GPF paid by BNZIS1 and BNZIS2 will be deductible by those companies, but not by BNZI. The guarantee in the Gen Re 1 transaction provides that any monies paid out by the parent under the guarantee shall be paid "to BNZI instead of to BNZIS1" (13/8059 – clause 1 of the guarantee). In the CSFB transaction the guarantee provides that any monies paid out under the guarantee shall be paid "to BNZI or such investor affiliate, as applicable" (21/14553). Thus, assuming the payment is gross income, it is derived by BNZI. However, the BNZ has not established the required nexus between the GAF and the derivation by BNZI of gross income. Mere assertion by BNZI that it held the securities in BNZIS1 and BNZIS2 on revenue account, with the consequence that any monies paid upon the guarantee would be gross income derived by BNZI, is not sufficient to establish that that is the case. [168] The Gen Re 2, Rabo 1, Rabo 2 and Lehman's transactions are different in that they did not have the GAF paid by BNZI to BNZIS1 and BNZIS2 in the Gen Re 1 and CSFB transactions respectively. It is the GPF paid by BNZI to the repo counterparty that the Commissioner has disallowed. The BNZ strikes the same initial difficulty that mere assertion that BNZ held the repo securities on revenue account does not establish that to be the position.[169] Further, I accept the Commissioner's point that the nexus is in fact between the payment of the GPF and the distribution. The transaction documents obligated BNZI to pay the GPF to the counterparty and the counterparty to pay the distribution to BNZI. [170] The BNZ's alternative argument also strikes a further, related obstacle. An expenditure is not an allowable deduction under s BD 2 to the extent that it is incurred in deriving exempt income: s BD 2(2)(b). The distributions derived by the BNZ from the counterparties were exempt from tax under the applicable FTC (for Gen Re 1) or conduit relief (for the other five transactions) provisions. [171] Even if, contrary to the view I have just expressed, there was a nexus between the GPF and the potential payments under the guarantee, then the BNZ needed to prove the extent of that nexus, because that is also the extent of the deductibility of the GPF. The transaction documents show that payments under the guarantee would have included any outstanding distributions, which were exempt income. To the extent it 'covers' those outstanding distributions, the GPF is non- deductible. The quality of those payments is not altered by their incorporation as part of the "repurchase price". [172] For those reasons I hold against the BNZ's alternative argument.Step 2 of Ben NevisThe legislative policies of the specific provisionsIntroduction[173] The BNZ made a detailed submission about these, contending that Ben Nevisconfirms the specific statutory provisions "must be front and centre stage in the analysis conducted by the Court, as they provide the framework that defines the scope of any tax avoidance inquiry". [174] I have already mentioned (in [136]) that the Commissioner criticises this approach as an attempt to resurrect the threshold argument. Mr Galbraith expresslydisavowed that this amounts to a rerun of the threshold argument rejected by the Supreme Court in Ben Nevis. In contrast to that argument, it was the BNZ's submission that, following Ben Nevis, once deductibility is established under the relevant specific provisions, s BG 1 requires the Court to consider the use of the specific provision in the light of the arrangement as a whole, and having regard to the commercial reality and economic effect of the use made of the specific provisions. Viewed in that way, are the tax consequences of the arrangement consistent with what Parliament contemplated? [175] The Commissioner made a detailed response to the BNZ's submissions as to the legislative policies behind the applicable specific provision. As I indicated (at [141]), I consider opposing arguments at this stage, before turning to a more detailed factual analysis of the transactions.The deductibility of expenditure in the transactionsIntroduction[176] With the exception of the GAFs and GPFs, the Commissioner accepts that the expenditure deducted by the BNZ met the tests under the applicable specific provisions. That expenditure comprises the Bank's funding costs of the transactions, and the net expenditure it incurred on the interest rate swap in each transaction. That net expenditure was the difference between the interest rate on the fixed leg of the swap and BBR, which the Commissioner's assessments equate with the Bank's funding costs. The Bank does not challenge that aspect of the Commissioner's assessments. It did not match fund any of these transactions, but funded all of them from its general borrowings on the New Zealand money market. It does not know its precise cost of funds and is not in a position to challenge the Commissioner's assumption that they equated to BBR. [177] As for his objections to the Bank's entitlement to FTCs for the Gen Re 1 transaction, the Commissioner has disallowed the Bank's expenditure in the transactions on the basis that it is not within the "scheme and purpose" of theapplicable provisions, when interpreted in tandem with s BG 1. I will return to this after detailing the Bank's argument on this aspect.The legislation[178] The Bank pointed out that there is no longer a requirement that interest expenditure relate to particular income in order to be deductible. The previous tracing requirements were abandoned in legislation introduced in 2001, but with retrospective effect from the 1997-98 income year. In and from that income year, the deductibility of interest has been limited only by the thin capitalisation and EIA regimes. [179] The only exceptions to this are some restrictions on the deductibility of expense incurred in deriving exempt income. Not only are those restrictions not applicable here, but what is applicable is s DD 1(3) and (4) which expressly provide that expenditure on interest to derive exempt income in the form of dividends is an allowable deduction for a company. The Bank submitted:Parliament has therefore expressly approved of the very asymmetry that lies at the heart of the Commissioner's allegation of tax avoidance.[180] These provisions are:DD 1(3) Subject to section DB 1(1)(e) and despite subsection (1)(b), expenditure on interest is an allowable deduction of a company.DD 1(4) In subsection (3) – A company does not include – (c) A company within a wholly-owned group of companies if one or more companies within the group derives exempt income, unless all of the exempt income is one or more of the following: (i) Dividends; or [181] With effect from 1 July 2005, Parliament substituted a new thin capitalisation regime for banks, replacing the existing thin capitalisation and EIA regimes. Theycontinue to apply to non-banking entities. The amending Act inserted new sections FG 8B to FG 8J into the Act, and made consequential amendments. [182] The new regime substitutes a net equity approach for the debt percentage approach under the thin capitalisation and EIA regimes. Although the on-lending concession was not carried through to the new rules, they nevertheless stipulate different levels of equity funding for different assets: a) Most assets, including the bank's loan book, must be equity funded to the extent of 4% of their "regulatory value", as defined in s FG 8F(2). b) Certain equity investments in a non-resident entity, including any potentially qualifying for conduit or full FTC relief, must be 100% equity funded. [183] This new thin capitalisation regime leaves the conduit regime, the FTC regime, and the provision allowing a deduction for expenditure on interest incurred in deriving exempt dividends intact.Opening contentions[184] Based on its review of the relevant specific provisions, BNZ made three submissions. First, the legislation expressly provides that the receipt by a company of exempt dividend income does not restrict its ability to deduct interest incurred in deriving that income. This reflects a conscious decision by the legislature. The Court can take it that Parliament has expressly approved "the very asymmetry that lies at the heart of the Commissioner's allegation of tax avoidance". [185] Secondly, Parliament enacted the thin capitalisation rules and, of more direct relevance here, the EIA rules, specifically to deal with the very situation that arises in these proceedings – conduit-relieved income derived from deductible interest expenditure. The Commissioner accepts the BNZ was within the "safe harbour" provided under both regimes. Applying Ben Nevis, it is inconsistent with the legislative policies to attack the BNZ within the "safe harbours" Parliament hasprovided. Parliament cannot have intended that, given the interest deductions arise in respect of arms length borrowings, and in order to finance an investment with a third party. [186] Thirdly, the Bank submitted that the new thin capitalisation regime for banks enacted in 2005 was "remedial legislation", in the sense that it was Parliament's response to transactions of the type in issue here. Even so, that legislation does not proscribe or "close down" (the Bank's phrase) transactions of this type. It certainly makes them more expensive for a bank, in terms of the cost of capital, but it does not oust them. [187] The Commissioner's response to these submissions was in general terms, and was centred around this passage in the judgment delivered by William Young P for the Court of Appeal in Accent Management Ltd & Ors v Commissioner of Inland Revenue (2007) 23 NZTC 21,323 at [126] (I am citing a slightly longer passage than did the Commissioner in his submissions): When construing specific rules and looking for their scheme and purpose, it is necessary to keep general anti-avoidance provisions steadily in mind. On this basis, it will usually be safe to infer that specific tax rules as to deductibility are premised on the assumption that they should only be invoked in relation to the incurring of real economic consequences of the type contemplated by the legislature when the rules were enacted. Further, it also seems reasonable to assume that deductibility rules are premised on a legislative assumption that they will only be invoked by those who engage in business activities for the purpose of making a profit. [188] Although not submitting that the mere existence of a tax advantage rendering an otherwise unprofitable transaction attractive automatically gives rise to tax avoidance, the Commissioner made these points: a) It is unusual for tax consequences to enhance profit: tax policy typically involves the state taking from, rather than giving to, a taxpayer. Where (as here) transactions involve the state giving to the taxpayer, the question must arise: how do those transactions not frustrate the scheme and purpose of the legislation?b) Whenever a transaction is more commercially attractive post-tax than pre-tax, it necessarily involves a favourable shift in the incidence of tax with the consequences that:• At least one of the components of the definition of "tax avoidance" will be present.• Any tax avoidance will inevitably be "more than incidental". c) The answer to the question posed in a) must involve exceptions – specifically envisaged and legislated for – to the "legislative assumption" referred to by the Court of Appeal in Accent Management. d) The only express legislative exceptions the BNZ has pointed to are that s DD 1 does not limit interest deductibility for companies to generating assessable income, and the limits on interest deductibility imposed by the thin capitalisation and conduit EIA regimes. Although the Commissioner accepts the effect of those provisions, all of them "fall far short of specific legislative endorsement of these transactions".Summary[189] The BNZ's points convincingly demonstrate: a) A company can deduct interest even if extended to derive dividends which are exempt from income tax. b) The deductibility of interest is not governed by the ability to trace it to particular income. c) The BNZ was within the income deductibility limits contained in the thin capitalisation and EIA regimes.[190] Those are cogent points, but the Commissioner is surely correct in submitting that they fall short of demonstrating legislative contemplation, let alone approval, of transactions structured as were the six in issue here. In particular, I do not accept that these points establish that Parliament contemplated transactions with the asymmetry fundamental to these ones.The FTC regime (utilised by the Gen Re 1 transaction)[191] At a 'technical compliance' level, the Commissioner conceded the BNZ's entitlement to the foreign tax credits it claimed for the Gen Re 1 transaction. The entitlement was because GFRT paid tax in the United States on BNZIS1's income from GRFT. However, based on the "scheme and purpose" of the FTC regime, the Commissioner advanced two objections to the BNZ's entitlement to the FTCs claimed for the Gen Re 1 transaction. These objections are appropriately viewed as based upon an interpretation of the FTC regime 'in tandem' with s BG 1 – the interpretative approach suggested at [103] in Ben Nevis. The Commissioner's two objections were: a) BNZIS1 was kept free of expenditure. b) The foreign tax paid by BNZIS1 was not really borne by either the BNZ or the counterparty because, although GRFT paid tax in the United States, GRCF took an equivalent tax deduction. [192] The background to the first of these objections is that an FTC is limited, by a combination of ss LC 1(5), LC 2 and LC 14, to the lesser of the overseas tax actually paid and the New Zealand tax that would otherwise be payable on the income. BNZIS1 was incorporated specifically for the Gen Re 1 transaction. Capitalised by BNZI, it was BNZIS1 which invested in the income units in GRFT, and received distributions of interest from GRFT. These distributions were taxable in BNZIS1's hands. The Bank accepts that the transaction was structured to keep BNZIS1 free of net expenditure, so that it could obtain the maximum FTC to set against the tax that would otherwise have been payable on the interest distributions from GRFT. For example, although BNZIS1 paid the GPF to GRFT, it was effectively reimbursed byBNZI by the GAR. Further, any monies due under the guarantee were payable to BNZI and not BNZIS1. [193] The Bank submitted that the transaction had been structured in this way to shield BNZI (an entity with substantial capital and assets) from the risk of being liable for debts of GRFT, in the case of the latter's insolvency. The Bank referred to the evidence of Mr Brown, who was called by the Bank. In the event of the insolvency of GRFT, Mr Brown stated that there was potential for the Trust's creditors to "look through" the Trust to BNZIS1. The Bank maintained that that potential also explained why the proceeds of the guarantee, if called upon, were to be paid to BNZI and not to BNZIS1. [194] Mr Brown was a retired partner of Debevoise & Plimpton LLP, an international law firm. He had a long career advising on commercial transactional work. There was no challenge to Mr Brown's evidence, which I therefore accept. However, unless I have overlooked it, there is no indication in the documentation relating to the Gen Re 1 transaction, that the risks explained by Mr Brown were a factor in the structuring of Gen Re 1. But there are plenty of indications that it was structured to enable BNZIS1 to obtain the maximum FTCs. Accordingly, I hold against the BNZ's submission that the structuring of the Gen Re 1 transaction was "consistent with the achievement of commercial (and in particular, risk mitigation) objectives". [195] Further, the BNZ submitted that the structuring of Gen Re 1 involved a legitimate use of separate entities to take maximum advantage of the available FTCs. I agree with the Bank that Ben Nevis [129] provides support for the first of those aspects: the use of companies and trusts as separate taxpayer entities will normally be an acceptable mechanism for taking advantage of concessions available under specific provisions, being within what Parliament must have contemplated in enacting them. In that context, the obligations will be those of the entities which incur them under any arrangements and not others such as shareholders [196] I also accept that the judgment of Smellie J in Charity Finance Ltd v CIR(1998) 18 NZTC 13,565 at 13,569 supports the view that it is legitimate for ataxpayer to arrange its affairs so as to maximise a FTC. I think that support is tangential at best, because Smellie J was not confronted with the present issue, or anything remotely resembling it. The situation in Charity Finance was that the Commissioner had disallowed the taxpayer's claim to a FTC because losses within the taxpayer group of companies had been offset so that the taxpayer company had no tax to pay. There was, therefore, nothing against which the FTC could be applied. It was in that context that Smellie J commented: The appellant company so arranged its affairs that there was no "New Zealand tax payable in respect of" the income earned on the Australian investment. Had the company not chosen to off-set losses equal to the income earned there would have been room for the (FTC) [197] The BNZ also took issue with what it termed the Commissioner's "group" approach to the FTCs claimed in Gen Re 1, because the Act is specific as to the situations in which a "group" approach to FTCs is permissible. The Bank referred to s LC 5 which permits transfer of an FTC from one to another company within a group in carefully prescribed circumstances, and to s LC 16 dealing with the availability of FTCs to members of a consolidated group of companies. [198] The Commissioner's second objection is more problematic. Although the Commissioner accepts that BNZIS1 "technically bore the cost of the US tax so as to entitle it to receive a foreign tax credit", he takes the point that GRCF took "an equivalent tax deduction". That occurred because, through the US IRS "check the box" process, GRFT was treated as a corporation and GRCF as the owner of the equity interest in GRFT. When GRFT made a distribution it was deemed to have been paid to GRCF, and GRCF was deemed to have made a deductible interest payment to BNZIS1. GRFT was capitalised by GRCF to a level of $US562 million to enable it to make the required interest distributions to BNZIS1. In the result, the only net tax paid by the Gen Re Group (including GRFT, which was part of the group for accounting purposes) was on the profit Gen Re made from the $NZ500 million funding provided by the BNZ. [199] This emerged from the evidence of Ms Miller. In answering a question from Mr Galbraith in her evidence-in-chief, Ms Miller explained that the US tax and accounting treatment made it "costless for this type of capitalisation (of GRFT byGRCF) to be implemented" by a multinational such as Gen Re (T 1507). In the course of cross-examination by Ms Scholtens for the Commissioner, Ms Miller described the US tax treatment of GRFT as a "kind of virtual reality" (T 1643). [200] The US tax treatment of the three US transactions, including Gen Re 1, was explained by Mr Shay. I summarised his evidence in [92]>. Although Mr Shay's explanation was admirably simple, the detail of the US tax treatment is not, as a glance at the tax worksheet produced as Exhibit F will demonstrate. [201] Central to the Commissioner's second objection is that the economic burden of the US tax on the gross interest distributions made by GRFT to BNZIS1 has not been borne (because it has not been paid) by any member of the Gen Re group or by BNZIS1 or any other BNZ entity. The Commissioner submits that this result does not comply with either the letter or the "evident policy intent" of the FTC regime. The "letter" relied on by the Commissioner is s LC 1(3A) which disallows an FTC: if and to the extent that a taxpayer or a person pays foreign tax, and the taxpayer, the person or a person associated with either of them receives a refund or repayment of the foreign tax or another amount (whether in money or money's worth) or a benefit of any kind (including the remission of a debt) which is determined, directly or indirectly, by reference to the amount of the foreign tax paid or any part of that amount The policy is simply that of allowing a New Zealand taxpayer credit for tax it has paid in a foreign jurisdiction i.e. the policy of avoiding double taxation - the same income being taxed in two jurisdictions. [202] The Commissioner supports his policy objection in two ways. First, by reference to the February 1995 Government discussion document "International Tax", which described (at para 3.1.6, p23) the objective of FTCs in these terms:The objective of foreign tax credits is to avoid double taxation of the foreign-sourced income of residents. If a resident derives foreign-sourced income and that income is subjected to foreign tax, New Zealand provides the resident with a foreign tax credit.Secondly, by reliance on s LC 1(1), the first general section in Sub-Part C of the Act – Foreign Tax:LC 1(1) Credit for tax paid overseas Where a person who is resident in New Zealand derives gross income from a country or territory outside New Zealand, income tax paid in that country or territory in respect of that income shall be allowed as a credit against the income tax liability of the person (Commissioner's emphasis)[203] The BNZ's responses are two-fold. First, the BNZ reiterates its point that the Commissioner disregards the separate legal entities involved, contrary to Ben Nevisat [129]. Secondly, the fact that GRCF took a deduction for expenditure incurred does not detract from the tax cost having been borne by GRFT. Deductions for expenditure incurred are not "equivalent" to refunds of tax, contrary to the Commissioner's assertion. [204] I resolve this point in the Commissioner's favour. Although there may not be "equivalence" between the tax deduction taken by GRCF and the tax cost borne by GRFT, I consider s LC 1(3A) applies, with the result that the FTC is not allowed. I regard GRCF as "a person associated with" GRFT, in the sense that the two entities are connected. If the s OD 7(1) test of association must apply, then GRFT and GRCF are "associated persons" as defined in s OD 7(1), in that GRCF held the class B interest in GRFT. The deduction (i.e. refund or repayment) received by GRCF was "determined, directly or indirectly, by reference" to the tax paid by GRFT. [205] Further, although it may risk merging legal entities, I consider there is substance in the Commissioner's point that no tax corresponding to the FTC was in fact paid (the operative word in s LC 1(1)) in the United States. I accept the Commissioner's submission that actual payment of foreign tax is the policy foundation of the FTC regime. Unless there is such payment, there is nothing against which to credit a FTC. That is essentially the point Smellie J made inCharity Finance.The conduit regimeBackground[206] I draw this largely from the BNZ's submissions. It is generally uncontentious. [207] Preceding but very relevant to the conduit regime was the thin capitalisation regime introduced as Sub Part FG of the Act with effect for the 1996-1997 income year. The purpose of the thin capitalisation regime, as expressed in s FG 1, was to prevent a foreign controlled group of companies from having a disproportionately high level of debt funding in its New Zealand group of companies, relative to the world wide group. The controlling mechanism was the New Zealand group's debt percentage i.e. the group's total debt as a percentage of its total assets. [208] Particularly applicable to banks was the on-lending concession in s FG 6 which excluded, from the debt calculation, loans made on arms-length terms by a group. [209] The conduit regime was introduced two years later, with effect for the 1998- 1999 income year. Conduit relief was introduced against the background of the controlled foreign company (CFC) rules. These deal with the taxation of foreign equity investments, with the aim of reducing the ability of New Zealand resident taxpayers to avoid or defer New Zealand tax by accumulating income in foreign companies. The CFC rules attribute the income of the off shore company to its New Zealand owner. Dividends received from a CFC were also taxed under the dividend withholding payment (DWP) rules, although there was relief (in subpart MF) to ensure that the same income was not taxed under both the CFC and the DWP rules. [210] In something of a departure from the international norm for equivalent legislation, New Zealand's CFC rules do not distinguish between active and passive income. The former is derived from commercial operations, the latter has a "passive investment quality" e.g. dividends or interest. [211] The conduit regime was introduced to provide relief from the CFC and DWP rules, in situations where a non-New Zealand person invested in a foreign company through a New Zealand company i.e. where New Zealand was used as a "conduit" for investment. The 'mischief' the conduit regime was intended to remedy wasneatly summarised at 55-56 in the officials' report to the Finance and Expenditure Committee in respect of the Bill:New Zealand's controlled foreign company (CFC) and foreign investment fund regimes tax foreign-sourced income earned by New Zealand residents as it accrues. Because tax is imposed irrespective of the extent to which a resident company may be owned by non-residents, the indirect effect is to tax non-residents on foreign-sourced income. This is referred to as the conduit issue. One implication of this is that the competitiveness of New Zealand-based multinationalists is harmed. This is because New Zealand companies must raise capital on world markets, and non-resident investors can make global investments without paying tax to intermediate countries. This makes New Zealand-based companies less competitive for making investments into other countries if they are to pay the world rate of return to their non-resident investors. The Government's view is, therefore, that conduit relief should be extended, provided this does not create tax avoidance opportunities for New Zealand residents or opportunities for non-residents to reduce tax on their New Zealand-sourced income.[212] The conduit rules relieve dividends on such "conduit" investments from New Zealand tax, to the extent of a taxpayer's non-resident shareholding in the New Zealand company. [213] When developing the conduit regime the Government recognised the risks it posed to the New Zealand tax base through companies allocating an excessive amount of interest expense to their New Zealand operations, relative to the foreign interests for which the conduit relief was provided. Interest expense allocation rules to address this concern were foreshadowed in the Government discussion document "The Taxation of Conduit Investment" in May 1997: To address interest allocation concerns in relation to conduit investment, it may be necessary to create some rules similar to New Zealand's existing thin capitalisation rules to apportion interest expense between a group of New Zealand companies and their related CFCs and [foreign investment fund (FIF) interests] The Government accepts that interest expense allocation may be unwelcome as it could reduce the benefits that companies expect to enjoy from the conduit income tax relief. However, the Government considers that some effective interest expense allocation rules are essential to protect the New Zealand tax base.[214] Upon its introduction as the Taxation (Remedial Provisions) Bill (No 2) 1997, the conduit regime included specific excess interest allocation (EIA) rules. The commentary to the Bill explained these rules:In paragraph 3.30 of the discussion document – The Taxation of Conduit Investment - the inappropriate allocation of interest expense was noted to be a significant risk in accurately measuring conduit income, and the allocation of interest expense within a group as the easiest way to shift profits between companies. The interest allocation rules in new subpart FH aim to ensure that companies do not allocate an excessive amount of interest expense to their New Zealand operations, relative to foreign interests for which conduit relief is to be provided. If an excessive interest allocation exists, the excess will be reallocated against conduit income, thereby reducing the amount of conduit relief to which the New Zealand company will be entitled.[215] The EIA rules used the group debt percentage (including the on-lending concession I referred to in [208]) to determine allowable levels of borrowing. Compared with the thin capitalisation regime, the EIA regime was more restrictive in two ways: a) It lowered the debt percentage to 66% (75% under the thin capitalisation regime). b) It calculated that percentage only on investments in grey list countries (which included the UK and USA). [216] The on-lending concession allowed a bank's loan book (i.e. on-lending by the bank to its customers) to be 100% debt funded. Apart from that concession, the EIA regime applied to banks and other lending intermediaries. In particular, it applied to all a bank's equity investments. Thus, if a bank, already at its 66% debt limit, made, in a grey list country, a $500 million equity investment which qualified for conduit relief, it could debt fund $330 million of it, but needed to fund the balance of $170 million by investing additional equity capital.Legislative policies underlying the relevant provisions of the conduit regime[217] The Commissioner points to parts KH and LG of the Act, submitting that they contemplate three events occurring when the conduit tax provisions are engaged: a) A New Zealand company would be liable for income tax under the CFC rules or the DWP rules, whether as a result of a passive or active investment in an offshore company (the "first tax liability"); b) A non-resident withholding tax ("NRWT") obligation of 15% arising when the New Zealand company distributes its profits to its non- resident owners (the "second tax liability"); and c) The first tax liability being relieved but not the second tax liability. [218] For its part, the BNZ submits that two policies relevant to this case emerge from ss MI 2 and NH 7: a) Conduit relief applies to all equity investments including those with loan-like characteristics. b) Conduit relief is not dependent on the dividends being passed through New Zealand to the foreign parent. [219] As to the first of these two policies, the BNZ submits that, although equity investments with loan-like characteristics (e.g. redeemable shares) are common, and are expressly recognised in ss 68-75 Companies Act 1993, they are not excluded from the conduit relief regime. Taxation legislation does differentiate in other areas. For instance, excluded securities (such as fixed rate non-voting shares) are not counted when measuring ownership interests in a company pursuant to ss OD 3(3)(c)(i) and OD 4(3)(c)(i). [220] Parliament endorsed the Select Committee's rejection of a submission that such excluded securities not be treated as equity capital (primarily for the purpose of calculating the percentage of foreign ownership) in the proposed conduit tax reform.In its report the Select Committee pointed out that such capital was equity capital for tax purposes. [221] The BNZ submits this first policy is a critical point in these proceedings because the Commissioner submits that an "in substance" loan does not qualify for conduit relief. The BNZ refers to paragraph 123 of the Commissioner's NOPA for the Rabo 2 transaction:The Commissioner considers that BNZ Group structured these transactions as equity investments in order to avail themselves of the relief accorded to them under the conduit regime. There is an artificiality and contrivance to the arrangement that indicates a more than merely incidental purpose or effect of tax avoidance.[222] Thus, the BNZ submits that the Commissioner is attempting to read into the conduit regime a qualification Parliament consciously omitted. This, it contends, is an impermissible application of s BG 1. [223] The Bank argues that the Commissioner's approach also overlooks the use in New Zealand since the 1980s of equity instruments such as redeemable preference shares (RPS) in structuring financing transactions for tax purposes. It referred to the domestic RPS transactions which banks have entered into, and also to the fact that New Zealand corporates continue to raise funds by issuing preference shares. The Bank's submissions referred particularly to evidence given for the Bank by Mr Kerr, a director of Asia-Pacific Risk Management Ltd, and an experienced adviser in New Zealand's financial markets. No disrespect, but I do not see a need to refer in any detail to Mr Kerr's evidence. I say something about the domestic RPS transactions entered into by banks at [244]>. The preference share issues that New Zealand corporates are still undertaking offer fully imputed dividends, generally in situations where, for whatever reason, those dividends are less valuable to the issuing company than to the investor. [224] Responding to the BNZ's first policy point, the Commissioner submitted that the nature of the investment (whether it was active or passive, or had loan-like characteristics) was irrelevant. What was relevant was whether the New Zealand company was acting, or could act, as a conduit – having its income/distributionsrelieved from the first tax liability until it made a distribution itself to its non- resident shareholders, attracting the second tax liability. [225] I turn to the second policy, that conduit relief is not dependent on the dividends being passed through New Zealand to the foreign owner. The BNZ says this emerges from s NH 7. I think I can spare readers a replication of this migraine- inducing provision. The BNZ's point, as I understand it, is that, although s NH 7 makes detailed provision as to the calculation of the dividend withholding payment, it does not require that such a payment be made. In other words, it does not require that the New Zealand subsidiary pass on the dividends to its foreign owner. [226] The Commissioner accepts that the Act does not make the claiming of conduit relief conditional on the dividend being passed on in the form of a distribution by the New Zealand subsidiary to its non-resident owner. That is why the Commissioner concedes the BNZ's entitlement to conduit relief in the latter five of the transactions in issue. [227] Rather, the Commissioner's argument is that a 'scheme and purpose' analysis of the conduit regime demonstrates that Parliament contemplated such 'passing-on' of the dividends: the step set out in [217]b) above. The Commissioner drew attention to the speech of Ms Belinda Vernon MP upon the second reading of the Bill:As a result, taxation will be limited to the existing 15 percent non-resident withholding tax on such income distributed by way of dividends. (1997 NZPD, 2 December 1997, 5821)[228] The Commissioner says this legislative contemplation is reflected in, and effected by, s NG(2)(1)(a)(iii) and (vi) and (c) which impose NRWT of 15% on conduit tax relieved dividends. Section NG 1(1) provides that this applies notwithstanding any other provision in the Act or the Tax Administration Act. [229] The regime 'tracks' the conduit relief given at the first step, set out in [217]a) in the following way:a) Relief is given under sub-part KH and NH 7 (for income tax); b) That relief is credited to the taxpayer's conduit tax relief account established under s MI 4(1); c) The conduit tax relief account is then debited when the taxpayer makes a distribution to its shareholders (s MI 5(1)(a)); d) The credit attaches to dividends paid to non-residents in the manner required by ss MI 7 and 8. e) Section MI 9 contains rules against improperly transferring conduit tax credits; f) The conduit relief available is adjusted when a taxpayer's non- resident shareholding changes between the first tax liability arising and the second tax liability (s MI 5(1)(e)); and g) The provisions of LG 1 require an additional dividend to be declared contemporaneously with a conduit relieved dividend to a non-resident. [230] The Commissioner points out that the five 'conduit' transactions here did not function as conduits of dividends from the foreign subsidiary through the New Zealand subsidiary to the foreign owner by way of distributions. In particular: a) No distributions of conduit-relieved income were made by the BNZ. The conduit tax relief account for the BNZI shows credits accumulating year by year, with no debits from the account. The account for the BNZ has a nil balance with no movements at all. (The Commissioner points out that this was because there was no possible income or gain for the BNZ to pass on to the NAB: these were loss making transactions, apart of course from the tax benefits they generated.)b) Consequently, no NRWT was paid on distributions by the BNZ to its off-shore owner, the NAB, in relation to distributions from the latter's conduit investment. [231] The Bank makes four points in response to the Commissioner's submission that there is a 'pass through' requirement. First, the Bank submits the 'conduit' refers to the investment flow not to the income flow. It refers to a foreign investor directing investment via its New Zealand subsidiary into a foreign subsidiary, not to the income flowing back from that foreign subsidiary. [232] I do not agree. I think the correct position is that the "conduit" refers to the New Zealand subsidiary which is so labelled in Figure 1 which appeared both in the discussion document and the Bill: [233] Certainly this figure shows investment flows through the conduit, but it also shows income flowing back to the conduit – the income the Government was proposing to relieve from tax. In short, investment flowed in one direction through the conduit, and income earned by that investment flowed back in the other direction through the conduit. [234] The BNZ's second and related point was that conduit relief applies to relieve tax on attributed income (i.e. income earned by a CFC or FIF as calculated under the accounting profits or branch equivalent methods), as well as tax on foreign dividends received. In the cases of attributed income, there may not be any dividend or othercash inflow to a New Zealand entity, and thus no cash to pass on to the foreign investor/owner. [235] Although the Commissioner did not respond directly to this point, I anticipate he would accept that there may well not be a match (in time and/or amount) between the conduit relieved income and the distributions passed on to the foreign owner, with the consequence that the taxpayer's conduit tax relief account may stand in credit at times. I anticipate that the Commissioner would also accept that some of the conduit relieved income may remain in New Zealand. But the requirements I have listed at [229]b)-g) are consistent only with Parliament contemplating that some of the conduit relieved income would in due course be passed on to the foreign owner. Otherwise those requirements are pointless. It is also difficult to conceive of a foreign owner not requiring some return on the foreign investment it made through its New Zealand subsidiary. [236] It may be a small point, but I note, at the time it was entering into the conduit relieved transactions, the BNZ shared the Commissioner's understanding of the way the conduit regime would work. For example, in giving its approval on 2 June 1999 to the Gen Re 2 transaction, BNZ Taxation stated:The transaction relies on the conduit tax relief provisions of the Income Tax Act. Conduit tax relief provides that the US dividend income from the US unit trust be exempt from income tax in New Zealand. This tax relief can ultimately be passed on to National Australia Bank Limited using conduit tax relief credits and paying supplementary dividends. The only New Zealand tax impost will be non-resident withholding tax when the dividend is paid out of New Zealand as is the case with all dividend payments. The conduit regime came into effect in the 1998 income year upon filing of the appropriate conduit and dividend withholding payment elections. (29/22075)The emphasis is mine: it is when, not if. [237] Third, the BNZ contended that the Government's decision to retain NRWT on distributions of conduit relieved income out of New Zealand (i.e., in Figure 1 in [232], by the New Zealand subsidiary back to its foreign owner/investor), suggests ambivalence as to whether the New Zealand subsidiary would distribute such income to its foreign owner. In the discussion document, in a section headed "Appropriaterate of NRWT", the Government advanced two reasons why it was desirable that NRWT should continue to apply to distributions of conduit income. It added: The NRWT represents an additional cost that must be weighed by a company considering transferring profits from New Zealand.[238] I agree with the BNZ's submission that this suggests the Government saw retention of profits in New Zealand as a positive outcome. That is consistent with the objective of attracting foreign investment to New Zealand, with the resulting benefits to the country. These are summarised in paragraphs 1.10-1.12 of the discussion document. NRWT at the rate of 15% (the general rate is 30%) on conduit relieved income was duly enacted, by amendment to s NG 2. [239] Fourth, the Bank submitted that the Commissioner's submission sought to use s BG 1 to reverse Parliament's conscious policy decision, in enacting the conduit tax reform, to substitute a current-based relief mechanism for the then existing repatriation-based relief mechanism. Under that repatriation-based relief mechanism, foreign dividend income incurred dividend withholding payments (DWP) when received by a New Zealand resident. DWP was a tax at the company rate, less an allowance for foreign tax credits. The resulting DWP credits attached to dividends passed on by the New Zealand resident company, the non-resident recipient receiving a refund for the balance. To demonstrate the working of this: a) Upon receiving a $100 fixed rate dividend from its foreign subsidiary, NZ subsidiary pays $33 DWP to IRD. b) NZ subsidiary pays $67 cash dividend to its Australian parent, with DWP credit of $33 attached. c) $15 of that credit is applied to meet the NRWT payable on the dividend, the IRD refunding the balance of $18 to the Australian parent. [240] In paragraph 3.22 of the discussion document, the Government expressed a preference for the current-based relief mechanism, which the document had described in this way:Current-based relief mechanism3.17 A comprehensive conduit relief mechanism would provide for relief on CFC and relevant FIF income that is attributable to non-resident shareholders as that income is derived (current-based relief). This would achieve the policy intent of the regime (relief of New Zealand tax on foreign-sourced income earned on behalf of non-resident shareholders of a New Zealand company) without requiring current distribution of foreign-sourced income. 3.18 The mechanism would work by:• relieving a New Zealand company deriving attributed income from the tax otherwise payable on that income, to the extent the company has non-resident shareholders; and• requiring the New Zealand company to pay the benefit of this relief out to its non-resident shareholders when it distributes dividends sourced from conduit income.[241] The Taxation (Remedial Provisions) Act 1998 broadly enacted the current- based relief mechanism. [242] I do not see much force in this last point. Certainly, the then existing repatriation-based relief mechanism was replaced with a current-based relief mechanism, namely the conduit relief regime. But 15% NWRT remained on dividends passed on to the foreign owner. Indeed, the second bullet point under 3.18 of the discussion document (set out in [240] and referred to by the Bank in its submissions) can only be read as supporting the Commissioner's view of the "scheme and purpose" of the conduit regime. As in [236], it is when, not if.Summary[243] I consider the scheme and purpose of the conduit relief regime was/is that some (at least) of the conduit relieved income would be passed on as dividends by the New Zealand subsidiary to its foreign owner, attracting 15% NWRT.Redeemable preference shares (RPS)[244] This is a convenient place to mention RPS, about which I heard a good deal of evidence. I accept that the transactions in issue here had the same corecharacteristic as the domestic RPS transactions that New Zealand banks commonly invested in through the 1980s and early 1990s, until the exemption that made them profitable (or much more profitable than conventional lending) was removed, effective April 1992. [245] In his evidence for the BNZ Mr McLeod identified that core characteristic as the ability for the bank (shown as Investor Co in the diagram in [248] below) to claim an interest deduction on its funding to subscribe for RPS in a subsidiary (Investor Sub) which would then subscribe for RPS in the ultimate issuer (Issuer Co), and for the bank to receive, via its subsidiary, tax exempted dividends. [246] That resulted from the characterisation, for tax purposes, of the Bank's investment in Issuer Co as an equity investment, rather than a loan. It also needed to be an investment in shares in a group company, thus the interposition of Investor Sub. [247] I accept also that the investment could have been pre-tax negative for the bank e.g. if the bank's funding cost was 14%, but the dividend only 13%. It was the tax treatment that converted this to profit:• Bank's after tax interest cost was 9.38% (14% x (1- 0.33)).• But its income was a tax exempt 13%.• Result, a post-tax profit of 3.62% (13 – 9.38%). [248] I do not see how this much progresses the BNZ's case here. That RPS investment was profitable for the bank because legitimate tax treatment applied to it. This was explained by the IRD in the Tax Information Bulletin it issued in 1990 (Volume 1, No. 8 of February 1990). That Bulletin contained this diagram to illustrate the structure it was describing:[249] It is convenient to quote now from Mr McLeod's evidence:The statement confirmed that as far as interest deductibility was concerned, Inland Revenue recognised that a bank (Investor Co) could capitalise a subsidiary (Investor Sub) in order to obtain an interest deduction rather than lend to or invest in the issuer directly. The statement contained the following explanation of why the general anti-avoidance provision would not apply to that aspect of the investment: Section 106(1)(h)(ii) provides a special deduction outside the general scheme of the Act, which has its own nexus, not between expenditure and assessable income, but between expenditure and the acquisition of shares in a group company. Investor Co has incurred genuine expenditure and both Investor Co and Investor Sub are members of the same group of companies. The terms of section 106(1)(h)(ii) are therefore satisfied. The Department considers that section 99 would not apply to this arrangement because the scheme and purpose of the relevant provisions of the Act have not been frustrated and Investor Co is involved in genuine commitments and expenditure.(PB 5.18) [250] Section 106(1)(h)(ii) provided:106. Certain deductions not permitted – (1) Notwithstanding anything in section 104 of this Act, in calculating the assessable income derived by any person from any source, no deduction shall, except as expressly provided I this Act, be made in respect of any of the following sums or matters: (h) Interest except so far as the Commissioner is satisfied that – (ii) It is payable by one company included in a group of companies in respect of money borrowed to acquire shares in another company included in that group of companies.[251] Mr McLeod is managing partner of Ernst & Young New Zealand, and a former Chairman of that firm. He has been a specialist tax practitioner in New Zealand for over 25 years. Like Mr Hagan, who also gave evidence, his credentials as a taxation expert are impeccable. [252] The salient points emerging from that statement are: a) The profitability of the investment depended on the "special deduction" provided by s 106(1)(h)(ii) Income Tax Act 1976, which had its own nexus between expenditure and the acquisition of shares in a group company. b) It affirmed the nexus between expenditure and assessable income in the general scheme of the Act (incidentally, precisely the nexus referred to by the Court of Appeal in Accent Management at [126], which I have cited at [187]. c) The IRD considered s 99 (the predecessor to s BG 1) did not apply because the scheme and purpose of the Act had not been frustrated. d) Those points do no more than underscore the following issues in these proceedings: e) Were the BNZ transactions within the "special" provisions they sought to take advantage of? f) Did those provisions contain their own nexus between the BNZ's expenditure and the exempt income it derived. g) Has the scheme and purpose of the relevant provisions been frustrated here?Factual analysis of the transactions[253] As I indicated in [175] I now turn to a more detailed analysis of what I consider are the key elements or features of the transactions. I do this as a basis for answering (in [526] and following) the question: are the transactions safely within the policy of the specific provisions as submitted by the BNZ, or are they caught by s BG 1 as the Commissioner contends?"Off the shelf" nature of the transactions[254] Amongst the nine characteristics of structured finance transactions identified by Mr Stanton (I list them in [21]), were the involvement of specialist transaction arrangers, and the replicability of the transaction. Elaborating on the first of these, Mr Stanton explained that "arrangers" are typically firms who seek to earn fees by introducing a particular structure and assisting in developing it for a particular market or particular bank (PB 2.40). Explaining replicability, to the extent a transaction structure has been devised and put in place, Mr Stanton said that a structured finance team will typically look to repeat the technique in the country it was first devised for, and see if the structure can be adapted for use in other jurisdictions and seek out potential counterparties (PB 2.42). [255] Professor Rosenbloom agreed: Transactions of this type may also be developed by investment banks, accounting firms, or other independent actors, who present them to interested parties with a view to earning a fee for their efforts. (PB 3.14)[256] In [58] I chronicled how the CSFB transaction structure was first introduced to the BNZ, noting that Mr Kyle was required to sign a confidentiality undertaking in respect of the "Proprietary Information" comprising the transaction structure. [257] Interlocutory stages of these and comparable challenge proceedings brought by other New Zealand banks have included arguments about the discoverability of "other bank" documents. Essentially, each bank has contended that transactionsentered into by other banks are irrelevant. The Commissioner put in evidence a table headed "Key Features of Other Bank Template Transactions". The A5 format of that (and even then, the print is small) precludes its inclusion in this judgment. However, having reviewed the "other bank" documents shown to him, Mr Stanton identified these similarities: a) The other banks had a tax liability which they sought to reduce through transactions almost identical to the six in issue here. b) The transactions undertaken by the other banks all incorporated a GPF typically set at or around 2.95% (Mr Stanton noted one exception to this, a transaction with a fee structure that operated differently). c) As with the BNZ, the other banks had not undertaken proper analysis of the company that was being guaranteed. d) The transactions incorporated an interest rate swap with similar characteristics. e) SPVs with similar characteristics were a feature of the other bank transactions. f) The distribution rate was determined by a formula, based off the fixed leg of the swap, the GPF and the agreed split of tax benefits. g) All contained similar early termination provisions. h) All involved varying degrees of risk layoff. (PB 5.70) Those eight similarities, which are consistent with the key features shown on the Commissioner's table, justify the Commissioner's categorisation of the transactions in issue in these proceedings, and those between the five other banks and the Commissioner, as "template transactions".[258] I have no evidence as to the circumstances in which other banks entered into these transactions. However, the time line I have included at [37] shows the terms of the transactions entered into by the five banks overlapped, and shows some common counterparties e.g. AIG, CSFB, Gen Re, Lehman's and Rabobank. [259] The relevance of this "template" feature of the transactions, when considered in conjunction with factors such as the BNZ's controlled use of its tax capacity, is that it indicates that the BNZ used the transactions for a "tax avoidance" purpose i.e. reducing or avoiding income tax it would otherwise have paid, or altering the incidence of income tax.How common were transactions such as these?[260] The BNZ contended these transactions were altogether distinguishable from the Trinity scheme implicated in Ben Nevis. While the Trinity scheme had no known cohorts, structured finance transactions such as these were commonplace. The Commissioner accepted this submission of structured finance transactions generally. His challenge was to the words "such as these". [261] As it is not contested that structured finance transactions have been a common – I think increasingly common – feature of banking and finance for the last 20 years or so, extensive reference to the evidence to that effect is unnecessary. It is also necessary to be selective, given the veritable mass of evidence I heard about this. [262] Dealing first with the US, Mr Hicks said: A sale and repurchase arrangement (or "repo") which is treated for tax purposes as a loan in the borrower's/issuer's country, but as an outright sale/purchase in the financier's country, is an example. Repo transactions of this type, and of which the BNZ transactions are typical, are commonplace. (PB 2.12)[263] Mr Hicks also opined that international tax arbitrage will always exist because tax systems have never been harmonised and never will be. He pointed outthat tax harmonisation is not a policy objective of the OECD (PB 3.1-3.3). He expressed also his belief that: it is very difficult for a government tax policymaker or tax administrator to distinguish generally between "good" tax arbitrage and "bad" tax arbitrage, in circumstances where tax arbitrage is an inevitable consequence of rational commercial behaviour I the increasingly globalized financial markets. (PB 3.9)[264] In a lengthy section of evidence, Professor Rosenbloom said that financing transactions taking advantage of cross-border tax opportunities are "normal and unobjectionable" (PB 3.1-4.11, but specifically at 4.11). Turning to the transactions in question here, the Professor said they were but examples of the utilisation of such opportunities, the BNZ Group and the counterparties "dealing at arm's length" (PB 5.17). In particular, Professor Rosenbloom said that none of the GPF, the fixed- income return for the BNZ or the interest rate swap was "extraordinary" (PB 5.19- 5.20). [265] The Professor concluded:The transactions in question here were designed to maximize and share benefits offered by the fact that the tax laws of the United States and New Zealand view sale-repurchase agreements differently. There is nothing abnormal, unusual, or malevolent about this. As noted previously, commercial transactions invariably reflect tax planning, and the benefits of planning are frequently shared by parties dealing at arm's length. Viewing these transactions as examples of tax avoidance is tantamount to faulting taxpayers for taking advantage of benefits that the tax laws plainly make available. And it is particularly striking, as far as I am concerned, that the core of these transactions- the benefit flowing from the sale-repurchase agreement itself – is not challenged or even noted by the IRD. (PB 5.26)[266] Professor Rosenbloom's evidence mixed fact with opinion, and spiced it with a degree of moral judgment. Given the caveat at the end of [102] in Ben Nevis I will not be engaging in the latter. [267] This was Mr Gross'summary of his evidence about this aspect:a) Cross-border structured finance is a well-developed form of capital raising.b) Tax based cross-border structured finance is a significant source of capital funding for large US and UK financial institutions. c) Taxed based cross-border financing is based on differences in tax treatment i.e. an arbitrage of tax differences between countries. d) Such transactions often include interest rate and currency swaps, and guarantees or other mechanisms for delivering credit enhancement. These features, while they have tax consequences, are not what gives rise to the tax benefit that forms the basis of the transactions. Rather they are included to mitigate risks, such as interest rate, currency and credit risks. e) The transactions are not contrived but are real providers of capital, which follow current market practices and regulations in the countries involved. (PB 4.41)To b) Mr Gross added that the Bank of America had raised in excess of US$50 billion through tax based cross-border structured finance transactions. [268] Turning to the UK, Mr Nias said that cross-border financing transactions taking advantage of arbitrage had been "utilised for decades" (PB 27). He went on to say that the UK Government had recognised and accepted arbitrage activity was a feature of the way businesses plan their operations to achieve lower cost of funds. He quoted from a UK Government statement when targeted anti-arbitrage legislation was introduced in 2005. The Paymaster General had stated: The Government sees nothing wrong with arbitrage itself, which is inevitable, given the variations between national codes and the existence of laws that apply to some companies and not others. (P B 32)[269] Dealing specifically with the transactions in this case, Mr Nias said the three UK transactions followed "a straightforward well-developed structure which has been a common feature of structured finance transactions" (PB 60). In the summary section of his evidence, Mr Nias rounded matters up in this way:For a UK banking institution, this type of transaction forms part of its existing banking trade in the all embracing nature of merchant banking and is part of the "organic growth" of banking business and not in any way separate from its existing banking business. (PB 73) Repo transactions have been common place in the structured finance industry for a number of years; a fact acknowledged recently in the Court of Appeal by Lawrence Collins LJ: "Repo transactions are widely used by institutions operating in the financial markets": HMRC Commissioners v Bank of Ireland Britain Holdings Ltd [2008] EWCA Civ 58 (a case involving a UK/Ireland cross-border repo). In my view, the Rabo and Lehman transactions are typical examples of such repo transactions which have been a regular feature of the structured finance industry in the UK for over 10 years. (PB 75)[270] Responding particularly to the evidence of Professor Rosenbloom and Mr Gross, Mr Stanton expressed the view:While a company's internal group tax department will endeavour to minimise tax on a cross-border transaction, it is going too far to say that transactions designed primarily to reduce a bank's tax liability are a generally acceptable activity. (PB 6.11.2)Level of disclosure[271] Mr Stanton listed 'absence of publicity' as the last of the nine characteristics of structured finance transactions. Elaborating, he explained:Structured finance transactions are ones that tend not to be publicised. This goes against market practice, where banks go to great lengths to advertise the corporate transactions they have completed. This is most often done by way of providing transaction information to be included in "league tables", which then rank banks in terms of the volume of transactions for which they have been responsible. Other public disclosure of transactions is by way of advertising the transaction in the financial press, where the type of transaction and the name of the bank, or banks, responsible for the transaction will be described. (PB 2.51)[272] In the regulations under the UK disclosure regime, to which I referred in [86], one of the descriptions of arrangements that must be disclosed is:Confidentiality: arrangements which a promoter or user might wish to keep confidential from the revenue authorities.[273] I also heard evidence that structured finance transactions were typically not tombstoned. Mr Birch did not have tombstones for these transactions on his desk,and nor did Ms Miller for the Gen Re 1 and Gen Re 2 transactions. No advertisements in the familiar format ("this advertisement appears as a matter of record only") recording the details of the completed transactions were placed in the financial press. [274] Although I regard this feature of the transactions as comparatively unimportant, two aspects warrant mention. First, the BNZ did not apply for a binding ruling on any of these transactions. In the course of evidence, various reasons were advanced for this: the ability to rely on the binding rulings obtained for the AIG 1 and Morgan Stanley transactions; the similarity of these transactions to RPS transactions long accepted by the IRD; the unacceptable delay in obtaining a ruling; pressure – even direction – from the counterparty; the certainty of the tax position. [275] Irrespective of whether some or all these reasons were genuine or justified, the fact is that the BNZ did not apply for a binding ruling on any of these (substantially tax negative for New Zealand) transactions. Applying for a binding ruling obviously discloses the transaction to the Revenue, and there was no such disclosure. [276] Secondly, Mr Birch was taken to task in cross-examination about the level of disclosure of these transactions by the BNZ in its financial statements. Re- examining Mr Birch, Mr Galbraith therefore took me in some detail to the relevant financial statements. I need not refer to the detail of this evidence. I am satisfied that the transactions were properly disclosed by the BNZ in its financial statements. It would be surprising if that were not the case, because of the regulatory supervision of the BNZ and the requirement that those financial statements be audited. Indeed, it transpired that it was the disclosure of the transactions in the Bank's financial statements, coupled with its ETR, that alerted the Commissioner to the transactions. [277] Mr Stanton directed some evidence to the level of disclosure. He stated:In my experience of accounting for tax driven transactions in special purpose subsidiaries, the description given is typically the minimum required under the relevant accounting regulations. In this context, I have reviewed the financial statements referred to by Mr Birch and whilst they do discloseaspects of the arrangements, in my view, it is not possible to discern the underlying nature of the transaction or its financial effect from those statements. (PB 6.39)[278] Mr Stanton then elaborated on this, with reference to the BNZ's financial statements, taking the financial statements for BNZIS1, BNZI and the BNZ's consolidated accounts as an example. This analysis confirmed that disclosure was to minimum requirements. [279] There is a degree of irony in the Commissioner's position on this aspect. On the one hand, he points to absence of publicity as a feature of structured finance transactions. The insinuation is obvious: "we do not want IRD to know about this". On the other hand, the Commissioner complains about the lack of disclosure of the transactions by the Bank in its financial statements. [280] To summarise, I find that the Bank did not publicise or disclose these transactions beyond regulatory and legal requirements.Complexity[281] Several witnesses commented about this. They agreed that the complex structuring of transactions such as these results from the use of SPVs, subsidiary companies and interest rate swaps to achieve the desired tax treatment. For example, Mr Nias said of the three UK transactions:Any association with complexity has more to do with the nature of the UK legislative provisions which the transactions are trying to satisfy than their commercial terms (PB 60)And Mr Shay noted: The complex structure preserves the US tax treatment of each transaction as a loan from the US borrower's perspective (PB 11)[282] The witnesses also agreed that, for all their complexity, these transactions were in substance straightforward funding or loan transactions. For example, Mr Choudhry stated, of the two transactions which he had selected for his analysis: For all their complexity, this is exactly what Gen Re 1 transaction and the Rabo 1 transaction consisted of: a straightforward loan in each case, albeit between a bank and a non-bank financial corporation in the case of the Gen Re 1 transaction. (PB 12.3)[283] Professor Choudhry put the complexity of these transactions in context in this evidence:The Transactions are at the more complex end of structured finance deals, but such deals have been and continue to be undertaken by many banks around the world. What it is unusual is for complex structuring to be involved in what are still solely funding deals; the Transactions were not (for example) securitisations, synthetic securitisations or project finance deals. Funding transactions can be the simplest of any type of deal in finance: the unsecured interbank loan between two banks for example is the simplest type of financial transaction. (PB 12.3)He added: they are very complex transactions that can be undertaken by only a very small minority of banks in the world. I would estimate that no more than 50 or at most 100 banks in the whole world could contemplate or create or structure these types of transactions. Given that there are many thousands of banks in the world, I would estimate that less than 1% of all banks in the world could actually create or structure these type of transactions; so they are in themselves very very complex structures, although at their heart they present a loan of unsecured cash from BNZ to the counterparty. (T 2579)In addition to some 20 years experience in financial markets, currently as Head of Treasury at Europe Arab Bank plc in London, Mr Choudhry (he was content to be so addressed - 2580) is Visiting Professor at the Department of Economics of London Metropolitan University, a Visiting Research Fellow at the ICMA Centre of the University of Reading and a Senior Fellow at the Centre for Mathematical Trading and Finance of the CASS Business School in London.[284] To summarise, complexity is typically a feature of structured finance transactions. But it is unusual in transactions that are in substance straightforward loans, as these were.The benefit split[285] Each transaction split the New Zealand tax benefits between the BNZ and the counterparty. This was expressed as the pre-tax equivalent return to each of the parties. For example, the tables I have set out at [84] and [87] show a pre-tax equivalent return of 2.43% for both the BNZ and CSFB from the CSFB transaction. The BNZ's awareness of how valuable a resource its tax capacity was increased, so the benefit splits increasingly favoured the BNZ over the counterparty. In summary, those splits were:Transaction Benefit Split BNZ/CounterpartyGen Re 1 50/50 CSFB (incl. Extn) 50/50 Gen Re 2 58/42 Rabo 1 66/33 Rabo 2 60/40 Lehman's 73/27[286] Giving evidence for the Commissioner, Dr Fitzgerald analysed this split of tax benefits in detail for each transaction. In respect of Rabo 2 he gave this explanation:The total amount of the tax benefit in each case will simply be 33% of the sum of the fixed swap rates (approximately at market levels) and the 2.95% GPF. So for Rabo 2 one has: RABO 2 0.33(7.6371 + 2.95) = 3.4937% The division of the tax benefit is then entirely controlled by the level of the distribution rate relative to the sum of the swap rate and the GPF. In Rabo 2, the distribution rate is 8.8446 and the sum of the swap rate and GPF is 10.5871. Hence the Rabobank share of the benefit is (10.5871-8.446) or 1.7425, and the benefit to BNZ is (3.4937-1.7425) or 1.7512. Once again, it is demonstrably the case that the level of the distribution rate is the determinant of the benefit split between Rabobank and BNZ (PB 7.22-7.23)Rabobank's benefit of 1.7425 is its 40% share referred to in the table in [285]. The BNZ's 60% share is the 1.7512 which, grossed up, is 2.6137. [287] Dr Fitzgerald has held Associate Professorships at Princeton and New York Universities in the United States, as well as the Chair in Finance at Strathclyde University in Britain. He holds a Doctorate in Finance from Manchester University. Prior to 1988 Mr Fitzgerald was Chief Economist at Credit Lyonnaise Securities, and then spent five years as Director and Head of Arbitrage at Mitsubishi Finance International plc, the securities arm of Mitsubishi Bank. He is currently Chairman and CEO of Equitable House Investments Ltd, a small firm specialising "in the design of arbitrage and volatility trading strategies" for clients such as banks, investment managers and hedge funds. [288] Mr Stanton gave a detailed explanation of the benefit split mechanism, illustrating its operation in a series of cash flow tables (PB 4.4-4.15). He said that his analysis showed:• It is the reductions in the bank's tax liabilities which are being split and that this split is being put into effect by a fixed rate benefit payment and an adjustment to the distribution rate; and• By making a fixed payment to the counterparty, and claiming a deduction on this payment, the tax reductions achieved by the bank (and shared with the counterparty) are increased. (PB 4.16)[289] In terms of the overall transaction returns, Mr Stanton explained:The fixed leg of the swap is an important leg of the transactions because the higher the figure used, the higher the reductions in BNZ's tax liability. This is largely a mathematical consequence since the size of the tax deductions generated by these transactions was derived from the sum of the fixed leg of the swap and the guarantee procurement fee. On the assumption that the guarantee procurement fee is 2.95% (as this was always used in the disputed transactions), and that a benefit share percentage has been set (i.e. at 50/50 etc), then whatever level the fixed leg of the swap is set at, the distribution rate will be the last figure set, and set so to achieve the desired benefit split. (PB 4.19)[290] The interrelationship of the components that produced the benefit split was also analysed by Professor Evans. I refer to this aspect of his evidence in [410]>.[291] In summary, once agreed, the division (or split) between the parties of the tax benefits generated by the transaction was controlled by the level of the distribution rate relative to the sum of the fixed swap rate and the GPF. The split progressively moved in favour of the BNZ, reflecting its increasing awareness of the value of its tax capacity.The distribution rate[292] Evident from the previous section, and confirmed by Mr Stanton (PB 4.13), is that the three main elements of each of the transactions are the distribution rate, the interest rate swap and the GPF. I deal with each of these in turn. The question in relation to the distribution rate, is whether it represented a 'market' or commercial return for the BNZ. In his evidence Dr Fitzgerald explained:Because the distribution rates on the Trust units or the preferred stocks are set in relation to the combination of fixed swap rates and GPF, in order to determine the parties' share of the tax benefits, it is, in my view, self-evident that they cannot be considered as either market related rates or determined by a standalone commercial or economic rationale. (PB 7.31)[293] Dr Fitzgerald demonstrated this by comparing the distribution rates with the applicable 5 year swap rate for each of the six transactions. He prefaced this comparison by explaining that the swap rate would be that payable by the marginal bank active in the NZD swap market. As such a bank could be expected to have a credit rating around AA-, it made no commercial sense for entities such as Gen Re (rated AAA), Credit Suisse (AA-) and Rabo Bank (AAA) to pay distribution rates well above the market swap rate:Date Transaction Distribution Rate 5 year Swap Rate Margin over Swap Rate (bps) 1.7.98 Gen Re 1 9.8694 7.6150 225 21.8.98 CSFB 9.87 6.9850* 289 2.6.99 Gen Re 2 7.8349 6.8050 103 22.11.99 Rabo 1 9.1699 7.7500 141 18.7.00 Rabo 2 8.8446 7.4200 142 26.7.02 Lehman's 9.40 6.6820 272• This was a 3 year swap rate.[294] Dr Fitzgerald pointed out that this table shows that it was both the absolute and relative levels of these distribution rates that were out of line with market. Between July and August 1998 New Zealand market rates fell around 80-90bps, yet the margin as between the Gen Re 1 and CSFB transactions increased by some 60bps. Further, because the income of the US trusts was floating, the equivalent fixed rate fell by about 80bps between 1.7.98 and 2.6.99, as did the fixed NZD rates. This contrasts with a fall in the distribution rates of around 200bps between the two Gen Re transactions. [295] Similar disparities existed for the later transactions. Between the Gen Re 2 and Rabo 1 transactions (both parents AAA rated) the distribution rates rose over 130bps (from 7.83 to 9.17%), while 5 year swap rates only rose 62bps. Between Rabo 1 and Rabo 2 the distribution rate fell 30bps, while the 5 year swap rate fell approximately 100bps. A still further, obvious, disparity is the drop from 225 to 102bps over swaps in the distribution rates for the Gen Re 1 and Gen Re 2 transactions respectively. Yet Gen Re remained AAA rated throughout. [296] Dr Fitzgerald concluded: the lack of relationship between the distribution rate movements, and market rate movements and the levels of the rates, makes it impossible, in my view, to argue that the fixed distribution rates were set in an arms length commercial negotiation. (PB 7.33)[297] Dr Fitzgerald pointed out that setting the distribution rates at non-market levels had implications for the level of rates on the investments made by the SPVs in the various securities issued by the counterparties of the BNZ and/or the amount of principal investment required. It was the income generated by those investments that funded the distribution. In the CSFB transaction the investment was the Loan Note with a face value of over NZ$1 billion carrying a fixed coupon of 7.5923% p.a. referred to in [74]. In Gen Re 2 the comparable investment was USD271.8 million paying LIBOR plus 55bps in USD. Dr Fitzgerald considered that AAA rated Gen Re could have funded itself in the money market at LIBOR, and probably less. He concluded: Hence to pay Libor + 55 basis points would represent a rate that cannot be explained as a transaction on normal commercial or economic grounds. (PB 7.39)[298] Irresistible is Dr Fitzgerald's conclusion that the distribution rates in the six transactions "neither reflect the then current market conditions, nor the credit status of the counterparties, nor any normal commercial or economic levels" (PB 7.35). Doubtless, the reason for this is that mentioned in [286]-[287] : that the setting of the distribution rate was critical to achieving the desired benefit split. None of this was seriously challenged by the BNZ in its evidence or submissions, although at least one of its witnesses (Mr Nias, PB 69) suggested otherwise.The GPFIntroduction[299] Two interrelated, factual issues arise about the GPF: a) Was the GPF a contrivance? b) Was the pricing of the GPF within market parameters? [300] As I pointed out in [147], in the Gen Re 1 and CSFB transactions it was the GAF paid by BNZI to BNZIS1 and BNZIS2 respectively, by way of reimbursement, that was disallowed by the Commissioner. That does not affect this part of the judgment, which is concerned with the GPF paid to the counterparty in each transaction, to procure its parent's guarantee. [301] Before returning to these issues, I set out the four main propositions advanced by the BNZ, and state my view on them. First, a parent guarantee was a fundamental commercial requirement. Without the credit of the parent, these transactions would not have proceeded. Secondly, the parent's guarantee had value. [302] There was much evidence, from witnesses called by the BNZ and the Commissioner alike, supporting both these points. The nub of this evidence issuccinctly captured in this exchange in the course of Mr Galbraith's cross- examination of Mr Stanton, one of the Commissioner's witnesses:Q. As I read your evidence, you accept that the guarantee would have value in the present transactions. Is that fair? A. Yes. Q. And undoubtedly the transaction wouldn't have proceeded without a guarantee? A. Yes. (T 3412)Mr Birch also stated:We decided at almost the beginning of our review of the (AIG 1) transaction that a direct guarantee from AIG would be necessary. (PB 3.29)Mr Birch added that AIG's subsequent decision to use an unrated SPV to undertake the transaction "simply reinforced in our mind the importance of obtaining a guarantee from AIG" (3.31). I accept the Bank's first two propositions. [303] Thirdly, a guarantee can be paid for separately. The Bank's primary expert witness on the economics and market values of credit enhancement, Dr Mackay, explained the implicit/explicit options in building the cost of a guarantee into a financing transaction:Putting aside for the moment the question of market practice, it is undeniable from an economic perspective that paying a gross interest rate of, say, 13 percent while receiving a guarantee fee of, say, 3 percent is economically equivalent to paying a net interest rate of 10 percent. The implicit reduction in the net interest rate paid is simply the explicit value of the guarantee. From an economic perspective, then, the two approaches to paying for a guarantee are equivalent and are consistent with the underlying economics of the transactions. (RB 2.19)[304] I accept, upon the evidence of several of the expert witnesses (primarily Professor Schwartz, Dr Mackay and Messrs McCormick and Gross), that the trend in credit markets in recent years has been toward breaking out the cost components infinancing transactions e.g. toward attributing a discrete and explicit value to any guarantee. That trend is perhaps explained by the huge growth in the securitisation of credit instruments. The only one of those four experts I have not yet introduced is Mr McCormick. For 28 years until he retired, Mr McCormick was with the international law firm Freshfields Bruckhaus Deringer, for 22 of them as a partner. He specialised in banking and finance. He is a senior research fellow and visiting professor at the London School of Economics. [305] Fourthly, the pricing of the GPF, although it never varied, was within market parameters for each of the six transactions. If that submission refers to the open market for credit enhancement, then I accept it. But, for the reasons I will develop at [350] and following, I do not accept that a comparison with open market pricing is appropriate for the GPF in these transactions. In closing for the Bank, Mr Galbraith submitted that, "the Commissioner placed little emphasis on challenging the objective justification that the pricing of the guarantee fee fell within a market range". I anticipate that my answer to the first issue, to which I now turn, explains why the Commissioner did not dwell on the pricing of the GPF.Was the GPF a contrivance?[306] Quite apart from challenging the deductibility of the GPF, the Commissioner challenged it as a contrivance, the only real object of which was to increase the BNZ's deductible expenses. [307] The genesis of the GPF in the first of the nine BNZ transactions – AIG 1 – can be traced through the contemporary documentation. It appears that the initial proposal prepared by Allco Finance Group for the NAB, for what became the AIG 1 transaction, was forwarded by the NAB to the BNZ on 3 July 1995 (5/2600 >). This proposed (at 2603) a "Procurement and Guarantee Fee" of 1.5%. With the prescience he demonstrated throughout his evidence, Mr Birch had noted against this in handwriting "could be taxable – pay entirely as procurement fee". Mr Birch had other concerns about this fee, for example that it extended to "procuring the issuance of the B Class shares".[308] The BNZ ALOC memorandum prepared for the AIG transaction in September 1995 (2651) states "this transaction will be covered under a specific guarantee from AIG". The structure diagram attached to that ALOC as Appendix A (2675) depicts that "Specific Guarantee" by an arrow curving down from American International Group (AIG) (AAA rated) to American International Group Financial Products (AIGFP) (also shown as AAA rated). [309] In a fax to Mr Birch on 19 September 1995 a senior tax manager at Ernst & Young stated (2666) that only the "fee for AIGFP guaranteeing the payment of the dividend to BNZ" is "in fact a guarantee". The faxed advice suggested that fee be split off as it will be "at risk of being taxed as an off shore insurance premium". [310] It appears that the guarantee fee transmuted into a procurement fee following that advice, because the draft application to the IRD for a binding ruling sent to the BNZ by Ernst & Young on 29 September 1995 refers (at 2688) to a "Procurement Fee", although, interestingly, it was stated to be "for AIG-FP procuring the issuance of the B class shares by AIG-FPNZ to BNZF Sub". [311] Consistent with the parent guarantee being a given from the outset of consideration by the BNZ of the nine transactions, there was no evidence of any negotiation over the amount of the fee. The Commissioner pointed out that the only occasions on which the implications of a variation in the fee were discussed were within the BNZ, in the context of the impact of that on the consequential tax benefits. There are two examples of this. The first is in the ALOC memorandum dated 22 March 1996 for the AIG transaction:Economically, from a US tax perspective, this fee has the effect if improving the US benefits available from the transaction by allowing an increase in the size of the dividends paid by AIG-NZ to BNZ Sub i.e. the greater the fee, the greater the level of dividend, and the greater the available benefits. As an illustration of the economic impact of the fee, we estimate that a reduction in the fee by 1.00% (to 1.95%) would likely reduce the pre-tax margin available to BNZ Sub from the transaction by 0.29%, or NZD7,250,000 over the 5 year life of the transaction. From a New Zealand tax perspective, the fee is assessable to AIG-NZ and deductible to BNZ Sub and therefore its effect is neutral. (2921 – AIG NZ is explained at 2918)[312] The second is in a fax Mr Birch sent Mr Brandon of NAB Group Credit on 22 July 1997 in relation to the (at the time, proposed) Gen Re 1 transaction:Attached are calculations showing the impact of a change in the guarantee procurement fee on the benefits available from the investment. I have used the example of a 100bps increase in the level of the fee. Increasing the fee by this amount, with no change to the dividend rate, provides GRC with a 100bps increase in yield at BNZI's expense. However, by adjusting the dividend rate at the same time, the return to both parties can be equalised at a yield which is approximately 20bps higher. (58/45157)[313] "Procurement" carries the connotation explained by Dr Fitzgerald: In my view, the expression "guarantee procurement fee" inevitably implies the idea of an external agency or third party being paid to obtain a guarantee from the counterparty parent corporations. This is patently not the case with the transactions under discussion, where the recipient of the procurement fee is the parent itself through a wholly owned subsidiary. In my view, paying firms like (and he listed here GRCF and its counterparts in the other transactions) as agents to procure a credit guarantee from the parent groups, is an artificial construct, which does not reflect normal commercial practice. (PB 7.50)[314] The Commissioner also pointed out that there was no evidence of actual procurement. He argued that this reality – the parent's credit being a given from the outset - was at odds with the 'sequential scenario' implicit in the evidence of the Bank's expert witnesses i.e.:• The BNZ wishing to finance the unrated counterparty;• But becoming concerned about its creditworthiness;• Thus, asking the counterparty to obtain a guarantee from its parent;• The counterparty responding by 'procuring' that parent guarantee;• Resulting in "enhancement" of credit risk for the BNZ (Mr McCormick PB 7.26) or, conversely, "substantially reducing" the BNZ's credit risks (Mr Gross PB 6.3).[315] Under cross-examination, the Bank's witnesses tended to accept that no such sequential scenario had occurred. For example, questioned about his suggestion of credit "enhancement", Mr McCormick accepted that the BNZ would never have entertained advancing finance to the unrated counterparty (T 1749). In its closing submissions, the Bank disowned any suggestion that the GPF came later in the development of the transaction, in response to the BNZ's concern about the creditworthiness of the proposed counterparty. [316] Mr Smith, called by the Commissioner, was particularly adamant that the three American transactions (which were the subject matter of his evidence) should be treated as financings made to the US parent. The GFP had no purpose other than to achieve a New Zealand tax benefit. In the course of cross-examination, this exchange occurred:Q. I would suggest that there's obviously a value to the entity in the group that's getting the funding from the advantage of having the Parent Guarantee? A. I disagree with that nobody picked this entity, like, I want to fund this entity, and then well, it isn't quite creditworthy enough, I'll get a guarantee; it was, I want to deal with Gen Re, Gen Re is a wonderful business opportunity, our New York office is dealing with them, I want to deal with Gen Re and I'm going to rely on the Gen Re credit. (T 1777-1778)Mr Smith practised corporate law for 40 years with the Boston based law firm Ropes & Gray. His principal focus was on commercial financing transactions, advising both lenders and borrowers. He was a partner of the firm from 1976 until his (mandatory) retirement in 2007. He continues with the firm as senior counsel. He developed and taught a course in commercial lending for the Boston University's Graduate Programme in Banking Law Studies. [317] Despite the breadth and depth of the experience of the witnesses called by the BNZ and the Commissioner alike, it appeared none of them had actually encountered a "Guarantee Procurement Fee" in unrelated transactions. Ms Miller gave evidence that somewhere between 50% and 65% of similar transactions entered into by Gen Re worldwide had GPFs, but promptly added "they were probably just calledguarantee fees". Ms Miller went on to explain that the word 'procurement' "just doesn't make the cut" in financial "lingo" in the United States, being associated with soliciting "a lady of the evening" (T 1627-1628). It became apparent that Ms Miller was simply referring to guarantee fees. [318] Mr Gross also gave evidence of his experience with guarantee fees. In his evidence-in-chief Mr Gross stated:6.6 I also do not think that the payment of a fee for a parent guarantee is in any way unusual in a structured transaction such as this. The payment of a fee for credit enhancement is common in financial markets. Common examples are credit default swaps, under which a periodic fee is paid in return for protection in the form of an amount payable by the counterparty in the event of certain adverse credit events arising. These types of transactions, in which credit risk is bought and sold for a fee, have become increasingly common over the past ten to fifteen years.Because of the connotation explained by Ms Miller, Mr Gross was also not familiar with the use of the word "procurement" in financing (T 3090). But he provided examples of guarantee fees being paid, including companies investing in Brazil paying a fee to the Bank of America to provide a guarantee on behalf of their Brazilian subsidiary (T 3104). [319] Mr Nias had not encountered a guarantee and collateral procurement fee, or even a guarantee fee (T 2775). [320] Mr Choudhry was dismissive of the whole concept of the BNZ paying a guarantee fee in these transactions. He said there was no commercial logic in having a GPF payable by the lender where the borrower is a wholly owned subsidiary of the guarantor. In such a case a parent company guarantee without charge would be the norm. Mr Choudhry said he had never heard of the party requiring the guarantee to have to pay for it, and he regarded the guarantee procurement fees "as extraordinary elements of the transactions" (PB 9.1, 9.2). At that point, Mr Choudhry added to his statement of evidence in this way:Q. Professor Choudhry did you wish to add a comment at this point, about the commercial logic of the guarantee?A. Yes, I would. What I would like to say is that, even in the year 1998 or 1999 the guarantee procurement fee would have had no logic because it's the equivalent of a lender paying for the privilege of being able to lend money; that's not something that a bank does, not that I've ever come across. Post the 2007/2008 credit crunch, the fantasy or the nonsense of that can be seen even more clearly. Money markets are very illiquid and even very highly rated large banking institutions would pay many basis points, and are doing so, are paying many basis points above LIBOR right now to borrow money. The idea of a lender, a bank, paying someone to be able to lend to it in order to get a guarantee would have been fantasy. So, as I said, it's equivalent to paying for the privilege of lending, when in fact it would be expected to be the other way round. (T 2577-2578)[321] Professor Choudhry was particularly critical of the GPF paid in the Lehman's transaction. This was the one transaction where the fee was described as one paid for the parent guarantee and the provision of the collateral. Mr Galbraith's cross- examination of Mr Choudhry included this:Q. Now I can take you to the documentation if you'd like to, but are you aware that the guarantee procurement fee in the Lehman's transaction is paid both for the provision of the collateral and the Parent Guarantee? A. I was aware of it. Q. So there is value? A. There is value in that it has a value ascribed to it in the transaction. What I'm trying to say is that if I am lending money to the subsidiary, an unrated subsidiary of an A rated parent and that A rated parent has put up AA rated collateral, as a logical rational non- time-wasting chap I wouldn't personally then require a guarantee from the A rated parent because he has already given me AA rated collateral. A guarantee from an A rated parent has no value when I've already collateralised it with a double A – I'm speaking in economic rational terms – AA rated collateral. Now, I agree that in the transaction, that value is ascribed to it because a fee is paid for it. But I just pointed out, in the rational world, in the real world as a real lender of cash I wouldn't go to the trouble of doing all this "to do" because I've already got AA rated collateral. (T2617-2618)When pressed by Mr Galbraith, Mr Choudhry said that a bank would manage the risk of the collateral falling in value by "what's called a haircut or a margin on the collateral" e.g. requiring collateral worth £200 to secure a loan of £100 (T 2619- 2620).[322] A further criticism made by Mr Choudhry was that the GPF did not vary, even where the BNZ held a letter of credit for a substantial part of the funds it had advanced. For example, the GPF in the Gen Re 1 transaction remained at 2.95% p.a. although GRCF was required to provide, at its cost, additional credit support, satisfactory to the BNZ, of at least NZ$200 million. GRCF did that by providing a letter of credit from the Toronto Dominion Bank in favour of the BNZ. That LoC was subsequently replaced by an agreement under which the Chase Manhattan Bank pledged highly rated collateral to a value of at least US$118 million. The Commissioner submitted that the curious feature that the collateral could be resorted to only if the parent guarantee failed, might have been an endeavour to deflect that further criticism by Mr Choudhry. [323] Mr Stanton's experience was the same as Mr Choudhry's: in transactions such as these the parent would provide a guarantee to the debt provider (the BNZ) without any separate fee being charged. Mr Stanton said he had never seen such a fee paid in this type of transaction. He stated: In my view there was no reason to pay such a fee (the 2.95% GPF) other than to provide a mechanism by which part of BNZ's tax reductions could be passed over to the counterparty and, at the same time, increase the size of the tax benefits being split. (PB 5.9)[324] All this supports my conclusion that the credit of the parent of the counterparty was fundamental to all the transactions. They were not going to happen without it. [325] Somewhat obviously, if the BNZ had advanced the $500 million direct to the parent, no guarantee would be required. But the transactions were differently structured, in particular by requiring the finance to be advanced to an un-rated SPV. I accept that that was dictated to some extent by tax considerations at the counterparty end. For example, in relation to the AIG 1 transaction, Mr Birch explained that there were US withholding tax considerations (PB 3.61-3.62). But another reason for the structure was clearly to necessitate – or at least to justify - a guarantee from the counterparty's parent. On the evidence, I accept that guarantee had value and could therefore be ascribed a value. I also accept the value ascribedcould be either explicit (in the form of a guarantee fee) or implicit (a lower distribution rate). I find that the guarantee was attributed an explicit value, in the form of the GPF, for tax advantage in two respects: a) As a discrete GPF was perceived to be a deductible expense, it increased the tax advantages of the transaction for the BNZ, in the form of the expenses deductible against the Bank's gross income. b) Because the GPF was a component of the formula upon which the transaction was priced, it was critical to the tax benefits produced by the transaction. As was candidly recorded in the ALOC I have quoted from in [311], the greater the GPF, the greater the benefits available from the transaction. In the case of the BNZ, those benefits took the form of the tax relieved distribution rate. No GPF, or a lower GPF, would have lowered that distribution rate. [326] In the early stages of the development of the structure of the transaction, the fee was termed a guarantee fee or a "procurement and guarantee fee". I find (indeed, this was essentially conceded) that the fee was re-labelled a GPF so that it could be contended it was for procurement services and did not attract the withholding tax due if it was treated as an off-shore insurance premium. [327] At one level – that favoured in the Bank's closing submissions - the GPF is commercially explicable and reasonable. But I am required to view these transactions in a commercially and economically realistic way. So viewed, I am unable to avoid the conclusion that the GPF was a contrivance, in that its genesis and primary function was tax advantage: to create a deductible expense, and to contribute to the income produced by the transaction (in the case of the BNZ, tax relieved income). Perhaps the most candid indication of this was in a draft of the application prepared by the BNZ for a binding ruling on the (at the time proposed) Gen Re transaction. This draft is marked #4 and dated 25 July 1997. Section 5.13 of the draft is headed "Guarantee Procurement Fee – Non Resident Insurer". The first sub-heading is directed to the non-application of s CN 4. This was the concern I have mentioned as to whether the GPF might be viewed as an off-shore insurancepremium attracting withholding tax. The draft includes this paragraph, with this deletion and handwritten comment against the asterisk:5.13.12 * GRCF cannot be an "insurer" for the purposes of section CN 4 because it is not a party named in a "contract of insurance" with a potential "insurance" liability to BNZI. GRCF is simply a party that will receive a form of commercial profit, which the parties have chosen to call a guarantee procurement fee, in respect of the preferred share transaction. Delete *How does this description fit with our deductibility/arms length assumptions?(16/11280) [328] When confronted with that in cross-examination, Mr Birch struggled somewhat. He accepted that the handwritten comment was his. His answers to Ms Scholtens' questions about the deletion and his comment included this: Well, that's wrong, that's why I presume it's crossed out. I'm not sure looking at it now how that comment of mine at the top applies to that crossed out bit. Yes, certainly I was concerned to make sure that the guarantee procurement fee was on an arm's-length basis. (T 460)[329] In my view that deleted description "a form of commercial profit which the parties have chosen to call a guarantee procurement fee" accurately describes the GPF. [330] Accordingly, I answer the first factual issue 'Yes – the GPF was a contrivance'.Was the pricing of the GPF within market parameters?[331] Mr Birch's evidence-in-chief about the way in which the level of the GPF was fixed is this:3.65 It was ultimately agreed between the parties that the level of the guarantee procurement fee would be set at 2.95%. The decision that it was appropriate to pay a guarantee procurement fee and then to set that fee at 2.95% was one that evolved as we became more familiarwith the proposed transaction. The figure itself was first suggested to us by Allco. 37 In terms of how we evaluated the level of the fee, I believe we reflected on what it would cost to upgrade BNZI's exposure to the repo counterparty (here the unrated AIG Financial Products subsidiary) from what was really a "junk" investment (meaning not investment grade), to a AAA rated investment. As I recall it, 2.95% was accepted by the parties as a fair assessment of an acceptable fee to pay for that level of credit enhancement at the time. ___________________________37 5/2831-5/2832)[332] A good part of the lengthy cross-examination of Mr Birch was directed to the GPF. The more pertinent parts of that cross-examination include these:Q. And do you recollect why Allco used that figure? A. I think at the time we were trying to work out what would be an appropriate rate for the guarantee procurement fee. I didn't have a lot of experience at working out those rates at that time, Allco did, I think they went away to have a think about it, came back with 2.95%. Q. What did AIG, how were they involved in the discussion? A. That rate would have been discussed with AIG at that point. Q. By Allco? A. No, by both of us I suspect, by Allco and ourselves. Q. But there are no documents about that? A. No, there's not. (T 1408-1409)[333] And this:Q. And the fee moved from 1.5% from your handwritten amendment in December 1995 to 2.95%? A. Well, yes, I think it was 1.5 in the very first document then I amended that from whatever it was we saw to 1.75, and that ultimately became 2.95. (T 1419)[334] Mr Birch also detailed the two inquiries he did make as to the appropriate level for a guarantee fee. These were: a) In relation to the proposed Gen Re 1 transaction, on 22 June 1998 Mr Birch emailed Mr Kilfoyle who was Vice-President, Structured Finance, in the New York office of the NAB. The fax read:In relation to the 2.95% Guarantee Procurement Fee for the GRCF repo transaction, I would appreciate some comment from you on where you would lend to an unsecured, unrated entity holding paper issued by a similarly unsecured, unrated entity. We will call tomorrow to discuss. (17/11984)Mr Kilfoyle responded on 26 June. He set out the broad detail of the Gen Re 1 transaction and then stated:Without the General Re Corporation Guarantee, the transaction is a blind investment into an unregulated Trust and is an investment the Bank would not enter into. Therefore, the General Re Corporation Guarantee defines the Bank's principal credit risks making the investment acceptable to the Bank and the 2.95% Guarantee Procurement Fee reasonable in my assessment. (17/11992)b) An inquiry made through NAB in London, which in turn obtained information from Lloyds Bank, "which referred to 300 to 400 points" (mentioned by Mr Birch at T 1429). [335] The expert evidence about the pricing of the GPF was, as might be expected, at a somewhat more sophisticated level than those inquiries. [336] With the exceptions of Dr Fitzgerald, Mr Stanton, and I think also Mr Choudhry, all commercially experienced witnesses agreed that the counterparties in these transactions had a sub-investment grade credit-standing – BB at best, and probably lower. For example, Mr Gross would have used a rating of B (PB 6.22). One of the more substantial counterparties was GRCF, but Ms Miller considered "it was clearly appropriate to treat it as sub-investment grade" (PB 6.29). Dr Mackaysaid that, when according the counterparties a BB rating, he had "incorporated a lift for the parent's rating. Absent that lift, I would have started with an alternative rating of B or even C" (PB 2.15). To an extent, Dr Fitzgerald endorsed Dr Mackay's justification for the "lift". He referred (PB 7.55) to the several references to the SPVs as "look through" entities, which he interpreted as indicating that the BNZ was looking through them to the parent to which ultimately the NZ$500 million was being advanced. [337] On this aspect of the pricing of the GPF, Mr Stanton explained that it would have been necessary, at the time the 2.95 GPF was agreed, for the BNZ to analyse the creditworthiness of the various companies being guaranteed e.g. GRCF, MadPar and so on. He said that there was nothing in the documents he had reviewed to indicate that the BNZ had undertaken any such analysis. He stated:Without any such credit analysis, or indeed any meaningful information upon which to prepare such an analysis, I think it clear that BNZ would not have lent to any of these companies on a stand alone basis – which makes the next question, namely what interest margin BNZ would have charged the various Company X's if it had lent them NZ$500m each, somewhat academic. (PB 5.17)[338] Addressing Mr Mackay's analysis, Mr Stanton pointed out that it rested on the assumption that the subsidiary being guaranteed would attract a BB credit rating. Mr Stanton outlined the type of information and analysis that would be needed to make such an assumption. Based on the documents he had reviewed, he pointed out that none of the basic information required appeared to have been obtained by the BNZ. Had Mr Mackay's assumption been a proper one, Mr Stanton accepted that his modelling was appropriate. But he pointed out that it was entirely predicated on there being an assumed credit rating, or borrowing margin, for each of the subsidiary companies being guaranteed. He stated: In my view, it is at this point the model breaks down, in that one cannot reasonably assume a credit rating for, or price a loan to, a company about which one has no information. (PB 5.23)[339] Putting this information deficit difficulty aside for the moment, witnesses (excepting again Dr Fitzgerald and Mr Stanton) agreed that the correct approach to pricing the GPF is to look at the market's assessment of the value of the difference (the spread) between the credit standing of the counterparty, and that of its parent. To take just one example, Mr Choudhry said: if you're pricing a guarantee without naming the guarantor you'd take the risk-free rate and you would take the spread over that payable by the unguaranteed entity If you then speak specifically, you would take the rate payable by the guarantor and the spread above that of the unguaranteed entity (T 2606-2607)[340] All the witnesses who were asked about the point also agreed that the costs to the parent of providing the guarantee were irrelevant to its pricing. For example, Dr Fitzgerald stated:If there's a market price, then the cost of creating that product is irrelevant. (T 3232)[341] The Bank's primary evidence on the cost of credit enhancement was Dr Mackay. He produced, as Figure 15 (he actually labelled it Exhibit 15) a graph showing the OAS (option adjusted spreads) of differently rated bonds. While the OAS of bonds rated AAA, AA or A are relatively close, those between BBB, BB and B are wider. Figure 15 also shows a rising trend in each of the BBB, BB and B indices, a trend underscored in the data I mention shortly. [342] During his evidence, Dr Mackay prepared, overnight, and produced next morning, Figure 15A on which he had plotted the six transactions against the monthly average OAS. When the Gen Re 1 transaction closed, the spread between GRCF, (if accorded a BB rating) and Gen Re (AAA rated) was 192bps. If GRCF were accorded a B rating, the spread was 331. By the time the Lehman's transaction closed on 13 August 2002 the comparable spreads were 631 and 859. Dr Mackay's Figure 15A showed that the 2.95% (295 bps) GPF used was within the monthly average OAS for every transaction if the counterparty was accorded a B rating, and was also within the spreads for all the transactions except Gen Re 1 (192), CSFB (221) and Gen Re 2 (240) if the slightly higher BB rating was adopted.[343] In his Figure (Exhibit) 18 Dr Mackay plotted the 295 bps GPF against the OAS, ASW (asset swap spreads) and CDS (credit default swaps) spreads of 5 to 7 year bonds from January to June 1998 for BB rated bonds underlying the Merrill Lynch US Corporate High Yield, Cash Pay Index. For Gen Re 1, this showed that the GPF fell in the 82 nd percentile of the OAS inter-quartile range, and in the 88thpercentile using the ASW inter-quartile range. In other words, it was at the higher end of the market – outside the 75 th quartile. Consistent with the rising indices shown in Dr Mackay's Figure 15, by the time the Lehman's transaction closed on 13 August 2002, its GPF fell in the 26 th percentile based on the CDS curves dataset provided by GFI Group Inc., which Dr Mackay used to assess the GPF for the Rabo 2, CSFB Extension and Lehman's transactions. [344] Dr Mackay summarised this part of his evidence as follows:6.84 My analysis of what market participants were actually willing to pay for similar levels of credit protection at the time of each transaction in the 1999 to 2002 period shows that the price of 2.95% per annum paid by the NZ Investor to reduce its credit risk was well within the commercially reasonable range, and indeed is towards the middle of that range. While for the 1998 transactions the 2.95% per annum price lies I the upper end of this range, it is still below the 90th percentile for the comparable OAS and/or ASW data. [345] The GPF never varied from 2.95% through the nine transactions with closing dates spread over some five years. The Bank's witnesses, expert on this aspect, readily agreed that that did not reflect changes in the market. Those changes emerged from the evidence of Dr Mackay, which I analysed in [341] to [343]. Dr Mackay was asked about the static nature of the GPF:Q. And would you agree that in the finance world that would be really recognising that the fee was not adjusting to changes in the systematic risk that were occurring in the marketplace? A. I think I've said this before; I mean, my understanding from Mr Birch's testimony was that for what he viewed as business reasons whatever benefit there might be from staying with an earlier rate that had been agreed to, he chose the counterparties, the two counterparties chose to leave the rate at 295 which in my view means that actually in the later periods they certainly could have justified a significantly higher fee for those transactions; but instead they didn't take advantage of that, they left it at 295. Q. So, is that a "yes" to my question?A. I think that was a "yes" to your question. The Court: As you started I thought, this is a "yes". (T 2264)[346] Mr Middleton gave reply evidence for the Bank. He has 34 years' experience in banking and finance, working for Bank of Scotland/HBOS plc. In his statement Mr Middleton acknowledged "that in theory the (GPF) could be expected to change over time as the market moved ". He then offered two reasons for the BNZ's decision to stick to the same rate in later transactions. One of those reasons was also the reason the Bank advanced in opening its case:The reason is that the Bank wished to adhere to the rate used in the AIG 1 transaction, which Inland Revenue had accepted as market and favourably ruled on. (Bank's opening 5.69)[347] In closing on the GPF, the Commissioner reiterated the points that I have dealt with in [45] in relation to the ruled transactions. The nub of these was that the AIG ruling could not, and did not, determine that 2.95% p.a. was an appropriate commercial rate. The Commissioner relied on s 91E(4)(a) and (j) Tax Administration Act 1994 which provide:Private Rulings 91E Commissioner to make private rulings on request (4) The Commissioner may not make a private ruling if – (a) The application for the ruling would require the Commissioner to determine questions of fact; or (j) The application for the ruling would require the Commissioner to form an opinion as to a generally accepted accounting practice or to form an opinion as to a commercially acceptable practice.[348] Consistent with those limitations, the Commissioner said that the reference to the GPF in the AIG ruling was stated to be an "assumption". I may have missed it, but I do not see the GPF amongst the assumptions A)-T) on which that ruling wasbased (7/4165-4167). The GPF is detailed under the heading 'Fees' at 7/4162. As the Commissioner pointed out, the inability of the BNZ to rely, in these proceedings, on the AIG ruling was noted by the Court of Appeal in Westpac Banking Corporation v CIR [2009] 2 NZLR 99, particularly at [84]:[84] To the extent to which this argument is based on the First Data ruling, we see it as inconsistent with the very limited effect given to rulings under the binding rulings regime. [349] To summarise to this point, the Bank's evidence about pricing pegged the 2.95% GPF to the market for credit enhancement. I mentioned (at [305]) that I did not accept that was the correct to the pricing of the GPF. If I am wrong in that, then I certainly accept the Bank's submission that its evidence establishes, albeit ex post facto, that the 2.95% per annum GPF for enhancing the BNZ's credit risk from BB to at least AA was within market range over the whole of the period of the transactions, and by the end of the period was at the lower end of the market. However, for the reasons explained by Mr Stanton, I do not accept that the Bank's inquiries at the time, or its evidence in these proceedings, established that BB was the appropriate credit rating for the counterparty entity that was guaranteed in each of the transactions. Nor do I intend my acceptance of the Bank's evidence, so far as it went, to detract from my finding in [330] above. A contrivance can be correctly priced. [350] I need now to explain why I do not accept the Bank's approach to the pricing of the GPF. It results from the evidence of Mr Stanton and Dr Fitzgerald, which I found convincing as to the pricing of the parent guarantee. First, Mr Stanton. In transactions structured as were these, Mr Stanton said the funding would be provided at the interest rate at which the parent could borrow (reflecting its credit strength), and the parent would provide a guarantee. There would be no separate fee payable by the credit provider either direct to the parent in return for its guarantee, or to the subsidiary for procuring the guarantee. Mr Stanton said he had never seen such a fee paid in this type of transaction (PB 5.5-5.6). He concluded:In my view, whilst the parental guarantee was required, the guarantee procurement fee (or the fee for the credit enhancement) was not. (PB 5.25)[351] In similar vein, Dr Fitzgerald said he was not familiar with the concept of a credit guarantee procurement fee paid to one entity within a financial services group to obtain the guarantee of the parent corporation. He said there was patently no element of real procurement here, because the recipient of the GPF was the parent itself through a wholly owned subsidiary. Dr Fitzgerald considered that paying an entity like GRCF or MadPar a fee to procure a credit guarantee from its parent was "an artificial construct, which does not reflect normal commercial practice". Dr Fitzgerald amplified his view:7.51 If one lays aside the internal complexities of the transactions, and whatever description is used to describe the fees, I believe it would also be readily apparent to any reasonably competent financial markets participant that the structure and level of those fees were out of keeping with market practice. Suppose one takes the CSFB transaction as an example. A third party is contemplating lending via a sale and repurchase agreement to a wholly owned, totally controlled subsidiary of CSFB. The third party states that it will only carry out the transaction if the subsidiary of CSFB is guaranteed by the Credit Suisse Group. Credit Suisse and CSFB may well indicate to the third party that the provision of a guarantee will impact upon the lending rate, but undoubtedly that impact would be driven by the additional capital cost, if any, to Credit Suisse, not by reference to credit spreads in the marketplace between Credit Suisse at AA- and an unrelated, non-investment grade entity.[352] Dr Fitzgerald then drew on his own experience. He said that sometimes a customer lending to or investing with Mitsubishi Finance International (which was unrated) requested a credit guarantee from the parent, Mitsubishi Bank. The quote from the Bank was generally around 10-20bps. That fee was either absorbed by Mitsubishi Finance International (if it still permitted a satisfactory return), or reflected in a lower return to the lender/investor. Dr Fitzgerald expressed the view:7.52 I believe that is exactly how the process would have worked with General Re, Credit Suisse, Rabobank and Lehman's Bros in a conventional and genuine commercial transaction.[353] Dr Fitzgerald then took the Credit Suisse Group to demonstrate his view of the correct approach to the pricing of a guarantee within (i.e. internal to) a financial services group. I start by replicating the table Dr Fitzgerald added as p 38A to his statement of evidence (T 3161; 3163-3164):INTERNAL COST OF PARENT CREDIT GUARANTEE - CREDIT SUISSE GROUPFull 8% capital weighting 4.5% capital weighting for B credit 2% capital weighting for BB credit 4.17% WACC 33bps 21bps 8-9bps 6.5% net CoE (11.5%-5%) 52bps 29bps 13bps 8.25% net CoE (11.5%-5%) (.65) 66bps 37bps 16.5bps[354] Dr Fitzgerald assumes the guarantee amount is NZ$500 million. His first line takes the weighted average cost of capital of the Credit Suisse Group of 4.17% as at September 2008 (his Exhibit 13). His first (left hand) column assumes the conventional regulatory capital requirement is 8%, and on that assumption applies a full (or 100%) capital weighting. In other words, the parent would need to set aside capital of NZ$40 million to cover the parent guarantee. The 4.17% WACC and 8% capital weighting yield a cost of providing the parent guarantee of 33bps. In practice, Dr Fitzgerald considered full capital weighting was unlikely; he anticipated the actual cost being substantially less. In 6.19 of his statement of evidence, Mr Gross set out in chart form the capital "charges" applied to insurance/reinsurance companies by the United States regulator, the National Association of Insurance Commissioners. For a B rated asset, the capital charge was 4.5%. Dr Fitzgerald applies this in his middle column, producing a cost of 21bps. In his third (right hand) column, Dr Fitzgerald combines the 2% capital charge for a BB rated asset (the rating assumed by Dr Mackay – [336] above) with Credit Suisse's 4.17 WACC, producing a cost of 8-9bps. [355] Dr Fitzgerald's second line addresses the possible criticism that, in using a WACC, he is not using a figure high enough because this type of guarantee will be backed entirely by equity capital. In 1998-1999, Credit Suisse's cost of equity capital (CoE) was 11-12% - fairly typical of participants in financial markets at that time. However, an 11.5% CoE is ameliorated by the return on that equity capital. For example, if it is invested in Treasury bills at 5%, the net CoE is 6.5% (11.5%- 5%). To that CoE, Dr Fitzgerald then applies the same 8%, 4.5% and 2% capital weightings producing costs of 52, 29 and 13bps respectively.[356] The third line adjusts for the fact that the 5% return earned by the equity invested in the Treasury bills will be taxed. Dr Fitzgerald applies the Swiss corporate tax rate of 35%, increasing the CoE to 8.25% (11.5-5% taxed at .35%). Applying the three successive capital weightings to that higher CoE produces costs of 66, 37 and 16.5bps. [357] Dr Fitzgerald summarised this approach in this way:So my essential conclusion from that is that, if you came to me and said what do you regard as a maximum reasonable cost for the provision of an internal guarantee by Credit Suisse, I would say, if you put it as 45 to 65 basis points per annum, that's as high as I could reasonably go. (T 3163)Comparing that to the GPF used in all the transactions Dr Fitzgerald said:7.53 It would be my view that the level of 295 basis points quoted as the credit guarantee procurement fee bears no relation to the cost of providing a parent guarantee that would be charged internally by a parent to a wholly owned and controlled subsidiary.[358] One of the factors in my concluding that the GPF was a contrivance was Mr Choudhry's view: The idea of a lender, a bank, paying someone to be able to lend to it in order to get a guarantee would have been a fantasy (even in the year 1998 or 1999). So, as I said, it's equivalent to paying for the privilege of lending, when in fact it would be expected to be the other way round. (T2578)[359] Neither Mr Stanton nor Dr Fitzgerald, both witnesses of wide practical experience, had encountered the concept of a credit guarantee fee (or procurement fee) being paid to one entity within a financial services group to obtain the guarantee of the parent corporation. I have indicated that I accept their evidence, the effect of which is that no fee was commercially justifiable here. But, even assuming there was a genuine and commercial justification for the lender (the BNZ) to pay the borrower (e.g. MadPar in the CSFB transaction) a parent guarantee, Dr Fitzgerald has demonstrated in a fairly convincing way that the internal cost -–as opposed to the open credit enhancement market cost – was in the range 45-65bps tops. That is far removed from the GPF of 295bps.The interest rate swap[360] Each of the transactions involved an interest rate swap. The swap involved the BNZ paying a fixed rate, and receiving a floating rate, on NZ$500 million for the duration of each transaction. Two points arise about the interest rate swap, both of them relating to the underlying issue of whether these transactions were tax avoidance arrangements: a) What was the purpose of the interest rate swap? b) Was the pricing of the swap (the fixed interest rate) within market parameters? [361] An interest rate swap involves A agreeing to pay B a fixed rate of interest, and B agreeing to pay A a floating rate of interest (or vice versa), for the duration of the swap in the currency stipulated and at the stipulated floating rate e.g. BBR. An interest rate swap is based on the market's perception of future interest rate movements i.e. the yield curve. When the swap is transacted, the present value of the two money streams (the fixed and floating rates of interest) net to zero i.e. the swap has an NPV of 0. Conventionally, the value of the swap thereafter depends on the movement of interest rates. That movement will dictate both the payments due during the life of the swap, and the cost (if any) of closing out the swap early. For example, if A swaps 5% paying fixed for BBR floating on $5 million for 5 years, and BBR two years later has risen to 10% with an upward yield curve, A is sitting on a valuable swap. A only has to pay 5% but is receiving 10% and rising on $5 million with 3 years to run. In an interest rate swap done on market (not the case with the swaps here), the bank or other financial intermediary transacting the swap will take a margin. [362] Broadly, the BNZ's position on the interest rate swaps was: a) As the detail of the swap in each transaction differed, each must be considered separately. The swaps do not lend themselves to generic criticism.b) The primary role of a swap is to hedge a party's cash-flows in the transaction. A swap reallocates the interest rate risk. Priced at market, a swap is truly an ancillary part of the transaction. c) Whatever the Commissioner's initial criticisms of the swaps may have been, the only challenge he made in his closing submissions was to the setting of the swap rates in advance. d) The use of interest rate swaps in structured finance transactions is common, almost standard, practice. e) In structured finance transactions it is also common practice to set the swap rate in advance. f) The pricing of the swaps in the transactions was within a reasonable market range of values except for the Gen Re 1 and CSFB transactions. The Bank concedes the pricing of those two transactions was outside market range, though not significantly: 55bps for Gen Re 1; 19 bps for CSFB. That resulted from the Bank wrongly adopting the 1 year swap rate for each of those transactions, rather than the 5 year swap rate for Gen Re 1, and the 3 year rate for CSFB which the Bank accepts were the correct ones, as each the transactions were for five and three years respectively. [363] In closing the Commissioner submitted the role of the interest rate swaps in the transactions was two-fold: a) To create one of the two fixed cash flows that collected a predetermined tax benefit from the New Zealand tax base. b) To fund, in combination with the GPF, both the tax free distribution and the fee (discount to LIBOR) to the counterparty. In combination with the GPF, the Commissioner submitted the swap fixed the tax benefits available to be shared between the parties. The fixing of the distributionrate then locked in the parties' shares of those benefits. The swap was accordingly an integral part of the structure of the arrangement, and of its tax avoidance goal. [364] These opposing submissions are a subset of the parties' opposing cases in these proceedings. The BNZ says the interest rate swap was a standard component of transactions of this type, which were commonplace, but concedes that the fixed rate used in two of the transactions was "off-market". The Commissioner responds that the swap was an integral part of these transactions which are tax avoidance arrangements – it was a vital cog of "the fantastical tax machine" (the Commissioner's description in closing) which the BNZ engineered with the counterparties. [365] In short, the relevant part of the Bank's submissions focus on the swaps per se while the Commissioner's argument is directed to the swap as a component of the transaction. Significantly, the Bank's closing submissions drew attention to Mr Gross's evidence about the role of the swaps and guarantees in the transactions. In his statement of evidence, particularly at 5.5-5.7, Mr Gross stated that the sine qua non for the transactions was their different tax treatment in New Zealand and the counterparties' jurisdictions. He stated:5.6 The swaps and guarantees are, in this context, incidental to the transaction. The transaction works economically without them, because the tax benefits, which drive the transaction, arise from the different characterization of the repo transactions for tax purposes as I have described. That difference in tax treatment relates to the repo, and does not depend on the swap or guarantee fee. The role of the swaps and the guarantees is to mitigate the interest rate and credit risks otherwise present in the transaction 5.7 The swaps and guarantees do of course have tax consequences in each transaction, because the costs involved were, I understand, tax deductible. However, as I have said above they are not the basis for the transaction. [366] Mr Gross then gave an example of how the transaction remained profitable for both parties without a swap or a GPF. His example (PB 5.8) took a margin of 1.25% on a BBR of 5.7%. Thus, [BBR + 1.25] (1-.33). This gave the BNZ a distribution rate of 4.66%, or a pre-tax equivalent yield of 6.95%. From the USissuer's perspective, the distribution rate of 4.66% was 104bps below BBR, swappable into USD at LIBOR minus 104bps. [367] Cross-examined by Ms Scholtens, Mr Gross said of this example: no swaps, no guarantees, and I show that it's still an economical deal what the swaps and the guarantees don't drive the transaction; what drives the transaction is the asymmetry. If you don't have your asymmetry, there'd be no deal. (T 3086)[368] I accept Mr Gross's point. However, it rather underscores two points the Commissioner made in closing: a) The swap was necessary to fix the return from the transaction in the form of the distribution, and the deductible costs of earning it. Absent the swap, the BNZ's floating cost of funds would have been matched by a floating distribution, minus the spread which represented the counterparty's share of the benefits. That matching would have been another way of removing the BNZ's interest rate exposure. The BBR of 5.7% Mr Gross took for his example was a floating rate. b) The method of setting the swap rate "enabled the parties to squeeze a bit more out of the tax base". This refers to the extent to which the fixed swap rate used in some of the transactions was above market. [369] Although I deal more fully with the formula underlying these transactions in [405], it is convenient to set it out here, emphasising the swap rate component:((guarantee procurement fee + fixed swap rate) x (1-tax rate))/(1-(tax rate x benefit split))Given that formula, it could not be – and was not – disputed by the Bank that the rate on the fixed leg of the interest rate swap was a component in fixing the shared tax benefits created by the tax asymmetry which drove these transactions. To that extent, the swap was a vital component of the transaction. It was (as the Commissioner would have it) a cog in the tax machine. If the arrangement was a taxavoidance arrangement, it follows that the swap was a vital component of that arrangement. [370] Unlike the GPF, I do not consider the swap was per se a contrivance. There was no real dispute that interest rate swaps are a commonplace, if not standard, feature of structured finance transactions. Here, by swapping away a fixed interest rate for the duration of the transaction in return for receiving a floating rate, the BNZ approximately matched its funding costs of the transaction. However, there are two points about this element of commerciality. First, it involves circularity because it was the fixing of the distribution rate that created the Bank's interest rate exposure. Had both the distribution rate and the Bank's funding costs remained floating, the Bank would not have had an interest rate exposure. I make this point in [368]a). [371] Secondly, any commerciality in the Bank hedging its funding costs through the swap, must be viewed in the context of the commerciality of the transaction of which that swap was but a component. Mr Choudhry made this point, and succinctly put the interest rate swap in its context in the transaction when he stated:11.5 Leaving aside any tax or accounting implications, the fixed rate amounts payable on the interest rate swaps in the Transactions were immaterial to the parties because they were merely one part of an overall net floating rate payment to BNZ set at a negative spread to BBR. [372] From the counterparty's viewpoint, the interest rate swap had the effect of transferring its fixed rate distribution obligation under the repo agreement into a floating rate obligation. Thus, the swap operated to hedge each party's payment obligations under, or related to, the transaction. [373] In its submissions the Bank asserted that the Commissioner's main criticisms concerned the pricing of the swaps and the pricing process. In terms of pricing, the Bank urged that the question was whether the rate used was within a market range. Mr Stanton made this point about the pricing of the interest rate swaps in these transactions:In my experience, when negotiating the rate for the fixed leg of a fixed/floating swap, one party will want to agree as high a rate as possible and the other will want to agree as low a rate as possible. This does notappear to have been the case in the pricing of the swaps in the disputed transactions. Rather, in my view, the pricing of the swaps reflects the fact that it was in both party's interests to set the fixed leg of the swap as high as possible, and that the normal commercial pricing tension between parties to a swap did not exist. (PB 4.20)Given that reality, I need to review the evidence as to how the rate in each transaction was fixed, as well as evidence as to how that rate compared to market. [374] In closing the BNZ identified the following as the issues focused on by the Commissioner as to the pricing of the swaps:a) Although the rates were set in advance of closing, the parties were not bound to those rates, and so Mr Das' analysis is irrelevant. (Mr Das was the Bank's primary witness on the swaps.) b) The "over-riding desire" of the Bank was to use the swaps to increase the tax benefits in the transactions, evidenced by: (i) the use of the 1 year rate in the Gen Re 1 and CSFB transactions; (ii) the re-setting of the rates on the Gen Re 1 transaction prior to closing; (iii) the failure to reset the rates on the Gen Re 1 and CSFB transactions on their first anniversary; (iv) the attempt to retain the existing rate in the CSFB extension; and (v) the addition of a liquidity premium in the CSFB extension and the Lehman's transactions.[375] That helpfully sets out all the points. I start by making three general points, which subsume some of the points set out in [374]. First, the Bank contended it was factually incorrect for the Commissioner to submit that the fixed rate payment swapped by the Bank had "somehow funded the payment of the distribution". The Bank made the point that swaps are concerned with the net position, after the fixed and floating leg "payments" have been set off. I accept that. Here, gross interest rate payments were only made by the parties in the Gen Re 1 transaction, via the escrow arrangements that were part of that transaction. Those gross fixed swap payments are shown in the "fixed swap" column of the revised cash flow schedulesGen Re sent Mr Birch at the BNZ on 30 June 1998 (17/12033-12034). In all the other transactions, only the net adjustment due was paid on the payment dates stipulated in the agreement. I do not consider that makes the Commissioner's submission, set out in [363]b), factually wrong. The cash flows on the fixed and floating legs of the swap may have been notional, to the extent that they were netted off, but they remained cash flows in the transaction. Certainly, in terms of analysing the transaction economics, the fixed interest rate the BNZ paid on the swap was included, in calculating both the percentage and dollar pre and post tax returns. An example of this is in the Credit Submission for the Lehman's transaction:7. TRANSACTION ECONOMICSThe transaction provides a pre-tax equivalent yield to the NAB Group of approximately 3.01% pa. based on a 5 year swap rate of 6.80% (Gross RoE of 51%, Gross RAROC of 466% and annual EVA of NZ$14m). This is indicated as follows:NZ Tax Rate Transaction Nominal pa 33% NZ$m 500.000 Distributions 8.55% 42.750 Tax impact (due to exempt dividends) 4.21% 21.056 Guarantee Procurement Fee -2.95% (14.750) Swap Funding Cost -6.80% (34,000)Pre Tax Equivalent Income 3.01% 15.056Tax -0.99% (4.968)Post Tax Return 2.02% 10.088(46/34213) (RoE = return on equity. It is essentially income divided by equity; RAROC = risk adjusted return on capital. It is a measure of the bank's risk return; EVA = economic value added. It is the dollar equivalent of the RAROC.) [376] Secondly, there is a difficulty, and an artificiality, in comparing the swap rates used in the transactions to "market". That is because the market in NZ$500 million swaps was illiquid. In fact, the evidence indicated it would be more accurate to say that the market for swaps of that size was non-existent. NZ$20-50 million was the normal range for market interest rate swaps. The Commissioner's witnesses were critical of Mr Das for incorporating a 30bps liquidity premium to allow for this situation. Dr Fitzgerald said this about Mr Das' liquidity premium:7.16 I am unable to understand, or accept the justification for, the 30 basis points liquidity premium. This is not a case where a party was seeking to do a swap of size NZD 500 million with an independent market counterparty, where such a premium might be justified. The interest rate swap was internal to the structured transaction, and in fact was necessary for it to go ahead. Indeed it is equally advantageous to both parties, so why would one pay a liquidity premium to the other to get it to go ahead. If the structure is terminated, then the NZD 500 million swap is automatically terminated without further payment, except for any net accrued interest at the time. So there are no liquidity or size issues at that point either. Hence I can see no justification for BNZ paying an extra 30 basis points to a Credit Suisse subsidiary on the fixed side. In my view, this is simply an artifice to increase modestly the total size of the tax benefit to be shared among the parties. The additional benefit is clearly 0.30 (0.33) or 0.099% shared equally. Although this may appear small, over a three year period the total equivalent cash flow is NZD1.49 million.[377] Mr Stanton considered that allowing any premium for the size of the swap involved pure speculation. His reason was the absence of any market examples of swaps of this size (PB 5.40-5.41; T 3429). Mr Birch was obliged to concede in the course of cross-examination that the .30% liquidity premium was nothing more than a fee which pushed up the fixed swap rate, and thus the resulting benefits for both parties. He certainly accepted that "the premium represented the thinness of an untested market" (T 867-869, particularly at 869). [378] Thirdly, and with one proviso, the counterparty had no real concern about any increase "over market" in the rate adopted for the fixed leg of the swap. There were three reasons for this: a) The counterparty received the higher rate. b) The counterparty paid a floating rate unaffected by the fixed rate. c) The counterparty was getting a share of the tax benefits created by the transaction. All the counterparties were sophisticated financial institutions. All of them would have been aware that the consequence of the formulaic pricing that underlay the transactions was that the higher the fixed rate of the swap, the greater the benefits to be shared. The proviso is that the counterparty may have been concerned that it be in a position to justify the fixed swap rate as a market one, if need be. This emerged from MsMiller's evidence (her statement 6.28, though primarily in relation to the GPF, and 6.35). However, Ms Miller was unaware (T 1534) of the sequence of setting and re- setting the fixed swap rate set out in [380] below. That was doubtless because this process was attended to by Ms Dooley of Gen Re, not by Ms Miller. In the course of answering Ms Scholtens, questions about re-setting swap rates, Ms Miller said "I mean, you weren't able to just change that (the rate that had been set)" (T 1535). [379] Before considering the points listed in [374], I summarise the setting of the fixed swap rate in each transaction, an analysis the BNZ submitted was required. [380] Gen Re 1 – 5 year transaction• 2.3.98 BNZ proposes to Gen Re fixing swap rate at previous Friday's closing one year swap rate of 8.6%.• 31.3.98 BNZ proposes new rate of 9.04% based on previous Friday's closing one year rate.• 11.6.98 BNZ proposes new rate based on previous day's closing one year swap rate of 9.46%.• 23.6.98 Gen Re agrees to 9.46%.• 1.7.98 Transaction closed using 1 year 9.46% rate. 5 year market swap rate is now 7.62%. (I think Gen Re was actually closed on 8.7.98.) [381] CSFB –3 year transaction, extended for a further 3 years• 10.4.98 Harcourt seeks confirmation from CSFB of swap rate of 9.03% (that day's closing one year swap rate). (At that stage the CSFB transaction was to be a "club deal" with the National Bank of New Zealand, through its subsidiary, Harcourt.)• 10.6.98 Harcourt seeks to confirm swap rate at 9.46% (that day's closing 1 year swap rate).• 11.6.98 BNZ seeks same confirmation – at 9.46%.• 3.8.98 BNZ confirms swap pricing at 9.46% i.e. 9.35% on a quarterly basis.• 13/14.8.98 BNZ and CSFB agree that swap rate should be re-set after 1 year.• 21.8.98 Transaction closes at 9.35% fixed quarterly swap rate. Market rate has fallen to 6.985%.• 23.3.99 BNZ (Mr Birch) calculates re-pricing CSFB at current swap rates would result in a pre-tax equivalent loss of $6.2 million for balance of transaction, assuming the rate is re-set on 26.8.99.• 30.9.99 Rates "re-set" but fixed swap rate remains unchanged at 9.3507%. [382] CSFB Extension• 8.8.01 BNZ (Mr Birch) presents ALOC requesting authorities to extend CSFB for 3 years at same (10.6.99) 9.35% fixed swap rate.• 16/21.8.01 CSFB supports extending transaction at same rates.• 22.8.01 BNZ (Mr Bourke) prepares memorandum stating that extending the transaction at existing fixed swap rate of 9.35% "reflect current arm's length market pricing".• 23.8.01 BNZ Taxation gives sign off to extension of transaction noting "the fixed swap rate has been re-set to a current market rate".• 24.8.01 Extension closed. Fixed swap rate recorded as 7.05% (being the 3 year closing swap rate on 23.8.01 of 6.746% rounded to6.75% plus a liquidity premium of .30%). On 23.8.01 the 1 year swap rate was 6.06%. [383] Rabo 1 – 5 year transaction• BNZ proposes fixed swap rate of 7.12% based on that day's closing rate.• 30.6.99 Rabobank (Mr Bowker) suggests "locking in" the pricing suggested by Mr Birch "so we are not exposed to rate movements".• 11.8.99 BNZ (Mr Birch) in e-mail to Rabobank (Mr Bowker) notes "interestingly, the five year swap rate is trending up, and closed at 7.48% today".• 29.10.99 BNZ (Mr Birch) suggests fixing swap rate at previous day's closing 5 year swap rate of 7.80%, giving a fixed quarterly swap rate of 7.7254%.• 3.11.99 Rabobank (Mr Bowker) confirms that pricing.• 22.11.99 Transaction closes at fixed quarterly swap rate of 7.7254%. Market rate is 7.75%. [384] Lehman's –5 year transaction• 29.10.01 BNZ to Lehman's confirming 5 year fixed swap rate of 6.8%, being 6.31% plus liquidity premium as agreed on 23.10.01.• 9.11.01 BNZ notes that rate is "not agreed by Lehman's".• 15.11.01 Lehman's to BNZ asking for explanation as to how the 6.8% swap rate was calculated.• 11.4.02 BNZ to Lehman's suggesting fixing the 5 year swap rate at 7.7% being the screen rate of 7.4% that day plus a liquidity premium of .3%. Agreed by Lehman's.• 26.7.02 Transaction closes using that agreed 7.7% fixed swap rate. Market rate that day was 6.68%. [385] It is evident from this analysis that all the swap rates were set well in advance of the transaction closing. In 5.47 of his statement of evidence, Mr Stanton included this table, based on the information in Appendix 7 to Mr Das's statement: Transaction Fixed Leg of Swap Agreed Gen Re 1 28 days in advance CSFB 76 days in advance Gen Re 2 28 days in advance Rabo 1 27 days in advance Rabo 2 35 days in advance Lehman's 124 days in advance [386] Mr Stanton agreed with Mr Das that this was out of line with normal commercial practice, under which the swap rate was fixed a few days before the transaction was closed (PB 5.48). Given the delays in closing these transactions, Dr Fitzgerald expressed the view that it would be normal commercial practice to re-set the rates in line with market rates at the actual transaction dates if the rates had changed significantly. He defined "significantly" as more than about 5-10bps. [387] Dr Fitzgerald agreed with Mr Das that not resetting the swap rates in line with market had impacted only minimally on the overall economics of the transaction, because the other rates were not changed either. But Dr Fitzgerald made this point:8.16 However, what Satyajit Das fails to take into account is the effect on the tax benefit being created. In a majority of the transactions, a resetting of the fixed swap rates to market rates at the time of the transactions would have had a significant downward impact upon the total tax benefit.[388] I conclude that the setting of the fixed swap rate well in advance of the transaction, and/or the failure to re-set it shortly before the transaction closed, was not in accordance with normal commercial practice. Given the analysis I have set out at [380]-[384], and the differences in the table at [396], I find this was contrived to increase the tax benefits flowing from the transactions. For example, when the rate was fixed for the Gen Re 1 and CSFB transactions, the NZ yield curve was significantly downward sloping i.e. shorter term swap rates were significantly higher than longer term swap rates. Thus, by taking the one year rate the BNZ artificially boosted the tax benefits from those two transactions. Mr Stanton calculated that this boosted the annual tax reductions obtained by the BNZ by NZ$2 million approximately in the Gen Re 1 transaction, and by NZ$1.5 million in the CSFB transaction (PB 5.42-5.43). [389] In terms of not re-setting the swap rate when the transaction closed, the high point was the CSFB transaction. Between 10 June 1998 when the rate for CSFB was fixed, and 21 August 1998 when the transaction closed, the one year New Zealand swap rate fell by some 270bps. The impact of re-setting to the market rate at closing would have reduced the tax benefit from 4.059% to 3.17%, impacting significantly on the post-tax economics of the transaction. (Fitzgerald 7.13). [390] All experts agreed, and the Bank conceded, that it was not commercially justifiable to use the one year fixed swap rate for the five year Gen Re 1 and three year CSFB transactions. I do not accept Mr Birch's justification for doing that, and I do not accept that he ever genuinely believed it was a commercially proper thing to do. I find that it was another contrivance to increase the tax benefits from the transactions. It follows that not re-setting the fixed swap rate on the first anniversary of the Gen Re 1 and CSFB transactions was hardly a failure. The failure was in using an inappropriate rate in the first place. Not re-setting on the first anniversary merely prolonged that failure for the balance of the transaction. [391] The collusion, apparently between the structured finance teams of the BNZ and CSFB, to retain the same fixed swap rate for the three year extension of the CSFB transaction is not evident from my summary at . Apparently following discussion, it involved the CSFB team drafting a proposed letter to the BNZ. It sentthe draft to the BNZ structured finance team for possible alteration of its wording or discussion. The covering e-mail stated "hope it works". The letter was subsequently sent by CSFB to the BNZ. The letter concluded:For the above reasons we feel that in order to provide the guarantee and avoid some difficult tax issues on our side which could affect the transaction, the only way to extend the transaction would be to keep the same rates and we request BNZ Investments Group seriously consider the transaction in this context.[392] The letter did not work: NAB Group Credit and BNZ Taxation both gave sign off for the CSFB Extension only on the basis that the fixed swap rate was re-set to current market rates. [393] I find that there was an unsuccessful attempt by the structured finance teams in the BNZ and CSFB to extend the CSFB transaction using the same fixed swap rate. That swap rate was, when the extended transaction closed on 24 or 25 August 2001, over three years and two months out of date, and was originally a one year rate. Its retention would have been indefensibly uncommercial. [394] As to the liquidity premium incorporated in the fixed swap rate used in the CSFB Extension and the Lehman's transactions, I find the reasoning of Dr Fitzgerald set out in [376] above convincing. I find that premium was another contrivance to increase the fixed swap rate and the benefits to which it contributed. [395] In 8.48-8.64 of his evidence, and in his Tables 8 and 9, Mr Das made a detailed analysis of the fixed swap rates to determine whether they were on or off market. After making various adjustments (for example, for delay and the size of the swap) and applying a tolerance margin (of 40bps either way), Mr Das concluded that the rate used in the Gen Re 1 transaction was off market by 55bps, and the rate in the CSFB transaction by 19bps. In both cases the reason was the adoption of a one year rate rather than the five year and three year rates appropriate to the respective transactions. In the case of the other four transactions, and the extension of the CSFB transaction, Mr Das concluded that the fixed rates were within market range. [396] Dr Fitzgerald did not accept Mr Das's 30bps size adjustment, and considered a more reasonable tolerance range was plus or minus 20bps. Having made thoseadjustments, he compiled this table of the differences between a 'tolerable' market range and the actual fixed rates used in the six transactions:Transaction Tolerance range Actual rate % DifferenceGen Re 1 7.4150-7.8150 9.3507 +1.5357 CSFB 6.7850-7.1850 9.3507 +2.1657 Gen Re 2 6.6050-7.0050 6.5049 -0.1001 Rabo 1 7.5500-7.9500 7.7254 0 Rabo 2 7.2200-7.6200 7.6371 +0.0171 Lehman's 6.4800-6.8800 7.7000 +0.8200(PB 8.21) [397] He then expressed his conclusion in this way:8.22 Based on this analysis, I would find that five out of the six fixed swap rates are outside what could be termed the on market range, although in the case of Gen Re 2 and Rabo 2 only by relatively limited amounts. Nevertheless, I would take the view that in these five cases, it would have been normal commercial practice to reset the rates. I would further add that in my view of the evidence, I have seen no indication that any resetting of the swap rates was contemplated, regardless of the size of any post setting rate movements.[398] For two reasons, I prefer Dr Fitzgerald's analysis. First, for the reasons Dr Fitzgerald gave ([376]) above, I do not agree with the 30bps size (i.e. liquidity) premium Mr Das applied. Secondly, I consider his 40bps tolerance too high, although I accept the appropriate tolerance is a matter for professional judgment. However, it is probably unnecessary to express a firm preference as between the evidence of Mr Das and Dr Fitzgerald. The point is that both of them found the fixed swap rates used in some of the transactions were outside market. That would not be the position if normal commercial practice had been followed. It is indicative of what I find was the contrived and uncommercial manner in which the fixed swap rates used in the transactions were arrived at. [399] Dr Fitzgerald considered the fixed swap rates used in the transactions were non-market in another respect. This was a consequence of what Dr Fitzgerald considered was the setting of the distribution rates in the transactions at non-market levels. To fund those distributions, the yields on the investments made by the SPVsin the various securities issued by the counterparties of the BNZ, and/or the amounts of those investments, had to be pitched at certain levels. The first three transactions sufficiently demonstrate this:• Gen Re 1: LIBOR plus 55bps in USD on USD 561,603,899.• CSFB: 7.5923% in NZD on NZD 1,000,002,000.• Gen Re 2: LIBOR plus 55bps in USD on USD 271,844,820. [400] In the absence of the tax benefits produced by the transactions, Dr Fitzgerald did not consider that Gen Re (rated AAA) would ever authorise GRCF to pay 55bps over LIBOR in USD to a US trust (the Gen Re 1 and Gen Re 2 Trusts). He considered that Gen Re would normally be able to borrow in financial markets at LIBOR, and probably sub-LIBOR. It was commercially and economically inexplicable for Gen Re to pay LIBOR plus 55bps. It was even more commercially inexplicable for Gen Re to pay the same rate in the two Gen Re transactions. Between July 1998 when Gen Re 1 closed and June 1999 when Gen Re 2 closed, although three month USD LIBOR fell by 30bps, the AAA spread (or margin) over equivalent Treasury yields increased from 61 to 85bps. [401] Dr Fitzgerald used the Bloomberg swap calculator to convert the Gen Re USD rate into the equivalent fixed NZD rate. Having done that, he compared the fixed rates with market swap rates on the relevant dates:GEN RE 1 CSFB GEN RE 2 1/7/98 21/8/98 2/6/99NZD FIXED RATE 8.1244 7.5923 7.3067 MARKET SWAP RATE 7.6150 6.9850 6.8050 DIFFERENCE +51 bps +61 bps +50 bps[402] Dr Fitzgerald did not consider that these differences (or spreads) were normal commercial margins. He considered CSFB (with an AA- rating) would have been able to enter into an NZD fixed or floating swap at a fixed rate of no more than 2 or 3bps above or below the mid market swap rate, and an AAA rated institution such as Gen Re at significantly better rates.[403] I conclude that the fixed swap rates used in two, at least, of the transactions were significantly off-market. [404] To summarise, I answer the points I set out in [360] as follows: a) The primary purpose of the swap was to facilitate a fixed distribution rate under the transaction, and thus fix the tax benefits shared by the parties. b) The manner in which the interest rate swaps were transacted was not in accordance with market practice in several respects. This had the consequence that the fixed interest rate in at least two of the transactions was well out of line with the market rate.The formula[405] In analysing the transactions for the Commissioner, three of his witnesses spotted that the pricing of the transactions was based on a formula. I heard a lot of evidence about this formula, including from those three witnesses. I see the relevance of the formula thus: how did it work? What was its effect? What do the answers to those questions indicate in terms of the purpose or effect of the transactions? [406] The formula was produced as Exhibit A:((guarantee procurement fee + fixed swap rate) x (1-tax rate))/(1-(tax rate x benefit split))[407] When this formula was put to Mr Birch he readily acknowledged it. I accept that he may not have been aware of the formula from the outset. But it is evident to me that he very rapidly grasped that some formula underlay the fixed rate referred to in the initial proposal which the NAB referred on to the BNZ. This is the proposal I referred to in [48]. When asked about his algebraic workings on the proposal (at 5/2604) Mr Birch said he thought he may have been trying to "reverse engineer thenumbers" (T 1374). I think this was an early attempt to derive the formula, or at least to understand it. [408] It might be protested of the formula, 'so what'? For example, when asked about the formula Mr Gross replied:All the transactions we (the Bank of America) did were based on a formula. In fact, all banking transactions are based on a formula. (T 3085)[409] In order to respond to the 'so what', and to answer the questions posed in [405], I need to review the evidence, and I start with Dr Fitzgerald. He expressed the view that the combination of the fixed swap rate, the distribution rate and the GPF was designed to produce specific tax benefits and to divide those benefits between the parties in an agreed manner. His evidence about this was:In the negative spread transactions, Gen Re 1 and Gen Re 2, the tax benefit would always be the fixed swap rate (prior to the negative spread) plus the credit guarantee procurement fee times the NZ tax rate. GEN RE 1 (6.9194 + 24313 + 2.95) (0.33) = 4.0592 GEN RE 2 (4.8849 + 1.6200 + 2.95) (0.33) = 3.1201 The tax benefit is divided with the negative spread being the gross benefit for General Re, and the tax benefit minus the negative spread being the net benefit to BNZ. TAX BENEFIT NEGATIVE SPREAD (GEN RE BENEFIT) DIFFERENCE (BNZ NET BENEFIT) GEN RE 1 4.0592 2.4313 1.6279 GEN RE 2 3.1201 1.6200 1.5001 In the other four cases, where there is no negative spread, the tax benefit is simply the sum of the fixed swap rate and the GPF times the NZ tax rate. CSFB (9.3507 + 2.95) (0.33) = 4.0592 RABO 1 (7.7254 + 2.95) (0.33) = 3.5229 RABO 2 (7.6371 + 2.95) (0.33) = 3.4937 LEHMAN'S (7.7000 + 2.95) (0.33) = 3.5145 The difference between the distribution rate and the sum of the fixed swap rate and the GPF represents the gross benefit to the counterparty. The remainder of the tax benefit is the net benefit to BNZ.COUNTERPARTY GROSS BENEFITCSFB (9.8700 – 9.3507 – 2.95 = 2.4307 RABO 1 (9.1699 – 7.7254 – 2.95) = 1.5055 RABO 2 (8.8446 – 7.6371 – 2.95) = 1.7425 LEHMAN'S (9.4000 – 7i.7000 – 2.95) = 1.2500 TAX BENEFIT COUNTERPARTY GROSS BENEFIT DIFFERENCE (BNZ NET BENEFIT) CSFB 4.0592 2.4307 1.6285 RABO 1 3.5229 1.5055 2.0174 RABO 2 3.4937 1.7425 1.7512 LEHMAN'S 3.5145 1.2500 2.2645 The formulaic nature of all this is self-evident. The fixed swap rate and the GPF determine the tax benefit. The negative spread, or the level of the distribution rate, determines the split of the tax benefit between the parties. Apart perhaps from the perceived need to argue that the levels are market related, it is entirely arbitrary as to what the levels of the swap rates and procurement fees have to be to produce a specific benefit. (PB 7.82-7.84)[410] Professor Evans dealt with this aspect in a lengthy section, paras 47-92 of his statement of evidence. He explained that the quarterly payments listed in [82] constituted a circular flow of funds between the BNZ and CS Group which had the function of utilising tax benefits and splitting them between the parties. He depicted these payments in his Figure 3 as a basis for a formulaic description of them:[411] He pointed out that the GPF (g) and fixed leg of the interest rate swap (f) sum to 12.3007%, which is equal to the distribution rate of 9.87% plus 2.4307%. That 2.4307% is equal to the CS Group's pre-tax equivalent return (prior to its cross currency swap with NBNZ and its letter of credit cost). Thus, the levels of the GPF and fixed leg of the swap are set to enable the BNZ Group to circulate the distribution of 9.87% back to the CS Group plus its agreed share of the tax benefits. The total payment of the GPF and fixed leg of the swap can be expressed in the mathematical formula g + f = d + (y-tz)/(2-t), where "t" equals the New Zealand corporate tax rate, and the other components are as shown in Professor Evans' Figure 3. [412] Professor Evans explained that the mathematical relationship underpins the circular flow of funds in each of the other five transactions, although the formula differs because the agreed benefit splits differ. [413] In his analysis, Professor Evans treated the fixed swap rate as exogenous, because it was set with reference to a particular market rate. However, he went on to demonstrate that the GPF and distribution rate in the CSFB transaction could both be increased in a way that increased the profitability of the transaction to both parties. Obviously, an increase in both the GPF and distribution rate would increase profitability for the CS Group provided the increase in the GPF was greater than theincrease in the distribution rate. Because the distribution rate was tax exempt, an increase in the distribution rate greater than 1 minus the tax rate, multiplied by the increase in the GPF, also increased profitability for the BNZ. [414] The effect of increasing the GPF and distribution rate would have been further to reduce the New Zealand tax base. As shown in Figure 3 in [410] above, the tax deductions generated for the BNZ by the CSFB transaction were 4.0988% of NZ$500 million. If the GPF had been increased to 5% (from 2.95%) and the distribution rate to 11.5% (from 9.87%) the tax deductions generated would have been 4.7753% of NZ$500 million (33% of (5% + 9.3507% + 0.1200%)). [415] Professor Evans demonstrated by three dimensional graphs that the GPF and distribution rate could have been increased over a wide range in such a way as to maintain the equal split of the transaction pre-tax equivalent returns. [416] Professor Evans said that the same analysis of the setting of the fixed variables applied to the remaining five transactions. However, Gen Re 1 and Gen Re 2 had a negative spread on the fixed and floating legs of the swap (he disregarded as "relatively immaterial" a 0.03% negative spread for the first year on the floating leg of the swap in the Rabo 1 and Rabo 2 transactions). In the two Gen Re transactions the distribution rate was set so that it equalled the sum of the GPF and the fixed leg of the swap less negative spread. The net fixed payment by the BNZ to Gen Re for each transaction thus equalled the negative spread on the floating leg of the swap. That spread equalled Gen Re's agreed share of the transaction pre-tax equivalent returns. That negative spread was an artificial construct which ensured that the distribution rate netted to zero with the GPF and the fixed leg of the swap less negative spread. That obviated the need for the exchange of any payments, other than the negative spread (as applied to the floating leg of the swap). [417] Mr Stanton also expressed the view that it was axiomatic that, in order to achieve the agreed benefit split, the distribution rate was set by way of a formula. He demonstrated that by deriving the distribution rate for the Rabo 2 transaction with its benefit split of 60/40 in the BNZ's favour (PB 4.22). He explained that the sameprinciple applied in the other transactions, although the process would be different. He gave (PB 4.26) this example:In Gen Re 1 for example, with its 50/50 benefit split, by setting the fixed leg of the swap at 9.35% and the guarantee procurement fee at 2.95%, the required distribution rate (d) would be: d = [(2.95 + 9.35) x (1-0.33)]/[1-(0.5) times (0.33)] from which d = 9.87%[418] Mr Stanton summarised the formulaic pricing of three legs of the transaction in this way:In order for the benefit mechanism to function effectively it was necessary that all three legs were set together. In terms of process, having set the swap rate and the guarantee procurement fee, the distribution rate would then be derived so as to achieve the agreed split. Were, for any reason, it to be agreed that the swap rate or guarantee procurement fee were to be set at a different level, then the distribution rate would need to be recalculated and reset. Put another way, the rates of the three legs were interdependent. (PB 4.30)[419] Finally, Mr Stanton said that the other bank transactions he had reviewed worked in the same way, with one exception relating to a transaction that had a different fee structure.The sub-LIBOR funding for the counterparty[420] As I understand it, the Commissioner's position on this aspect (his closing submissions 31-36), was that none of the counterparties had a "need" for the funds the BNZ provided to them. No counterparty approached the BNZ seeking funds. That would have been unlikely, because an AA- rated bank such as the BNZ was throughout, cannot lend profitably to an AAA rated bank or institution such as all the counterparties here were (except Lehman's (A)). [421] So the BNZ attracted the counterparties to these transactions by offering them funds at well below their normal cost of funds, as compensation for theirparticipation in an arrangement that provided very substantial tax benefits to the BNZ. That compensation took the form of an agreed share of those benefits. [422] In the closing submissions for the BNZ, that last point – that the counterparty received a share of the tax benefits generated by the transaction – is conceded:8.39 the point that is missed in the Commissioner's Closing, is that the level of reduction against LIBOR that is received by the counterparty is simply a function of the tax benefit available, and the prevailing level of interest rates. 8.42 Accordingly, once it is recognised that the funding benefit to the counterparty comes from the receipt of an exempt payment by the Bank, the level of reduction in funding costs as against LIBOR is understandable and not remarkable. [423] At peril of unnecessarily dealing further with this aspect, I think there is a point to be made. Its relevance is that it further evidences how exceptional the level of funding obtained by the counterparty was, compared with conventional returns on structured transactions. [424] The BNZ submitted that Dr Fitzgerald explained that "most financial institutions look to raise funding at a substantial deduction to LIBOR". To put numbers on this, what Dr Fitzgerald said was: Conventionally, on the borrowing side, the funding rate targets even for AAA or AA institutions have, in my view, been relatively low. A floating or fixed rate issue, whether plain vanilla or structured, that produced a net result of Libor minus 30/40 basis points or more would be attractive. I believe this view would be shared by most knowledgeable financial market participants, and is consistent with the funding rate targets set by major US and European institutions that I have been familiar with in structured finance activities. (PB 6.7)Dr Fitzgerald went on to compare this with the mean spread for the US Government of 43bps below LIBOR, and for the UK Government 38bps below LIBOR. His point was that AAA or AA rated institutions borrowing at LIBOR minus 30/40bps are borrowing at around Government rates.[425] The analysis carried out by Dr Fitzgerald which I have set out in [409] shows that the counterparties to these transactions received the NZ$500 million of funding at between 243 and 125bps below LIBOR (or its New Zealand equivalent BBR). [426] The evidence of the attractiveness of funding at those levels included this: a) BNZ (in a file note dated 8 June 2000 signed by Mr Rogers, Credit Controller, Specialised Finance, in relation to the Rabo 2 transaction): They don't come much better than this. (58/45328)b) Mr Birch:Extremely good yields. (T 861)c) Mr Choudhry: In my view, these rates represent an extremely cheap cost of funding, and almost certainly better than most banks could get. (PB 13.2)d) Dr Fitzgerald:Extraordinarily attractive levels. (PB 7.85) Obviously, if you come and offer them LIBOR minus 100, you're trampled to death in the rush of people wanting to do that sort of transaction, and if you offer them 243 basis points below LIBOR, I guess their first reaction would be, this is astounding and, you know, tell me more about it. (T 3182-3183) I think raising funds at LIBOR minus 240 is astronomically beneficial (T 3250)e) Mr Hodder: if you can borrow at less than Treasury, you're doing very well. You've got a very good rate. (T 2648)f) Mr Middleton:Attractive rates attractive margins. (RB 4.34; T 2988-2989) Q. What I want to suggest to you is that these were very good rates indeed, and I wonder if you have a view on – how good do you think the rate of cost of funds to Rabo was out of these transactions? Good, very good, exceptional? A. They would certainly have been good, they would have been pleased with the transaction. Q. Good? A. Yes. Q. Can I push you any further than good? A. I don't think I'm on auction here, are we? (T 3007)g) Ms Miller:Yes, it was advantageous low cost funding, yes. (T 1635)h) Rabobank (in an internal memorandum dated 29 November 1999):The transaction involves the Group raising funding from an independent third party (BNZ) at around NZ$ Bank Base Rate (BBR) less 1.50%. This compares with the bank's normal target funding cost in NZ$ of around BBR less 0.125%. The Bank has sourced one of a very few investors willing to do such a transaction at the right price. (34/24950)One comment, on the evidence of Mr Middleton. He was a Scottish banker. In addition to being expert in banking and finance, I regarded him as skilled at understatement. [427] Conventionally, returns are proportionate to risks. The counterparties bore no risk in these transactions [472]. The evidence set out in [426] needs to be considered in that light.Transactions profitable at 'market' pricing[428] Dr Fitzgerald points out that the transactions would still have been profitable had they been based on (what he considers was) 'market' pricing, and on the tax treatment assumed by the BNZ. In view of my finding about the pricing of the interest rate swaps and the reservations I have expressed about the pricing of the GPF, it is appropriate that I deal with Dr Fitzgerald's point. He demonstrated it in his Exhibit 27 which: a) Uses a GPF of 25bps. b) Uses market swap rates as of the transaction dates. c) Assumes an equal split of gross benefits. [429] Dr Fitzgerald's Exhibit 27 shows that funding for the counterparties continues at spreads under LIBOR ranging from 137 to 158 bps, and net benefits continue for the BNZ ranging between 92 and 106 bps. He observes (PB 7.89) that these "are still significantly in excess of conventional returns for structured transactions". [430] Rather than replicating the whole of Dr Fitzgerald's Exhibit 27, I have confined it to the Rabo 1 and Lehman's transactions, as they are the two demonstrating each end of the ranges referred to by Dr Fitzgerald:PRE AND POST TAX RETURNS USING MARKET VARIABLESRabo 1 Lehman'sPre tax Post tax Pre tax Post taxFixed distribution rate 6.4192 6.4192 5.5622 5.5622 Procurement fee (0.25) (0.1675) (0.25) (0.1675) Swap fixed side (7.7500) (5.1925) (6.6820) (4.4718) Funding cost (LIBOR) (0.67 LIBOR) (LIBOR) (0.67 LIBOR) Swap floating side LIBOR 0.67 LIBOR LIBOR 0.67 LIBOR Net -1.5808 1.0592 -1.3698 0.9175Early termination[431] Each of the transactions contained detailed early termination provisions. The Commissioner appended summaries of these to his closing submissions (Appendix 12). Although the conditions differ slightly, those in the Gen Re 2 transaction are typical:[432] Prominent amongst the non default reasons for termination were adverse tax changes or events. The BNZ criticised the Commissioner's summaries as incomplete and too heavily paraphrased. However, the summary I have replicated adequately demonstrates the prominence, amongst the non-default reasons for termination, of adverse tax changes or events. [433] Upon early termination, the payments due were only those accrued to termination date. In other words, there were not to be payments representing thepresent value of the balance of the transaction components. If the repo agreement was terminated early, then the interest rate swap terminated with it, also with payments on an accruals basis only. [434] Mr Fitzgerald's experience was that this was unusual: In general, in the case of an early termination of a structured transaction, one would expect to see termination payments by one or other party to reflect the mark to market value of the components of the transaction. Similarly if two parties have entered into an interest rate swap, then the swap will doubtless have developed a mark to market value with the passage of time and movements in market rates. Normally, such a positive or negative mark to market value would be paid by one party to another. (PB 7.91)[435] Mr Choudhry did not agree, at least in respect of the interest rate swap: It is usual for the termination terms of a swap to mirror the termination provisions of the underlying transaction to which the swap is a hedge. Therefore, while the early payment provisions would be odd if one looked at the interest rate swaps in isolation, they make sense when the swaps are seen as merely components of what are, essentially, economically unsecured loans. (PB 11.9)[436] I need not resolve this conflict, because the point about the early termination provisions is that they support the Commissioner's contention that these transactions were "tax machines". While that description is his, not mine, it is apt to this feature of the transactions. The "machines" were running on tax benefits. When those tax benefits ran out, the machines could – and would – be shut down.Use of tax capacity and ETR[437] Both in opening and in closing its case the BNZ submitted that these aspects had no impact on the tax effects of the arrangement, and were irrelevant. I do not altogether agree with that. Both factors reflect and point up the tax effects of the transactions for the BNZ.Tax capacity[438] Tax capacity (also referred to as tax shelter or tax shield) refers to whether or not a taxpayer is in a tax paying position. If a taxpayer has taxable profits, it has tax capacity. Specifically, in relation to structured finance transactions, the NAB, in a presentation in June 1998 on tax shelter utilisation, stated:Tax shelter is the level of taxable income generated from traditional business against which deductions generated by potential structured finance transactions may be taken to reduce the ultimate amount of tax payable. (58/45248; Mr Hooper was cross-examined about this at T 268>)In his evidence Mr Kyle described tax capacity succinctly:By "tax capacity" I mean that the funding costs and other deductions associated with the transaction were able to offset against gross income of the BNZ Group. (PB 3.15(l)) [439] Evident from these descriptions is the fact that transactions of this sort require, and consume, tax capacity. Although the BNZ considered it had a comparatively "large tax capacity" (noted as a "strength" in the BNZ Structured Finance Group's Strategic Plan 1997/98 at 50/36602), it recognised that this capacity was a "very limited and precious resource" to be "very carefully monitored" (Mr Hooper, T 213). [440] Throughout these transactions, the first call on the BNZ's tax capacity was always the requirement to "frank" dividend payments to its parent, the NAB. The BNZ's periodic "Summary of Tax Shelter" calculations always stated that "tax capacity must be allocated first to maximise the repatriation of dividends. The Summary thus contained a schedule along these lines:NZ$'m BNZ group taxable Income X New Zealand tax at 33% X Tax required to cover dividend payments to NAB (x) AVAILABLE TAX SHELTER X(This is in Mr Stanton's evidence at PB 2.25. An actual example is at 51/36924) [441] In his evidence Mr Birch described tax capacity as one of the limiting factors on the structured finance team undertaking transactions such as those in issue here. He explained: For these reasons, we actively managed the transactions we had in place involving this structure, to ensure that we obtained the best returns available from the use of our tax capacity. (PB 5.4)[442] Around the time the Gen Re 1 and CSFB transactions (together involving NZ$1 billion) were being finalised, NAB's Group Leadership Team issued a paper on "Tax Shelter Utilisation" (51/36900. It is dated 27 July 1998), and subsequently "Tax Shelter Allocation Guidelines" were introduced. These allocated tax shelter to New Zealand as follows:1998 1999 2000New Zealand (NZ$m) 173 198 262(51/36916) [443] The Guidelines also required monitoring of the use of tax shelter, in the form of quarterly reporting (51/36917 – the format of the report was Appendix 1 to the Guidelines). [444] At [272] I referred to the BNZ's growing awareness of how valuable a resource its tax capacity was, reflected in the increasingly favourable (to the BNZ) split of the tax benefits in the successive transactions, shown in the table in [272]. [445] Throughout the BNZ's documentation relating to these transactions are constant references to tax capacity. Examples include: a) A 27 March 2001 memorandum recognising that it has all but used up its allocated tax capacity and is unable to enter into any new structured finance transactions utilising tax capacity. The BNZStructured Finance Group was to ask NAB to agree to a dividend retention policy which would enable a further proposed £250m "funding deal" to proceed. (58/45342) b) Mr Birch presenting a "Business Plan 2003" at a BNZ Structured Finance Leadership Team conference on 28 and 29 October 2002: Significant aspects of the environment in which the Structured Finance New Zealand business will operate in 2003 include: a severe shortage of tax capacity (49/36060) Our ability to add to this book (of five existing 'repo' transactions) is seriously hampered by a lack of capacity (49/36062) [446] Reflecting the facts that the transactions used tax capacity and affected the BNZ's ETR, each required approval from BNZ Taxation Services. For example, the 19 November 1999 approval for the Rabo 1 transaction noted:The tax capacity of the BNZ following the transaction is expected to be sufficient to allow full imputation credits on dividend paid to NAB. The tax capacity will continue to be monitored to ensure sufficient capacity exists to fully impute the NAB dividends.33/24870) [447] Lastly, it was the combination of tax capacity constraints and the more favourable benefit split available, that caused the BNZ: a) To terminate Gen Re 1 (with a 50/50 benefits split) in order to undertake Rabo 2 (60/40) (Mr Birch PB 6.66); b) To substitute the Lehman's transaction (73/27) for CSFB (50/50) (Mr Birch PB 6.92); c) To convert the AIG 1 structure (which consumed tax capacity, because the intra-company tax exempt dividends it paid to the BNZwere utilised to shelter the BNZ Group's other taxable income) to the AIG 2 structure (which did not consume tax capacity, because the BNZ received a fully imputed dividend in respect of debentures issued under s FC 2 of the Act), for the new transaction.ETR[448] When these transactions were discussed at the BNZ, ETR was generally mentioned in the same breath as tax capacity. The BNZ's ETR was/is its tax expense expressed as a percentage of its gross profit. The comparison is to the corporate tax rate of 33%. [449] Although the Bank did not disclose its ETR in its annual report or financial statements, it was closely monitored and reported upon in conjunction with tax capacity, in relation to these transactions. That was because the effect of the transactions was to reduce the BNZ's ETR. For example, commenting on the proposed Gen Re 1 transaction on 2 June 1998, the Head of Accounting and Compliance at the BNZ noted that, as a result of the transaction: the BNZ Group's tax expense as a proportion of pre-tax income will decrease. (17/11919)[450] The Bank's evidence about these aspects was given primarily by Mr Hagan. For 25 years until August 2005 Mr Hagan was a corporate finance partner of Deloitte. For the last 10 of those years he was also Chairman of Deloitte. He is also a past Chairman of the Accounting Standards Review Board, a past Chairman of the Financial Reporting Standards Board and a past President of the Institute of Chartered Accounts of New Zealand. His credentials as an accounting expert are impeccable. Mr Hagan did not consider the Bank's ETR had any relevance to matters in issue in these proceedings. He said: I mean, to me it's a sideshow, and an irrelevant one (T 2288)[451] Mr Hagan had previously explained:I don't think that the Inland Revenue Department would be terribly interested in what the effective tax rate is. What the Inland Revenue Department would be interested in is how much tax has the company paid Their interest would be, has each of these companies paid the amount of tax that it should pay. The effective tax rate has got nothing to do with how much tax the company pays, its just a tax charge in the PNL (the income statement) (T 2286) [452] How then, did Mr Hagan explain the frequent reference in the BNZ documentation to the affect of these transactions on the Bank's ETR? Mr Hagan answered that in this way: the way I see the bank's interests; the bank wants to be seen as a good corporate citizen. Now, as an ordinary New Zealander I know that the headline tax rate's 33%. So, if I saw a bank reporting that it was paying 2% income tax, I'd say, what's going on? So, the banks and these other companies, when they're reporting effective tax rates, they're trying to manage the expectations and the impressions of the relatively simple investment community (T 2287-2288)[453] That, indeed, appears to have been the BNZ's concern, because the contemporary documentation shows the Bank comparing its ETR with that of other banks and large corporates. Just one example sufficiently demonstrates this. It is in a 2 June 1998 memorandum sent by the Head of BNZ Taxation to his counterparts at the NAB, at a time when approval was being sought for the Gen Re 1 and CSFB transactions. Mr Papageorgiou advised:Effective Tax Rate (ETR)The effective tax rate for the BNZ Consolidated Group for the half year was 28% (at this stage we do not have comparatives of other Banks). If the two US proposals proceed the year end position is expected to be 25.5% : 20.1% (1999) and 23.4% (2000), if only one proceeds the rate will be 26.8% : 24.0% (1999) and 26.20% (2000). Under either scenario the rates are within the bounds of acceptability when regard is had to the comparative rates set out below for the 1997 financial year: BNZ 27.02% ANZ 28.41% National 29.61% Westpac 32.04% Tower 06.00% AMP (1996) 06.90% National Mutual 18.70% Mobil Oil NZ 29.00%FCL – Paper 19.40% FCL – Forest 14.60% Carter Holt 28.50% (49/36280)[454] Concern about the Bank's ETR was also the reason for the BNZ's concerted effort, during the negotiations for each of the six transactions, to consolidate the issuer in the Bank's accounts. These efforts were successful in the case of the Gen Re 2 and Rabo 2 transactions. Broadly, consolidation had the effect of maintaining the Bank's ETR, because the tax paid by (or on behalf of) the Bank in the counterparty's jurisdiction was comprised in the reported tax paid. Consolidation had no affect on the New Zealand tax base: the transactions remained equally fiscally negative for New Zealand. [455] BNZ Taxation's approval for each of the Gen Re 2 and Rabo 2 transactions noted that the issuer was able to be consolidated. I take the Gen Re 2 approval on 2 June 1999 as an example:We recommend that the transaction proceeds on the following grounds: • Group Accounting have confirmed that the Trust is able to be consolidated for financial reporting purposes and this accounting treatment ensures that there is no deterioration in the effective tax rate of the BNZ.(29/22074)Transactions pre-tax negative[456] In opening its case the BNZ acknowledged:without that tax relief for the distributions received, the transactions would not have been profitable for the Bank and would not have been entered into; (Opening submissions 1.11(b))[457] The Bank also made an opening submission that New Zealand Courts have never held that the fact a transaction is unprofitable absent the tax benefits is per se an indication of tax avoidance: Challenge at 549, Petersen v CIR [2006] 3 NZLR433 at 444. I accept that the Privy Council in Petersen said that transactions should not be struck down merely because they were influenced by the prospect of obtaining a tax advantage. [458] All witnesses who gave evidence about this aspect agreed that transactions driven by tax (in the sense I use the term in [23] are commonplace in financing, and that they make business sense, provided of course the parties have got "the tax treatment right" (Mr Choudhry T2637). [459] Mr McLeod summed the position up nicely:Investments resulting in a pre-tax negative return will inevitably depend on the tax treatment for their viability. That follows from the fact that the tax treatment of an amount will affect the net return to the investor, and therefore the decision as to whether or not the investment is viable. (PB 6.23)[460] Consistent with that, witnesses agreed that businesses, in assessing the profitability of proposed transactions, generally look at them on a post tax basis. That, in turn, reflects that business views tax as a cost. This exchange occurred during the cross-examination of Professor Saunders:Q. I take it you accept Professor that tax is a business cost? A. Well, tax can produce tax shelters which is a benefit as well as a cost, yes. (T 3281)[461] In closing the BNZ suggested that, faced with this evidence, the Commissioner had not really pursued this aspect, and he did not. [462] In Petersen at [44] the Privy Council stated the legal position in this way:Tax relief often makes the difference between profit and loss after tax is taken into account; and the transaction does not become tax avoidancemerely because it does so (my emphasis)[463] It follows that the fact that the BNZ provided the NZ$500 million funding in each of these transactions at substantially less (up to 2.5% less) than its cost of funds is a factor for me in deciding whether these transactions were tax avoidance arrangements. It certainly is not conclusive. It is best approached as one aspect of viewing the transactions in a commercially and economically realistic way, and I now do that.The transactions "viewed in a commercially and economically realistic way"[464] The quote is from the main judgment in Ben Nevis at [109]. [465] Contrasting the transactions here to the Trinity scheme in Ben Nevis, the BNZ stressed they were "driven by real obligations incurred by the bank", obligations that were intended to be performed and were performed. Mr Galbraith focused upon two fundamental obligations under each transaction: The market borrowing by the BNZ; the funding provided by the BNZ to the counterparty. Each involved a $500 million exposure for the BNZ. These points emerged particularly from the BNZ's opening 1.7 and 1.16 and its closing submissions at 1.22 and 4.28>. [466] The Commissioner accepted that the transactions created obligations, as do I. Indeed, almost equally fundamental was the corresponding obligation of the counterparty to repay the funding to the BNZ i.e. to repurchase the repo securities at the agreed $500 million. [467] But does not Ben Nevis require an answer to the question: what was the business purpose of these transactions? Or, to tailor the question to the BNZ's 'obligations' point, what was the purpose of creating those obligations? The BNZ answers: profitable investment of $500 million with the counterparty. Those profits were all tax benefits: a combination of the benefit of expenses deductible against the BNZ's other income, and the tax exempt income that expenditure earned. That reality will, shortly, lead me back to the application of s BG 1. In terms of the business or commercial rationale of these transactions, I consider Dr Fitzgerald's evidence exactly captures the position:All in all it is my view that, in the absence of the taxation benefits, none of the transactions, taken either as a whole or in terms of the components, are reasonable or defensible from a commercial or economic point of view. Even taking the taxation aspects into account, it is my opinion that the use of offmarket rates and GPFs boosted the tax benefits considerably above what they would otherwise have been. In my view, the six transactions' all in terms were not determined as part of an arms length market driven process. (PB 5.12) Equally clearly and self-evidently, in the absence of tax effects, it is my opinion that the transactions cannot be considered commercially or economically rational from the point of view of BNZ, since they would result in an effective loan to the counterparties at a floating rate in New Zealand dollars of anything up to 250 basis points below their cost of funding. Even taking taxation effects into account, the transactions are only rational for BNZ if it has a sufficient tax shelter requirement arising from its other businesses. The transaction would not work without that sufficient tax shelter. (PB 7.86)[468] In terms of economic effect, I found the evidence of Professor Evans definitive. Beyond a general challenge to its relevance, it was not contested. GivenBen Nevis [109], I consider Professor Evans' evidence about economic effect relevant. [469] First, as I mentioned in [13], Professor Evans accepted that, by better matching funds and risks across investors, structured finance transactions can lower the cost of capital, and thus facilitate real economic activity (PB 40-43). In the Professor's view, none of these transactions had any effect on "real" economic activity. That was because the BNZ was not a net borrower, and the transactions had negligible effect on the BNZ's risk profile, apart from tax risk. Therefore it cannot be the case that the transactions lowered the cost of capital to the BNZ itself (PB 106). [470] Under cross-examination by Mr Galbraith Professor Evans said:If there was any real economic activity it would be the transference of some liquidity between the two, i.e. the BNZ transferring some liquidity namely $500 million worth of it to the other counterparty, to the counterparty. Now that would be the only connection to real economic activity. (T 3495)[471] Professor Evans went on to explain that the effectiveness of that liquidity was constrained by the termination clauses in the agreements – by the fact that the NZ$500 million was effectively on-call after the first year: The termination clauses significantly limited the liquidity attached to the loan. (T 3496) but what this means is that the termination clauses mean that it isn't as though liquidity was provided for the period of the transaction. (T 3497)[472] Risk is invariably a factor of business transactions. After careful analysis of the CSFB transaction, Professor Evans concluded that it held only tax risk for the BNZ, although that was considerable (PB 116). There were no real risks for the Credit Suisse Group (PB 120). The risk position was substantially similar in the five other transactions, although the mechanisms used to mitigate risk differed (PB 121). [473] Professor Evans assessed the PV of the CSFB transaction for the BNZ at its closure (i.e. commencement) on 26 August 1998 at NZ$21,689,215. That calculation used a discount rate of 6.49% adjusted to a tax paid basis, and excluded the cost of the work on the transaction by BNZ employees. Thus, the transactions returned high yields for the BNZ with no risks other than tax. That is the antithesis of the usual commercial equation. [474] On the assumptions, and calculated on the basis, he outlined in his evidence, Professor Evans calculated that the New Zealand only domestic social cost of the CSFB transaction at closure was $3,259,495. Adding the benefits transferred to the United States i.e. transferred to CSFB over the duration of the transaction (benefits that were totally lost to New Zealand), the cost rose to NZ$36,952,859 (PB 135- 136). [475] Professor Evans calculated that the six transactions enhanced the value of the BNZ Group to its owner (i.e. to its shareholder, the NAB) by NZ$238.6 million as at 30 June 2005 (termination of the last two transactions). As at the same date thetransactions had a total cost to New Zealand society of NZ$335.6 million (PB 151 and table 3 at p82). [476] Professor Evans noted that the rationales claimed for the transactions were: primarily their profitability; strengthened relationship with a large financial institution; enhanced reputation of the BNZ Group in structured finance transactions (PB 105). He explained: It is common for the private costs and benefits of private sector transactions to be a close proxy for social costs and benefits that may arise on a forward-looking basis, including investments in reputation and relationships over time. However, any investment in reputation and relationships in the current case is effectively funded out of the tax base, rather than privately by the BNZ Group. I do not consider that the reputation for skills in structured financial transactions and the strengthened relationship between BNZ Group and CS Group to be a benefit to New Zealand of any significance. (PB 133)Accordingly, in calculating the social costs set out in [474] and [475], Professor Evans did not net off any of the benefits claimed for the transactions. He did, however, acknowledge: that the transactions revealed to the counterparties the ability and willingness of the BNZ Group to design and participate in bespoke transactions that utilised its tax capacity and thereby the New Zealand tax base. (PB 156)[477] This was Professor Evans' conclusion about the economic effect/costs of the transactions, and the relationship of those to the commerciality of the transactions:Each of the six BNZ Group transactions that I analysed utilised the BNZ Group's tax capacity and thereby the New Zealand tax base. They relied on reduction in the New Zealand tax base for their profitability. Taken as a whole the transactions were very profitable to the BNZ Group and imposed significant economic costs on New Zealand society. These costs are quite unusual for commercial transactions. (PB 159)ConclusionsDeductibility of the transaction expenses[478] Compliance with the applicable specific provisions is conceded by the Commissioner, except in the case of the GAF/GPF. I have found in favour of the BNZ on its primary argument, that the GAF/GPF was deductible pursuant to s DD 1(3). [479] The Bank's point that the Act no longer requires that deductibles be expended in earning assessable income is a point well made. The abandonment of the need for a nexus was based on sound reasons, namely the fungibility of money. [480] In my view the general control on deductibility is now s BG 1, because of "a legislative assumption that (deductibility rules) will only be invoked by those who engage in business activities for the purpose of making a profit". I take that from [126] of William Young P's judgment for the Court of Appeal in Accent Management Ltd v CIR (2007) 23 NZTC 21,323 (cited at [187]) The President added:Further, schemes which come within the letter of specific tax deductibility rules by means of contrivance or pretence are candidates for avoidance.[481] The GAF in the Gen Re 1 transaction, which effectively transferred the deductible expense of the GPF from BNZIS1 to BNZI, qualifying the former for the maximum FTC, is a striking example of structuring contrived to defeat the legislatively assumed symmetry referred to by the Court of Appeal (the FTC claimed by the BNZ in the Gen Re 1 transaction was limited to tax otherwise payable). [482] The Bank relied upon s DD 1, which does not limit interest deductibility for companies to generating assessable income, and to the limits on interest deductibility imposed by the thin capitalisation and conduit excess interest allocation (EIA) regimes. The BNZ was within the income deductibility limits contained in both those regimes.[483] Although those were powerful points, I consider they fell short of establishing that the fundamental asymmetry that drove the transactions in issue here was within legislative contemplation, let alone was expressly approved by the scheme and purpose of the Act.The foreign tax credit claimed in Gen Re 1[484] Gen Re 1 was a re-design of the proposed Gen Re 0 transaction, in the face of US IRS Notice 98-4. It is significant that the parties switched the transaction so that what the US IRS had signalled it intended regarding as an "abusive tax-motivated transaction with a purpose of generating foreign tax credits that can be used to shelter low-taxed foreign sourced income" was visited on New Zealand. [485] Gen Re 1 was structured so that: a) GRFT paid tax in the US on behalf of the BNZ, as owner of the Class A income unit in the Trust (it was this payment of tax that generated the FTC). b) There was a corresponding deduction elsewhere on the Gen Re side of the transaction, so that the net US tax consequence for Gen Re was that it paid tax only on its agreed share of the tax benefits i.e. 2.43% of NZ$500 million. [486] The BNZ maintains the US tax position is irrelevant. But is it? The essence of the FTC claimed by the BNZ was that tax on the distribution it received had been paid in the US. The evidence is that there was an off-setting deduction claimed, though by another Gen Re entity (GRCF). The result, viewing the whole arrangement in a commercially and economically realistic way as required by Ben Nevis, is that tax was not paid in the US, at least not in the amount of the FTC claimed by the BNZ or anything remotely approaching it. Gen Re 1 was not just structured to avoid double taxation in the US and New Zealand, it was structured to avoid tax being paid on the distribution in either jurisdiction.[487] I conclude that the FTC claimed by the BNZ in Gen Re 1 was both: a) Not allowed, by the terms of s LC 1(3A). b) Not within the scheme and purpose of the FTC regime.Conduit relief[488] I have a short, and a somewhat longer, conclusion about the conduit relief claimed by the BNZ in the last five transactions. [489] In [243] I expressed the view that the 'scheme and purpose' of the conduit relief regime was to allow a pass through New Zealand of foreign sourced income by a New Zealand subsidiary (the conduit) to its foreign owner, free of tax except for the 15% NRWT when the income was distributed to the foreign parent. [490] That has not happened here and it cannot happen because there is nothing for the BNZ to pass on. These transactions generated no income or gains to pass on. They generated only tax benefits for the BNZ which it cannot pass on to its parent the NAB. [491] The inevitable, and short, conclusion is that these transactions were not within the scheme and purpose of the conduit regime. [492] I reach the same conclusion by a slightly longer route. The aim of the conduit regime was to encourage multinationals to use New Zealand as an investment base by relieving, from New Zealand income tax, income earned overseas by the New Zealand subsidiary. [493] The conduit referred both to the investment flowing from the foreign multinational via its New Zealand subsidiary to the overseas subsidiary, and to the income returning along the same route.[494] I have accepted that there was no express requirement that the New Zealand subsidiary pass the income through to its foreign owner. But that was undoubtedly the Legislature's intention and expectation. It was the whole point of the regime, and it is indicated by the requirement for conduit tax relief accounts. [495] How do these transactions stack up against that legislative intention? First, in one way they did encourage the NAB to "invest" through its New Zealand subsidiary, the BNZ. The evidence is that the NAB switched the transaction from one of its UK subsidiaries (Clydesdale Bank) to the BNZ when the NAB's UK tax position became uncertain. But these transactions hardly involved real foreign "investment" by the NAB through the BNZ. The commercial and economic reality was that these were unsecured loans (of money sourced from the New Zealand money market) by the BNZ to the foreign counterparty (I refer to the counterparty parent). [496] Secondly, the NZ$500 million was in substance lent for a fixed term and was effectively on call after the first year. That is hardly foreign investment. Thirdly, the investment returns (the fixed distribution) came back to the BNZ in New Zealand but has gone no further, and cannot go any further. This is the point I made in [230] and again in [490]. Fourthly, rather than operating as a conduit for investment through New Zealand by a multinational, to the benefit both of the New Zealand economy and the New Zealand tax base, these transactions have operated at a substantial cost to both. The unchallenged evidence of Professor Evans was that the transactions had a total cost to New Zealand society of NZ$335.6 million, much of that from the deadweight cost of moneys transferred out of New Zealand, and permanently lost to the New Zealand economy. That result is the antithesis of the legislative intention in introducing the conduit regime. The consequence is that the answer to the ultimate question posed in Ben Nevis both at [5] and [109] is: "no, the transactions, viewed in a commercially and economically realistic way, do not use the conduit regime in a manner consistent with Parliament's purpose in enacting it". [497] The BNZ points out that Parliament has still not proscribed transactions such as these. The tighter thin capitalisation rules introduced in 2005 (the BNZ referred to these as the "remedial legislation"), only made the transactions more expensive, interms of the cost of capital. That submission is well made, so far as it goes. In a tax system that has a GAAR, there is limited force in submitting that what is alleged to be a tax avoidance arrangement has not been specifically proscribed, or that one or more of its elements has not been proscribed.The interest rate swap[498] There was commercial sense in the BNZ matching its cost of funds from the money market to its return from its "investment" in these transactions. But that could have been achieved by a floating distribution rate. [499] It was the formulaic pricing of the transaction which required fixed components, including a fixed distribution rate which was a function of the (fixed) GPF and fixed leg of the swap, that created the need for a swap. To argue that the swap was needed to match the cost of funds is to argue in a circle. [500] The evidence was that interest rate swaps are a common feature of structured finance transactions and the function of/reason for the interest rate swap in these transactions demonstrates why that is so. [501] As with the GPF, the parties had a common interest in setting the rate paid on the fixed leg of the swap as high as possible. There was a complete lack of commercial tension ensuring market pricing. That was compounded by two things. First, the fixing of the rates on the fixed leg of the swaps well in advance of the closing of the transactions, and not re-setting the rates to market upon closure. Secondly, using a one year swap rate instead of the applicable three or five year rates in the Gen Re 1 and CSFB transactions. [502] The result of that combination of factors was that the pricing of the interest rate swaps was significantly off market for some of the transactions, as demonstrated by the (ex post facto) analysis given in evidence.[503] The bottom line of all this is that the interest rate swaps were an integral part of transactions designed solely to generate tax benefits, and the swaps were priced to maximise those benefits.The GAFs/GPFs[504] I have held that the GAFs/GPFs were expenditure deductible under s DD 1(3) of the Act. [505] I have accepted that a parent guarantee was needed. But I have found that these transactions were always effectively with the counterparty parent i.e. with Gen Re Corporation or the CS Group. The SPV lent to/invested in was created specifically for the transaction, to give the required tax result. Given the deeply sub- LIBOR pricing, a guarantee was always going to be forthcoming from the counterparty parent, and at no fee. Otherwise, the parent was not going to get the irresistibly cheap funding ("this is as good as it gets") funding the BNZ provided to it. [506] Because of the formulaic relationship of the three pricing components, the pricing of the BPF was integrally wound up in generating the shared tax benefits. The BNZ accepts that the higher the GPF was, the greater those tax benefits were. Both parties knew that. Instead of commercial tension between one party trying to negotiate the GPF down as low as possible, and the other trying to get it up as high as it could, there was a common interest in the latter position. [507] Although I accept that a contrivance can still be correctly priced, it is difficult to separate the pricing of the GPF from the fact that it was a contrivance. [508] On Mr Mackay's credit enhancement market based approach to the pricing of the GPF, it was within market parameters. But the credit enhancement here was not being purchased in the market which Mr Mackay analysed. [509] I accept that the BNZ did take steps to ensure that the GPF was within market bounds. Two somewhat rudimentary inquiries were made at the time it was firstfixed at 2.95% p.a. As it turns out, if Mr Mackay's approach is accepted, that pricing was within market parameters, and well within them for the last transactions. [510] But I have concluded that the commercial reality, had these been arm's length negotiated deals, was that a parent guarantee would have been provided without any cost (the evidence of Messrs Choudhry and Stanton) or at 45-65bps maximum (Dr Fitzgerald's evidence). [511] I conclude that the GPF was a contrivance, substantially overpriced, to increase the tax benefits generated by the transaction.Commerciality[512] Putting aside the tax benefits they generated, these transactions had no commercial rationale, logic or purpose for the BNZ. They involved lending/investing at a substantial loss. As the BNZ accepted in closing (at 1.17(a)), that is a "classic indicator" of tax avoidance: Miller and Dandelion. The transactions involved the BNZ lending/investing at a substantial loss. The BNZ accepts that; it is an undeniable fact. [513] The BNZ counters that many business transactions are pre-tax negative. Even accepting that that is so, those are business transactions. Amongst the many examples of such transactions mentioned in evidence were finance leases. For example, shipping company A needs two new bulk carriers for its business. It cannot afford to purchase them outright for cash. It cannot operate them profitably if it borrows the purchase price from its banker B. Because of the tax benefits/advantages that finance leasing provided (the tax position has changed), A can lease the two ships from B profitably. So bank B purchases them and leases them to A. That profitability is a function of the splitting of the tax benefits of leasing, the lease payments being lower than the loan interest payments would have been. Leasing turns what would otherwise (if funded by a conventional loan) have been unprofitable business into profitable business.[514] Unless lending or funding at a substantial loss is termed banking business, there was no business to be done here. This was the basis for the Commissioner's submission that the business had followed the tax, rather than the tax the business. In short, the tax was the business, and that tax business comprised: a) Generating expenses deductible against other income of the BNZ; and b) Applying that deductible expenditure to generate tax exempt income. [515] A feature of these transactions was that the higher the costs (fixed swap rate and GPF), the higher the profits (tax benefits) generated by the transactions. In commercial terms, that result is antithetical. It points up the lack of business substance in these transactions. [516] I have found that the nine BNZ and 16 'other bank' transactions were template transactions. Although a standard form transaction can be entirely commercial (e.g. a finance lease such as that referred to in [513]), a template replicated for different businesses can indicate a tax avoidance purpose, as did the 'Russell' template in Miller. [517] Structural finance transactions are long and well established in banking finance. So are repos, including cross border repos which take advantage of tax arbitrage opportunities. In that sense, I accept that these were 'mainstream' rather than one off transactions of the Trinity scheme variety. That is certainly a point in favour of the BNZ's case. However, as Mr Galbraith confirmed of Mr Stanton in cross-examination, each transaction is different in its detail:Q. Its correct also, isn't it, that in the last 20 years or so – forgetting about these structures to some extent – there's been a growth in the variety of financing structures which have developed both in the public and private markets? A. Yes. Q. And each structure has its own features obviously? A. Yes. (T 3398-3399)[518] Thus, while some of the concepts or broad structuring of these transactions is familiar, the detail is not. Beyond the 22 transactions that are in issue between the Commissioner and the five New Zealand banks in the current raft of proceedings, I heard no evidence of any transaction comparable in its detail. [519] Finally, these transactions involved a substantial (NZ$335.6 million) cost to New Zealand society. As Professor Evans pointed out, that is "quite unusual for commercial transactions".Economic effect[520] I accept the evidence, in particular of Mr Choudhry and Professor Evans, that these transactions were, in economic substance, unsecured loans. Significantly, that is the tax treatment they received in the counterparties' jurisdictions, the US and UK. [521] The transactions did not lower the cost of capital for the BNZ, which could have been a worthwhile economic outcome and was generally the aim of structured finance transactions. That was because the BNZ was not a net borrower and the transactions had negligible effect on the BNZ's risk profile, apart from holding significant tax risk. [522] I have found that the only economic effect of the transactions was to transfer some liquidity to the counterparties, the effectiveness of that transfer considerably constrained by the fact that the funds were effectively on call after the first year of each transaction. [523] The transactions held no risks for either party, other than tax risk for the BNZ. [524] The transactions enhanced the value of the BNZ by $238.6 million, but imposed an economic cost on New Zealand society of $335.6 million, a significant part of that being the dead weight cost of the monies transferred off shore and permanently lost to the New Zealand economy.[525] In the counterparties' jurisdictions, the transactions were treated as a loan of NZ$500 million to the counterparty. Taking the CSFB transaction as an example, the net US tax consequence was that CSFB paid tax on 2.43% of NZ$500 million. In the UK the result was essentially the same; only the accounting profit earned by the UK counterparty was taxable.Does s BG 1 catch the transactions? (Issue 1)[526] Drawing together the conclusions I have set out in [478] to [525], I reach the conclusion that all six of the transactions in issue are caught by s BG 1. They are arrangements falling on the avoidance "side of the line" referred to at [112] in Ben Nevis. I have reached that conclusion for the following principal reasons, which I list in order of importance: a) The transactions had the purpose or effect of substantially altering the incidence of tax for the BNZ. This was not a "merely incidental" purpose or effect of the transactions. b) The transactions had no commercial purpose or rationale. Shorn of the tax benefits they were anticipated to generate, they involved the BNZ providing funds to the counterparties at a substantial loss. Their only purpose was to use the Bank's tax capacity to generate exempt income. c) Because no tax corresponding to the FTC claimed by the BNZ in the first (Gen Re 1) transaction was paid in the US, that FTC was not within the scheme and purpose of the FTC regime. Nor was the FTC within the terms of s LC 1(3A) of the Act. d) The later five transactions did not make use of the conduit relief regime in the manner contemplated by Parliament when it enacted that regime. That was because the BNZ, as the New Zealand subsidiary of its foreign parent (the NAB), did not, and cannot, pass on to the NAB the distributions it received in those transactions. Inshort, the BNZ was unable to act as the conduit contemplated by the conduit relief regime. e) The transactions generated the claimed deductible expenses in a contrived or artificial way:• The 2.95% p.a. GPF was a contrivance. A guarantee from the parent of the counterparty would have been forthcoming for no fee, or for a fee of no more than 0.65% p.a.• The BNZ contrived to set the fixed rate on the interest rate swap at the highest rate it thought defensible, and in the case of some of the transactions this was significantly outside market parameters. f) The returns on the transactions to the BNZ and counterparties alike were substantially in excess of what could have been expected from a risk free (except for tax risk for the BNZ) investment via a structured finance transaction, negotiated at arm's length.Is the way in which the Commissioner counteracted the tax advantage correct? (Issue 2)[527] Although I set out s GB 1(1) in [113], it is convenient to set the section out here again, this time in full:GB 1 AGREEMENTS PURPORTING TO ALTER INCIDENCE OF TAX TO BE VOID GB1(1) Adjustment of income Where an arrangement is void in accordance with section BG 1, the amounts of gross income, allowable deductions and available net losses included in calculating the taxable income of any person affected by that arrangement may be adjusted by the Commissioner in the manner the Commissioner thinks appropriate, so as to counteract any tax advantage obtained by that person from or under that arrangement, and, without limiting the generality of this subsection, the Commissioner may have regard to – (a) Such amounts of gross income, allowable deductions and available net losses as, in the Commissioner's opinion, thatperson would have, or might be expected to have, or would in all likelihood have, had if that arrangement had not been made or entered into; or (b) Such amounts of gross income and allowable deductions as, in the Commissioner's opinion, that person would have had if that person had been allowed the benefit of all amounts of gross income, or of such part of the gross income as the Commissioner considers proper, derived by any other person or persons as a result of that arrangement.[528] The approaches to s GB of the Bank and the Commissioner differed sharply. The Commissioner's reconstruction involved his denying the BNZ: a) Its funding costs. b) Its net swap costs i.e. the fixed payment it made, less the floating payment it received, under the interest rate swap. c) The GARs or GPFs. d) The risk participation fees. [529] As the BNZ and Commissioner both equated a) to the floating rate the BNZ received under the interest rate swap, a) + b) = the fixed rate the BNZ paid under the interest rate swap. [530] The Commissioner contended this reconstruction removed "any tax advantage obtained by (the BNZ) from or under (the arrangements)". That tax advantage was the difference between the pre-tax results of the arrangements and the post-tax results:Pre-tax result + tax advantage = post-tax resultConsequently, the Commissioner submitted his reconstruction exactly removed the tax advantage to the BNZ from and under the arrangements. [531] Although it seems a make weight point, the Commissioner pointed out that the "tax advantage" he had disallowed was less than what the BNZ had regarded asits tax benefit from these transactions. That "tax benefit" or "tax gross-up" refer to the difference between the BNZ's pre-tax and its "pre-tax equivalent" position. The latter is the post-tax result divided by 0.67. Thus 1/.67 = 1.5 (1.4925 to be precise). [532] The BNZ contended the Commissioner's approach was wrong in two main respects. First, the BNZ's funding costs were not part of the arrangements. They were part of the Bank's ordinary fund raising activities, unrelated to the terms of the transactions. On that basis, the deductibility of those funding costs was not a tax advantage obtained "from or under" any tax avoidance arrangement. [533] Secondly, if the pricing of the GAR/GPFs and swaps was outside a reasonable range of values, the prices should only be disallowed to the extent that they were outside market parameters. If I found that the guarantee fees were not an expense incurred as a matter of "commercial reality", the BNZ conceded they should be disallowed, but that should not affect the deductibility of the Bank's funding costs. [534] The BNZ also argued that pricing issues should not be addressed under s BG 1, other than in exceptional cases. If the Commissioner was concerned about the pricing of the guarantee fees and swaps, then he should have invoked the targeted provision, s GD 13. I confess I did not follow this argument, certainly not when included at this last stage of the case. I think the Commissioner was justified in responding that it confused the test or consideration required under s BG 1, with that required under s GB 1. [535] The Bank's reliance was on McGechan J's judgment in this Court in BNZ Investments Ltd v CIR (2000) 19 NZTC 15,733:[200] I have no doubt (s GB 1) is intended to counteract tax advantages obtained out of avoidance, but not otherwise. Where tax advantages are increased through avoidance over a base level which would have existed in any event, it is that increment above base level which is to be counteracted, not the legitimate base level itself. That is all preservation of the tax base – the purpose of the section – requires.And the Privy Council in Petersen at [35]:The critical question is whether the tax advantage which they obtained amounted to "tax avoidance" capable of being counteracted by (s BG 1), for the Courts of New Zealand have long recognised that not every tax advantage comes within the scope of the section; only those which constitute tax avoidance as properly understood do so.[536] It sought to distinguish cases such as Miller and Dandelion, which both lacked any element of commercial dealing. Similarly, it distinguished Ben Nevis, although pointing out that the Commissioner there had not challenged the deductibility of the costs of planting and tending the Douglas Fir trees, those being expenses which the investors in the Trinity scheme had actually paid. [537] The BNZ's submissions were particularly forceful in respect of its funding costs. It urged:The only characteristic that could distinguish these costs from the rest of BNZ's aggregate interest expenditure and hedging costs, is that in this case the funds borrowed resulted in the derivation of tax-relieved income.The Bank reiterated that legislative policy not only allowed, but positively affirmed, the deductibility of interest incurred by companies in deriving exempt dividends. In short, it reiterated its conduit relief arguments. If any funding costs were to be denied, the Bank contended it should be only 34%, since the 2005 "remedial legislation" (assuming it applies) expressly sanctioned conduit relieved income financed up to 66% by borrowings. [538] In my view the Commissioner's counteraction under s GB 1 is correct, and the BNZ's contended alternative approach incorrect. [539] I have held that all the deductions disallowed by the Commissioner were integral parts of the tax avoidance arrangements correctly treated by the Commissioner as void under s BG 1. It is thus the whole of the arrangements and each of the parts comprising them that are void. The position is as stated by the Court of Appeal in Accent Management:[155] The effect of s BB 9 and GB 1 is that the scheme is void as against the Commissioner. Under that void scheme, the taxpayers claimed deductions to which they were not entitled. The entirety of the deductions was thus illegitimate and their extent provides the measure of the tax advantages which the Commissioner must counteract. The counterfactualenvisaged by the s GB 1(a) is the position "if that arrangement had not been made or entered into". There is thus no need for the Commissioner (or court) to conjure up an alternative and more effective scheme into which the taxpayers might have entered. (my emphasis)[540] I do not accept the Bank's specific submission in relation to funding costs, which I have outlined in [537]. I consider another distinguishing characteristic is that the funds borrowed, the costs of which the Bank deducted, were deployed by the Bank in tax avoidance arrangements. [541] A last point about the application here of s GB 1. In seeking to distinguish the transactions here from, for example, the Trinity scheme in Ben Nevis, Mr Galbraith stressed what he termed the two fundamental obligations the BNZ had undertaken. First, a NZ$500 million exposure to the New Zealand money market, second a similar exposure to the counterparty it had funded in each transaction. Mr Galbraith emphasised that those obligations were intended to be performed, and had been performed. I reiterate that I accept that. The Commissioner's counteraction under s GB 1 does not affect those two fundamental obligations.Result[542] The BNZ's challenges to the Commissioner's assessments fail. Specifically, I decline: a) To make the order sought in paragraph (a) of the plaintiffs' prayer for relief in their consolidated statement of claim dated 5 February 2009, cancelling all the assessments there detailed. b) To make the order sought by the plaintiffs in paragraph (b) of the prayer for relief in the plaintiffs' consolidated statement of claim dated 5 February 2009, varying or reducing the disputed assessments. [543] Costs are reserved, for submission failing agreement.Solicitors: Russell McVeagh, Wellington for Plaintiffs Crown Law Office, Wellington for Defendant