Citation: 2013 TCC 404
Date: 20131213
Dockets: 2008-2949(IT)G
2008-3471(IT)G
BETWEEN:
MCKESSON CANADA CORPORATION,
Appellant,
and
HER MAJESTY THE QUEEN,
Respondent.
REASONS FOR JUDGMENT
TABLE OF CONTENTS
1. Overview
2. The Financing
Transactions
a) The RSA Discount
Formula
(i) The
Yield Rate
(ii) The
Loss Discount
(iii) The
Discount Spread
3. The
TDSI Opinions on Arm’s Length Terms & Conditions and Pricing
a) Eligibility Criteria
b) Termination Events
(i) Delinquency
Ratio Trigger
(ii) Loss
Ratio Trigger
c) Discount Rate
(i) The Yield
Rate
(ii) The Loss
Discount
1. Designated Obligors
2. Other
Obligors
(iii) The
Discount Spread
1. Servicing
Discount
2. Prompt Payment Dilutions Discount
3. Accrued
Rebate Dilutions Discount
4. Interest
Discount
4. TDSI’s
Supplemental Report on Servicing Fees
5. The Law
(a) GlaxoSmithKline
(b) Reasonableness
(c) Relevant Series
of Transactions
(d) Scope of
Adjustments Permitted Under Paragraph 247(2)(c)
(e) Factors that Exist
Only because of the Non-Arm’s Length Relationship
(f) The Rule in Browne
v. Dunn and Opinions Within a Witness’ Expertise
(g) The Court’s
Analytical Approach to be Followed in this Case
6. The Position
of the Appellant
7. The Position
of the Respondent
8. The Witnesses,
the Expert Reports and the PwC Report
(a) Mr. Brennan
(b) Ms. Hooper and the TDSI Reports
(c) The PricewaterhouseCoopers Report
(d) The Frisch Expert Report
(e) The Reifsnyder Expert Report
(i) Factors One and Two –
Servicing Fees and Prompt Payment Discounts
(ii) Factor Three – Credit Risk
(iii) Factor Four - $900MM Commitment to Finance
(f) The Becker Expert Report
(g) The Finard Expert Report
(h) The Glucksman Expert Report
(i) The Other Expert
Reports
9. The Appropriate Methodology
10. Analysis of
Transfer Pricing Issue
a) The Discount Rate
(i) The Yield
Rate
(ii) The Loss
Discount
(iii) The Discount
Spread
1. Servicing
Discount
2. Prompt Payment Dilutions Discount
3. Accrued
Rebate Dilutions Discount
4. Interest
Discount
b) Summary of Court’s
Estimate of Discount Ranges
11. Conclusion on Transfer Pricing Adjustment
12. Timeliness of
Part XIII Assessment of McKesson Canada
a) The Issue
b) The Provisions of the Income
Tax Act and the Treaty
c) Position of the
Parties
d) The Interpretation of
Treaties
e) Analysis
f) Conclusion re: Part XIII
and the Treaty
13. Dismissal of Appeals
Boyle J.
[1]
The principal appeal by
McKesson Canada Corporation (“McKesson Canada”) concerns the amount of the
transfer pricing adjustment made by the Canada Revenue Agency (“CRA”) to its income
under paragraphs 247(2)(a) and (c) of the Income Tax Act (the “Act”)
in respect of financial transactions involving McKesson Canada and several of
its non-arm’s length and related non-Canadian affiliates during its 2003
taxation year. The related appeal involves the secondary issue of McKesson Canada’s liability under the Act for its failure to withhold and remit to CRA an
amount equal to the Part XIII non-resident withholding tax from the disallowed
amounts paid by it to its non-resident parent.
[2]
As described in greater
detail below, in 2002 it was decided by the McKesson Group that McKesson Canada
would sell the receivables owing to it from its customers to a related non-resident
McKesson Group entity at a discount. A facility was put in place pursuant to
which the receivables would be transferred by McKesson Canada daily at a discount from the face amount of each transferred receivable.
1. Overview
[3]
McKesson Canada is the principal Canadian operating company in the McKesson group of companies owned by the U.S. multinational McKesson Corporation (“McKesson U.S.”). McKesson U.S. is a United States public
company and is the 15th ranking largest public company in the
Fortune 20 list of companies. Its annual revenues are an excess of US$100
billion. It is the largest U.S. health care company. It has over 32,000
employees worldwide. It has been said that McKesson U.S. is the biggest company
no one has ever heard of.
[4]
Worldwide, and in
Canada, the core business of the McKesson group of companies (“McKesson Group”)
is the wholesale distribution of over-the-counter and prescription
pharmaceutical medicine products. This accounts for about 97% of its revenues.
Its other related business is that of hospital software technology.
[5]
The McKesson Group’s
wholesale pharmaceutical business has an impressive market share. The McKesson
Group delivers one-third of all medicare to the public in the US. In the years in question, McKesson Canada had about one-third of the Canadian market. It
distributes the products of a large range of pharmaceutical companies, and
sells to drug store chains large and small, to large grocery store and
department store chains that have pharmacies and/or sell over-the-counter
medicinal products, to independent pharmacies, to hospitals, and to long-term
care institutions.
[6]
In the year the
receivables facility was put in place, 2002, McKesson Canada had sales of $3 billion, profits of $40 million, 2,400 employees and the largest
share of the Canadian market. Its Canadian customers included a number of Canada’s largest retailer grocers and drug store chains. It had credit facilities available
to it in the hundreds of millions of dollar range with major financial
institutions. Its public ultimate parent, McKesson U.S., had a solid investment
grade credit rating and the interest rates on the available lines reflected
that.
The McKesson Group had a very considerable cash surplus built up in its Irish
affiliate.
[7]
At that time, and in
the years leading up to it, McKesson Canada had its own successful and sizeable
credit department which managed its credit and collection policies and practices.
Credit and collections results were trending favourably. McKesson Canada’s receivables were managed with considerable success, having a roughly 30 day
payment average,
and a 0.043% bad debt experience
with its customers overall. That is, 99.96% of its receivables proved to be
good and collected when managed by McKesson Canada’s credit department applying
McKesson Canada’s credit and receivables collection policies. This was very
important to McKesson Canada’s success given that the wholesale drug business
was low-margin – in the range of 2% - on high volumes.
[8]
There was no evidence
that there was any pending imminent or future change expected, anticipated or
planned for in the make-up, nature or quality of McKesson Canada’s receivables
or customers, although there was always the future risk of unforeseen adverse
change.
[9]
At that time, 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. There was a
so-called double-dip Nova Scotia Unlimited Liability Company or ULC financing
which was coming to maturity and would need to be recapitalized in some
fashion; this was for a fraction of the amount of the new receivables facility.
McKesson Canada did not approach its traditional lenders or conventional
financial institutions (nor anyone else) before entering into its own non-arm’s
length receivables facility and related transactions. The McKesson Group had
previously put in place a tax-effective international corporate structure and
inter-group transactions that allowed it to amass very large amounts of cash in
Ireland. The non-Canadian members of the McKesson Group were able to use this
money to finance all of the purchases of McKesson Canada’s receivables under
the facility.
[10]
The non-Canadian
McKesson Group company that purchased the receivables had the right to put
non-performing receivables back to McKesson Canada for a price equal to 75% of
the face amount, later readjusted to the amount actually collected on it by
McKesson Canada. The purchaser did not otherwise have recourse to McKesson Canada for unpaid purchased receivables.
[11]
The non-Canadian
McKesson Group entity that purchased the receivables borrowed all the money
needed from another non-Canadian McKesson Group entity. The borrower’s
obligations to the lender under the loan was fully guaranteed by yet another
non-Canadian McKesson Group entity, which also indemnified the borrower for any
shortfall between what the borrower received from McKesson Canada’s receivables
and what it needed to pay on the loan.
[12]
As described below, the
non-resident affiliate also paid McKesson Canada to continue to have McKesson Canada’s credit and collections department manage the receivables applying McKesson Canada’s credit and receivables collection policies and practices. Under the agreements
these policies and practices could not be changed without consent. Similarly,
McKesson Canada could only continue to grant other discounts or rebates in the
ordinary cause of its business and in accordance with its usual practices when
the facility was entered into.
[13]
Most of the proceeds of
the initial $460,000,000 receivables sale were returned by McKesson Canada to its non-resident shareholder affiliate, a portion was loaned for a period to another
Canadian corporation to permit its tax losses to be used, and about 1% of the
proceeds were used by McKesson Canada for its general corporate purposes.
[14]
The CRA has challenged
these related party transactions for McKesson Canada’s 2003 taxation year on
the basis that the amounts paid to the non-Canadian McKesson entity pursuant to
the receivables purchase transactions differ 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 the revolving facility was a 2.206% discount from the face amount. While this
discount rate and the overall transactions between the parties are considered
in greater detail below, this discount rate for receivables that on average
were expected to be paid within about 30 days can be restated as an annual financing
cost payable by McKesson Canada for its rights under the facility in the range
of 27% per annum.
[15]
A direct result of
these discounts was that McKesson Canada ceased to be profitable for its 2003
taxation year and reported a tax loss in the year in issue in this appeal.
McKesson Canada’s profits in later years were similarly significantly reduced.
[16]
The taxation year of
McKesson Canada under appeal ending March 29, 2003 was a short taxation year of
approximately three and a half months, having started upon its amalgamation as
part of a Canadian restructuring of the McKesson Group’s Canadian interests. Its
taxation and financial year ends on the last Saturday in March of each year.
Its financial year is divided into a 13 four week Accounting Periods. CRA’s
2003 transfer pricing adjustment was approximately $26,610,000, reflecting a
1.013% discount for the purchased receivables. No transfer pricing penalty was assessed.
[17]
The receivables
facility was a five-year revolving facility. As detailed below, the purchaser
had several rights to terminate the agreement in the event of any anticipated
deterioration in the quality of receivables generated in McKesson Canada’s business.
[18]
As discussed further
below, 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 McKesson Canada’s predominant purpose; they were results
of the transactions.
[19]
This trial was a very
lengthy and hard fought 32 day trial heard over a period of five months from October
2011 to February 2012. Formidable groups of lawyers represented each of the
Appellant and the Crown. The Court heard from two material witnesses and five
expert witnesses. Reams and reams of documentation were entered into evidence,
including further expert reports whose authors did not testify. After oral
argument both parties made further written submissions and further responding
submissions. Following the Supreme Court of Canada’s decision in Canada v. GlaxoSmithKline
Inc., 2012 SCC 52, [2012] 3 S.C.R. 3 in October 2012, both parties made
further written submissions.
2. The Financing Transactions
[20]
McKesson Canada and its Luxembourg immediate parent company (“MIH”) entered into a Receivables Sales
Agreement (the “RSA”) and a Servicing Agreement effective December 16, 2002.
[21]
Under the RSA, MIH
agreed to purchase all of McKesson Canada’s eligible receivables as at that
date (about $460,000,000) and committed to purchase all eligible receivables
daily as they arose for the next five years unless earlier terminated, and
subject to a $900,000,000 cap.
[22]
Eligible receivables
were generally trade receivables owing by an arm’s length customer who was not
in default on other receivables and whose receivables would not, except in
specific circumstances described below, represent in the aggregate more than 2%
of the then outstanding receivables pool. The 2% concentration limit on
eligibility did not apply to McKesson Canada’s handful of its largest named
customers who accounted for about one-third of sales and whose receivables each
already exceeded 2% of McKesson Canada’s outstanding receivables pool (the
“Designated Obligors”). All hospitals were also defined to be Designated
Obligors. The RSA has specific provisions, including discount rate calculation
considerations, applicable to these Designated Obligors. The RSA contemplated
the 2% concentration limit being waived or additions being made to the list of
Designated Obligors with MIH’s consent.
[23]
The RSA provided that
if a termination event occurred, MIH could direct McKesson Canada to advise its customers of the sale. In accordance with the related Servicing Agreement, McKesson
Canada would continue to service and collect the receivables in accordance
with its credit and collection policies and practices which were not to be
changed without the consent of MIH.
[24]
The Servicing Agreement
provided that McKesson Canada was to be the initial servicer but could be
replaced upon the occurrence of a termination event under the RSA. MIH agreed
to pay a fixed annual fee of $9,600,000 to the servicer to service the
outstanding transferred receivables regardless of the amount outstanding.
[25]
McKesson Canada did not warrant or guarantee the collectibility of the receivables or any portion
thereof. MIH had the right to put a defaulted receivable back to McKesson
Canada for an amount equal to the lesser of (i) 75% of its face amount, and (ii)
the amount ultimately collected on it. When exercised, McKesson Canada was to pay the 75% amount to MIH and any ultimate downward adjustment was to be made
subsequently.
[26]
MIH could terminate its
obligations to purchase any further McKesson Canada receivables upon the
occurrence of certain defined termination events, generally designed to
identify or anticipate deteriorating creditworthiness of McKesson Canada or its pool of customers generating the receivables. These events included financial
defaults of McKesson Canada or its affiliates, increases in the delinquency ratio
or loss ratio of the receivables beyond specific thresholds, a downgrade in the
credit rating of McKesson U.S., McKesson Canada’s name being changed to drop
the word McKesson, McKesson Canada ceasing to be controlled by McKesson U.S.,
McKesson U.S. ceasing to guarantee McKesson Canada’s bank and commercial paper
lenders, and any event occurring which materially adversely affected the enforceability
or collectibility of the receivables or MIH’s rights under the agreements. It can
be noted that the termination events were not limited to things in McKesson Canada’s control, and included events in the control of its direct and indirect
shareholders/parent corporations.
[27]
McKesson Canada continued to collect the receivables in the ordinary course. While ownership of the
receivables was transferred daily, settlement (i.e. payment by the purchaser
MIH) was more or less monthly.
McKesson Canada was not required under the RSA or the Servicing Agreement to
segregate the funds collected on MIH’s behalf as they came in, unless MIH
required it following a termination event. Each month the amount collected on
the receivables for MIH’s benefit would be used first to pay McKesson Canada for newly generated receivables and any balance would be remitted to MIH. If there
was a shortfall because newly originated receivables exceeded amounts
collected, MIH was required to put McKesson Canada in funds.
[28]
While the RSA was for
up to a five-year term, it was clearly a revolving facility. Purchased
receivables could be expected to be collected by MIH within about a month. MIH
could be expected to know within a short period of time if any obligor’s
payment history or prospects were declining or deteriorating, or if McKesson Canada’s creditworthiness was declining or deteriorating, and could take immediate steps to
protects its interests and its future exposure, without waiting five years.
[29]
The RSA provided that McKesson
Canada would pay MIH’s costs and expenses related to the transactions,
including the costs of an inter-company transfer pricing study.
[30]
The RSA provided that
the purchase price payable by MIH to McKesson Canada for each receivable would
be at a formulaically determined discount from its face amount. This is
described in greater detail immediately below.
[31]
Any dilution or
reduction to the face amount of a receivable resulting from discounts, rebates,
disputes or returns, or by way of set-off under the terms of the receivable or
otherwise, were deemed collections by McKesson Canada as servicer and to be
accounted for to MIH as such. This did not apply to prompt payment discounts
for reasons that were not directly explained in the evidence. The scope and
nature of how the dilution risk relating to prompt payment discounts is dealt
with under the agreements is in issue in this appeal.
[32]
During the term of the
RSA it was discovered that interest received on the transferred receivables had
not been accounted for and was retained by McKesson Canada. This was not
consistent with the RSA.
The parties agreed not to account for the past error or to correct it going
forward. In 2005, the parties agreed in writing that the interest and late
payment charge obligations on the transferred receivables were never intended
to be transferred – something that is difficult to reconcile with the RSA
language or on any arm’s length basis.
[33]
The agreements are
governed by Luxembourg law.
[34]
MIH, McKesson Canada’s direct parent, borrowed all of the money in Canadian dollars to purchase the
receivables from an Irish company in the McKesson Group (alluded to above),
that was one of its indirect parents. The purchase of receivables under the RSA
was MIH’s stated use of the funds in its loan agreements. The interest payable
by MIH was a direct function of the discount enjoyed by it under the RSA. MIH’s
obligation to repay its borrowings to its Irish affiliate was fully guaranteed
by its indirect parent, another related Luxembourg company (“MIH2”). In
addition, MIH enjoyed an indemnity from MIH2 under a Memorandum of
Understanding for any amounts payable in accordance with the RSA that were not
received from McKesson Canada in order to allow MIH to fully re-pay its
borrowings from its Irish affiliate.
MIH, McKesson Canada’s counter-party to the RSA and the Servicing Agreement,
did not take any financial risk under this group of contemporaneous,
inter-woven agreements all of which were financially and legally linked and
related. All such risk was borne by other entities in the McKesson Group. That
risk ultimately remained economically with McKesson U.S., everyone’s ultimate
parent company, both before and after the RSA transactions.
a) The RSA Discount Formula
[35]
The amount payable for a
purchased receivable under the RSA was the product obtained from multiplying
(i) the face amount of the receivable and (ii) one minus the Discount Rate
expressed to four relevant digits.
By way of illustrating the method in which this formula worked, if the Discount
Rate (as defined itself by a further formula) worked out to 0.0150, MIH would
pay $98.50 for every $100 of receivables, buying them at a 1.5% discount from
face.
[36]
The Discount Rate is
defined in the RSA to be the sum of (i) the Yield Rate, (ii) the Loss Discount,
and (iii) the Discount Spread.
(i) The Yield Rate
[37]
The Yield Rate was the
30 day Canadian dollar bankers’ acceptance (BA) rate, or CDOR, on the
first business day of the relevant settlement period. This operated as the
floating base rate. This was intended to reflect a current, risk-free market
rate of return. It is not challenged as an appropriate floating base rate for
this Canadian dollar, Canadian obligor transaction.
(ii) The Loss Discount
[38]
The Loss Discount was
intended to reflect the credit risk of the McKesson Canada customers who were
the receivables debtors. The Loss Discount was made up of two parts: (i) a Loss
Discount component applicable to the Designated Obligors (whose receivables
each exceeded 2% of the pool) and (ii) a Loss Discount component applicable to
the other smaller Obligors making up the more diversified majority of the
receivables pool.
[39]
The RSA expressed a
fixed Loss Discount of 0.23% applicable to the initial purchase of receivables
in 2002 to the end of 2003. This applied to the year under appeal.
[40]
For the remaining term,
the Loss Discount was to be recalculated each year starting January 1, 2004.
Also, if at any time MIH felt that the ratio of Designated Obligors’
receivables to other Obligors’ receivables in the pool was materially different
than originally calculated for a year, MIH could require the Loss Discount to
be recalculated that month. McKesson Canada as seller did not have any similar
right. The result was that the RSA required the Loss Discount to be recalculated
annually, and permitted only MIH to require it to be recalculated as often as monthly
in the event of such a material change in risk.
[41]
For the larger
Designated Obligors, a schedule to the RSA fixed their Individual Loss Discounts
for the entire five-year period. In computing the aggregate Loss Discount for
Designated Obligors, the weighted average of these fixed Individual Loss
Discount amounts (weighted by each Designated Obligor’s share of the
receivables pool as at the end of the prior year) was used. Since the Individual
Loss Discounts were intended to reflect the unique credit risk of each
Designated Obligor, there was a considerable range within the scheduled amounts
(ranging from approximately 0.04% to 0.35% - a nine-fold range).
[42]
For each of the other Obligors
(comprising about two-thirds of the pool), the RSA fixed an Individual Loss
Discount at 0.2380% for the entire five-year term. The mechanics of how this
number was arrived at was not apparent from the RSA as the agreement simply
stipulated a fixed number for the entire duration. In computing the Loss
Discount for these other Obligors, the weighted average of this fixed 0.2380%
(weighted by the total of these other Obligors’ share of the receivables pool
as at the end of the prior year) was used.
[43]
The Loss Discount was the
sum of the weighted Individual Loss Discount for all Obligors.
[44]
The initial fixed Loss
Discount of 0.23% applicable to the first full year was in fact calculated on
this same basis. The amount thereof is significantly in issue in this appeal.
(iii) The Discount Spread
[45]
The RSA fixed the Discount
Spread at 1.7305% for the entire duration of the agreement. Since this was
fixed, the agreement does not describe how this number was arrived at. The
evidence is that generally this relates to (i) the risk that McKesson Canada’s
creditworthiness deteriorated significantly, and receivables debtors might set
off their rebate entitlements in such event, (ii) the risk that McKesson
Canada’s customers might increase their take-up of available prompt payment
discounts, (iii) the risk that MIH might decide to appoint a new servicer
following any termination event who might require a greater servicing fee than
provided for successor servicers in the Servicing Agreement and (iv) the need
for the Discount Rate to fully cover MIH’s cost of funds.
[46]
There was no corresponding
consideration given to the possibility of McKesson Canada’s creditworthiness
improving, customers taking less advantage of prompt payment discounts, or the
impact of more prompt payments on the DSO. These imbalances were never
explained.
[47]
This 1.7305% amount, it
being a fixed amount, and the extent of MIH’s exposure to McKesson Canada credit risk under the agreements in the circumstances, are also significantly in
issue in this appeal.
3. The TDSI Opinions on Arm’s Length Terms &
Conditions and Pricing
[48]
The RSA and all related
agreements were first signed as of December 16, 2002. The conception,
structuring, planning and drafting was under way for an unknown amount of time
before that. This process seems to have been lead primarily by the
Vice-President of Taxes of McKesson U.S., together with the tax and banking
lawyers at Blake, Cassels & Graydon LLP (“Blakes”). Some general transfer
pricing advice on approaches to, and issues in, structuring such a transaction was
obtained in the summer of 2002 from Horst Frisch, a U.S. consulting company
specializing in transfer pricing.
McKesson Canada’s role was limited to providing support and information
regarding such things as its customers, its receivable portfolio, its
projections, and its credit and collection policies et cetera.
[49]
In the weeks before the
signing of the agreements, probably around December 1, 2002, Toronto Dominion
Securities Inc. (“TDSI”) was retained by Blakes to provide advice on certain
arm’s length aspects of certain of the terms and conditions of the RSA and of
certain components of the discount calculation. The TDSI engagement letter was
sent to TDSI on December 3rd. It is clear from TDSI’s advice that 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
these to reflect any advice or information given by TDSI. This is consistent
with the testimony of Mr. Hooper of TDSI.
[50]
It can be noted that in
2002 the Act included contemporaneous arm’s length transfer pricing
documentation/analysis requirements to defend the 10% transfer pricing penalty
provisions. In fact, the TDSI opinions were relied on as the only
contemporaneous basis to successfully contest CRA’s pre-reassessment proposal
to impose transfer pricing penalties.
[51]
TDSI’s advice was
initially sought on (i) receivables eligibility criteria (ii) termination
events/triggers and (iii) the appropriateness of the discount pricing. Somewhat
oddly and not explained, Blakes’ engagement letter specifically identifies and
raises the possible need to address the effect of a potential replacement
servicer under the Servicing Agreement as part of the discount.
[52]
McKesson Canada had a pre-existing business relationship with the Toronto Dominion Bank group. The
full scope of that was not put clearly in evidence, however, several years
earlier, McKesson Canada had done a receivables factoring transaction with TD
Factors. The TD Factors transaction was an entirely tax-driven year-end
short-term transaction designed to avoid Canadian federal capital tax and seems
to have been priced accordingly. It is entirely unreasonable to suggest this
was a truly comparable transaction for arm’s length pricing purposes to the one
in issue in this appeal.
[53]
Barbara Hooper is the
person at TDSI that Blakes chose to contact. She was known to be a senior
member of TDSI’s securitization group. Her advice was sought notwithstanding that
everyone knew the RSA and related transactions were not structured as securitization
transactions, were not intended to be securitization transactions, and that the
purpose, objective and characteristics of the RSA transactions were
significantly and materially different than a securitization transaction. No
advice or information was sought from anyone other than the TDSI securitization
group (and the TD bond traders briefly and casually consulted by Ms. Hooper’s
group.)
[54]
Barbara Hooper was and
is clearly a recognized professional and experienced expert (as a business
person would use that word) in securitization matters, including trade
receivables securitizations. She testified in this hearing as a material
witness and not a qualified expert witness. Since her role as a material
witness in offering the TDSI opinions involved her exercise of her professional
judgment, she was allowed to fully explain in her testimony those opinions,
including her reasons, and her supporting information, bases and subsidiary
opinions.
[55]
Clearly Ms. Hooper’s experience
with trade receivables securitizations qualified her to give valuable advice to
the McKesson Group entities participating in the transaction and to the Court.
She is certainly very knowledgeable about the risks associated with transfers
of trade receivables, how those risks can be identified, and how those risks
can be minimized in a receivables securitization transaction. Her experience
and expertise did not however extend to pricing those risks if the risks were
to be transferred, nor did it extend to market discount rates applicable to
outright non-recourse or limited recourse sales or factoring of trade
receivables. She did not hold out or suggest otherwise in the TDSI opinions or
in her testimony.
[56]
Her testimony and
TDSI’s involvement in the RSA transactions have been helpful to the Court. The
TDSI opinion followed the conceptual approach dictated by the RSA presented to
it by the McKesson Group. As is often the case with expert and other opinion
evidence, the Court found much of the detailed explanation and reasoning behind
the opinions, as well as some of the data and information supporting the
opinions, helpful notwithstanding that the Court does not arrive at entirely the
same conclusion in the end.
a) Eligibility Criteria
[57]
The TDSI opinion deals
with this in a single short three-sentence paragraph. It concludes, without
explanation or analysis, that the definition of eligible receivables in the RSA
is within the range of normal in an arm’s length transaction of this nature.
While it uses the words “transaction of this nature”, I can only conclude this
is a reference to a securitization transaction involving receivables given the
description of the experience TDSI brought to bear.
[58]
It goes on to identify
that the exposure to the receivables pool concentration levels associated with
McKesson Canada’s Designated Obligors would not be present in a securitization,
will need to be addressed by TDSI in addressing the Discount Rate, and that
this component of the Discount Rate will need to be dynamic, reflecting
possible changes in the relevant balance of Designated Obligor receivables in
the pool from time to time.
b) Termination Events
[59]
TDSI is satisfied that
the triggers in the RSA definition of termination event are within the range of
normal in an arm’s length transaction “of this nature”. I repeat my earlier
observations about her use of this phrase in the TDSI opinion.
[60]
The TDSI opinion makes
specific reference to the role of such termination triggers as protection for
poor performance of receivables or declining creditworthiness of the seller. It
identifies McKesson Canada’s creditworthiness as seller as relevant in part
because of its obligations to remit collections to MIH. TDSI is of the express
view that “because [McKesson Canada] is so closely tied and important to
[McKesson U.S.], it is reasonable to use the public debt ratings of [McKesson
U.S.] as an indication of [McKesson Canada’s] creditworthiness”.
[61]
The TDSI opinion goes
on to specifically consider i) the receivables pool’s delinquency ratio trigger
in the RSA, and ii) the receivables pool’s loss ratio trigger in the RSA.
(i) Delinquency Ratio Trigger
[62]
TDSI considered the
improving two year historical trend in the delinquency ratio of McKesson Canada’s receivables and the recently maintained 1.0% rate. The 2.5% trigger rate in the
RSA would, in TDSI’s opinion, represent a significant adverse deviation from
the current steady state of 1% and so considered reasonable. TDSI highlighted
the importance of the dynamic four-month rolling average approach to measuring
the delinquency ratio in the RSA, and uses this approach in its analysis. TDSI
confirmed that this is consistent with the three to six month periods generally
used for such purposes.
(ii) Loss Ratio Trigger
[63]
TDSI looked at three
years of historic bad debt experience on McKesson Canada’s receivables
portfolio. TDSI identified the difference between accounting write-offs and the
90 day delinquency definition of losses for purposes of the loss ratio in the
RSA, with the result that the latter ratio could be expected to exceed the
former. TDSI opined that a dynamic loss ratio, which measured a four-month
average 90 day delinquency, and with a trigger of 0.25%, appeared reasonable
given that, although write-offs to sales on a monthly basis at times reached
this level, it had never exceeded 0.10% on a four-month rolling average.
c) Discount Rate
[64]
The TDSI Report says
its assessment of the appropriate compensation was set by (i) where possible,
looking at pricing of comparable risks in the market and (ii) “where pricing
comparables were unavailable, we assessed the potential cost to [MIH] of
assuming the risk”.
[65]
In its summary
paragraphs on the total discount at the end of its report, TDSI addressed the
total discount following the RSA definition of Discount Rate as the sum of (i)
the Yield Rate, (ii) the Loss Discount, and (iii) the Discount Spread. In
summarizing the TDSI Report on the Discount Rate, it is easiest for the Court
to follow this same order in its Reasons rather than the different order more
loosely followed in the body of the TDSI Report.
(i) The Yield Rate
[66]
TDSI identifies that
the RSA used the 30 day Canadian dollar CDOR/BA Rate as the floating base rate
component and that the 30 day CDOR Discount Rate needs to be adjusted to reflect
the receivables’ DSO.
[67]
Notwithstanding the
definitions of Discount Rate and Yield Rate in the RSA (even after it was
amended and restated, and after being further clarified), the Court observes
that the Yield Rate component of the Discount Rate in fact needs to be adjusted
to reflect that the 30 day CDOR Rate is expressed as an annual rate and
therefore needs to be adjusted to reflect the receivables pool’s DSO by
multiplying it by the DSO and then dividing by 365.
[68]
TDSI selected the receivables
pool’s three year average monthly DSO it calculated as 32. There was no discussion
of why a shorter period’s average monthly DSO was not used, nor was there any
discussion in the TDSI Report of a dynamic rolling average approach to DSO.
[69]
TDSI recognized that
the DSO for the initial purchase of approximately $460,000,000 of receivables
would very significantly overstate the expected payment term for these
receivables as they were a pool of mature short-term receivables (i.e. some
would be paid the following day because they had been outstanding for between 1
and 30 days or more).
[70]
TDSI’s approach was to
instead use a 16 day DSO for the initial purchase. However, it did not in fact
calculate the Discount Rate for the initial December 16, 2002 purchase using a
16 day DSO; it instead averaged the “missing” 16 days across the entire five
years of the RSA term, and set a fixed DSO of 31.73 days for all Discount Rate
calculations for all receivables purchases under the RSA. It is obvious that
the effect of this was to 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
in issue.
(ii) The Loss Discount
1. Designated Obligors
[71]
TDSI begins its Loss
Discount analysis by determining the monthly write-offs to sales from the 3 to
4 years of data provided to it for McKesson Canada, together with the same
numbers computed on a three and 12 month rolling average basis. They conclude
that, viewed from this dynamic perspective, McKesson Canada’s write-offs to
sales are “very low”.
[72]
TDSI identifies a
concentration risk issue associated with the larger Designated Obligors that
losses on their receivables have an increased likelihood of deviating from
historical levels (presumably not necessarily adversely). TDSI does not attempt
to quantify that increased likelihood, nor most importantly, does it analyze
the historic data associated with Designated Obligors separately to try to
validate the increased risk.
[73]
TDSI looked at the
Designated Obligors separately from the other Obligors and looked at each
Designated Obligor individually (treating all hospitals as a single Designated
Obligor), as contemplated by the RSA.
[74]
TDSI thought it
appropriate to consider each Designated Obligor’s public debt rating, or if a
Designated Obligor did not have one of its own, its parents’ rating without
adjustment. If neither was rated, the TDSI Report assumed a non-investment grade
credit rating. TDSI then looked at each Designated Obligor’s 180 day risk
credit spreads in the public debt market. TDSI thought it was appropriate to
treat MIH’s risk under the RSA as 180 day risk which would allow for TDSI’s
estimate that it would take 90 days to wind down the pool and liquidate the
portfolio upon termination.
[75]
TDSI estimated the Individual
Loss Discount for each Designated Obligor using this market based discount,
first adjusted for the DSO and then applied to (multiplied by) its total receivables
at that time (December 2002). These numbers were then fixed in the RSA’s
Schedule D of Designated Obligors’ Individual Loss Discounts.
2. Other Obligors
[76]
The TDSI Report uses
one month historic data to assess the Loss Discount attributable to the smaller
Obligors comprising the majority of the receivables pool. The report gives no
reason for choosing the one month historic average instead of either the three
or 12 month rolling averages also computed and set out in this portion of the TDSI
Report (or, for that matter, the four month rolling average set out and used by
TDSI in the portion of its report opining on loss trigger termination events). The
TDSI Report does not appear to even break out the Designated Obligor portion of
the pool for this purpose to look only at the other Obligor performance. Neither
the report nor Ms. Hooper’s evidence provided any explanation.
[77]
TDSI goes on to note
that the multi-year data it has been provided does not cover a full economic
cycle of Canadian trough-to-peak-to-trough. The report does not say more
historic data was asked for, nor that it was unavailable. For this reason, it “suggests”
adding three standard deviations from mean (although TDSI only set out average
not mean in its charts). No reason is given for this suggestion that three
standard deviations be used. This suggestion, combined with the selection of
monthly numbers, had the result of driving the .03% actual monthly average loss
to 0.24% - that is an eight-fold increase. (Though the TDSI Report then uses
0.2380%; presumably more accurate four digit calculations had been done outside
the report).
[78]
It can be noted from
TDSI’s chart that, while the computed standard deviation was .07% for monthly
historic averages, it was a fraction of that for the rolling 12 month and three
month historic loss performance (.02% for 12 month and .04% for three months).
It can be observed readily that had this same approach of adding three standard
deviations to twelve month historic rolling average losses been taken, the twelve
month average of .04% would only increase to .1% - which would still be a two
and one-half fold increase (and still without any given reason for using three
standard deviations).
[79]
TDSI correctly points
out that this approach to the Loss Discount percentage applicable to other Obligors
is a function of losses on the dollar amount of receivables, not an annual or
other rate that is a function of time, and thus does not need to be adjusted
for the DSO.
[80]
The TDSI Report then
simply totals the pool’s weighted DSO adjusted credit spread for each
Designated Obligor consistent with the RSA, and the other Obligors’ 0.2380%
weighted to their share of the receivables pool owing by the other Obligors.
[81]
TDSI described the
annual (or earlier at MIH’s request) adjustment of the Loss Discount component
of the Discount Rate as appropriate. It did not discuss why it considered those
appropriate.
(iii) The Discount Spread
[82]
The set and fixed
discount spread in the RSA of 1.7305% appears from the TDSI Report to have been
built up by TDSI. The TDSI Report built up this number from four components (1)
a servicing discount, (2) a prompt payment dilutions discount, (3) an accrued rebate
dilutions discount, and (4) a credit spread interest discount.
[83]
It is not at all clear
from the RSA, the TDSI Report, or the witnesses why these four risks are
addressed by something called a discount spread. In any event, it is clear that
the parties to the RSA and related transactions, and their advisors, considered
these the material risks associated with the RSA transactions other than the
risk of loss on the receivables themselves resulting from Obligors’ financial
defaults.
1. Servicing Discount
[84]
The TDSI opinion begins
its analysis with the points that (i) the RSA requires MIH to service the
purchased receivables, (ii) under the Servicing Agreement the servicer (including
the initially appointed servicer, McKesson Canada) will be paid a fixed fee of
$9,600,000 annually, paid monthly, to service the receivables pool regardless
of size, and (iii) the Servicing Agreement contemplates the prospect of MIH
choosing to, or needing to, appoint a replacement servicer.
[85]
TDSI ties the “very
possible” need for a replacement servicer to the fact that MCC is not a highly-rated
credit on a stand alone basis. The connection is not described further and is
not obvious.
[86]
TDSI then set out the
replacement servicer pricing it obtained from a single source (believed to be perhaps
a major accounting firm). This indicated collection fees for current accounts in
the range of 1.0% to 3.0% of the face amount of the receivables. TDSI did not
seek replacement servicer pricing from TD Factors.
[87]
TDSI observed again that
a replacement servicer would only be needed for a short period of time once the
agreement was terminated and no further receivables were purchased by MIH.
[88]
TDSI then chose to use
a 2% replacement servicer fee for the reason that it was the midpoint of the 1%
to 3% range.
[89]
TDSI observed that the approximately
$800,000 fee payable monthly under the Servicing Agreement would be in the
range of 0.2% assuming that approximately one-half of the $900,000,000 cap, or about
the original purchased amount of receivables, was outstanding throughout the
term. No attempt is made in the report to explain or substantiate the “expected
sales” volumes.
No explanation is made of the relevance of a $900,000,000 cap that does not
relate to expected sales. No attempt is made to explain the tenfold difference
between the 0.2% charged by McKesson Canada and TDSI’s selection of 2% from the
range of 1% to 3% for a replacement servicer. Nor is this last issue dealt with
in TDSI’s supplemental report.
[90]
The TDSI Report then
notes that a Moody’s publication consulted by it indicated that, in the
previous year, 9.41% of companies with McKesson U.S.’ rating were downgraded to
the RSA trigger rating or had their rating withdrawn. From there, without any
further explanation, TDSI then used a “conservative” “assumption” of a 25%
“chance” that a replacement servicer would need to be appointed.
[91]
Using these numbers
alone, TDSI then multiplies 2.0% times 0.25, and 0.2% times 0.75, and adds
these to arrive at a “reasonable expected cost of servicing” of 0.65%. This
number, three and one-quarter times what McKesson Canada or a replacement servicer
is to be paid to service the receivables under the Servicing Agreement, is then
used to discount each receivable (including the $465,000,000 worth purchased
upon signing the agreement which appeared to have virtually every likelihood of
being paid in orderly fashion beginning the very next day, and including those
purchased on the next day, and the day after that, et cetera).
[92]
Several months after
the transactions were in place, TDSI was asked for follow-up advice on the
fixed fee payable to the servicer under the Servicing Agreement. TDSI’s
supplemental report is described further below.
2. Prompt Payment
Dilutions Discount
[93]
As described above, the
RSA does not treat prompt payment discounts enjoyed by McKesson Canada’s customers as deemed collections on their receivables. The TDSI Report accepts this
without question and without even addressing whether this would be normal in
arm’s length securitization transactions. It is not self-evident why prompt
payment discounts are not treated in the same manner as other dilutions such as
volume rebates to customers.
[94]
Thus, TDSI identifies the
risk in the RSA that there may be a change in the extent to which McKesson
Canada’s customers take up the prompt payment discounts offered by paying their
receivables earlier than past individual or overall experience might suggest.
For example, if MIH purchases $100 of receivables and the purchaser promptly
pays $98 in full satisfaction, McKesson Canada does not have to account to MIH
for the $2. The RSA pricing therefore had to somehow address the fact that prompt
payment discount rights would be exercised by McKesson Canada customers. MIH would be overpaying if the embedded take-up was underestimated and
McKesson Canada would be underpaid if the embedded take-up was overstated. TDSI
computed McKesson Canada’s prompt payment discounts historically on an annual
basis as 0.5% of sales and observed this is “very consistent”.
[95]
TDSI then suggested
that a 20% “buffer” be added to bring the 0.5% to 0.6%. The TDSI Report does
not attempt to explain why a 20% buffer was chosen. It does not acknowledge,
much less address, the fact that if up to 20% more customers start enjoying
prompt discount payments on the terms already offered, there would be a
corresponding favourable impact on the DSO of the receivables pool which would
have a correspondingly material impact on risk and pricing as MIH would be paid
materially more quickly.
[96]
Unstated and
unexplained, but implicit in the TDSI Report, is that TDSI is fine with the
prompt payment discount risk being estimated at the outset, fixed throughout at
historic levels, plus an unexplained buffer, instead of it being reflected in
its rolling average actual performance throughout the RSA’s term, and even
though it is to be incorporated in the MIH RSA risks instead of being treated
as a deemed collection.
3. Accrued Rebate
Dilutions Discount;
[97]
McKesson Canada’s volume rebates are paid separately to its customers and are not enjoyed at point
of purchase or time of payment. Rebates are instead paid directly by McKesson Canada to its customers periodically. The risk that transferred receivables would be
subject to a reduction for rebates was therefore not transferred to MIH nor
accounted for under the RSA. However, there was nonetheless the risk that a
customer would choose to set-off its McKesson Canada payables by the amount of
an anticipated or earned but yet unpaid rebate owing to it from McKesson Canada. TDSI speculated this may happen if customers became concerned about McKesson Canada’s financial situation.
The RSA expressly and clearly requires such a set-off risk to be fully borne
by, and indemnified by, McKesson Canada. Thus, TDSI identifies that the RSA has
MIH taking McKesson Canada (MIH’s direct subsidiary) credit risk for McKesson Canada’s indemnity obligation.
[98]
TDSI computed the
three-year historic rebates as a function of sales numbers. The average was
3.8%. The minimum was 2.13% and the maximum was 5.53%. Each of these percentages
is a total of all accrued rebates. Thus, without further adjustment, using
these raw numbers would reflect the situation where all of McKesson Canada’s customers decide it is time to set-off rebates. Without any explanation other than
to be conservative, TDSI selected the highest maximum number of 5.53% of the
receivables balance as the amount at risk for set-off. No suggestion is made by
TDSI that a dynamic rolling average approach should be taken or considered
through the term of the RSA. TDSI does not address the prospect that MIH might
be expected to have and exercise termination rights if McKesson Canada’s financial situation were to become such that MIH was concerned about its financial
viability. The TDSI Report does not say they are aware that any such set-off
has ever been made, claimed, threatened or even asked for.
[99]
To this maximum set-off
amount, TDSI applies a factor selected by it to reflect McKesson Canada credit risk. For this purpose they assessed a 5.25% credit spread (that is the
amount charged above a lender’s floating fixed base rate) for McKesson Canada on the basis that, without its own credit rating, they should use non-investment
grade credit spreads as a proxy. (Non-investment grade borrowers’ bonds in the
bond market are also referred to as high yield bonds or junk bonds). This, even
though several times elsewhere in the TDSI Report, they use unrated Obligors’
parent ratings as their proxy, and accept McKesson U.S.’s credit rating risk as
reflective of McKesson Canada’s creditworthiness risk in the discussions of
termination event and triggering event. The TDSI Report does not explain the
relationship of its estimated 5.25% credit spread to McKesson Canada’s actual rates available to it from its available lenders and facilities which, as
described elsewhere, are very different. Notably, given that the RSA clearly
did not require McKesson Canada to segregate collections (absent a termination
event), the McKesson Group and MIH had little apparent concern about McKesson Canada’s creditworthiness or financial situation. After all, the risk of collections not
being passed through to a purchaser of receivables is one of the significant
risks in a factoring transaction.
[100]
The TDSI Report does
not address MIH’s rights under the RSA to have the funds segregated and not commingled
at any time there has been a termination event including a material adverse
change. Nor does it address whether the failure to segregate absent a
termination event and the willingness to accept commingling risk in the RSA
about termination is consistent with market practices in arm’s length negotiated
agreements such as those involving securitizations or arm’s length servicers.
[101]
TDSI multiplies the
5.53% maximum historic set-off risk by the 5.25% non-investment grade borrower
credit spread (adjusted for the DSO because credit spreads are expressed as an
annual rate) to arrive at a fixed discount of 0.0244% for the accrued rebate
dilutions discount for the entire term of the RSA. There is no explanation
offered for accepting a fixed approach.
4. Interest Discount
[102]
This aspect is dealt
with in a single paragraph of the report. It is not a risk- based assessment
but reflects TDSI’s observations that “in addition to being compensated for the
risks it is assuming, [MIH] must also cover its cost of capital out of the
discount.”
[103]
TDSI then says that
because MIH is “exceptionally thinly capitalized” it is
again appropriate to use non-investment grade bond market credit spreads as a
proxy.
[104]
These statements may be
correct from a profitable business transaction point of view, however TDSI does
not address the possibility that the transaction exactly as structured could
not be done profitably on arm’s lengths terms and conditions, or the fact that
cost of funds/source of financing is an expense to a buyer that does not
generally increase or decrease the value of a seller’s assets, or any other
issues raised thereby.
[105]
TDSI was not made aware of MIH’s
sources of financing, and was therefore unaware it also enjoyed a full
guarantee and indemnity from its parent company in the McKesson Group.
[106]
The TDSI Report then goes on to
add a five-year swap spread for non-investment grade bond (i.e. junk) issuers
of 5.25% (as described above with respect to rebates) to the 30 day CDOR rate
because of the five-year term of the agreement.
[107]
This non-investment grade/high
yield/junk bond rate is thus factored into the discount pricing twice by TDSI,
once to reflect McKesson Canada’s creditworthiness with respect to the maximum
rebate exposure potential, and again to reflect MIH’s needed return to cover its
cost of the full amount borrowed by it.
[108]
TDSI does not explain why it uses
five year pricing for funds needed by MIH in a monthly settled revolving facility
with a DSO of about a month and which TDSI estimates (twice) in its report as
perhaps as much as 180 day exposure (which includes a 90 day allowance to wind
up).
[109]
As with the other components of
the Discount Spread, the TDSI Report does not address the appropriateness of
the Interest Discount credit spread being at a fixed rate throughout the five-year
term.
[110]
Overall, it appears clear that
each of the four components of the Discount Spread have clearly been computed
using the maximum numbers even arguably justifiable (not necessarily being, or
clearly being, the maximums within a reasonable range), and then fixed each of
those at those maxed out numbers for the entire five-year term without
adjustment or recalculation on any basis even though there are dynamic monthly
and annual components (some of which are very similar) and material adverse
change or MAC rights elsewhere in the RSA agreements. This lack of balance in
favour of MIH is not expressly identified, and not discussed at all in the TDSI
reports or evidence.
4. TDSI’s
Supplemental Report on Servicing Fees
[111]
In April 2003, after
the execution and implementation of the RSA and related transactions, TDSI was
asked to report to McKesson Canada, MIH and Blakes on whether the fixed monthly
fee of $800,000 paid under the Servicing Agreement to administer and collect
all the receivables in the pool was within the range of normal for this type of
arrangement.
[112]
TDSI again begins by
being clear that its expertise in this area is based upon TDSI’s experience arranging
Canadian trade receivables securitization transactions, which are structured
without a separate servicing fee being paid or negotiated and “accordingly, we
do not have ready knowledge of comparable situations”.
[113]
The TDSI servicing
report carries on:
The level of the servicing fee must be assessed relative to
the resources required to effectively service the portfolio. The resource
requirements will depend upon the size and composition of the receivables
portfolio.
[114]
It then notes, however,
that under the Servicing Agreement, McKesson Canada is paid a fixed fee each
month regardless of the size of the portfolio. TDSI computes that the fixed
annual fee at $9,600,000 works out to somewhat in excess of a 1% per annum fee
on the $900,000,000 limit and somewhat in excess of a 2% fee on the then
current $460,000,000 range pool. TDSI’s investigations turned up servicing fees
of 1% per annum recorded by Bell Canada in a $1,000,000,000 receivables
securitization and 2% per annum by Telus in a $650,000,000 receivables
securitization. TDSI notes that these receivables would involve primarily Telus
and Bell’s retail customers’ phone bills.
[115]
TDSI then looked at
the Servicing Agreement fee as a function of the dollars to be collected and
determined this was approximately 0.1% of the maximum eligible pool size of
$900,000,000 and a 32 day DSO, and approximately 0.2% based on the receivables
balance at closing. TDSI’s investigations found only one U.S. transaction with relevant information disclosed publicly which reflected a servicing
fee equal to 0.1% of collections.
[116]
Finally, TDSI
contacted two large rating agencies with considerable involvement in rating
securitization transactions. According to Moody’s, negotiated servicing fees
between buyers and sellers are typically in the range of 1% to 2% of the
average receivables balance. According to Standard & Poor’s, servicing fees
are usually 1% per annum of the average receivables balance. Note that both
rating agencies describe it as a dynamic fee, based on average receivables
balances during the year.
[117]
In its closing
opinion, TDSI again notes it is unusual that the servicing fee is fixed. Based
on the amount of receivables transferred at closing, TDSI concludes that the
fee is somewhat higher than the few comparables. It is clear that by somewhat,
TDSI meant up to 100%.
[118]
TDSI’s servicing
opinion is dated April 25, 2003. The 2003 year under appeal ended in March
2003. I assume from the TDSI Report and discussion that the size of the
receivables pool did not increase materially between the mid-December 2002
initial closing and the end of the year under appeal in March 2003.
5. The Law
Subsection 247(2) of the Act provides as follows.
|
(2) Transfer pricing
adjustment -- Where a taxpayer or a partnership and a non-resident person
with whom the taxpayer or the partnership, or a member of the partnership,
does not deal at arm's length (or a partnership of which the non-resident
person is a member) are participants in a transaction or a series of
transactions and
|
2) Redressement -- Lorsqu'un
contribuable ou une société de personnes et une personne non-résidente avec
laquelle le contribuable ou la société de personnes, ou un associé de cette
dernière, a une lien de dépendance, ou une société de personnes dont la
personne non-résidente est une associé, prennent part à une opération ou à
une série d’opérations et que, selon le cas :
|
|
(a) the terms
or conditions made or imposed, in respect of the transaction or series,
between any of the participants in the transaction or series differ from
those that would have been made between persons dealing at arm's length, or
|
a) les modalités conclues ou imposées, relativement à l’opération
ou à la série :
|
|
(b) the
transaction or series
|
b) les faits suivants se vérifient relativement à l’opération ou à
la série :
|
|
(i) would
not have been entered into between persons dealing at arm's length, and
|
(i) elle n’aurait pas été conclue entre personnes sans lien de
dépendance,
|
|
(ii) can
reasonably be considered not to have been entered into primarily for bona
fide purposes other than to obtain a tax benefit,
|
(ii) il est raisonnable de considérer qu’elle n’a pas été
principalement conclue pour des objets véritables, si ce n’est l’obtention
d’un avantage fiscal,
|
|
any amounts that, but for this
section and section 245, would be determined for the purposes of this Act in
respect of the taxpayer or the partnership for a taxation year or fiscal
period shall be adjusted (in this section referred to as an “adjustment”) to
the quantum or nature of the amounts that would have been determined if,
|
Les montants qui, si ce n’était le présent article et l’article
245, seraient déterminés pour l’application de la présente loi quant au
contribuable ou la société de personnes pour une année d’imposition ou un
exercice font l’objet d’un redressement de façon qu’ils correspondent à la
valeur ou à la nature des montants qui auraient été déterminés si :
|
|
(c) where
only paragraph (a) applies, the terms and conditions made or imposed, in
respect of the transaction or series, between the participants in the
transaction or series had been those that would have been made between
persons dealing at arm's length, or
|
c) dans le cas où seul l’alinéa a) s’applique, les modalités
conclues ou imposées, relativement à l’opération ou à la série, entre les
participants avaient été celles qui auraient été conclues entre personnes
sans lien de dépendance ;
|
|
(d) where
paragraph (b) applies, the transaction or series entered into between the
participants had been the transaction or series that would have been entered
into between persons dealing at arm's length, under terms and conditions that
would have been made between persons dealing at arm's length.
|
d) dans le cas où l’alinéa b) s’applique, l’opération ou la série
conclue entre les participants avait été celle qui aurait été conclue entre
personnes sans lien de dépendance, selon des modalités qui auraient été
conclues entre de telle personnes.
|
[119]
The Minister’s
reassessments rely upon paragraphs 247(2)(a) and (c) to make a
transfer pricing adjustment in respect of the RSA. These subparagraphs apply if
a taxpayer (McKesson Canada) and a non-resident person with whom the taxpayer does
not deal at arm’s length (MIH) are participants in a transaction (each
of the RSA and the Servicing Agreement) or a series of transactions (the RSA,
the Servicing Agreement, and MIH’s loan agreement with its Irish indirect
parent, the related guarantee thereof from MIH2 to the Irish company, and the
related indemnity of MIH2 in favour of MIH regarding McKesson Canada’s obligations under the RSA), and the terms or conditions thereof differ from those
that would have been made by arm’s length persons. If the “terms and
conditions” do so differ, then the “amounts” that would otherwise be used by
the taxpayer for purposes of the Act shall be “adjusted” to the “quantum
or nature” of the amounts that would have been determined had the “terms and
conditions” been those that arm’s length parties would have agreed to.
(a) GlaxoSmithKline
[120] The Supreme Court of Canada in GlaxoSmithKline
had occasion to address the scope of the review of the relationships and
circumstances that a Court is to undertake in a transfer pricing appeal:
(1) A judge is to take into account all
transactions, characteristics and circumstances that are relevant (including
economically relevant) in determining whether the terms and conditions of the
transactions or series in question differ from the terms and conditions to
which arm’s length parties would have agreed.
(2) The transfer pricing provisions of
the Act govern and are determinative, not any particular methodology or
commentary from the OECD Guidelines, or any source other than the Act.
I would add the observation that OECD
Commentaries and Guidelines are written not only by persons who are not
legislators, but in fact are the tax collection authorities of the world. Their
thoughts should be considered accordingly. For tax administrators, it may make
sense to identify transactions to be detected for further audit by the use of economists
and their models, formulae and algorithms. But none of that is ultimately
determinative in an appeal to the Courts. The legal provisions of the Act govern
and they do not mandate any such tests or approaches. The issue is to be
determined through a fact finding and evaluation mission by the Court, as it is
in any factually based issue on appeal, having regard to all of the evidence
relating to the relevant facts and circumstances.
(3) Arm’s length prices are established
having regards to the independent interests of each party to the transaction.
In this appeal, this means that the RSA transactions must be looked at from
both the perspectives of McKesson Canada and of MIH.
(4) Other arm’s length transactions can be
relied upon as comparables in a transfer pricing analysis only if either there
are no material differences that would affect pricing, or if reasonably
accurate adjustments can be made to eliminate the effects of such differences.
(5) Quoting from GlaxoSmithKline:
61 As
long as a transfer price is within what the court determines is a reasonable
range, the requirements of the section should be satisfied. If it is not, the court
might select the point within a range it considers reasonable in the
circumstances based on an average, median, mode, or other appropriate
statistical measure, having regards to the evidence that the court found to be
relevant. I repeat for emphasis that it is highly unlikely that any comparisons
will yield identical circumstances and the Tax Court judge will be required to
exercise his best informed judgment in establishing a satisfactory arm’s length
price.
(b) Reasonableness
[121]
While GlaxoSmithKline
involved the former subsection 69(2) transfer pricing rule, which was worded
differently, I see no compelling reason to depart from the Supreme Court of
Canada’s approach and comments in GlaxoSmithKline. While section 247
does not use the words “reasonable in the circumstances” or “fair market
value”, arm’s length persons should generally be assumed for purposes of
section 247 to act neither irrationally nor unreasonably having regard to all
relevant circumstances. Similarly, arm’s length persons should generally be
expected to transact for products and services at amounts within the range of
their fair market value having regards to all relevant circumstances. This is
not inconsistent with the wording of section 247.
(c) Relevant Series of Transactions
[122]
The Supreme Court of
Canada in Copthorne Holdings Ltd. v. The Queen, 2011 SCC 63, mandates an
expansive approach to the issue of series, given the inclusive nature of the
meaning to be given to “series of transactions” in subsection 248(10). The
starting point is the common law series in which each transaction in the series
is pre-ordained to produce a final result. Then, subsection 248(10) deems any
related transaction completed in contemplation of a series to be part of that
series.
[123]
The determination of the existence
of a series and its constituent transactions is a question of fact to be
determined on a balance of probabilities. While only amounts under the RSA have
been challenged, clearly the Servicing Agreement, MIH’s Loan Agreement, and MIH2’s
guarantee and indemnity are also all part of a series of transactions that
included the RSA. In addition, their existence and terms also each meet the
threshold of being relevant to a consideration of the RSA as described by the
Supreme Court of Canada in GlaxoSmithKline. Each of these agreements are
contemporaneous and interdependent in fact and in law. These agreements were
needed to fund and to repay the RSA between McKesson Canada and MIH. They
relate directly to the trade receivables of McKesson Canada’s business.
McKesson Canada may not be a party to each of these agreements, but they each expressly
refer to the terms and conditions of the RSA, or the sale, servicing, or
collections on those receivables, et cetera.
[124]
Given that all of these
transactions meet the threshold test of relevance in GlaxoSmithKline for
purposes of considering the transfer pricing within the RSA itself, and given
that the Crown has only challenged the amount of the discount used in the RSA,
the Court does not need to rely further on the fact that these transactions
together constitute a series of transactions for purposes of section 247.
(d) Scope of Adjustments Permitted Under Paragraph
247(2)(c)
[125]
A reassessment under
subparagraphs 247(2)(a) and (c) does not permit a recharacterization
of the transactions entered into by non-arm’s length parties, nor can another
different transaction entirely be substituted therefor. This would only be
permitted under subparagraphs 247(2)(b) and (d) which have not
been pleaded and the Crown is not relying upon. A transfer pricing recharacterization
is only permitted under those provisions if arm’s length parties would not have
entered into the transaction chosen by the non-arm’s length parties even with
different terms and conditions and amounts, and if the only bona fide
primary purpose of the transaction was to obtain a tax benefit.
[126]
However, it is clear
from the provisions of section 247 that under subparagraphs (a) and (c)
the Court is not limited to making adjustments with respect to the quantum of
an amount in a term or condition that incorporates an amount. I do not accept
the taxpayer’s submission that I am so limited. Paragraph 247(2)(a) is
triggered when terms or conditions differ from those terms and conditions that
arm’s length parties would agree to. There is no such limiting restriction on
the phrase terms and conditions. Paragraph 247(2)(c) then mandates an
adjustment to the quantum or nature of an amount used by the taxpayer for
purposes of the Act to reflect the quantum or nature of that amount that
would have been used had the “terms and conditions” conformed to what arm’s
length parties would have agreed to.
[127]
Perhaps there is a point
at which the extent of changes to the agreed non-arm’s length terms and
conditions needed to reflect arm’s length terms and conditions in a transaction
can constitute an effective recharacterization of the transaction only
permitted to be affected under paragraph 247(2)(d) and only in the
circumstances described in paragraph 247(2)(b) which provisions are not
engaged in this appeal. Perhaps there also may be some terms and conditions in
a transaction that are so fundamental that any particular change thereto could
constitute in effect a recharacterization of the transaction. The Court does
not need to venture anywhere close to that line in disposing of this appeal.
That can be left for another day. In this case the Court is able to limit
itself to a consideration of terms and conditions which it finds to not be on
arm’s length terms and that directly relate to pricing.
(e) Factors that Exist
Only because of the Non-Arm’s Length Relationship
[128]
Within a transfer
pricing review, the question arises whether factors that exist only because of
the non-arm’s length relationship are assumed away in the notional arm’s length
analysis or remain relevant characteristics and circumstances.
[129]
This question may not
arise to any extent in the context of a single purchase at a fixed price. The
question does appear significant in the context of a long-term commitment to do
things over a period of time. For example, in transactions such as those
involving the RSA, does the Court assume a notional arm’s length MIH would
still enjoy the benefit of the Irish company loan supported by the MIH2
guarantee and indemnity? In looking at transactions like the RSA, does the
Court assume the notional arm’s length MIH still has the power throughout the
term of the notional arm’s length contract to change McKesson Canada’s name, sell McKesson Canada, or do something else in order to trigger a termination event at
will? Does the Court assume that the notional arm’s length purchaser still has
the right to cause McKesson Canada to agree to change terms as they apply to
future transactions under the agreement? Does the Court assume that the
notional arm’s length MIH still has access to all of the financial information
of McKesson Canada and information regarding its receivables portfolio and its
entire business even though it may not be specified or required in the RSA?
[130]
This issue was addressed
by Justice Pizzitelli in Alberta Printed Circuits v. The Queen, 2011 TCC
232:
It
is important to note that factors or circumstances that exist solely because of
the non-arm’s length relationship of the parties should not be ignored,
otherwise the reasonable businessman would not be standing entirely in the
Appellant’s shoes …
…
In General Electric, the Federal Court of Appeal confirmed that no error
of law was made in taking into consideration the Appellant in that case, as a
sub of its larger parent company, stood in the position of having an implicit
guarantee by its parent of its bank debts.
[131]
Based on this, all
circumstances, including those that arise from, derive from or are rooted in
the non-arm’s length relationship should be taken into account.
[132]
I think the better view
is therefore that the Court can and should consider notional continued control
type rights in appropriate circumstances when looking at term or executory
contract rights. Not to do so would be to not look at all of the relevant
characteristics and circumstances of the relationships. If these were to be
ignored by a Court, companies within wholly controlled corporate groups could
enter into skeletal agreements conferring few rights and obligations to the
non-resident participant, (such as financial information disclosure, use of
funds, financial covenants et cetera), all with the view to obtaining a more
favourable transfer price to reduce Canadian taxes. Not approaching this issue
this way would seem entirely inconsistent with this Court and the Federal Court
of Appeal in G.E. Capital having focused on implicit unwritten,
unenforceable guarantees of the parent company of the borrower. However, in
this case, I do not need to do so in order to fully dispose of the appeal with
respect to the proper transfer pricing adjustment, as detailed below. This too
can be left for another day.
(f) The Rule in Browne
v. Dunn and Opinions Within a Witness’ Expertise
[133]
The rule in Browne
v. Dunn, (1893), 6 R. 67 (H.L.) generally requires counsel to give notice
to those witnesses that the cross-examiner intends later to impeach. This
allows the witness the opportunity to explain, and reflects a sense of fair
play and fair dealing with the witness and his or her evidence.
[134]
In response to an
objection by Appellant’s counsel, the Court had to rule in the course of the
hearing on the applicability of the rule in Browne v. Dunn to the
opinion and supporting evidence of the Appellant’s experts in respect of which
the testimony of one of the Respondent’s experts expressed different views. The
Court again had to rule on the same issue in response to a further objection by
Appellant’s counsel as it related to one of the Respondent’s experts having
different views from the TDSI Report and Ms. Hooper’s testimony regarding her
opinion and her reasoning expressed therein. In both cases the Court denied the
Appellant’s objection to the evidence of the Respondent’s expert.
[135]
The rule in Browne
v. Dunn is not a fixed absolute. The extent of its application is within
the discretion of the trial judge after considering the circumstances of the particular
case per the Supreme Court of Canada in R. v. Lyttle, [2004] 1 S.C.R.
193 at paragraph 65.
[136]
The rule in Browne
v. Dunn may understandably have different practical application in the
particular case of material witness on a question of fact, than in the
particular case of an expert witness on his or her opinion evidence and
information supporting that opinion, and on the contents of the expert’s
rebuttal report responding to the other party’s expert reports. In this regard,
I would add that Ms. Hooper’s testimony relating to the TDSI Report was
effectively testimony regarding her professional opinions within her area of
expertise, and I would note that the opinions, reasoning and supporting information
set out in the TDSI Report were the subject of considerable comment in the principal
and rebuttal reports of the Respondent’s experts.
[137]
At the outset of the
hearing it was agreed that each expert would testify as to the contents of both
their principal report and their rebuttal report when called, notwithstanding
that the Appellant’s rebuttal reports responded to principal expert reports of
the Respondent’s experts before the Respondent’s experts had a chance to
explain them to the Court. There was a right to recall any witness in the event
this approach was considered to be found wanting on any point.
[138]
There could have been
no doubt in the Appellant’s mind, nor the minds of its experts, that the
Respondent’s experts challenged the opinions and reasoning of the Appellant’s
experts and the TDSI Report.
[139]
In these circumstances
of this case, there was absolutely no element of attempted surprise, much less
any attempt to deny anyone an opportunity to fully explain. This was fully covered
off by an opportunity to recall any witness if a party felt further explanation
was needed. This was a standing offer that the Appellant chose not to take
advantage of, even after the offer was re-extended as part of denying the
objections.
[140]
What was in issue in
the circumstances of each of the objections raised in this particular case
related to the witness’ professional opinions without any suggestion or
inference that he or she was not speaking the truth or was a witness unworthy
of credit. This was in reality only a matter of competing and differing professional
opinions.
[141]
This approach to the
resolution of these objections finds support in R. v. Union Carbide
Canada Ltd., [1991] O.J. 1213 (Ont. Court of Justice).
[142]
An attempt to force
blind adherence to the rule in Browne v. Dunn should not be allowed to
interfere with an orderly ordering of witnesses. Nor should it require that
further rebuttal reports be filed in response to rebuttal reports when the mischief
addressed by the rule is not present in the particular circumstances.
(g) The Court’s Analytical Approach to be
Followed in this Case
[143]
The Court’s analysis of
the pricing of the Discount Rate in the RSA will proceed on the basis that
transactions described above are the transactions McKesson Canada and others in the McKesson Group chose to enter into. The real task in this transfer
pricing appeal is for the Court to determine whether or not the terms and
conditions of the transactions resulted in a Discount Rate committed to in the
RSA by McKesson Canada that was within the range of what McKesson Canada and
MIH would have agreed to had their transactions used terms and conditions that
affect the pricing of the Discount Rate which persons dealing at arm’s length
would have used.
6. The Position of the Appellant
[144]
The Appellant’s
position with respect to the transfer pricing adjustment can be generally
summarized by me as follows:
Methodology:
[145] The appropriate transfer pricing analysis
methodology is not one of the four methodologies named in the OECD Guidelines
and, therefore, an “other method” should be used. The taxpayer called an expert
witness, Dr. Horst Frisch of Horst Frisch, in support of this view.
[146]
Based upon the evidence
presented, I agree. (I am not sure that the Respondent substantially disagreed).
An OECD “other method”:
[147]
An “other method” that
would be appropriate is precisely that followed by Barbara Hooper of TDSI, set
out in the TDSI Report, and explained in her testimony at trial. This
analysis takes the RSA as the parties structured it and does not introduce or
rely on any recharacterization or additions to the RSA. She expertly and
accurately identified the risks transferred to MIH with its purchase of the
receivables under the RSA. The TDSI analysis and report was contemporaneous.
While Ms. Hooper did not testify as an expert witness, she and TDSI clearly had
considerable expertise in trade receivables transfers in a securitization
setting. That allowed an accurate identification of the risks inherent in the
terms and conditions of the RSA. She was able to use her experience and
knowledge in reviewing and commenting upon the pricing of the Discount Rate in
accordance with the terms and conditions of the RSA.
[148]
I agree that this is an
appropriate “other method” to be considered in reviewing the terms and
conditions of the RSA to determine whether those that impact on pricing under
the RSA are on arm’s length terms.
Another OECD “other method”:
[149]
Another “other method”
that would be appropriate was that set out in the expert report of Mr. Jeremy Reifsnyder.
This approach did not follow the parties’ chosen structure set out in the RSA. This
approach 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 that to reflect certain issues and characteristics particular to
the RSA.It is fair to say that the use of this “other method” was the Appellant’s
principal position.
[150]
The Reifsnyder
methodology is described in greater detail below along with my reasons for
finding it would not be appropriate to follow it as an acceptable “other method”
in this case. That is not to say that parts of Mr. Reifsnyder’s analysis and
the opinions that he expressed, and information that he relied on, were not
helpful to the Court. Indeed, as described below, apart from (i) his build up
being based upon an index I find far from acceptably comparable or appropriate,
and (ii) the magnitude of a number of his adjustments, I found his testimony
and his differing “other” approach to be at times helpful and informative.
[151]
The Appellant’s only
witnesses were Mr. Brennan, the Vice-President of Taxes of McKesson U.S. (and also on the Board of MIH and MIH2), Ms. Hooper of TDSI, and its experts Dr.
Frisch and Mr. Reifsnyder. No one was called from McKesson Canada.
[152]
While Dr. Frisch suggested
that he believed that arm’s length parties did non-recourse receivables
purchase transactions on comparable terms and conditions and Mr. Reifsnyder
suggested there was no reason why arm’s length persons could not do such
transactions, no further evidence was given or called by the Appellant to
substantiate that. Absent any supporting evidence, I am unable to give any
weight or relevant consideration to the suggestions that there were such
comparable arm’s length transactions being done.
[153]
Also, as Appellant’s
counsel stressed and as is clear from the RSA and all of the evidence, one of
the largest and most significant benefits to McKesson Canada of its receivable
sales to MIH other than the cash purchase price rights, was the
transfer of the risk of loss on the receivables related to the Obligors’ credit
and finances. According to the Appellant’s Objection to the reassessment,
McKesson Canada effectively bought insurance against these risks by having them
assumed by MIH as purchaser. There is credit risk or 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.
However, the Court was given no evidence to be considered by it of the
practical availability (or non-availability) or effectiveness, or pricing/costs
for such. I found it somewhat surprising that neither side tendered any such evidence.
I certainly think such evidence may have provided the Court with other helpful
information with which to try to price the risk transference with information
beyond McKesson Canada’s available historic loss and performance and
projections for its receivables portfolio, and risk spreads in the public bond
market.
7. The Position of the Respondent
[154]
It is fair to say that
the position of the Respondent shifted a fair bit through the trial. After the
close of evidence and hearing the Appellant’s argument, the Crown conceded that
the sale of any particular receivable under the RSA was non-recourse to
McKesson Canada (the only exception being the limited right described above for
MIH to require McKesson Canada to repurchase the receivable, initially at 75%
of its face amount but ultimately only for what it was able to collect).
[155]
The Crown recognized
that all of the other risk mitigation factors in the RSA in favour of MIH only
allowed MIH to stop purchasing more receivables, or to recalculate the Loss
Discount, on short notice in the event of expected or continued declining or
deteriorating quality of the receivables, the Obligors, McKesson Canada or the McKesson
Group, or material adverse change which could include market changes generally.
There was no McKesson Canada covenant or assets supporting the collectibility
of any receivable after it was transferred to MIH. This was recognized as
fundamentally different than either a securitization or a secured loan. The
Crown therefore resiled from its positions that would introduce structural
reserves into the notional arm’s length RSA which were advocated, at least in
the alternative, by two of its three experts. I think this was a wise decision.
[156]
In argument, the
Respondent’s position with respect to the transfer pricing adjustment can be
generally summarized by me as follows:
[157]
It is possible and
appropriate to make adjustments to the Discount Rate pricing and terms and
conditions in the RSA to reflect the terms and conditions that arm’s length
parties would have agreed to, both by adjusting amounts and by adjusting the
method or formula by which such amounts are directly determined or
re-determined.
[158]
If the Court is
concerned that the adjustment of a formula or criterion (e.g. fixed versus
floating, yearly versus monthly et cetera) may not clearly be permitted under
the language of paragraphs 247(2)(a) and (c), 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. Such an adjustment would be a permitted adjustment
under paragraphs 247(2)(a) and (c) and would not be a recharacterization
of the RSA transaction itself described in 247(2)(b) and (d).
[159]
The appropriate
adjustments are all to the risk factor mitigation issues and provisions described
in the TDSI Report. The non-arm’s length terms and conditions can be identified,
and the appropriate adjustments can be determined, having regard to the
evidence from the experts in this trial and from TDSI (including Ms. Hooper’s
evidence).
[160]
The Respondent called
three expert witnesses to testify: Dr. Brian Becker, Mr. Joel Finard and Mr.
Myron Glucksman. Each of these experts filed expert reports as well as rebuttal
reports to the other side’s expert reports. Additional expert reports were
filed by the Respondent with the Appellant’s consent and without viva voce testimony.
[161]
Dr. Becker first used a
risk identification and assessment “build up approach” similar to those of TDSI
and PwC generally following the structure of the RSA pricing provisions.
[162]
Dr. Becker also used an
alternate comparable transaction approach focused entirely on McKesson Canada’s prior receivables factoring transaction with TD Factors. As described below, I do
not accept that there is any helpful comparability between the RSA transactions
and the prior TD Factors year end capital tax driven factoring.
[163]
Mr. Finard’s first
approach was to use an “attribute analysis” identifying the risks inherent in
the RSA from a credit and operations perspective and trying to quantify
appropriate arm’s length pricing for each. As part of this approach, Mr. Finard
sought to introduce a cash reserve to further protect against loss and thereby
reduce the credit risk portion of the Discount Rate. While such a reserve would
reduce the risk of loss to MIH on the transactions, it would do so by reducing
the risk of loss on receivables after they were purchased by MIH which simply
was not the deal in the RSA. As stated below, I am not sure that such is
appropriate as a comparability matter, nor, without deciding it, am I sure that
would be a permitted notional arm’s length term or condition to support an
adjustment under paragraphs 247(2)(a) and (c).
[164]
Mr. Finard describes
his second alternate approach as a “structured finance analysis”. In this
approach Mr. Finard priced the credit risk portion of the Discount Rate by
comparing McKesson Canada’s long-term historic loss experience on its
receivables portfolio to rating agencies’ published loss rates by company debt
rating, and used the public rating of companies whose public debt had a similar
historic loss experience to identify the public debt market’s credit spread for
a company with such a rating. I found this to be helpful and informative,
recognizing, as with Mr. Reifsnyder’s approach, that public markets are only
one particular market.
To the extent there is comparability for the RSA transactions in the public
debt market, Mr. Finard’s approach and identification of a rating somewhere
between A and Baa makes much more sense than Mr. Reifsnyder’s approach using
non-investment grade/high yield/junk bond rating status as a starting comparability
point.
[165]
Lastly, Mr. Finard went
through TDSI’s analysis, factor by factor.
[166]
The Respondent’s third
expert witness, Mr. Glucksman, went through the credit risks identified in the
TDSI reports and commented on their approach to pricing each of these as a
component of the appropriate Discount Rate under the RSA. However, Mr.
Glucksman’s approach dealt with the credit risk of the receivables’ Obligors
differently than provided for in the RSA. He instead incorporated reserves into
his arm’s length notional comparable transaction terms and conditions as a
deferred purchase price which would be paid to the seller only as the
transferred receivables performed. He then estimated the cost (operationally
and time value of money et cetera) to the parties of maintaining such reserves.
While it may be acceptable in paragraph 247(2)(a) and (c)
adjustment of amounts reflecting arm’s length terms to look at pricing/cost of
an alternative work-around or input for comparison purposes, the Glucksman reserves
would have functioned very much like a reserve or hold-back or
over-collateralization in a securitization transaction, or like the security on
a secured loan, and leave the credit risk effectively with the seller, thereby
making it effectively a recourse transaction. I have the same concerns with Mr.
Glucksman’s reserves as I do with Mr. Finard’s reserve. I am able to fully
dispose of the appeal without needing to decide to what extent, if any, the
costs associated with such reserves, could help in a subparagraph 247(2)(a)
and (c) analysis.
[167]
The Crown did not
directly or indirectly raise any fair share or fiscal morality arguments that
are currently trendy in international tax circles. It wisely stuck strictly to
the tax fundamentals: the relevant provisions of the legislation and the
evidence relevant thereto. Issues of fiscal morality and fair share are surely
the realm of Parliament.
8. The Witnesses, the Expert Reports and the
PwC Report
(a) Mr. Brennan
[168]
Mr. Brennan testified
on behalf of the Appellant, McKesson Canada. Mr. Brennan was the Vice-President
of Taxes at McKesson U.S. throughout the relevant time. He was
also a director of MIH, the Luxembourg parent of McKesson Canada and of MIH2. He was neither an officer nor director of McKesson Canada. As mentioned above,
no one from McKesson Canada testified at the hearing notwithstanding that it
had a finance department and a large credit and collections department, each
headed by accountants. No reason was given by the Appellant for the CFO and VP
of Finance of McKesson Canada not testifying in this matter other than that his
role was really only to provide financial information to TDSI to prepare its
report. This was also the case for the head of Credit and Financial Services at
McKesson Canada.
[169]
Within the McKesson
Group, corporate finance decisions were made by McKesson U.S. and dictated to McKesson Canada. All of the available evidence is that all of the relevant
transactions were decided upon without any material input from McKesson Canada and solely by McKesson U.S., the 100% controlling shareholder of both parties to the RSA. The
evidence is that these decisions were all made by the Treasury and Tax
functions at McKesson U.S. in consultation with Blakes and TDSI.
[170]
The McKesson U.S. Tax
department is responsible for tax compliance, (audits, filings and financial
reporting) and for tax planning. The Tax department has been a rapidly growing
group at McKesson U.S., having increased more than tenfold to 63 persons since
Mr. Brennan took over the reigns in 2000.
[171]
Mr. Brennan described
the $173,000,000 “double dip” ULC financing structure previously used since 1998
to finance McKesson Group’s Canadian holdings on a very tax efficient basis,
effectively allowing both McKesson Canada and McKesson U.S. to deduct the same interest amount in both Canada and the U.S. for tax purposes. This double
dip structure was paid out with a portion of the $460,000,000 received by
McKesson Canada under the RSA in December 2002 from the initial receivables
purchase.
[172]
Mr. Brennan also
described the McKesson Group’s use of the Irish structure to hold all of
McKesson U.S.’ non U.S. operations. He described this as very, very common
Microsoft and Apple type international tax planning that relies upon the
favourable tax treaties of countries such as the Netherlands and Luxembourg. It permits large multi-nationals like McKesson U.S. and its McKesson Group to
amass large amounts of cash in Ireland by restructuring to avoid taxes that
otherwise would have been paid in other countries.
[173]
MIH is a Luxembourg company that holds the shares of McKesson Canada and buys McKesson Canada’s receivables. In addition to Mr. Brennan, MIH’s board includes McKesson U.S.’ in house counsel as well as in house counsel of a McKesson Group UK company. In the year in
question, MIH had a single employee. It does not appear from Mr. Brennan’s
evidence that MIH had a second employee or leased any premises in Luxembourg until some time after the years in question.
[174]
Mr. Brennan testified
that the McKesson Canada DSO was in the 30 to 31 day range at the time of the
RSA and remained at around 30 days.
[175]
According to Mr.
Brennan, MIH would look at the actual aging of McKesson Canada’s receivables regularly. MIH would concentrate on the Designated Obligors because “that’s
really where risk of this company is at”. MIH would look at defaults of
receivables and would decide whether or not it would exercise its rights under
the RSA to sell them back to McKesson Canada as described above. It is not
clear why MIH would not always put defaulted receivables to McKesson Canada; it does not appear to be consistent with arm’s length behaviour.
[176]
Mr. Brennan testified
that at the time of the RSA, McKesson Canada had access to US$100,000,000 to
$150,000,000 of a US$550,000,000 McKesson U.S. one year facility. McKesson U.S. also had access to another US$550,000,000 standard credit facility which apparently could have
been used by McKesson U.S. to finance McKesson Canada. McKesson U.S. also had access to a one year securitization program, which renewed annually, and could also
have been used to finance McKesson Canada. These were not so used because it
would not make sense and would be inefficient to borrow money at interest from
third parties when all this surplus cash had been amassed by McKesson Group in Ireland for this very purpose. Later in his testimony, Mr. Brennan added that withholding
tax on interest would also cause further inefficiency.
[177]
Mr. Brennan’s evidence
emphasized the significantly increased cash flow to McKesson Canada from the RSA. As noted previously however, since the McKesson Canada receivables
pool had a DSO in the 30 day range, and the RSA settlement period was also
about 28 days, the cash flow impact would only arise upon and from the initial
December 2012 sale of the pool partway through a settlement period and only in
respect of a portion of that amount. Otherwise, McKesson Canada largely continued to receive cash on a similar schedule throughout each roughly 30
day period thereafter, that is, as its customers paid their bills. This largely
one-time pick up would not be insignificant, and would effectively be permanent
for the remainder of the term.
[178]
Mr. Brennan described
how he did a short, quick calculation of the net tax benefit to McKesson Group
of implementing the RSA and Servicing Agreement. The hand written calculation
he had prepared for McKesson U.S.’ CFO was put into evidence. It showed that
there would be a Canadian tax reduction of about US$4.5 million in the short
year ending March 2003, Luxembourg tax of US$29,000 for the year ending March
2003, and costs associated with maintaining MIH of US$300,000 of bank charges
and US$35,000 for accounting fees. The net tax benefit to McKesson Group was
over US$4.1 million in the three and one-half months remaining of McKesson Canada’s 2003 year and about US$15,000,000 annually thereafter. These numbers assumed only US$250,000,000
of McKesson Canada receivables were generated monthly. The net Canadian tax
savings would be corresponding larger since C$460,000,000 of receivables were
sold in December 2002.
It appears this analysis was not shared in any way with the CFO of McKesson Canada.
[179]
Mr. Brennan did not
read the TDSI Report before closing the RSA transactions. He is not sure he saw
the TDSI engagement letter. He simply got their Discount Rate number and used
that to do his tax savings calculations. Mr. Brennan testified that McKesson US
had looked at financing alternatives internally but just did not ever write
them down. He said there was no internal review done of TDSI’s numbers, nor did
McKesson Group do any sensitivity analysis to assess the reasonableness of any
alternative discount rate.
[180]
Mr. Brennan could not
recall why the RSA had a five-year term. He did say it was on the advice of
TDSI. This is not supported by the TDSI Report or Mrs. Hooper’s evidence. He
could no longer recall why the Servicing Agreement set an annual fee for the
servicer of $9,600,000.
[181]
Mr. Brennan spoke
several other times in his testimony about the role of TDSI in structuring the
transaction. He said TDSI reviewed McKesson Canada’s credit and collection
policies. Ms. Hooper testified afterwards that it had not; nothing in the TDSI
Report suggests it did. He also described TDSI as having been retained to tell
McKesson U.S. what the RSA Discount Rate should be as they did not have the
banking expertise to do this in-house. Not only does this appear to be an
overstatement of Ms. Hooper’s testimony and the language of the TDSI Report, it
also appears odd that McKesson U.S. not having the relevant expertise, turned
to TDSI who stressed in their reports and in Ms. Hooper’s testimony that the
RSA was also outside its expertise. Further, Mr. Brennan said it was TDSI that
recommended the RSA be a $900,000,000 facility even though that significantly
exceeded the amount of receivables at the time. This is also unsupported by the
TDSI reports and inconsistent with Ms. Hooper’s testimony. I accept Ms.
Hooper’s testimony in all of these regards. She was much less self-interested and
displayed more candour in her overall testimony. Her version was consistent
with the TDSI reports.
[182]
I conclude that Mr.
Brennan’s memory on these key aspects of these transactions has proven, for
whatever reason, faulty or lost. This does not make my task any easier, especially
since he was the only witness from the McKesson Group.
[183]
Mr. Morgan of Horst
Frisch advised Mr. Brennan in the summer of 2002 that such an RSA transaction
would be so unique that he recommended that an expert in the area should be
retained for a transfer pricing study. This appears to have resulted in Ms. Hooper
at TDSI’s role and the TDSI reports. However, Mr. Brennan testified in cross
examination that neither McKesson U.S., nor McKesson Canada sought a formal
transfer pricing study.
(b) Ms. Hooper and the
TDSI Reports
[184]
I have already fully
described the TDSI reports.
[185]
Ms. Hooper described
the original TDSI retainer as a one-off situation, that she had never done
something like this previously. She had never previously priced nor been
involved in pricing risks in receivables transactions because in
securitizations risks are avoided not purchased. She said the same was true of
the TDSI Servicing Agreement report, that she had no prior experience doing
such an analysis. She also acknowledged she had no experience with traditional
factoring of receivables in a non-securitization environment.
[186]
In her testimony, Ms.
Hooper identified the material transaction risks as (i) dilutions, including
prompt payment discounts (ii) losses – primarily credit losses and (iii)
servicing risk, that the servicer collects but does not remit to the buyer.
[187]
Ms. Hooper clarified in
evidence that, while the TDSI Report says that where possible it looked at
pricing of comparable risks in the market, TDSI did not really do that beyond
obtaining credit spread numbers from their bond traders. It appeared from the
TDSI reports and Ms. Hooper’s overall testimony that this may have been because
of her being a securitizations expert entirely unfamiliar with the pricing of
such risks in the market. To the extent that is the case, one can assume that
the McKesson Group and its advisors were aware of that.
[188]
She explained that the
DSOs used in the TDSI Report were computed by TDSI from historical data
received from McKesson Canada. TDSI’s DSO number simply divides the receivables
balance at the end of a McKesson Canada Accounting Period by the sales in that
period, and multiplies that by 30.
DSO numbers serve as a proxy for accounting purposes for how long receivables
can be expected to take before payment in full. However, a DSO number simply
compares the receivables amount at the end of the period with the sales during
that period. Thus, it does not actually reflect the time it takes for any
particular receivable, nor the receivables pool on average, to be paid. The DSO
will by definition increase if sales go down in a period and will decrease if
sales go up in a period, even if nothing in fact changes with respect to
payments on receivables in that period. No one questions that it is nonetheless
an appropriate and satisfactory proxy.
[189]
Ms. Hooper also
explained that the historic average DSO would not have been a good proxy for
the initial receivables pool transferred in December 2002, as these were not
transferred on the day receivables originated as would be the case thereafter
under the RSA. To address this, TDSI instead estimated it would take half as
long to collect the initial mature receivables pool and spread the “missing” 16
days over the five-year term. Neither the TDSI Report nor Ms. Hooper’s
evidence acknowledge or discuss that this effectively overstates the Discount
Rate in MIH’s favour and that the underpayment is only recouped by
McKesson Canada over five years without interest. It can also be noted this
gives McKesson Group a five year timing benefit in its Canadian income taxes.
[190]
With respect to the 20%
buffer added by TDSI to the historic prompt payment discount multi-year average
numbers, Ms. Hooper could not recall or explain how that 20% number was arrived
at. She speculated that TDSI may have just looked at the possibility that 100%
of Obligors would exercise their prompt payment discount rights. Again, the
fact an increase in prompt payment could be expected to have a significant
impact on DSO calculations was overlooked.
[191]
Ms. Hooper confirmed in
her testimony that she did not have any experience calculating loss discounts
in her securitizations area of expertise either. This is because, in
securitizations, credit loss risk is addressed through structuring credit enhancements
and other risk mitigators, not through price.
[192]
With respect to the
TDSI Report’s statement that the historic information provided on write-offs to
sales in considering the Loss Discount did not cover a full Canadian economic
cycle, it does not appear from her testimony that TDSI asked McKesson Group for
further data nor was told it was unavailable.
[193]
Ms. Hooper had
difficulty, even in direct examination, giving a satisfactory, responsive,
logical or complete explanation of the interest discount portion of the
Discount Spread relating to MIH’s cost of funds given it was a thinly
capitalized company. With respect to MIH’s cost of funds interest discount, she
testified that TDSI had simply assumed MIH was 100% debt financed even though
their information was that it was thinly capitalized. She has no recollection
the McKesson Group ever told her of the source or terms of MIH’s funds used to
purchase the McKesson Canada receivables.
[194]
Ms. Hooper added that
the delinquent portfolio performance trigger serves as an early warning system.
Typically with receivables one will see delinquencies increase in advance of
seeing losses increase. For this reason, she explained one wants the trigger
rate to permit termination of the agreement early enough for there to not be
material losses. She clearly understood that when the transaction could be
terminated would have a fairly significant affect on the overall risk that was
being transferred. According to Ms. Hooper, portfolio performance triggers,
delinquency and default rate triggers, were designed to limit the ultimate
losses to the purchaser by ceasing the acquisition of new receivables that
might not be expected to perform as well as receivables originated previously.
[195]
Ms. Hooper explained in
her testimony that in a typical securitization transaction one uses historical
performance to try to get comfortable with how performance might be expected to
unfold in the future.
[196]
Ms. Hooper said in her
testimony with respect to the servicing discount that, while the likelihood of
the delinquency or default termination ratios being triggered can be assessed
relative to historical performance, it is more difficult to quantify in any
reliable fashion the likelihood of other termination events occurring. She
could not explain how TDSI got from the 9.41% credit rating migration risk
number of a potential downgrade of McKesson U.S.’s credit rating (which would
give rise to a termination event) to the report’s 25% total chance of a
termination event occurring. Nor did she explain why the TDSI Report assumed
that MIH would exercise its rights to terminate 100% of the time a termination
event occurred no matter what the circumstances.
[197]
Ms. Hooper believed
TDSI continued to be involved, after the two TDSI reports in evidence, in assisting
with the annual review of the calculation of the RSA Discount Rate required by
the RSA.
[198]
Ms. Hooper could not
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. She could only repeat that TDSI chose the maximum number to be
conservative as mentioned in its report. She added that the accrued rebate
dilutions discount did not make up a very significant portion of TDSI’s overall
Discount Rate and, perhaps if it had been more significant, they might have
considered whether average rebate numbers would have been more appropriate.
[199]
The TDSI Report was
qualified with soft opinion language. That indicates to me that it was written
by TDSI and understood by McKesson Group to be primarily advocacy. The
shortcomings in the report and in her testimony described above served to
confirm that.
(c)
The PricewaterhouseCoopers Report
[200] McKesson Canada and its advisors had
PricewaterhouseCoopers (“PwC”) prepare a transfer pricing report relating to
the RSA in December 2005. The PwC Report was used to respond to CRA’s review of
the RSA transactions.
[201]
PwC followed much the
same approach as TDSI but without any supportive testimony at trial. I can give
it little weight except to the extent it supports the use of the TDSI approach.
I do find some of the data/information relied on or referred to in the PwC
Report to be corroborative of, or complementary to, some of the numbers/ranges/approaches/observations
used by others. I have identified below where I considered any such information
useful.
[202]
It is clear from the PwC
Report that it considered factoring receivables to a commercial financial
market player could be a potential comparable. It said that PwC had looked at
the range of factoring yields of third party factoring companies to support its
view that there should properly be a range for an arm’s length Discount Rate.
Later, it looked to a range of participants in the North American factoring
market to support its view that, as a highly leveraged entity, MIH’s cost of
funds should be borne by McKesson Canada as discussed further below. Yet, the
PwC Report did not develop the concept of arm’s length factoring and the
pricing or other terms and conditions thereof as a comparable transaction to
the RSA, nor did it address why that would not be possible or appropriate. This
is a significant shortcoming and causes one to think that the PwC Report was
primarily a piece of advocacy work, perhaps largely made as instructed.
[203]
The PwC Report notes
that the RSA fixes a DSO number instead of using a dynamic floating DSO. As
part of this, PwC looks to the historical actual performance of McKesson Canada’s pool of receivables to inform its views. The report specifically notes that, with
respect to the DSO number in the RSA, the interests and exposure of both
McKesson Canada and MIH need to be considered and balanced. However, PwC’s
conclusions are then limited to MIH’s potential exposure to adverse actual
performance of the DSO and PwC has “taken the position” and “assumed” a 20%
cushion should pad the fixed DSO number to incorporate a 38 day DSO. PwC did
not address in any way the flipside of the coin – McKesson Canada’s risk in fixing the DSO at any number for the term that its DSO actually continued to
improve. Nor did PwC address why incorporating a variable dynamic DSO which
would almost fully protect both MIH and McKesson Canada would not be more
appropriate than simply increasing the amount of the already fixed DSO.
[204]
By adding a 20% cushion
to the fixed DSO, the PwC approach makes MIH favourable upwards adjustments to
all of the factors comprising the RSA Discount Rate that are determined by
reference to annual rates, including the RSA’s risk free 30 day CDOR baseline
Yield Rate, by 20%.
[205]
It can be noted that PwC
looked at one year, not five year, securities when looking at the Obligor Loss
Discount factor. This suggests PwC did not equate the RSA’s five-year term with
the risk associated with the five-year term financings for credit risk
purposes.
[206]
The PwC Report refers
to debt issues with a below investment grade rating as junk bonds.
[207]
The PwC Report also
looked at the actual historic loss performance of the McKesson Canada
receivables pool as part of its analysis. However, it did not do so as a starting
point to projecting the credit loss risk associated with the pool. It assessed
credit loss risk by first assigning a Loss Discount attributable to the
Designated Obligors equal to the credit spread on public bonds issued by what
they believed were similarly rated corporate issuers in the public bond market.
This they then adjusted for their cushioned DSO. The PwC Report did not try to
reconcile its chosen approach to the actual known performance of the
receivables pool.
[208]
However, in dealing
with the other Obligors credit risk, PwC did analyze the receivables pools’
historic performance, but only to demonstrate that the other Obligors’
receivables historically had losses exceeding that of the Designated Obligors.
PwC then used this to justify notching all of the others Obligors’ credit risk
down two grades below the public debt rating PwC had assigned to the Designated
Obligors (before also adjusting it for their cushioned DSO). This picking and
choosing, mix and match as it suits approach to the relevance of the actual
performance of the receivables pool makes for transparently poor advocacy, and
even more questionable valuation opinions.
[209]
The TDSI Report had assumed
a 25% chance that a replacement servicer would need to be appointed when it
addressed the servicing discount portion of the Discount Spread. TDSI relied
upon Moody’s one year number for a potential downgrade of McKesson U.S. occurring (which would be a termination event). PwC raised this to a 40% probability
relying upon Standard & Poor’s five year downgrade number. As with
the TDSI Report, the PwC Report did not address why a change of servicer would
be a requisite necessity upon the occurrence of such a termination event. The
RSA and Servicing Agreement do not terminate McKesson Canada’s appointment as servicer upon a termination event. They merely give MIH the right
to appoint a new servicer if it chooses. Given that (i) the particular
termination event of a credit rating downgrade of McKesson U.S., which both
TDSI and PwC rely upon in support of their 25% and 40% numbers, would not
necessarily involve any change to McKesson Canada’s ability to service its
receivables, and (ii) given that MIH had the right in the agreements to end
commingling and require segregation of funds, thereby effectively removing
McKesson Canada credit risk, these probability numbers remain inadequately
explained, questionable and, to my mind, unreliable advocacy or posturing.
[210]
It can be noted that PwC’s
replacement servicer fees were in the range of 0.79% to 1.23% of the face
amount of receivables. The midpoint of this range is 1.01%.
[211]
It should also be noted
that PwC looked at and relied upon McKesson Canada’s four year historic prompt
payment discount when considering the prompt payment dilutions discount
component of the Discount Spread. While TDSI added an unexplained 20% buffer to
the historic actual 0.5% experience of McKesson Canada’s receivables pool, PwC
“assumed” that an arm’s length purchaser would want a 5% cushion. This
assumption was similarly unexplained and, tellingly, is an upwards adjustment
that only addresses MIH’s risk that more Obligors pay more quickly. It does not
address McKesson Canada’s risk that fewer Obligors take advantage of their
prompt payment discount. Nor did PwC account for the impact that more early
payments from Obligors would have upon the DSO, and which would also benefit
only MIH.
[212]
At this point in its report,
PwC summarized and arrived at a range of arm’s length Discount Rates for the
RSA of 1.3179 to 1.4823, still quite a distance from the rate agreed to in the
RSA between McKesson Canada and its parent MIH, and that in the TDSI Report –
indeed much closer to that used by CRA in the reassessment.
[213]
However, the PwC Report
then went on to add a further factor to that range that is essentially TDSI’s
interest discount, which effectively imposes on McKesson Canada the full amount of an assumed cost of funds of MIH. PwC does not address the fact that MIH is
borrowing all of the money at a cost of funds determined by reference to the
RSA Discount Rate. PwC assigns a credit spread based upon public debt issuers
which are 90% capitalized with debt, which are below investment grade rated
companies by Standard & Poor’s. This factor alone then drives the PwC range
for an arm’s length discount rate from the above numbers to a range of 1.9698
to 2.2646.
[214]
PwC’s sole expressed
reason for thinking that MIH’s cost of funds would be agreed to be borne by a
notional arm’s length McKesson Canada was because five-year terms are not
generally agreed to by large participants in the factoring market and that this
premium to the Discount Rate could therefore be demanded by MIH. PwC ignored
the presence and availability of large well-capitalized financial institutions
in the factoring market that it had already described in its report. PwC looked
instead at their smaller and much weaker, poorly capitalized competitors as
though McKesson Canada was driven by market forces to use one of these latter
players. McKesson Canada probably did have to use MIH, but only because of
McKesson Group’s control, not because of any market, economic or financial
reasons that are in evidence. There was no evidence that McKesson Canada or
McKesson Group was even interested in considering factoring its receivables to
any arm’s length financial institution player in factoring markets, presumably
because profits would then have left the McKesson Group.
[215]
It should also be noted
that PwC could only get this factor so high by assuming that MIH, the poorly
capitalized player, should also nonetheless command the level of premium
yields, net of its high cost of funds, realized by the well-capitalized large
players described in the PwC Report. The PwC Report numbers were not reflective
of the lesser yields earned by poorly capitalized players as they themselves
described the market. More unsupported, selective picking and choosing.
(d) The Frisch Expert
Report
[216]
Dr. Frisch has a PhD in
Economics from Harvard University. He is an expert in transfer pricing
methodologies. In his opinion, none of the named, recognized OECD Guidelines’
methodologies could be applied in a reliable manner to the RSA transactions. He
was of the opinion that an “other methodology” needed to be developed, as contemplated
by paragraph 2.9 of the 2010 OECD Guidelines. He volunteered that a transfer
pricing economist such as he did not have the expertise to develop such an
“other method”.
[217]
Based upon what
evidence the parties chose to put before me, I agree with Dr. Frisch’s opinion that
the named OECD methods are not appropriate to rely upon in this case, and that
an “other method” should be used.
[218]
I am surprised that Dr.
Frisch thought that developing or opining on an “other method” was outside the
expertise of expert transfer pricing economists. Clearly, the Respondent’s
expert, Dr. Becker, did not share the same reticence.
(e) The Reifsnyder
Expert Report
[219]
The Reifsnyder Report,
after downward corrections and amendments made in the course of trial, opines
that arm’s length parties “would have agreed to” a discount rate to be applied
to the face amount of the receivables in the range of 1.7326% to 2.4360%.
[220]
Mr. Reifsnyder’s
approach was to consider, estimate and add up four factors:
(i) Factors One and Two – Servicing Fees
and Prompt Payment Discounts
[221]
Mr. Reifsnyder began by
determining the anticipated cash flows to the purchaser in the transactions. In
addition to the cash flows expected by the purchaser from the face amount of
the receivables, he identified his first factor as the fixed monthly outbound
servicing fee payable under the Servicing Agreement, and his second factor as
the cash flow reductions that could be expected from prompt payment discounts
enjoyed by McKesson Canada’s customers. With respect to prompt payment
discounts he used McKesson Canada’s historic three-year average prompt payment
discount numbers. He did not see the need to be concerned with the risk of
change since, in his experience, early payment discounts were not usually
volatile in respect of trade receivables. In his experience, it was not
uncommon for prompt payment discounts to be borne by a buyer in a receivables
purchase transaction.
He does not identify any cash flow reduction potential from McKesson Canada’s customers setting-off rebate entitlements against the receivables.
(ii) Factor Three – Credit Risk
[222]
Mr. Reifsnyder then
developed a third factor to reflect the credit risk on the receivables portfolio.
This third factor focused on the credit risk associated with the Obligors, not
McKesson Canada.
[223]
His report accepts the
Yield Rate component of the Discount Rate in the RSA, being the CDOR 30 day BA
rate as the appropriate baseline risk-free rate of return to which an
appropriate spread should be added.
[224]
He next determined the
appropriate credit risk spread by starting with the spread on a particular
Canadian bond index fund that he viewed as sufficiently comparable, provided a
number of upwards and downwards adjustments were made to it, to reflect what he
thought an arm’s length potential purchaser would seek if presented with the
RSA.
[225]
Recognizing that the
RSA was a five-year agreement, dealing with Canadian dollar denominated
receivables, with some large concentrations in the pool with the Designated
Obligors, with some security but no credit enhancements, and with the majority by
value of Obligors not being rated, Mr. Reifsnyder selected a particular
Canadian high yield bond fund index. He described a bond fund index as an
aggregation of financial assets that share common and similar characteristics.
[226]
The credit spread on
his chosen Merrill Lynch Canadian Dollar High Yield Index for the six reported
months prior to the RSA reflected a spread of over 13% per annum (and almost
14% per annum for the last reported month) above the rates of return on risk-free
Canadian Treasury obligations. As of November 2002, that is when the RSA was
being priced, this index was composed of 26 bonds issued by 18 issuers, each of
which was rated below investment grade – that is the index was comprised of
high yield bonds also referred to as junk bonds. The minimum bond amount was
$50,000,000 and the minimum term to maturity was one year. A number of the
individual bonds making up this index were trading at discounts in excess of
50% and one was discounted 98%. A number had effective yields in excess of 30%,
the highest being 499%.
[227]
For the reasons
detailed below, I do not accept that the Merrill Lynch Canadian Dollar High
Yield Index was an appropriate starting point.
[228]
The Reifsnyder Report
then made a number of upward and downward adjustments to the High Yield Bond Fund
Index spread to reflect the particulars of the McKesson Canada receivables pool
as compared to the bonds making up the bond fund index.
[229]
The first adjustment
was to reflect the greater single industry concentration and correlation risk
in the McKesson Canada receivables pool. This was an upwards adjustment to the
bond fund index spread of 15% to 20%.
[230]
The second adjustment
was to reflect the greater proportion of small business Obligors in McKesson Canada’s receivables pool compared with the small number of public companies in the bond
fund index. A further upwards adjustment of 10% to 15% was made to the spread.
There was no discussion of whether small going concern businesses could really
be considered riskier than larger ones that, in the case of this index, had to
be rated junk status to be in the index.
[231]
The third adjustment reflected
that a portion of the Obligors in the McKesson Canada receivables pool were
large investment grade rated companies, and others were hospitals. Mr.
Reifsnyder made a McKesson Canada favourable downwards adjustment to the bond
index spread of 15%, roughly the approximate makeup of such Obligors in the
pool when the RSA was entered into.
[232]
The fourth adjustment
reflected the fact that the Obligors in the pool could change throughout and
that there was no limit in the RSA on the concentration risk of the Designated
Obligors. In his opinion, this contrasted unfavourably with the bond fund index
and warranted a further upwards adjustment of 20% to 40%.
[233]
The fifth adjustment of
5% to 15% upwards reflected the fact that the adjusted duration to payment or
maturity of the bonds in the index was just under three years whereas “[s]ince
the outstanding balance on the facility under the RSA could be as a high as the
facility limit [$900,000,000] up to the Termination Date in 2007, the duration
of the [McKesson Canada] portfolio is considered to be five years.” This is a non-sequitur;
there is simply no obvious or described causal relationship of the facility
limit to its term. Also, as mentioned previously, the term of the RSA as an
agreement and contractual obligation should not be confused or conflated with
five-year term money. Neither Mr. Reifsnyder nor any other of the Appellant’s
witnesses explained why a notional arm’s length McKesson Canada would want a $900,000,000 facility in its particular business circumstances, present
or projected.
[234]
A favourable downwards sixth
adjustment of 25% to 40% was also made to reflect his view that conventional
wisdom is that short-term trade obligations like the receivables are less risky
than long-term loans like the bonds making up the index. There was no attempt
to reconcile this thought with his previous adjustment’s position.
[235]
His seventh and final
adjustment was a further upwards adjustment of 5% to 10% to reflect his view
that a trade supplier’s security on inventory compares unfavourably to pledges
of assets by bond issuers in the index. He does not address the probability of
senior ranking security on inventory as compared with the improbability of
security or probability of numerous prior secured creditors of junk bond
issuers. The evidence is that McKesson Canada’s credit policies did require
specific security in appropriate cases and could even require seats on a
customer’s Board of Directors.
[236]
These seven adjustments
in total were between 95% and up to as high as 155% of the starting point of
his chosen bond fund index. Two adjustments were up to 40% each. While his
report gave some explanation for some of the numbers he assigned, and some more
was given in his testimony, these percentages were largely based upon his
judgment and experience – though he admitted to having no experience pricing a trade
receivables financing transaction off a high yield bond index. None of
this prevented him from opining on an arm’s length discount rate range to five
significant digits, surely an extreme example in false precision (to use a
phrase from his own report).
[237]
The Reifsnyder approach
resulted in a credit risk component of the Discount Rate in the range of 0.85%
to 1.6%. This is a broad range and is up to almost 40 times higher than the 0.43%
historic loss performance of McKesson Canada’s receivables pool, and more than
six times the 0.25% Loss Ratio termination event trigger in the RSA.
(iii) Factor Four - $900MM Commitment to
Finance
[238]
Lastly, Mr. Reifsnyder
added a factor to his arm’s length Discount Rate for the RSA to reflect that
the RSA committed MIH as purchaser to purchase up to $900,000,000 of eligible
receivables over the five-year term. In his opinion, this warranted a standby
charge/commitment fee type component in the Discount Rate to reflect the unused
portion of the facility at any time during the five years in order to
compensate MIH as purchaser for McKesson Canada’s option to obtain increased
funding up to the full $900,000,000.
[239]
His report notes that
the bank market annual fee to a low investment grade rated borrower for five
years has been “in the range of .00375% to .0050%”. He opines that an
appropriate rate for the RSA commitment would be “at least 1%” given the risky
nature of the receivables pool. His written report’s “at least 1%” for the RSA
is 200 to 300 times greater than the report’s “range of .00375% to .0050%” in
bank markets for low investment grade borrowers. One might therefore have
expected greater explanation in the report than just more risky.
[240]
His report does not address
the fact that MIH is a very thinly capitalized, non-bank, non-financial
institution that is not regulated as to its capital adequacy, that otherwise
has zero presence in the financial sector or financial markets, and whose
principal asset is its somewhat illiquid shares of McKesson Canada itself.
[241]
His report does not
address why a notional arm’s length McKesson Canada would pay for a standby
facility that was almost double the amount needed.
[242]
There was no evidence
to suggest that an arm’s length counterparty to an RSA done in the
commercial markets with a member of the McKesson Group would be such a
non-investment grade entity.
[243]
Overall, I do not
accept that there is any merit or value on the facts of this case to using Mr.
Reifsnyder’s methodology or his assigned inputs thereto to help determining an
appropriate arm’s length Discount Rate for the RSA. That is not to say, however,
that I reject his evidence completely.
[244]
In short, I reject his
chosen high yield bond fund index as an appropriate starting point because of
the make-up of that index. The scope and amount of the adjustments he needs to
make to account for its differences from McKesson Canada’s receivables pool
further confirms that it is not sufficiently comparable to serve as an
appropriate starting point. In my view, the appropriate analysis places no
weight on his methodology and approach as I find it not to be an appropriate
“other method” to determine an arm’s length RSA Discount Rate. I do not
therefore place any weight on his opinion of the appropriate range for an arm’s
length Discount Rate.
[245]
My specific detailed
concerns with Mr. Reifsnyder’s approach and opinion are as follows:
(a) I do not have
sufficient evidence to accept that there is any significant degree of
comparability of such a bond fund index to a pool of inventory receivables owed
by going concern customers to a significant going concern supplier. Mr.
Reifsnyder had never priced a trade receivables transaction off any high yield
bond fund index, nor could he say he had ever heard of or seen others do so. He
was not ever sure he had ever before looked at the Merrill Lynch high yield
bond fund indexes relied upon in his report. Similarly, none of the other
expert witnesses asked could say they had ever heard of it being done. Mr.
Reifsnyder did not testify that he had ever himself priced, or seen another
price, any structured finance transaction (other than perhaps a bond, bond
fund, or a bond fund index) off a bond fund index. The Respondent’s experts who
were asked said they had never heard of one.
(b) The high yield index
is made up of bond issues, corporations that have raised money in the public
markets generally for general corporate purposes. In contrast, the Obligors’
receivables in McKesson Canada’s receivables pools receivables are owing to one
of their key suppliers for the inventory they need to remain in business as a
going concern.
(c) Ms. Hooper of TDSI
testified that in Canada there was not at the time, and especially in 2003, a
very liquid market for sub-investment grade bonds.
(d) There are only a
handful of issuers in the high yield index. McKesson Canada has many multiples
that number of Obligors.
(e) Regardless of the
quality of the issuer in the index when they first issued their bonds, by
definition their bonds are only added to the high yield index if they are of junk
status. In extreme contrast, McKesson Canada’s Obligors at the time the RSA was
entered into had regularly paid their receivables in full about 99.96% of the
time. While there was actually a risk this default rate could change, there was
no evident expectation that it would. In December 2002, these would be two
drastically different snapshots virtually incapable of being more different
absent actual default.
(f) McKesson Canada’s Obligors
were monitored and rated by McKesson Canada’s Credit and Collections
department, and the terms of credit, if any at all, offered to them could be
changed by McKesson Canada at a moment’s notice. In contrast, the bonds
reflected in the chosen high yield index were for a term with either a lump sum
payment upon maturity years later, or perhaps a series of annual payments
through maturity. That is, a bond holder paid its money, took its chances and
had to wait and see or sell. According to Moody’s, the historic credit loss on
five year junk bonds is in the range of 27%. This is almost 700 times the 0.043%
historic loss on the McKesson Canada receivables pool.
(g) The individual
discounts on particular issuers’ bonds in the index were as high as 98%. The five
year bond issues mentioned above that traded at less than fifty cents on the
dollar and had effective yields in excess of 30% comprised 8.5% of the index in
November 2002. There is absolutely no evidence to even suggest that a single
one of McKesson Canada’s Obligors, including the Designated Obligor whose
financial situation was detailed during the trial, was in even remotely
comparable financial circumstances.
(h) The high yield bond
fund index spreads fluctuated very significantly. In the six months prior to
the RSA, the index spread’s performance had fluctuated by almost 50%. In
contrast, in the same period, the McKesson Canada receivables pool’s
performance remained at a roughly similar .04% write-offs to sales number and a
DSO within the 30 day range. Further, there was no evidence given to Mr.
Reifsnyder or to the Court of any significant change in the public ratings of
those of McKesson Canada’s customers that were rated, nor to the credit ratings
or evaluations of all of its customers by its own Credit and Collections
department. This volatility and fluctuation in the range of discount spreads on
the high yield index was such that had the bond fund index spread for a
different month been used as the starting point for Mr. Reifsnyder’s approach,
his approach’s arm’s length Discount Rate would have been materially different
even though there is no evidence to suggest the quality of McKesson Canada’s
receivables pools was materially different throughout the period. Mr.
Reifsnyder did not try to relate this index’s OAS volatility to fluctuations in
interest rates or credit spreads during the same period.
(i) the credit risk spreads on the bond fund indexes are
properly known as Option Adjusted Spreads or OAS. The OAS is widely viewed as
the credit risk premium over a risk-free rate that applies to issuers in a
similar class. OAS are not simple mathematical calculations. They are
calculated using quantitative models and quantitative assumptions. Mr.
Reifsnyder has never calculated an OAS and could not at all describe the models
or assumptions used either generally or in the case of his chosen index.
(j) There was no evidence
that any of McKesson Canada’s Obligors that had public ratings, or those whose
parent company was rated, had ever had their rating downgraded at all, much
less to anything comparable to high yield junk status.
(k) Mr. Reifsnyder acknowledged that in the case of
McKesson Canada’s rated Obligors, using their individual and particular bond
rating credit spreads could be more accurate than his approach – he just did
not know and did not explain why he did not consider that even though the
results would be very different.
(l) Notwithstanding the
fluctuating volatile spread of the high yield index OAS in the prevailing
market at the time, Mr. Reifsnyder’s approach used a fixed credit risk discount
spread throughout the five-year RSA term and did not use a dynamic floating
spread for the index to adjust for or reflect this underlying volatility. His
position was that while a dynamic approach to credit losses would be more
intuitive, they would constitute the use of hindsight. I could not
disagree more. A dynamic approach during the term of an agreement and provided
for in an agreement is either not the use of hindsight as that term is used in
circumstances such as these, or is at least not an inappropriate use of
hindsight. He did acknowledge that there is a continuing dynamic adjustment in
credit risk assessments in securitizations to reflect the most current
performance throughout the term of the transactions.
(m) The adjustments to the
high yield index proposed by Mr. Reifsnyder to make it comparable to McKesson
Canada’s receivables pool and the RSA were of orders of magnitude in the
aggregate (up to 155%), and at least some individually (two up to 40%) which belie
any starting point suitability for pricing comparability purposes.
(n) As noted above, the
explanations and premises for several of his proposed adjustments to the high
yield fund index were inadequately supported and/or insufficiently explained.
(o) Mr. Reifsnyder’s “at
least 1%” stand-by commitment fee for the undrawn portion of the $900,000,000
facility limit under the RSA can not be supported or accepted at all (i) absent
any explanation for his 20,000% to 30,000% adjustment to his 0.00375% to
0.0050% range data point for low investment grade borrowers in bank markets,
(ii) absent any understanding of why the $900,000,000 facility limit would be
of value to McKesson Canada throughout its term and that was reasonably
supported in the evidence, and (iii) absent any adjustment for the fact that
MIH was not a bank but an unregulated private company with virtually no capital
or financial market presence, nor any explanation for the absence of need for
any such adjustment. This component of Mr. Reifsnyder’s opinion of an arm’s
length RSA discount rate is not only wholly rejected, it raises further
concerns with his overall report.
(p) Notwithstanding his
extensive knowledge, experience and presentations on credit insurance for
structured finance transactions, and his reference to its availability in his
testimony, Mr. Reifsnyder seemingly never considered the cost of insuring the
receivables in his pricing approach, nor to test the results of his approach.
Nor did he explain why he did not do so.
(q) While Mr. Reifsnyder
maintained that markets should not price risk off historical performance, he
could not say that markets did not. His view is at odds with that of TDSI and
Ms. Hooper who refer several times to their need to consider historical
performance in terms of defaults, delinquencies, losses, dilutions, payment
times and reserves. It is also at odds with the rating agencies’ publications
on the significance of historic performance in trade receivables financing structures
and the evidence of other experts who testified. Historical delinquency and
write-off performance is generally considered to be the best indicator of
portfolio credit quality in trade receivables transactions. Mr. Reifsnyder
acknowledged that rating agencies look at five years of historic
performance data when reviewing a portfolio of financial assets. The Reifsnyder
Report and testimony try to downplay the relevance of the historic performance
of the McKesson Canada receivables pool in favour of the actual, then current
credit spreads of its chosen high yield index. The irony is that the high yield
index credit spread is itself a reflection in part of its historic performance
and that of the issues and issuers it comprises.
(r) Mr. Reifsnyder came
across as in large measure a partisan advocate quick to point out the specks in
the Respondent’s expert reports, and downplaying, if not refusing to
acknowledge, the weak points in his own.
[246]
Overall I can say that
never have I 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. One could only
assume that the Appellant knew full well the weaknesses of the TDSI Report and
this was the best method it could use to support the Discount Rate used by the
McKesson Group in the RSA.
(f) The Becker Expert
Report
Build up approach:
[247]
The first valuation
approach in the Becker Report relied upon by the Respondent, which was
described as a build up approach, essentially considered the same factors and
issues as the TDSI Report to arrive at a Discount Rate that reflects both the
time value of money and the potential risk of not collecting the full face
value of all of the transferred receivables.
[248]
The Becker Report
accepts the 30 day CDOR rate as the appropriate risk-free rate, noting it is
common to use government obligations of similar maturities as a bench mark
risk-free rate for the time value of money.
[249]
In this approach the
Becker Report assessed the risk of Obligors defaulting at the three to four
year historic McKesson Canada write-off experience of 0.0440% of sales. Similarly,
in this approach the Becker Report assessed the prompt payment dilution risk at
the three to four year historic McKesson Canada prompt payment discount
experience of 0.5324% of sales.
[250]
Adding these three
factors together (after first adjusting the 30 day CDOR Rate for the McKesson
Canada DSO over the same period) the Becker Report arrives at an arm’s length
discount of 0.8073%.
Comparable Transaction Approach:
[251]
The second valuation
approach in the Becker Report is a comparable transaction approach. For this
purpose only one arm’s length transaction is considered, that involving
McKesson Canada and TD Factors. Dr. Becker considered that an actual arm’s
length transaction involving McKesson Canada factoring its receivables to an
arm’s length party had a much greater degree of reliability as a starting point
than third party agreements or other McKesson Canada financing agreements could
have since it was so much closer to the RSA in nature and risk.
[252]
The Becker Report
calculates the net discount rate in McKesson Canada’s TD Factors transaction as
0.3376%. Since the TD Factors agreement did not specify that TD Factors took
any prompt payment discount dilution risk, an upwards adjustment equal to
McKesson Canada’s three to four year historic prompt payment discount experience
of 0.5324% was made. The Becker Report’s comparable transaction approach
concludes that an appropriate arm’s length Discount Rate for the RSA using the
TD Factors transaction as the comparable benchmark starting point discount rate
would be 0.8700%.
[253]
The Becker Report
places the appropriate range for an arm’s length discount rate for the RSA
between 0.81% and 0.87% (being the discount rates of each of its two approaches),
and identifies the midpoint as 0.8386% which is then used as the single
discount rate. The report observes that, had this rate been used in the RSA,
McKesson Canada’s pre-tax profit margins would have remained within their
historic ranges.
[254]
I am not at all
satisfied that the TD Factors transaction was sufficiently comparable to the
RSA to serve as an appropriate starting point for the Becker Report’s
comparable transaction approach, much less to be the only comparable
transaction even to be considered. While it was a transaction reflective of
McKesson Canada’s creditworthiness and it did involve a transfer of some of
McKesson Canada’s receivables, it was for a much shorter term, implemented for
a very specific and different purpose, and did not involve the purchase by TD
Factors of McKesson Canada’s entire receivables pool. For these reasons, I do
not rely in any way upon the Becker Report’s comparable transaction approach.
(g)
The Finard Expert Report
Structured Finance Approach:
[255]
The Finard Report’s
structured finance analytical approach to estimating an arm’s length Discount Rate
for the RSA was based upon a review of available public bond market rating and
credit default loss data. Moody’s back-tests the accuracy of their ratings over
a period exceeding 20 years. These show a high degree of correlation between
credit ratings and credit default loss rates over periods ranging from one to
five years.
[256]
The Finard Report
looked at one year numbers even though the RSA has a five year term. This
recognized that there are several risk mitigation provisions in the RSA that
would permit an earlier termination of the RSA. Also, the Moody’s data shows
risk of default increases significantly between year one and year five; hence
this approach is more conservative.
[257]
The McKesson Canada receivables
pool’s five year average write-offs to sales was 0.043%. Conservatively using
the five-year McKesson Canada write-offs to sales figures, the Finard Report identified
that this number, 0.043%, fell between Moody’s one year credit loss percentage
by rating for A rated corporate bond issuers and Baa rated corporate bond
issuers (or more than five credit ratings above Moody’s “speculative grade”
rating which had a credit default history about 90 times higher).
[258]
The Finard Report then
uses TDSI’s data on credit risk spread by rating and notes that the credit risk
spread for an A rated issuer in December 2002 was 0.50%, and for a Baa rated
issuer was 1.00%. The Report comes up with a weighted average of 0.68%.
Attribute Analysis Approach:
[259]
In the Finard Report’s
Attribute Analysis approach to estimating an arm’s length Discount Rate for the
RSA, a dynamic rolling average approach to the actual and ongoing performance
of the receivables was taken to be normative market practice in arm’s length
financing transactions that are similarly asset based. The Finard report’s
opinion is that the terms and conditions of an arm’s length transaction would have
a variable approach to computing DSO and to address prompt payment discounts.
Each should be computed based on a four settlement period rolling basis.
[260]
In computing a
servicing discount, the Finard Report’s Attribute Analysis approach used the
$9,600,000 annual fee paid by MIH under the Servicing Agreement and calculated
a servicing discount each settlement period based upon the average amount of
receivables computed dynamically on a four settlement period basis.
[261]
The Finard Attribute
Analysis rejected the idea of any discount attributable to MIH’s funding costs
(the so-called interest discount) or to the accrued rebate dilutions risk.
[262]
The Finard Attribute
Analysis approached the issue of loss discount credit risk by (i) taking a
dynamic approach to the Loss Discount based upon a rolling 12 month average of
write-offs to sales of McKesson Canada’s receivables, and (ii) by introducing a
general reserve of 21% of the purchase price of the receivables to deal with
the risk of incremental risk beyond the loss discount in the RSA.
[263]
The use of a 12 month average
of write-offs to sales for the loss discount was chosen because it is
consistent with the RSA providing that the loss discount is to be recalculated
each January 1st.
[264]
I am 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 RSA. There may be significant reserves in securitizations, in secured loans,
and in other asset backed loans, but the RSA gives no such security or
protection to MIH once it has purchased any particular receivables. I therefore
do not see how such a reserve could be appropriate from an arm’s length
comparability point of view. I am also somewhat troubled as to whether this can
be done in support of an adjustment under paragraphs 247(2)(a) and (c).
Given these doubts, which I do not have to resolve in this case given the rest
of my reasons, I will not be relying upon the results of the Finard Report’s
Attribute Analysis approach.
(h)
The Glucksman Expert Report
[265] In addition to critiquing the Discount Rate
approaches in the TDSI Report (and in the PwC Report), the Glucksman Report
computes an Affirmative Estimate of an arm’s length Discount Rate for the RSA.
[266]
The Glucksman Affirmative
Estimate approach dealt with credit risk by introducing an 18.5% reserve. The
report expressly regarded the RSA as best analyzed relative to accounts receivables
securitizations. It selected 18.5% as the appropriate reserve based upon that having
been used in a US$950,000,000 McKesson U.S. receivables securitization.
[267]
For the same reasons as
described above with respect to the Finard Attribute Analysis’ introduction of
a reserve that is not directly or indirectly reflected in the RSA, I will not
be relying upon the results of the Glucksman report’s Affirmative Estimate
approach to estimating an arm’s length Discount Rate for the RSA. Similarly,
the Glucksman Report’s Affirmative Estimate approach is viewed entirely through
the lens of accounts receivable securitizations and, for that reason, I will
not be relying upon any portion of the Glucksman Report’s Affirmative Estimate
approach. For these reasons, I will dispense with summarizing the rest of the
report’s Affirmative Estimate approach.
(i) The Other Expert
Reports
[268]
Further expert reports were
filed on consent, without the Court hearing testimony from their authors. As
with the PwC Report, little weight can be given to the contents of expert
reports written by persons who did not testify in the proceedings. To some
small extent they might provide corroboration for the approaches or data of
others.
9. The Appropriate Methodology
[269]
As is often the case
where there is more than one expert and expert report, the Court does not
accept the conclusions of any of those experts or their reports in their
entirety. However, while some approaches and some information, estimates or
components of their analysis may be unpersuasive or rejected, overall the Court
is informed by all of their testimony and the information they provided
relating to the opinions that they arrived at and, as can be seen, has relied
upon parts of their opinions and some of the factual information they relied
upon.
[270]
The Court is of the
view that the most appropriate and proper approach in this particular case is
to follow the structure of the RSA that the McKesson Group chose to enter into
and to approach the pricing issues largely as TDSI (and PwC) did, and consider
whether the terms and conditions which affect the Discount Rate pricing differ
from what arm’s length terms and conditions would be expected to provide, in
order to adjust the amounts best as I can to reflect the helpful evidence on
these issues, including the expert opinion evidence, before me.
10. Analysis of Transfer Pricing Issue
[271]
This was not a
securitization transaction. A securitization is generally 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. The Appellant’s evidence is 100% consistent that
this was not a securitization transaction, nor was there any apparent financial
or business reason for McKesson Canada to be interested in a securitization
transaction. Any comparison of any aspect of this transaction with a securitization
needs to be cautiously approached with this in mind.
[272]
The RSA was a five year
facility. This was not five year money any more than it was simply 30 day
money. Any suggestion that any aspect of it could be equated to comparable
terms of five year debt, medium term debt, or other long term debt needs to be cautiously
approached.
[273]
Similarly, the RSA had
a maximum receivables pool at any time of $900,000,000. Any suggestion when
making comparisons to other transactions that this represents that the RSA
reflected $ 900,000,000 of exposure, unless and until MIH ever hit that maximum
would similarly need to be very cautiously approached and thought through.
[274]
I find as a fact that
the predominant purpose and intention of McKesson Canada participating in the
RSA and related transactions with the other McKesson Group members was not to
access capital or to lay off credit risk. Those were results of the
transactions but did not motivate them. The purpose was to reduce McKesson Canada’s Canadian tax liability (and therefore McKesson Group’s worldwide tax liability) by
paying the maximum discount under the RSA that McKesson Group believed it could
reasonably justify. For the McKesson Group this appears to have been much more
of a tax avoidance plan than a structured finance product. No reason was ever
given for wanting to transfer risk to Luxembourg.
[275]
There is certainly nothing wrong
with taxpayers doing tax-oriented transactions, tax planning, and making
decisions based entirely upon tax consequences (subject only to GAAR which is
not relevant to this appeal). The Supreme Court of Canada reminds us regularly
that the Duke of Westminster is alive and well and living in Canada. However, the primary reasons and predominant purposes of non-arm’s length
transactions, whatever they may be in any given case, form a relevant part of
the factual context being considered. For example, if neither side has a
business purpose or need to do a particular non-arm’s length transaction, it
will probably not be particularly persuasive to try to argue that particular
terms, conditions, provisions, or approach reflect the particular business need
of either party.
[276]
The maximum amount
deductible in Canada by McKesson Canada is limited to what an arm’s length
person would agree to pay for the rights and benefits obtained. The Appellant says
it did not exceed that limit. The Respondent says they exceeded it by more than
100%. This is the only question that the Court is called to decide.
(a) The Discount Rate
[277] As already described, the RSA provides that
the Discount Rate for each purchase of receivables is the sum of (i) the Yield
Rate on the first business day of the relevant settlement period, (ii) the Loss
Discount (iii) the Discount Spread.
(i) The Yield Rate
[278]
The Yield Rate is the
only fully floating component of the Discount Rate in the RSA. In contrast,
both the Loss Discount and the Discount Spread are fixed. The Loss Discount is
recalculated annually or earlier at MIH’s request.
[279]
There is no dispute on
the evidence that the 30 day CDOR rate is the appropriate base line risk-free
rate. I accept that, and I accept that using that rate as of the first business
day of each 28 day settlement period (ignoring for the moment that the RSA was
signed on December 16, 2002, nine days after the end of McKesson Canada’s
Accounting Period 9 of 2003, on December 7, 2002, leaving only a 19 day initial
settlement period) is well within the range of what two arm’s length parties,
entirely adverse in interest on pricing and risk-related terms and conditions,
would agree to as both acceptable and reasonable.
[280]
The CDOR rate as of
December 16, 2002 was 2.79% per annum. There was no evidence that this changed
materially in the remaining two or three relevant Accounting Periods in
McKesson Canada’s 2003 year. For purposes of this appeal of McKesson Canada’s 2003 year, I will assume the CDOR rate did not change materially.
[281]
The CDOR rate is
expressed as an annual rate and this needs to be adjusted to reflect that the
receivables can be expected to be collected over a much shorter period than a
365 day year. I accept that using the accounting concept of Days Sales Outstanding
or DSO for the collection period of McKesson Canada receivables is an
appropriate proxy or measure for this purpose.
[282]
However, given my
observations above on the calculation of DSO by McKesson Canada and TDSI for
the RSA, I do not accept that using a fixed DSO of 31.73 days throughout the
term without regard to changes to McKesson Canada’s actual DSO at the relevant
time (whether resulting from changes in McKesson Canada’s sales or changes in
its customers’ payment patterns) is what two arm’s length parties, adverse in interest
as to pricing and risk-related terms and conditions, would agree to.
[283]
I find that arm’s
length parties would choose to incorporate a floating approach to DSO averaged
over some period, say three to four months or Accounting Periods and
would not accept the risk of fixing the DSO for the entire term of the RSA.
Notably the very concept of DSO is measured and tracked over successive periods
of time. This is also supported by the evidence of experts Becker and Finard.
Given that the RSA uses a four month rolling average in its definitions of Loss
Ratio and Delinquency Ratio, I find a rolling four Accounting Period average
appropriate in this appeal. In any event, I do not have any helpful evidence to
support a fixed approach as the Appellant has not adequately explained how the
risk of change in the DSO was factored into the Discount Rate pricing, why the
risk of change in the DSO is not relevant, nor provided other supporting
evidence of how this Court can price or adjust for the risk of DSO change.
[284]
TDSI calculated
McKesson Canada’s DSOs for each Accounting Period of McKesson Canada’s 2001, 2002 and 2003 fiscal years as part of its engagement. The average DSO per
period over this term prior to the RSA as computed by TDSI was 32.00 days.
[285]
While this is stated accurately in
the text of the TDSI Report, the TDSI backup schedules state that the average
was 31.73 days. This is not correct; if one works the average out from the
schedule of numbers it is 32.00 days. The fact that the average shown in the
schedule is 31.73 days, the same number TDSI uses after its rough adjustment
for the significantly shorter DSO on the maturing receivables pool of $460,000,000
initially purchased, certainly raises questions.
[286]
TDSI then recognized, in
helping the Appellant fix its DSO for purposes of the RSA computations, that
the DSO of 32.00 days would not be appropriate given that the initial
$460,000,000 purchase of the existing receivables pool on December 16, 2002 was
a mature pool of receivables. TDSI estimated that a mature pool in the
circumstances should be expected to be paid within one-half of the existing DSO,
or within sixteen days. That seems to be a sensible and reasonable estimation.
[287]
However, rather than
use a 16 day DSO for the initial receivables purchase’s Discount Rate, TDSI
instead averaged these “missing” sixteen days across the five-year term and
simply reduced the 32 day DSO to 31.73. I do not accept that is appropriate nor
that it is an arm’s length approach to this issue. This allowed MIH to underpay
McKesson Canada at the outset by virtue of a significantly overstated Discount
Rate, and it would only accrue back to McKesson Canada over the five years and
without interest. I am not persuaded that arm’s length parties to a financial
transaction would agree to deal with the issue on that basis alone. As has been
noted, this also created a corresponding five year Canadian income tax timing
benefit to McKesson Group.
[288]
I conclude that arm’s
length parties to a financial transaction would accurately account for the
difference between the initial purchase’s estimated 16 day DSO and TDSI’s 32
day DSO for new receivables. This will materially impact each of the components
of the Discount Rate that are annual rates that need to be DSO adjusted. These
are the Yield Rate and components of each of the Loss Discount and the Discount
Spread. This would not be an insignificant transfer pricing adjustment for
McKesson Canada’s 2003 year under appeal.
[289]
I do not, however,
accept TDSI’s computation of a 32 day DSO or its backup schedules to its
report. I prefer the McKesson Canada DSO schedules which show that the DSO as
computed by the company in its ordinary course averaged less than thirty for
the two financial years (2002 and 2003) preceding the RSA, and in only one
Accounting Period of the 18 preceding the RSA exceeded 30 (being 30.5). This is
consistent with Mr. Brennan’s evidence that the DSO was and remained in the 30
day range. I conclude that arm’s length parties would use a four month rolling
average DSO and that the best evidence of this was 30.0 days throughout.
[290]
Adjusting the 30 day CDOR
rate of 2.79% per annum for a 30 day DSI results in a Yield Rate of 0.2293
(2.79% x 30 ÷ 365).
[291]
The corrected or
adjusted Yield Rate for the initial purchase would be half of this. The Yield
Rate for the 2003 year in issue, which has three settlement dates and full DSO
cycles, will be only 0.1911 (the sum of 0.2293 plus 0.2293 plus 0.1147, all
divided by 3). This would spread the “missing” 15 days over the three month
period following the initial purchase, and all within the same tax year being
the year under appeal.
(ii) The Loss Discount
[292]
The Loss Discount in
the RSA is intended to account for the risk that the Obligors do not fully pay
their receivables. As outlined above, McKesson Canada’s multi-year receivables
performance numbers resulted in an average collection of 99.96% of its
receivables. This represents 0.04% write–offs to sales.
[293]
As described above,
the Loss Discount was a fixed 0.23% throughout the year in question and to
December 31, 2003.
[294]
The Loss Discount was
to be recalculated in accordance with the RSA each year thereafter or whenever
MIH felt the Designated Obligor receivables ratio had materially changed since
last calculated. Effectively, only the Loss Discount attributable to Designated
Obligors could be changed under the terms and conditions of the RSA as the Loss
Discount attributable to other Obligors was fixed at 0.2380% for the full
five-year term.
[295]
The evidence confirmed
that the fixed 0.23% Loss Discount for 2003 was calculated by the Appellant and
TDSI on the same basis as the annual recalculation provided for in the RSA.
[296]
The Court does not
accept that the terms and conditions of the RSA relating to the Loss Discount
of Designated Obligors, or the approach of the McKesson Group and TDSI to
computing the Loss Discount of other Obligors, reflect arm’s length terms and
conditions. The Appellant has not satisfied the Court on the evidence presented
that these terms and conditions of the RSA reflect what arm’s length parties,
adverse in interest as to pricing and risk, would agree to in similar
circumstances.
[297]
Firstly, the evidence
does not permit or support any conclusion being reached on the extent to which
a rated company’s bond rating is a reasonably accurate market approach to
assessing risk of default on the company’s trade payables for inventory owing
to a key dominant supplier. There was little if any evidence to support a
sufficiently direct or comparable correlation, though I do accept that one
might reasonably expect a degree of relevance. Some evidence I heard confirmed
the contrary. The Appellant did not tender any supporting evidence for the
credit analysis, ratings or scores assigned by McKesson Canada’s Credit and Collections department of its rated Obligors to support or explain its
position. In the circumstances of this trial, I can only conclude that was
intentional.
[298]
Secondly, I can not
reasonably conclude that a company that does not have a bond rating can be assumed
to be hiding a bad implicit rating from the public. Many private companies,
large or small, do not obtain public bond ratings simply because they have no
need or desire to raise money in the bond markets as conventional lenders and
sources of funds work fine for them. There is no rational basis supported in
the evidence or in common sense that all unrated companies are equivalent
credit risks to non-investment grade bond issuers.
[299]
The Court’s concerns
are further confirmed from a common sense point of view by the fact that the
approach taken by McKesson Group and TDSI assigns a going forward credit risk
to the receivables of McKesson Canada’s customers from time to time that is
many, many, many times higher than the multi-year historic performance of these
receivables.
[300]
The Court does not
accept that the Loss Discount attributable to Obligors other than Designated
Obligors would be fixed by arm’s length parties in the manner it was, nor fixed
at that same number for a five-year term.
[301]
The Court does not
accept that an arm’s length seller, would agree to terms and conditions that
resulted in a Loss Discount that was almost 600% of its historic receivables
write-offs without any significant projected, planned or reasonably anticipated
material risk of deterioration of its business, its customers, its receivables,
or the Canadian or world economies generally, of which there was no evidence.
[302]
The Court does not
accept that arm’s length parties would agree that the Loss Discount of
Designated Obligors could be recalculated at any time if the buyer thought that
the mix of Designated Obligors to other Obligors had changed, but not if the
seller thought the mix had changed in its favour, absent some off-setting
concession or quid pro quo, of which there was none.
[303]
The Court does not
accept that arm’s length parties would agree that the RSA’s Loss Discount
terms, nor their underlying computations by McKesson Group and TDSI, would not
directly take into account at all neither the historical or actual ongoing performance
of the receivables pool.
[304]
For all of these
reasons, I can not accept that the RSA terms and conditions regarding Loss
Discount, and the computations of the quantum of the Loss Discount by McKesson
Group and TDSI, were arm’s length terms and conditions or resulted in the
appropriate arm’s length amounts. It is therefore the role of this Court to
estimate what the appropriate amount or range of the Loss Discount should be
for purposes of computing the Discount Rate under the RSA. In doing so, this Court
is limited to the evidence available to it. This Court can only use an
estimation method that is able to be used with the available evidence to arrive
at a number or range of numbers. The parties’ choice of evidence may be a
constraining factor which may well preclude an arguably more effective or
appropriate method for the Court to estimate an arm’s length amount.
[305]
There is no magic about past
historical data. We can not consider future data, only make reasonable
predictions. After the fact we may also now have to assess whether a past
assumption about the future, had it been made, would have been reasonable. In
doing so, we may be cognizant of what would then have been future data but
which is now equally historical. All must be approached carefully by a court.
None are determinative, but none are entirely irrelevant considerations.
[306]
The RSA was signed at a
time when the receivables pool’s write-offs to sales performance had been in
the range of 0.04%. This was well known and tracked by McKesson Canada and McKesson Group. The RSA gave MIH an immediately exercisable termination right in
the event the pool’s Delinquency Ratio or Loss Ratio increased by specific
measures.
[307]
Ms. Hooper’s evidence
was that these two termination event triggers in particular were designed to
effectively stop the transfer of additional receivables once the portfolio does
not perform as well as it did in the past. The Delinquency Ratio was designed
as an early warning system. Given that delinquencies can be expected to
increase in advance of seeing losses increase, the termination right was
designed to occur early enough that one is not going to have very material
losses. According to Ms. Hooper’s testimony, the Loss Ratio and Delinquency
Ratio combined should allow the purchaser to stop acquiring additional receivables
in time to not suffer materially higher losses than expected based on past
performance. Ms. Hooper testified that she and her team at TDSI looked at both
the historical loss and delinquencies in the McKesson Canada receivables pool
as part of its engagement in preparing the TDSI Report.
[308]
I do not necessarily
accept the TDSI Report’s opinions on the reasonableness, normalcy or arm’s
length nature of these two termination triggers in the RSA. Indeed, I would
expect they might suffer from the same shortcomings as affects the rest of the
TDSI Report, which is primarily that the RSA is not a securitization and is in
that respect outside the expertise of Ms. Hooper and her group. In any event,
given that these two ratios are defined in the RSA to include McKesson Canada
financial information that is not in evidence, or at least certainly not
adequately explained in the evidence, and that these defined ratios and their
volatility leading up to the RSA were not put in evidence, I can not reach the
conclusion that I am satisfied with the TDSI’s Report’s conclusions on their
terms.
[309]
However, I fully accept
Ms. Hooper’s explanation of their purpose and effectiveness as designed. That
is, I find that the purpose and effect of the Delinquency Ratio termination
event trigger and the Loss Ratio termination event trigger in the RSA were
designed and fully expected to limit MIH’s risk of purchasing any day’s
receivables from McKesson Canada that could be expected to have materially higher
losses than had been experienced on the pool historically.
[310]
The historic loss
performance on McKesson Canada’s receivables pool was in the range of 0.04%. I
conclude from all of this that a notional arm’s length MIH would have been able
to and would have terminated its obligations under the RSA before it was
obligated to purchase receivables that would have a materially higher credit
loss risk than something in the range of 0.04%.
[311]
Allowing for a 50% to
100% increase as an extremely generous interpretation of what Ms. Hooper could
have meant by material (in part to compensate for the lack of elegance in this
approach), I find that a notional arm’s length MIH’s credit loss risk on its
continuing purchase of receivables is that, at some future point in the RSA
term (but not in the very short term), it could have purchased about four
months of receivables with an anticipated write-offs to sales number in the
range of 0.06% to 0.08%. These lesser quality receivables would only be
expected to have been purchased in the last four months of the RSA prior to
termination. Those bought on December 16, 2002 and for the other months prior
to the four months preceding termination could continue to be expected to be of
a better quality.
[312]
Using this approach, the
Court concludes that a Loss Discount component of the Discount Rate in the
range of 0.06% to 0.08% is at the generous end of what a notional arm’s length
MIH and McKesson Canada would agree to.
[313]
This range is
consistent with the number arrived at by Mr. Finard’s structured finance
approach. That approach identified that the 0.04% historic write –offs to sales
number for McKesson Canada’s receivables pool was comparable to Moody’s
published information for companies rated between A and Baa, which in turn had
credit risk spreads according to TDSI of 0.50% and 1.00% per annum, and was
computed on a weighted average basis by Mr. Finard at 0.68%. Once adjusted for a
DSO of 30 days, a 0.68% annual credit spread reflects a discount of 0.06%.
[314]
For these reasons, the
Court finds based upon what evidence was provided that an arm’s length Loss
Discount for purposes of the RSA would be in the range of 0.06% to 0.08%.
(iii) The Discount Spread
[315]
The discount spread in
the RSA was calculated by McKesson Group and TDSI as the sum of four different
components.
1. Servicing Discount
[316]
The Court does not
accept the TDSI Report’s position regarding the discount needed to reflect the
potential cost of choosing to engage a new servicer to replace McKesson Canada as servicer of the already purchased receivables.
[317]
The Servicing Agreement
sets out a fee payable of $800,000 monthly to the servicer and appoints
McKesson Canada as the initial servicer. By its terms, the Servicing Agreement
applies to a replacement servicer, assuming a replacement servicer would agree
to it.
[318]
Mr. Reifsnyder was of
the opinion that the $9,600,000 annual fee provided in the Servicing Agreement
was substantially rich enough that it could possibly fully cover servicing
costs even if a replacement servicer needed to be brought in for a short period
of time at that level. Mr. Reifsnyder had been involved in transactions where
servicers needed to be replaced.
[319]
I do not accept TDSI’s
unexplained 25% chance that a replacement servicer would need to be appointed.
That is entirely unsubstantiated. TDSI begins from a less than 10% prospect of
a rating downgrade termination event.
[320]
Even upon the
occurrence of a termination event followed by termination, I do not accept that
it is a given that a notional arm’s length MIH would invariably exercise its
right to appoint a replacement servicer. The evidence is that McKesson Canada
had very good and successful collection policies, practices and results. A
number of the termination events listed in the RSA could occur without any
related impact upon McKesson Canada’s continuing servicing abilities.
[321]
The appointment of a
replacement servicer for a short period of time, a period of between 30 days (the
DSO) and 90 days, being how long TDSI thought would be needed to wind up the portfolio
once no new receivables were being purchased, would have to be weighed against
disruption in the customer relationships which may further delay and hinder
payment, learning time and inefficiencies for a new servicer to get up to speed,
and cost considerations (hard and soft costs, direct and indirect). The
prospect of having to appoint a replacement servicer would have been assessed
by a notional arm’s length MIH at much less than the 25% assumed by TDSI or the
40% assumed by PwC. Based upon all of the evidence above, I would estimate that
the likelihood would have been significantly less than 10%.
[322]
I accept Mr.
Reifsnyder’s opinion based upon his experience with replacement servicers and
servicing in securitizations, that it was quite possible the servicing fee set
out in the Servicing Agreement was sufficient to fund a replacement servicer if
one was needed.
[323]
McKesson Group and TDSI
calculated the servicing discount using a replacement servicer cost of 2% of
the face amount of receivables. The only support for this was the range of 1%
to 3% of face obtained by TDSI from one third party provider, perhaps a major
accounting firm. The other evidence on the cost of replacement services according
to the TDSI Report, is that Bell Canada recorded 1% and Telus 2% on their
Canadian receivables securitizations. PwC, a major accounting firm, used
replacement servicer fees in the range of 0.8% to 1.2%. Moody’s and Standard
& Poor’s both put replacement servicers in the 1% range of the face amount
of receivables.
[324]
I find that a
reasonable maximum replacement servicer fee would be 1% of the receivables to
be collected in the case of a notionally arm’s length RSA.
[325]
MIH was paying
$9,600,000 annually to the servicer under the Servicing Agreement which works
out to $800,000 per month. Given that the DSO of the purchased receivables was
also in the one month range, this fee represents about 0.17% of the
$460,000,000 RSA utilization in 2003 by McKesson Canada. This fee under the
Servicing Agreement has not been challenged or reassessed and is not directly
in issue in this appeal. On the basis of the fees payable under the McKesson
Group’s Servicing Agreement, which were not challenged by CRA, and which were
supported by TDSI in its supplemental report, I accept that a notional arm’s
length RSA and Servicing Agreement would allow a basic servicing discount to be
financed out of MIH’s discount under the RSA equal to 0.17%.
[326]
If a new replacement
servicer was appointed, this fee would no longer be paid to McKesson Canada and would be paid to the new third party servicer, in accordance with the Servicing
Agreement.
[327]
According to Mr.
Reifsnyder, this may require no additional outlay beyond the Servicing
Agreement amounts. Following the McKesson Group and TDSI approach, a replacement
servicer, if appointed, would only be paid once to collect and wind down the
receivables pool in orderly fashion following termination of the RSA by MIH.
Assuming the receivables pool might have reasonably been expected to have
increased to $500,000,000, a 1% replacement servicer fee payable by the
Purchaser would be $5,000,000. In my opinion, that is the absolute, outside
maximum total dollar amount of discount that a notionally arm’s length McKesson
Canada would be agreeable to pay a purchaser for the replacement servicer
risk.
[328]
After accounting for
the $800,000 already available each month, MIH would need to pay no more than
an additional $4,200,000 upon termination. If this additional $4,200,000 were
to be simply collected over the five-year term, this would only require less
than a .02% discount assuming constant RSA utilization at $460,000,000. This
would not account for MIH’s risk that a replacement servicer might need to be
appointed before the last period of year five. On the other hand, it also would
not account for the fact that there is much less than a 100% chance of a
replacement servicer being needed.
[329]
If the $4,200,000 is to
be fully collected in the first 12 months, the additional discount needed for
replacement servicer risk under the McKesson Group/TDSI approach is 0.08%. If
this was to be fully collected over the first half of the RSA term, being 30
months, the additional discount needed would be .04%.
[330]
Even if a notional
arm’s length MIH wanted to recover this $4,200,000 all in the first three
settlement periods of the RSA, being those occurring in McKesson Canada’s 2003 taxation year, this means it would need to enjoy an additional discount of
$1,400,000 in each period. Assuming the projected RSA utilization remained at
$460,000,000 for those three months, this would account for a replacement
servicer cost discount of 0.3% for those three months only.
[331]
Based upon the numbers
above, I would estimate that it is reasonable to conclude that a notional arm’s
length McKesson Canada and a notional arm’s length MIH would agree to allow MIH
to fully fund and cover the potential additional cost of a replacement
servicer, should one need to be appointed, over a period ranging no shorter
than from the first 12 months of the RSA (which would require an additional discount
of .08%) and to no longer than the first 30 months of the RSA (which would
require an additional discount of .04%).
[332]
This approach to using
the RSA discount to fund the servicing fee under the Servicing Agreement, and
to fund the potential replacement servicer risk, does not constitute anything more
than changing the quantum of a term or condition of the parties’ RSA given that
the Discount Spread only consists of a numerical amount. It is certainly well
within the range of what is permitted under paragraphs 247(2)(a) and (c).
[333]
Adding the servicing
fee discount based upon the Servicing Agreement of 0.17% which Mr. Reifnsyder
believes might be sufficient to also cover off any replacement servicer risk,
to the maximum of the McKesson Group/TDSI approach’s range of replacement
servicer fee discounts of .04 to .08%, the Court estimates, based upon what
evidence was tendered, that the appropriate range of servicing discount in the
Discount Spread of a notional arm’s length RSA would be in the range of 0.17%
to 0.25%.
2. Prompt Payment
Dilutions Discount
[334]
It is not clear why the
parties to the RSA provided that prompt payment discounts are not treated as
deemed receipts but are instead at the purchaser’s risk. Regardless, that is what
the RSA provides and I accept that arm’s length parties might agree to such a
term in a similar receivables financing transaction. The issue is therefore
only whether the prompt payment dilutions risk has been accounted for in the
Discount Spread component of the Discount Rate on arm’s length terms.
[335]
The historic prompt
payment discount levels are very consistent according to the evidence at 0.5% or
0.53% of sales. TDSI tested this on an annual basis over several years and came
back with 0.5%. The Becker Report independently arrived at a three to four year
prompt payment discount experience of 0.5324% of sales.
[336]
I do not accept that
arm’s length parties would agree to a fixed discount spread for the five-year
term to address in a balanced fashion the risk of change to prompt payment
discounts taken by McKesson Canada’s customers. Specifically, I do not accept
that arm’s length parties adverse in interests as to risk and pricing would
agree to the 20% TDSI buffer or to the PwC 5% cushion.
[337]
If arm’s length parties
were to agree to a transfer of prompt payment discount risk (upside and downside)
to the purchaser, I conclude that for the initial purchase in December 2002
this would be based upon the historic 0.5% to 0.53%of sales performance to
date. I do not accept that an arm’s length seller would walk away from the
possibility of favourable variance if it was agreeing to give the buyer a 5% or
20% cushion. I do not accept that it would be fixed for the five-year term. I
do not accept that either party would not require the other party to also
factor the effect of more prompt payments upon DSOs for all purposes of the Discount
Rate.
[338]
In my estimation, based
upon the evidence, arm’s length parties would instead agree to virtually remove
the risk of change during the term of the RSA in the levels of prompt payment
discounts by adopting a three or four month, or annual floating dynamic prompt
payment component to the Discount Spread component of the Discount Rate to
fully capture and reflect the risk of change. This approach is supported by the
evidence of Dr. Becker and Mr. Finard. The Reifsnyder Report was not concerned
with risk of change to prompt payment discount levels because trade receivables
were not considered variable as to prompt payment discount participation. This
would again be permitted by paragraphs 247(2)(a) and (c).
[339]
Using this approach,
the Discount Spread component attributable to prompt payment dilutions would be
0.5% to 0.53% for the initial December 2002 receivables purchase. Since this
number had proven to be very consistent, and there was no evidence to suggest
that was expected to change, or had changed, in any material way in the
following few months, I further estimate that arm’s length parties would have
agreed to use a number in the range of 0.5 to 0.53% to reflect that prompt
payment discounts were borne by the purchaser under the RSA for the initial
purchase in December 2002 and for the remainder of McKesson Canada’s 2003 year.
3. Accrued Rebate
Dilutions Discount
[340]
I do not accept that
the accrued rebate dilutions risk warrants any material discount. The
Appellant’s expert reports did not support one. This thought was that of
McKesson Group and TDSI. Their approach was followed in the PwC Report which
went on to describe the accrued rebate risk as an expected loss that MIH would
suffer. Glucksman did consider it only reluctantly preferring instead to
substitute a reserve or pledge type approach. I have noted my thoughts on the
shortcomings of TDSI’s approach to this above.
[341]
I certainly can not
imagine that McKesson Canada would agree with a notional arm’s length purchaser
(absent a significant corresponding concession or quid pro quo of which
there was none in evidence) to a discount that either (i) reflected a full
recovery for the notional arm’s length purchaser based upon an assumption that
all of McKesson Canada’s customers would exercise a claim to set off their
accrued rebate entitlements, and this at a time when the total accrued rebates
were at an historic high, or (ii) that assigned a credit risk spread to
McKesson Canada equivalent to junk bond issuer status.
[342]
There was no evidence
that McKesson Canada’s customers had ever claimed, threatened or even asked to
set off their accrued rebates against their outstanding payables to McKesson Canada.
[343]
In the circumstances
described herein and the evidence before the Court, it is my estimation that
McKesson Canada would not agree with a notional arm’s length purchaser to any
material discount to reflect the mere possibility of a rebate set-off claim
being made and for which McKesson Canada would not in fact indemnify the
receivables purchaser. There would thus be no need to assign a McKesson Canada
credit risk spread to such an eventuality which is unsupported on the evidence.
[344]
This is supported by the
express evidence of Messrs. Becker and Finard and is consistent with accrued
rebate risk not being identified, much less quantified, as a cash flow dilution
risk of the RSA in the Reifsnyder Report.
[345]
This is also further
supported to an extent by the fact that McKesson Canada was not required to
segregate collections, and was permitted to commingle collections under the RSA
(absent termination). That is, no credit risk was recognized in respect of
McKesson Canada’s ability to pay amounts owing, even very significant amounts.
The McKesson Group did not consider there to be a material risk of McKesson Canada insolvency or bankruptcy, nor any other financial risk arising from the commingling
of funds.
4. Interest Discount
[346]
The interest discount
used by McKesson Group in the RSA 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. I find this
completely unacceptable, unreasonable, unsupported on the evidence, and a term
that would not be agreed to by McKessson Canada in a similar receivables financing
transaction with an arm’s length party adverse in interest as to risk and
pricing.
[347]
As a general rule, the
value of an asset to be sold is not generally affected by a particular
purchaser’s cost of funds. Generally, a business or an investor with cash or a
low cost of funds can profitably make less risky investments with a lower nominal
return on investment than can a person with a high cost of funds. A purchaser’s
cost of funds does not decrease the value of the asset it wishes to buy or the
investment it is considering. Rather, it simply determines whether that
particular purchaser can make the purchase or investment profitably, and if so,
how profitably.
[348]
There was no
satisfactory evidence tendered that would suggest that McKesson Canada was
driven to seek receivables financing from a high cost of funds/high cost
factoring company and not a better funded/lower yield/lower cost major
financial player described in the taxpayer’s own evidence. I was not, however,
provided with evidence of the cost of capital associated with receivables
factoring by major well-funded players.
[349]
In the circumstances of
the McKesson Group RSA in this case, and based upon what evidence I do and do
not have, I find that McKesson Canada would not agree to sell its receivables
at a Discount Rate that incorporated an interest discount to reflect its
notional arm’s length purchaser’s cost of funds at the level set by McKesson
Group in the discount spread component of the RSA’s discount rate of 0.4564%.
Based upon the evidence in this case, I estimate that, for the year in issue,
in a notional arm’s length transaction, a notional arm’s length McKesson Canada
would only agree to an interest discount of between 0.0% and 0.08%, which
latter number reflects a 30 day DSO adjusted rating-derived credit risk spread for
a company having the same rating as McKesson U.S., according to TDSI.
(b) Summary of Court’s Estimate of Discount Ranges
[350]
Tabulating the above
Discount Rate components arrives at the following:
Yield Rate: 0.2293%
Loss Discount: 0.06
– 0.08%
Servicing Discount: 0.17
– 0.25%
Prompt Payment Dilutions Discount: 0.5
– 0.53%
Accrued Rebate Dilutions Discount: 0
Interest Discount: 0
– 0.08%
_______________________________ ________________
Total Discount Rate Range: 0.959%
to 1.17%
[351]
Using the adjusted
Yield Rate of 0.1911 for the 2003 year in issue to reflect the much shorter
expected repayment period for the initial December 2002 $460,000,000
receivables purchase,
this Discount Rate range is reduced to 0.92% to 1.13%. This range will be even
further reduced once the 15 or 16 day DSO is also applied to the DSO adjusted
components of the Loss Discount and Discount Spread.
11. Conclusion on Transfer Pricing Adjustment
[352]
The Court has concluded
that its best estimate of the range of Discount Rate to which arm’s length
parties to a notional arm’s length RSA would agree is between 0.959% and
1.17%.
[353]
The taxpayer has not
been able to establish with sufficient credible and reliable evidence, that the
RSA Discount Rate of 2.206% was computed based upon arm’s length terms and
conditions.
[354]
The evidence does not
show that the Discount Rate used by the Minister of National Revenue (the
“Minister”) in the reassessment of 1.013% was below a Discount Rate computed on
arm’s length terms and conditions for a notional arm’s length RSA. The Court
can not conclude on all of the evidence that the reassessment was incorrect as
it was within the arm’s length range determined by the Court.
[355]
In any event it is not
necessary to fix a particular point within the determined range as the arm’s
length transfer price as, importantly, the taxpayer’s evidence does not
rise to the level of making out a prima facie case that “demolishes” the
key assumptions of fact made by the Minister that support the reassessments.
[356]
Assumption (v) in
paragraph 28 of the Amended Reply is that the Discount Rate that would have
been agreed to had the Appellant and MIH been dealing at arm’s length would
have been established of a rate no greater than 1.0127%. The taxpayer has not
been able to discharge the burden and onus upon it of showing that the amount
of the reassessment is incorrect. The shortcomings of the TDSI Report, the
Reifsnyder Report, and the supporting testimony regarding both, were obvious
and apparent and did not require contrary evidence from the Respondent to make
them evident. For this reason, the taxpayer’s appeal with respect to the
transfer pricing adjustment is dismissed.
[357]
This is an appropriate
result. It would not be appropriate for this Court to order the Minister in a
case such as this to reconsider and to reassess at the high point of the range
of arm’s length Discount Rates (1.17%). That would reward overreaching taxpayers
who would then count on the court process to ensure they enjoyed the highest
permissible transfer price. This would encourage the poor use of public
resources and expenditures. In contrast, in transfer pricing disputes which, as here, often involve
very large amounts, the taxpayer’s costs can be less than the value of even a
slight variance in the underlying price of the inputted asset or service.
Taxpayers would be economically encouraged to use the Court to ensure they get
their maximum transfer price by choosing one that is likely to exceed it.
[358]
Further, the Discount Rate range with respect to the year in
issue is less than 0.959% to 1.17%. Estimating that the midpoint of the 2003
range is the appropriate arm’s length Discount Rate, and after making the
further needed DSO adjustments to components of the Loss Discount and the
Discount Spread described in paragraph 288 above, it appears that in any event
the arm’s length Discount Rate for the 2003 year in issue as determined by the
Court is less than the rate used by the Minister in the reassessment.
12. Timeliness of Part XIII Assessment of
McKesson Canada
a) The Issue
[359]
The vicarious
assessment of McKesson Canada is for its failure to withhold and remit to CRA
on the benefit it paid to its parent (and sole shareholder), MIH, via the
transfer of receivables at an overstated Discount Rate which resulted in it
having given away some of its assets to its parent/shareholder.
[360]
The amount of this
benefit is deemed to have been a dividend paid by McKesson Canada to MIH upon which MIH is subject to Canadian Part XIII non-resident withholding tax. Under
the Canada-Luxembourg Treaty
(the “Treaty”) (which Respondent only admits for purposes for this appeal does
apply) the non-resident withholding tax rate payable on dividend income received
from a Canadian by a Luxembourg resident is reduced from 25% to 5%.
[361]
Distinct from MIH’s
liability for Part XIII tax under the Act on its Canadian-sourced
dividend income, McKesson Canada as payor is obliged under the Act to
withhold from MIH as payee and remit to CRA on behalf of the non-resident, an
amount equal to the amount of non-resident withholding tax payable by the
non-resident payee. The Canadian payor, McKesson Canada, is itself liable under
Part XIII of the Act for an amount equal to the amount that should have
been withheld from the non-resident, and remitted to CRA by it but was not. The
withholding obligation exists to facilitate enforcement and collection in Canada against Canadian payors without needing to pursue non-resident payees. If this
collection mechanism is not complied with, the subsection 215(6) vicarious
direct liability of the Canadian payor for an equivalent amount then further
serves this same purpose. The Act provides that the Canadian can seek indemnity
from the non-resident.
[362]
The transfer pricing
reassessment of McKesson Canada for additional Part I income tax under the Act
was issued by CRA on March 25, 2008.
[363]
The Part XIII
assessment of McKesson Canada for its vicarious liability for an amount equal
to the amount it should have withheld from, and remitted on behalf of, MIH was
issued by CRA on April 15, 2008.
[364]
CRA did not ever assess
non-resident withholding tax against MIH with respect to the non-arm’s length
RSA Discount Rate benefit by imposing Part XIII non-resident withholding tax on
MIH’s Canadian sourced dividend income.
[365]
There was no evidence
that there was a material amount of Luxembourg tax payable by MIH on its
profits under the RSA. The only evidence is that in its 2003 short year the
Canadian tax avoided by McKesson Group was US$4,500,000 and that some form of Luxembourg tax was expected to be payable in the amount of US$29,000. I was similarly not
given any evidence that there was a material or any amount of Luxembourg tax payable by MIH on its deemed dividend under the Canadian Act.
[366]
The Treaty has an
express and specific five calendar year limitation for assessing tax on certain
transfer pricing adjustment income in specific circumstances. CRA was mindful
of this five-year period in issuing the transfer pricing adjustment
reassessment of McKesson Canada, beating that date by a few days. For some
reason that was not explained, the Part XIII assessment was issued to McKesson Canada by CRA three weeks later and outside the five-year period, assuming the Treaty applied
to such an assessment. The Part XIII issue in this appeal is whether Article 9
of the Treaty applies to the assessment of McKesson Canada by the Canadian tax
authorities for its failure to remit to CRA the amount it was required by the Act
to withhold from MIH. It does not involve an assessment by CRA of MIH for
Canadian tax.
b) The Provisions of the Income Tax Act
and the Treaty
[367]
The relevant portions
of the provisions of the Act are as follows:
Part 1, subsection 15(1):
15(1) Benefit
conferred on shareholder -- If, at any time, a benefit is conferred by a
corporation on a shareholder of the corporation, … , then the amount or value
of the benefit is to be included in computing the income of the shareholder,
member or contemplated shareholder, as the case may be, for its taxation year
that includes the time …
Part XIII, paragraph 214(3)(a):
214(3) Deemed payments -- For the purposes of this Part [XIII],
(a) where section 15 or subsection 56(2) would, if Part I were
applicable, require an amount to be included in computing a taxpayer's income,
that amount shall be deemed to have been paid to the taxpayer as a dividend
from a corporation resident in Canada;
Part XIII, subsection 212(2):
212(2) Tax on dividends -- Every non-resident person shall pay an
income tax of 25% on every amount that a corporation resident in Canada pays or
credits, or is deemed by Part I … to pay or credit, to the non-resident person
as, on account or in lieu of payment of, or in satisfaction of,
(a) a taxable dividend … or
(b) a capital dividend.
Part XIII, subsection 215(1):
215(1)
Withholding and remittance of tax -- When a person pays, credits or provides,
or is deemed to have paid, credited or provided, an amount on which an income
tax is payable under this Part [XIII], … , the person shall, notwithstanding
any agreement or law to the contrary, deduct or withhold from it the amount of
the tax and forthwith remit that amount to the Receiver General on behalf of
the non-resident person on account of the tax and shall submit with the
remittance a statement in prescribed form.
Part XIII, subsection 215(6):
215(6) Liability for tax -- Where a
person has failed to deduct or withhold any amount as required by this section
from an amount paid or credited or deemed to have been paid or credited to a
non-resident person, that person is liable to pay as tax under this Part on
behalf of the non-resident person the whole of the amount that should have been
deducted or withheld, and is entitled to deduct or withhold from any amount
paid or credited by that person to the non-resident person or otherwise recover
from the non-resident person any amount paid by that person as tax under this
Part on behalf thereof.
Part XIII,
paragraph 227(10)(d):
227(10) Assessment -- The Minister may at any time assess any amount
payable under
…
(d) Part XIII by a person resident in Canada,
…
[368] Article 9 of the Canada Luxembourg
Treaty provides:
|
Article
9
Associated Enterprises
1. Where
(a) an enterprise of a Contracting State participates directly or indirectly in the management, control or capital of
an enterprise of the other Contracting State, or
|
Article
9
Entreprises associées
1. Lorsque
a) une entreprise d’un État contractant participe directement ou
indirectement à la direction, au contrôle ou au capital d’une entreprise de
l’autre État contractant, ou que
|
|
(b) the same persons
participate directly or indirectly in the management, control or capital of
an enterprise of a Contracting State and an enterprise of the other Contracting State,
|
b) les mêmes personnes participent directement
ou indirectement à la direction, au contrôle ou au capital d’une entreprise
d’un État contractant et d’une entreprise de l’autre État contractant,
|
|
and in either case conditions
are made or imposed between the two enterprises in their commercial or
financial relations which differ from those which would be made between
independent enterprises, then any income which would, but for those
conditions, have accrued to one of the enterprises, but, by reason of those
conditions, have not so accrued, may be included in the income of that
enterprise and taxed accordingly.
|
et que, dans l’un et l’autre cas, les deux
entreprises sont, dans leurs relations commerciales ou financières, liées par
des conditions convenues ou imposées, qui diffèrent de celles qui seraient
convenues entre des entreprises indépendantes, les revenus qui, sans ces
conditions, auraient été réalisés par l’une des entreprises mais n’ont pu
l’être en fait à cause de ces conditions, peuvent être inclus dans les
revenus de cette entreprise et imposés en conséquence.
|
|
2. Where a Contracting State includes in the income of an
enterprise of that State — and taxes accordingly — income on which an
enterprise of the other Contracting State has been charged to tax in that
other State and the income so included is income which would have accrued to
the enterprise of the first-mentioned State if the conditions made between
the two enterprises had been those which would have been made between
independent enterprises, then that other State shall make an appropriate
adjustment to the amount of tax charged therein on that income. In
determining such adjustment, due regard shall be had to the other provisions
of this Convention and the competent authorities of the Contracting States
shall if necessary consult each other.
|
2. Lorsqu’un État contractant inclut dans les
revenus d’une entreprise de cet État — et impose en conséquence — des revenus
sur lesquels une entreprise de l’autre État contractant a été imposée dans
cet autre État, et que les revenus ainsi inclus sont des revenus qui auraient
été réalisés par l’entreprise du premier État si les conditions convenues
entre les deux entreprises avaient été celles qui auraient été convenues
entre des entreprises indépendantes, l’autre État procède à un ajustement
approprié du montant de l’impôt qui y a été perçu sur ces revenus. Pour
déterminer cet ajustement, il est tenu compte des autres dispositions de la
présente Convention et, si c’est nécessaire, les autorités compétentes des
États contractants se consultent.
|
|
3. A Contracting State shall not
change the income of an enterprise in the circumstances referred to in
paragraph 1 after the expiry of the time limits provided in its national laws
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 conditions
referred to in paragraph 1, have accrued to that enterprise.
|
3. Un État contractant ne rectifiera pas les
revenus d’une entreprise dans les cas visés au paragraphe 1 après
l’expiration des délais prévus par sa législation nationale et, en tout cas,
après l’expiration de cinq ans à dater de la fin de l’année au cours de
laquelle les revenus qui feraient l’objet d’une telle rectification auraient,
sans les conditions visées au paragraphe 1, été réalisés par cette
entreprise.
|
|
4. The provisions of paragraphs 2 and
3 shall not apply in the case of fraud or wilful default.
|
4. Les dispositions des paragraphes 2
et 3 ne s’appliquent pas en cas de fraude ou d’omission volontaire.
|
c) Positions of the Parties
[369]
It is the Appellant’s
position that Article 9(3) of the Treaty applies and the assessment was barred
by Article 9(3) as it was issued outside of the five-year period.
[370]
It is the Respondent’s
position that, while CRA’s Part XIII assessment for McKesson Canada’s vicarious
liability for an amount equal to the amount that should have been withheld and
remitted to CRA by it when it paid MIH is the same as the amount of Canadian tax
that would have been payable by MIH on its deemed dividend income, the
standalone obligation of McKesson Canada under subsection 215(6) of the 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 resulting from the benefit
conferred and the resulting deemed dividend.
[371]
The Respondent also
argues that the description of income in Article 9 of the Treaty does not
extend to deemed dividend income. It argues that a deemed dividend that accrues
precisely because of the non-arms’ length relationship can not be considered to
be income described in Article 9(1). The Crown argues that neither a benefit
nor a deemed dividend could have accrued to MIH if the non-arm’s length
conditions were removed from the RSA that is, if the Discount Rate had been the
appropriate arm’s length discount rate. This would be the case only if the RSA
Discount Rate were computed on arm’s length terms and the resulting higher
purchase price was paid by MIH for the transferred receivables, or if less than
all of the receivables portfolio had been transferred to reflect the RSA’s
understatement of value. In either case, there would be no benefit conferred
under the adjusted transactions with the result that there would be no deemed dividend
and there could only be dividend income from McKesson Canada arising to MIH if
McKesson Canada paid a dividend or conferred a benefit outside the RSA once
adjusted to reflect arm’s length terms. The Crown argues that a real dividend
could have accrued to MIH if the non-arm’s length conditions were removed from
the RSA but only if McKesson Canada had separately declared a dividend for that
particular given amount.
d) The Interpretation of Treaties
[372]
The Vienna
Convention on the Law of Treaties provides that a treaty is to be
interpreted in good faith in accordance with the ordinary meaning to be given
to the terms of the treaty in their context and in light of its object and
purpose. It also authorizes regard to subsequent practice in the application of
the treaty in certain circumstances and for certain purposes, as well as the
use of other supplementary means of interpretation when the interpretation of
the treaty otherwise leads to a result which is manifestly absurd or
unreasonable.
[373]
In The Queen v.
Crown Forest Industries Limited et al., 95 DTC 5389, the Supreme
Court of Canada wrote: “In interpreting a treaty, the paramount goal is to find
the meaning of the words in question. This process involves looking to the
language used and the intentions of the parties.” The Court went on to quote
approvingly from Addy J. in Gladden Estate v. The Queen,
85 DTC 5188, wherein he wrote at p. 5191:
Contrary
to an ordinary taxing statute a tax treaty or convention must be given a
liberal interpretation with a view to implementing the true intentions of the
parties. A literal or legalistic interpretation must be avoided when the basic
object of the treaty might be defeated or frustrated insofar as the particular
item under consideration is concerned.
[374]
Both the Vienna
Convention and the Supreme Court of Canada in Crown Forest confirm that
“literalism has no role to play in the interpretation of treaties”: Coblentz
v. The Queen, 96 DTC 6531 (FCA).
[375]
In Crown Forest
the Supreme Court of Canada 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.
[376]
The Preamble to the
Canada-Luxembourg Treaty sets out its purposes of avoiding double taxation of
income earned by a resident of one country from sources in the other country,
and of preventing fiscal evasion. In Crown Forest the Supreme Court of
Canada held that the purposes of the Canada-U.S. Treaty also included the
promotion of international trade between the two countries and the mitigation
of administrative complexities arising from having to comply with two
uncoordinated taxation systems.
[377]
In The Queen v.
Prévost Car Inc., 2009 FCA 57, 2009 DTC 5053, the
Federal Court of Appeal affirmed the possible relevance of the OECD
Commentaries to the OECD Model Convention, including commentaries subsequent to
a particular treaty being entered into.
e)
Analysis
[378]
Paragraph 1 is the
primary transfer pricing adjustment paragraph of Article 9 of the Treaty. For
purposes of McKesson Canada’s appeal, it provides that if either (a) MIH
controls McKesson Canada or (b) McKesson U.S. participates directly or
indirectly in the management or control of both MIH and McKesson Canada, and
(c) the conditions of their financial or commercial relations differ from those
conditions which would be made between independent parties, then: (d) any
income which would have accrued to McKesson Canada but for those differing
conditions but did not so accrue because of those conditions, may be included
in McKesson Canada’s income and subject to Canadian tax.
[379]
Article 9(1) is in
question in this appeal because it is referenced in paragraph 3 of Article 9.
It can be noted that Article 9(1) clearly permits either state, Canada or Luxembourg, to tax either company if the preconditions are triggered. That is, in
appropriate circumstances Canada is permitted by Article 9(1) to make a
transfer pricing adjustment to MIH’s income subject to tax in Canada, such as if MIH carried on business in Canada in which it entered into non-arm’s length
transactions on non-arm’s lengths terms with non-arm’s length parties. This
makes obvious sense. An issue raised in this part of McKesson Canada’s appeal is whether Article 9(1) also addresses Canada indirectly adjusting the amount of
deemed Canadian sourced dividend income of MIH resulting from its shareholder
benefits or appropriations from McKesson Canada by virtue of the overstated
Discount Rate in the RSA.
[380]
Paragraph 2 is the
corresponding adjustment paragraph of Article 9. For purposes of McKesson Canada’s appeal, it provides that if Canada includes a transfer pricing adjustment amount in McKesson
Canada’s income and MIH has already paid Luxembourg income tax on that amount,
then Luxembourg shall make the appropriate corresponding adjustment to the Luxembourg income tax paid by MIH.
[381]
Article 9(2) is not at
all in question in this appeal. It can be noted that Article 9(2) clearly only
permits a state to adjust the tax paid to it by the enterprise of that state.
This also makes obvious sense.
[382]
In this appeal there
was no evidence that MIH paid any Luxembourg income tax on the income generated
by it under the RSA from McKesson Canada’s receivables. The only evidence was
Mr. Brennan’s handwritten note that some undescribed tax could be expected to
be payable to Luxembourg as a result of the purchase and collection of McKesson
Canada’s receivables under the RSA. This was not described as an income tax.
There was also no evidence that any adjustment to MIH’s Luxembourg tax was needed to relieve any double tax, or whether such relief was either requested by
MIH or granted by Luxembourg.
[383]
Paragraph 3 of Article
9 is the paragraph which imposes a five-year limitation period for making
certain described transfer pricing adjustments. For purposes of McKesson
Canada’s appeal of its Part XIII assessment for the amount it should have
withheld from MIH and remitted to CRA in respect of MIH’s Part XIII Canadian
tax liability on the benefit (or deemed dividend) of the overstated Discount
Rate and resulting underpayment by MIH to McKesson Canada under the RSA for the
transferred receivables, paragraph 3 provides that Canada shall not change the
income of MIH in the circumstances referred to in paragraph 1 after a specific
five-year period. That period is five years from the end of the year in which
the income of MIH sought to be adjusted by Canada would have accrued to MIH but
for the conditions referred to in Article 9(1).
[384]
Clearly Article 9(3)
provides a maximum five-year limit (except in cases of wilful default or fraud) for
either state to make an Article 9(1) transfer pricing adjustment. It also
clearly provides the same time limit on the other state having to make the
corresponding adjustment on its counterparty under Article 9(2).
[385]
It is clear that Article
9(3) can apply to an assessment by either country of either the Canadian or Luxembourg party since it refers to the circumstances referred to in Article 9(1) which can
so apply.
[386]
McKesson Canada argues that Article 9(3) also prevents Canada from assessing it under subsection 215(6) after
March, 2008. In order for this argument to prevail, the following requirements
of Article 9(3) (and by cross reference, Article 9(1)) must be met:
(i) the subsection 215(6)
assessment of McKesson Canada must be a change in the income of MIH.
(ii) that adjustment of
MIH’s income must be in the circumstances referred to in Article 9(1), namely:
(a) MIH controls McKesson Canada or both MIH and McKesson Canada are indirectly managed or controlled by McKesson U.S.;
(b) The conditions of the
financial or commercial relations between MIH and McKesson Canada differ from the conditions which would have been made between independent parties;
(c) The income adjustment
is income which would have accrued to MIH, not McKesson Canada, but for those differing conditions in their financial and commercial relations; and
(d) Canada sought to add the income adjustment to MIH’s income and taxed it accordingly.
(iii) a period of five
years must have passed since the end of the year in which the income of MIH
sought to be changed would, but for the conditions which differ from what
independent parties would agree to, have accrued to MIH.
[387]
In my view this
argument can not prevail because Canada’s subsection 215(6) assessment of
McKesson Canada does not satisfy all of these requirements.
[388]
Firstly, I question
whether a subsection 215(6) vicarious assessment of a Canadian payor for
failure to remit and withhold tax is a change by Canada of MIH’s income
(requirement (i) above), or constitutes Canada seeking to add a transfer
pricing adjustment amount to MIH’s income and tax that increased amount
(requirement (ii)(d) above).
[389]
I am more inclined to
see it as an enforcement and collection provision than a tax charging
provision. Subsection 215(6) permits CRA to assess the Canadian payor an amount
determined by reference to the amount it should have remitted to CRA but did
not, which withholding amount is in turn determined by reference to the amount
of Canadian tax that would have been payable by the non-resident payee. In the
circumstances, however, I do not have to decide this point to dispose of this
appeal.
[390]
It does, however,
appear clear that an assessment of McKesson Canada for its failure to withhold
and remit does not constitute Canada adding the transfer pricing income
adjustment to MIH’s income and then taxing it accordingly (requirement (ii)(d)
above). Adding it to MIH’s income and taxing it accordingly requires that Canada sought to tax MIH.
[391]
Secondly, these
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 RSA transactions (requirement
(ii)(c) above). 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 RSA had 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.
[392]
Had the RSA used an
arm’s length discount rate and not the non-arm’s length Discount Rate actually
used, the adjustment permitted by Article 9(1) can only be the additional
McKesson Canada income. There would have been no excess benefit to, or
appropriation by, MIH to be taxed as a deemed dividend, and there would not
have been any actual dividend either unless a dividend was declared and paid by
McKesson Canada to MIH which was also not the case. Clearly, Article 9(3) can
not be read to apply to MIH’s deemed dividend arising from the non-arm’s length
Discount Rate having been paid. For this reason alone, the taxpayer’s appeal
can not succeed in respect of the Part XIII assessment.
[393]
Thirdly, this same
fatal problem arises equally clearly yet again in respect of the Article 9(3)
requirement (described in (iii) above) 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 in the RSA,
have accrued to MIH. Again, had an arm’s length Discount Rate been used in the
RSA in McKesson Canada’s year ending March 31st, 2003, the
additional income would have accrued to McKesson Canada not MIH, as MIH would
have paid McKesson Canada more for its receivables. Clearly, the “but for”
wording of the Treaty requires the arm’s length conditions be substituted for
the non-arm’s length conditions and, if this is done, the Article 9(1) and 9(3)
adjusted income amounts can only be read as amounts that would have accrued to
McKesson Canada, not MIH. This is a third independent reason why Article 9(3)
can not relieve McKesson Canada from its liability under the Part XIII
assessment for its failure to withhold and remit upon transferring its
receivables to its non-resident parent for less than their value after agreeing
to an excessive Discount Rate in the RSA.
[394]
If I may use the term
primary transfer pricing adjustment to describe an Article 9(1) adjustment such
as Canada’s addition of additional income to McKesson Canada, and the term corresponding
transfer pricing adjustment to describe any corresponding downwards adjustment
that may be made by a treaty partner to the counterparty’s income from
non-arm’s length transactions by virtue of Article 9(2), I could only describe
any taxation by the country making the primary transfer pricing adjustment of
the excess amount of money wrongly appropriated by the counterparty in the
other country, whether by way of deemed dividend or otherwise, as a secondary
adjustment relating to the primary transfer pricing adjustment but not itself
capable of being a primary transfer pricing adjustment described in Article
9(1) or thus by Article 9(3). Such secondary related adjustments can never meet
the requirements of these paragraphs of Article 9 and therefore these
shareholder benefits and appropriations of their subsidiary’s cash or valuable
assets can not benefit from the provisions of these paragraphs.
[395]
This does not appear to
be an inappropriate result when looked at from a Treaty purpose or a policy
point of view. There was virtually no evidence of double taxation of the same
income. At most, US$29,000 dollars of Luxembourg tax may have been payable on
some basis by MIH as a result of the RSA. That was not on any shareholder
benefit or appropriation or on any dividend or deemed dividend income. That
amount is minuscule when compared with the millions of dollars of Canadian tax
sought to be avoided. The double tax, if any, may have been the subject of a
request by MIH for a corresponding adjustment from Luxembourg. I can assume
that, if circumstances in Luxembourg warranted a corresponding adjustment, the
McKesson Group’s Tax department would have applied for it.
Further, if McKesson Canada has a complaint that it has effectively had to pay
MIH’s Canadian tax on MIH’s deemed dividend income arising from its shareholder
benefit or appropriations from McKesson Canada, the Act gives McKesson Canada the right to seek indemnity from MIH. Further, given their common control within the
McKesson Group, MIH might be expected to simply indemnify McKesson Canada without McKesson Canada pursuing MIH.
(f) Conclusion re: Part XIII and the
Treaty
[396]
In conclusion, the five
year limitation period in Article 9(3) of the Treaty does not apply to the
assessment of McKesson Canada’s vicarious liability for the amount of Part XIII
tax payable by MIH which results from McKesson Canada’s failure to withhold and
remit such amount. As shown above that is the result of the clear wording of
Article 9(3), consistent with the overall context of Article 9, and consistent
with the purposes of Treaty. Paragraph 227(10)(d) of the Act otherwise
permits a subsection 215(6) assessment to be made at any time. For these
reasons, the taxpayer’s appeal of its Part XIII assessment is also dismissed.
13. Dismissal of Appeals
[397]
The taxpayer’s appeals
are dismissed, with costs.
Signed at Edmonton, Alberta this 13th day of December 2013.
"Patrick Boyle"
Email this Content