Tax Court of Canada reduces Discount Rate between Related Parties by Half
The Tax Court issued in late December an exhausting 105-page transfer pricing decision, reducing by half the discount at which McKesson Canada sold its receivables to its Luxembourg parent starting in 2003. McKesson Canada Corporation v. The Queen, Dec. 13 2013, 2008-2949(IT)G. The complex receivables transaction was designed solely in an attempt to reduce the taxes on the Canadian unit of this pharmaceutical supplier. The excess discount was treated as a taxable shareholder benefit subject to withholding tax, for which McKesson Canada was made liable outside of the treaty limitations period for transfer pricing assessments. The poor taxpayer also has to pay the Crown’s appeal costs.
It isn’t until the postscript in the 82d footnote that the true meaning of the case comes to light, in the honorable Court’s apology: “I can do no better than quote from a 2013 address by Lord Neuberger of Abbotsbury, President of the UK Supreme Court (entitled ironically Justice in an Age of Austerity): “We seem to feel the need to deal with every aspect of every point that is argued, and that makes the judgment often difficult and unrewarding to follow. Reading some judgments one rather loses the will to live – and that is particularly disconcerting when it’s your own judgment that you are reading.””
McKesson U.S. is the 15th ranking largest public company in the Fortune 20 list of companies, with annual revenues in excess of US$100 billion. In Canada, the core business of the McKesson group is the wholesale distribution of pharmaceutical medicine products. The McKesson Group delivers one-third of all medicare to the public in the US. McKesson Canada had about one-third of the Canadian market. In 2002, McKesson Canada had sales of $3 billion, profits of $40 million, 2,400 employees and the largest share of the Canadian market.
In 2003, McKesson Canada entered into an agreement to sell its current receivables at a discount to its Luxembourg immediate parent company (“MIH”). MIH purchased all eligible receivables as at that date (about $460,000,000), and committed to purchase eligible receivables daily as they arose for the next five years, subject to a $900,000,000 cap.
The discount rate selected greatly exceeded an arm’s length rate, resulting in a large shortfall in what would have been taxable income. Under absurd prevailing Canadian tax tendencies, this entirely-tax-motivated transaction is in itself valid, unless affected by specific anti-avoidance provisions in the Income Tax Act, such as the requirement in section 249 that related parties deal on arm’s length terms.
A secondary issue was McKesson Canada’s liability for failure to withhold Part XIII non-resident tax from the disallowed amounts paid to its non-resident parent. Assessment of the Part XIII liability would have been outside of time limitations, if it had been treated as a transfer pricing adjustment, rather than a secondary effect of an adjustment.
In 2003, 99.96% of McKesson Canada’s receivables proved to be good and collected when managed by McKesson Canada’s credit department. McKesson Canada had no identified business need for a cash infusion or borrowing, nor did McKesson Group need McKesson Canada to raise funds for another member of the group.
The McKesson Group had amassed large amounts of cash in Ireland. MIH used this money to finance the purchases of McKesson Canada’s receivables under the facility. MIH had the right to put non-performing receivables back to McKesson Canada for 75% of the face amount, readjusted to the amount actually collected by McKesson Canada.
MIH borrowed all the money needed from the Irish entity, directly or indirectly. MIH’s obligations were guaranteed by another non-Canadian McKesson Group entity. MIH paid McKesson Canada to continue to have its credit and collections department manage the receivables.
Most of the proceeds of the initial $460,000,000 receivables sale were returned by McKesson Canada to MIH. A portion was loaned to another Canadian corporation to permit its tax losses to be used.
The Revenue Agency challenged these related party transactions on the basis that the amounts paid to the non-Canadian McKesson entity differed from those that would have been paid between arm’s length persons transacting on arm’s length terms and conditions. The discount upon the purchase of the receivables in accordance with this revolting revolving facility was a 2.206% discount from the face amount. This discount rate for receivables expected to be paid within 30 days could be restated as an annual financing cost of 27%.
As a direct result of these pernicious discounts, McKesson Canada ceased to be profitable for its 2003 taxation year and reported a tax loss. McKesson Canada’s profits in later years were similarly significantly reduced.
The predominant purpose of McKesson Canada entering into the transactions was the reduction of its Canadian tax on its profits. None of the raising or freeing up of capital, reducing credit risk from its customers, nor improving its balance sheet was the company’s predominant purpose; they were results of the transactions. There were “belts and braces” inserted to ensure that the receivables facility did not leave the purchasing entity holding any financial or credit risk associated with McKesson Canada or its receivables.
The Court noted sarcastically that, in the weeks before the signing of the receivables agreements, Toronto Dominion Securities Inc. (“TDSI”) was retained by taxpayer’s tax law firm Blake Cassels, to provide advice on certain so-called arm’s length aspects of the receivables agreements. By the time that TDSI was consulted, the structure and pricing approach and formulae were largely settled. It is not clear that any significant changes were made to reflect any advice or information given by TDSI.
The Income Tax Act provides for contemporaneous arm’s length transfer pricing documentation/analysis requirements to defend the Act’s transfer pricing penalty provisions. The Court noted that last minute, rushed, not fully informed, paid advocacy that was not made available to the Canadian taxpayer and not read by its parent, could not easily satisfy the contemporaneous documentation requirements.
McKesson Canada had a pre-existing business relationship with the TD Bank group. Several years earlier, McKesson Canada had done a receivables factoring transaction with TD Factors to avoid Canadian federal capital tax. The Court found it entirely unreasonable to suggest that this was a comparable transaction for arm’s length pricing purposes to the one in issue.
TDSI’s bizarre approach in the current agreement was to use instead a 16-day discount period for the initial purchase, averaging the “missing” 16 days across the entire five years of the RSA term. The bank set a discount period of 31.73 days for all discount rate calculations for receivables purchases under the RSA. The effect of these strange and unnatural practices was double this portion of the discount rate for the December 2002 purchase, thereby providing a significant timing benefit to the McKesson Group in respect of McKesson Canada’s tax reduction for the 2003 year.
TDSI made numerous other design decisions in its structured monstrosity, with which the Court took issue. TDSI identified a concentration risk issue associated with the larger obligors, that losses on their receivables had an increased likelihood of deviating from historical levels. TDSI did not attempt to quantify that increased likelihood or validate the increased risk.
TDSI noted that the multi-year data it has been provided did not cover a full economic cycle of Canadian trough-to-peak-to-trough. The report does not say that more historic data was asked for, nor that it was unavailable. Thus, three standard deviations from mean that TDSI suggested adding, seemed unreasonable. TDSI’s bright suggestions, combined with the selection of monthly numbers, had the result of driving the .03% actual monthly average loss to 0.24% – an eight-fold increase.
TDSI tied the “very possible” need for a replacement servicer, and the associated risk factoring, to the fact that McKesson Canada was not a highly-rated credit on a stand-alone basis. The connection was not obvious. This component of the discount rate was also unreasonable.
It was not self-evident why prompt payment discounts were not treated in the same manner as other dilutions such as volume rebates to customers. The prompt payment discount risk was appallingly estimated at the outset, fixed throughout at historic levels, plus an unexplained buffer, instead of being reflected in its rolling average actual performance throughout the agreement’s term. The TDSI Report overlooked other obvious related risks. No explanation was offered for accepting a fixed approach.
The determination of interest discount was flawed. Because MIH was exceptionally thinly-capitalized, TDSI found it appropriate to use non-investment grade bond market credit spreads as a proxy. This was correct from a profitable business transaction point of view, but TDSI failed to address the possibility that the transaction exactly as structured could not be done profitably on arm’s lengths terms and conditions, and failed to address other basic considerations. TDSI was unaware (hello?) that MIH enjoyed a full guarantee and indemnity from its parent company. This lack of balance in favour of MIH was not identified or discussed.
TDSI’s report on the grossly excessive servicing fees admitted that it did “not have ready knowledge of comparable situations”.
Under section 247 of the Income Tax Act, where a taxpayer and a non-resident person with whom the taxpayer does not deal at arm’s length are participants in a series of transactions, and either
(a) the terms imposed between the participants differ from those that would have been made between persons dealing at arm’s length, or
(b) the series would not have been entered into between persons dealing at arm’s length, and can reasonably be considered not to have been entered into primarily for bona fide purposes other than to obtain a tax benefit,
any tax-relevant amounts are adjusted to the quantum or nature that would have been determined, if the terms imposed had been those that would have been made between persons dealing at arm’s length, or, where paragraph (b) applies, the series of transactions entered into had been the series that would have been entered into between persons dealing at arm’s length.
The Tax Court noted that there may be a point at which the extent of changes to the agreed non-arm’s length terms and conditions can constitute an effective recharacterization of the transaction only permitted under paragraph (b) above. It was unnecessary to resolve this issue in disposing of the appeal.
The Court emphasized that, in disposing of a transfer pricing appeal, it should consider “notional continued control type rights” when looking at term or executory contract rights. This refers to implicit unwritten, unenforceable guarantees of the parent company in the group.
The appropriate transfer pricing analysis methodology would not be any of the four methodologies named in the OECD Guidelines and, therefore, an “other method” should be used.
An “other method” that would be appropriate was that followed by TD Bank, taking the receivables transaction as the parties structured it, and not introducing any recharacterization or additions, but accurately identifying the risks transferred to MIH, unlike what TD naughtily actually did. Another expert’s report did not follow the parties’ chosen structure, but rather relied on a comparability and adjustment approach which began by looking at the public bond market discount rates on a five-year non-investment grade Canadian bond fund index and then making a number of significant adjustments to reflect issues peculiar to the receivables transaction. That methodology was inappropriate.
While some experts suggested that arm’s length parties did non-recourse receivables purchase transactions on comparable terms, absent any supporting evidence, the Court was unable to give any weight or relevant consideration to the suggestion.
The Court noted in despair that one of the largest and most significant benefits to McKesson Canada was the transfer of the risk of loss on the receivables. There is credit default insurance available in the market from arm’s length commercial players in the financial markets, both direct insurance and synthetic or derivative structured products. The Court lamented that it was given no evidence of the practical availability (or non-availability) or effectiveness, or pricing/costs for such. Such evidence might have provided the poor Court with other helpful information with which to try to price the risk transference, beyond McKesson Canada’s available historic loss and performance and projections for its receivables portfolio, and risk spreads in the public bond market.
The adjustment of a formula or criterion (e.g., yearly versus monthly) may not clearly be permitted under the language of ITA section 247. The Court held that the OECD Guidelines’ commentary on “realistically available options/alternatives” would support a broader reading of which amounts’ quantum can be adjusted to reflect arm’s length terms and conditions. This bold approach contradicts some of the Court’s earlier comments giving priority to the Act over the Guidelines and other international soft law, but is otherwise a reassuring tendency.
In her testimony, the TDSI officer identified the material transaction risks as (i) dilutions, including prompt payment discounts (ii) losses – primarily credit losses and (iii) servicing risk, i.e., that the servicer collects but does not remit to the buyer, however unthinkable on the present facts.
While the TDSI Report said that it looked at pricing of comparable risks in the market, the expert acknowledged glumly that TDSI did no more than obtain credit spread numbers from their bond traders. The expert had difficulty, even in direct examination, in giving a satisfactory, responsive, logical or complete explanation of the interest discount portion of the discount spread. She couldn’t even provide a very thorough or satisfactory answer as to why the TDSI Report, when assessing the accrued rebates dilution risk, assumed that 100% of McKesson Canada’s customers would set off their full accrued rebates. She could not explain how this was reasonable.
Another taxpayer expert report (PwC) had insufficient grounds for notching all of the others obligors’ credit risk down two grades below the public debt rating that PwC had assigned to taxpayer’s largest customers. This picking and choosing, mix and match as it suits approach, to the relevance of the actual performance of the receivables pool made for “transparently poor advocacy, and even more questionable valuation opinions.”
A credit rating downgrade of McKesson U.S. would not necessarily involve any change to McKesson Canada’s ability to service its receivables. The related probability numbers were inadequately explained, questionable and, to the Court’s mind, unreliable advocacy or posturing.
One of the taxpayer experts inexplicably never considered the cost of insuring the receivables in his pricing approach, nor to test the results of his approach. That hapless expert came across as a partisan advocate quick to point out the specks in the poor Crown’s expert reports, and downplaying, if not refusing to acknowledge, the weak points in his own.
The beleaguered Court was troubled by the introduction of a general reserve as enhanced credit protection to MIH on purchased receivables which MIH does not enjoy directly, or even indirectly, in the receivables sale. There may be significant reserves in securitizations, in secured loans, and in other asset backed loans, but this agreement gave no such security or protection to MIH once it purchased any particular receivables. The Court thus did not rely upon the results of two related expert reports. The Court emphasized that this was not a securitization transaction, but its definition is far too vague to be useful: “an off-balance sheet debt financing, via a thinly capitalized special purpose entity, that accesses low rate investment grade financing via a structured finance product that incorporates risk minimization features including support from the seller of the existing cash flow stream.”
There was no McKesson Canada covenant or assets supporting the collectibility of any receivable after it was transferred to MIH. This was viewed as fundamentally different from either a securitization or a secured loan. No mention was made at that point of the valuable putback of non-performing debt enjoyed by MIH for 75% of the face amount.
The Court dismissed comparables based on the maximum receivables pool of $900,000,000, since exposure was never expected to reach that amount.
The 30-day Canadian dollar bankers’ acceptance rate, or CDOR, as of December 16, 2002 was 2.79% per annum. The Court did not accept that using a fixed “days sales outstanding” (DSO) of 31.73 days throughout the term without regard to changes to actual DSO, was an arm’s length term. Arm’s length parties would incorporate a floating approach to DSO averaged over some period, say three to four months, and would not accept the risk of fixing the DSO for the entire term.
The Court also did not accept the special loss discount computed for taxpayer’s largest clients, and which was based on rated companies’ bond ratings, with various offensive or illogical assumptions. Under the receivables agreement, the loss discount for more-than-2% customers could very suspiciously be recalculated at any time, if the buyer MIH thought that the mix of large customers to other customers had changed, but not if McKesson Canada thought that the mix had changed in its favour, for lands’ sakes.
The Court disagreed that the loss discount attributable to the smaller customers would be fixed by arm’s length parties in the manner it was, or fixed at that same number for a five-year term. The loss discount was almost 600% of what would be expected from taxpayer’s historic receivables write-offs, without any significant projected, planned or reasonably anticipated material risk of deterioration of its business, its customers, or the economy generally. The loss discount did not even directly take into account the historical or actual ongoing performance of the receivables pool.
The discount spread in the receivable agreement was calculated as the sum of four components: a servicing discount, a prompt payment dilutions discount, an accrued rebate dilutions discount, and an interest discount.
The appalling issue with the prompt payment dilutions was whether the risk had been accounted for in the discount spread on arm’s length terms. The Court held that arm’s length parties would remove the risk of change during the term of the agreement in the levels of prompt payment discounts, by adopting a floating dynamic prompt payment component.
The Court did not accept that the accrued rebate dilutions risk warranted any material discount. The taxpayer would not agree with a notional arm’s length purchaser to a discount that either assumed that all of the taxpayer’s customers would exercise a claim to set off their accrued rebate entitlements, or that assigned a credit risk spread to the taxpayer equivalent to junk bond issuer status.
The interest discount was intended to provide MIH with a return from the discounted purchase of receivables, in addition to all of the above amounts, equal to an assumed cost of funds (that it did not in fact bear) equal to the cost to a below-investment grade borrower that borrowed 100% of the receivables purchase price by issuing its junk bonds in the market. The Court found this element completely unacceptable, unreasonable, unsupported on the evidence, and a term that would not be agreed to by with an arm’s length party.
Overall, the horrified Court could only say that it had never “seen so much time and effort by an appellant to put forward such an untenable position so strongly and seriously. This had all the appearances of alchemy in reverse.”
The total of these adjustments resulted in a discount range of 0.959% to 1.17%, compared to the 2.206% agreed to by the parties. This adjusted rate was comparable to the 1.013% reassessed by the Minister of National Revenue.
It was not necessary to fix a particular point within the determined range, as the taxpayer’s evidence did not rise to the level of a prima facie case that “demolishes” the key assumptions of fact made by the Minister that support the reassessments.
The Court felt obliged to try and translate the bizarre new direction shown by the Supreme Court in recent years, protecting the corporate tax shelter industry: “The Supreme Court of Canada reminds us regularly that the Duke of Westminster is alive and well and living in Canada.”, a reference to the outdated, foreign IRC v Duke of Westminster, (1936) AC 1. The Court noted that there was certainly nothing wrong with taxpayers doing tax-oriented transactions, tax planning, and making decisions based entirely upon tax consequences, subject only to the general anti-avoidance rule in section 245 of the Act. GAAR was not inexplicably relevant to this appeal, presumably because the Crown thought it safer to rely on more narrowly-worded anti-avoidance provisions. The Crown could have done both, but is presumably subservient to the errant recent Supreme Court doctrine.
The Part XIII withholding tax issue was whether the Revenue could charge McKesson Canada for failure to withhold on the deemed benefit transfer to parent. The Revenue assessed McKesson Canada just outside of the 5-year limitation for transfer pricing adjustments under the Canada-Luxembourg treaty. The Court held that the derivative liability of McKesson Canada was not protected by the treaty limitation clause.
The Court viewed a benefit as paid by McKesson Canada to its parent MIH via the transfer of receivables at an overstated discount rate, which resulted in giving away some of its assets to its parent corporation. All shareholders are broadly taxable under a general provision in section 15 of the Act. The amount of this benefit is deemed by the Act to be a dividend paid by McKesson Canada to MIH, subject to 5% withholding under the Canada-Luxembourg Treaty.
In in its 2003 short year the Canadian tax avoided by McKesson Group was US$4,500,000, and some form of Luxembourg tax was expected to be payable in the amount of US$29,000.
Article 9 of the Canada-Luxembourg Treaty provides that, any income which would, but for (bizarre and unnatural) conditions existing between associated enterprises that would not have existed between the independent, have accrued to one of the enterprises, but, by reason of those conditions, has not so accrued, may be included in the income of that enterprise and taxed accordingly. A Contracting State may not change the income of an enterprise in these circumstances, after the expiry of the time limits in national law and, in any case, after five years from the end of the year in which the income which would be subject to such change would, but for the (bizarre and unnatural) conditions above, have accrued to that enterprise.
The Revenue properly asserted that the standalone obligation of McKesson Canada under the Income Tax Act as a Canadian payor who fails to remit is distinct for Article 9 purposes from a change in the income of MIH for tax purposes.
Both the Vienna Convention and the Supreme Court of Canada in The Queen v. Crown Forest Industries Limited et al., 95 DTC 5389 (SCC), confirm that “literalism has no role to play in the interpretation of treaties”, Coblentz v. The Queen, 96 DTC 6531 (FCA, 1996). In Crown Forest, the Supreme Court also held that, in ascertaining the purposes of a treaty article, a court may refer to extrinsic materials which form part of the legal context, including model conventions and official commentaries thereon, without the need to first find an ambiguity before turning to such materials.
The Tax Court in McKesson held that an ITA subsection 215(6) vicarious assessment of a Canadian payor for failure to remit and withhold tax is not a change by Canada of MIH’s income, and does not constitute Canada seeking to add a transfer pricing adjustment amount to MIH’s income and to tax that increased amount. The Court was understandably more inclined to see it as an enforcement and collection provision than a tax charging provision.
The Court also found that the requirements are more clearly not met because the only transfer pricing adjustment in Article 9(1) is income which, but for the related party conditions, would have accrued to MIH under the receivables transactions. While the amount of MIH’s taxable benefit and deemed dividend may be the same as this transfer pricing adjustment, it is not an amount of income that, had the discount rate been an arm’s length discount rate, would have accrued to MIH. On the contrary, the transfer pricing adjustment is income that but for the non-arm’s length terms and conditions would have accrued to McKesson Canada. This reasoning is superficially correct but not in keeping with the broad interpretative approach mandated by the OECD Guidelines. The honorable Court should not have been so thrilled to have happened upon the argument.
The Court noted that this same problem arose equally clearly in respect of the requirement that a five-year limitation period can only begin to run from the end of the year in which the income of MIH sought to be changed would, but for the non-arm’s length discount rate used, have accrued to MIH. Again, had an arm’s length discount rate been used in the receivables agreement, the additional income would have accrued to McKesson Canada, not MIH.