|
Citation: 2003TCC544
|
|
Date: 20030905
|
|
Docket: 1999-3593(IT)G
|
|
BETWEEN:
|
|
CIT FINANCIAL LTD.,
|
|
Appellant,
|
|
and
|
|
|
|
HER MAJESTY THE QUEEN,
|
|
Respondent.
|
REASONS FOR JUDGMENT
BOWMAN, A.C.J.
A. Introduction
[1] This appeal is from an assessment
for the 1993 taxation year of Commcorp Financial Services Inc.
("Commcorp") one of a number of companies that
amalgamated to form the appellant, which was formerly called
Newcourt Financial Ltd.
[2] Originally, there were several
issues but the parties have settled all but two. After the
remaining issues have been dealt with by the Court, counsel have
agreed to prepare a draft judgment in which all of the issues are
disposed of.
[3] The remaining issues have to do
with capital cost allowance ("CCA") on software
acquired by Commcorp in 1993. The Minister of National Revenue
characterized the series of transactions and each transaction
within the series whereby Commcorp acquired the software and then
leased it back ultimately to the person who had developed it and
used it in its business as an avoidance transaction within the
meaning of the General Anti-Avoidance Rule ("GAAR") and
determined the tax consequences to Commcorp to be that it was not
entitled to claim any CCA on the software.
[4] The alternative position is that
Commcorp acquired the software from a person with whom it was not
dealing at arm's length and the cost to it should be reduced
to the fair market value ("fmv"), which the respondent
says was not greater than $13,100,000.
[5] The appellant concedes that
Commcorp acquired the software from a person with whom it was not
dealing at arm's length, but contends that the value at the
date of the acquisition was $33,091,255.
[6] The Crown has abandoned the
arguments contained in subparagraphs 13(a)(i), (ii) and
(iii), that the software was not acquired by Commcorp, that if
acquired it was not acquired for the purpose of gaining or
producing income or that the transaction or documents were a
sham.
[7] This leaves then the question of
the application of GAAR and the question of the fmv of the
software as well as the question of reasonableness under
section 67.
B. The transactions - April 1, 1993
[8] The transactions that took place
in April, 1993 are described in a Partial Statement of Agreed
Facts ("PSAF"), to which are attached the relevant
documents. Attached to the PSAF are two charts, Exhibits A and B,
in which the legal relationships and flow of funds are set out.
These two charts are attached to these reasons.
[9] The series of transactions can be
summarized briefly.
[10] BHP New Zealand Steel Limited (BHP) is
a steel manufacturing corporation resident in New Zealand. It
owned computer applications software (the "software")
that it used in operating its production function at its
integrated steel-making facility at Glenbrook, New Zealand and
stored data in relation to the financial management of the
operation.
[11] Commcorp received an appraisal of the
software from MACC and Partners Australia Limited at NZ
$50,000,000. The Canadian dollar equivalent was $33,091,255. For
convenience I shall from time to time in these reasons call this
figure $33,000,000.
[12] BHP sold the software to Macquarie
Leasing (NZ) Limited ("MLL") a subsidiary of Macquarie
Bank Limited ("MBL"), an Australian Bank, for
Cdn.$33,091,255 for which a promissory note was given.
[13] MLL sold the software to 1004583
Ontario Ltd. ("1004583") an Ontario company
incorporated for that purpose by Commcorp's lawyers for
$33,091,255 for which a promissory note was given.
[14] 1004583 leased and licensed the
software to Eagle Financial and Leasing Services Limited
("Eagle"), a Cayman company and a subsidiary of
Barclays Bank PLC ("Barclays"), a UK bank.
[15] The lease was for a term of 11 years
expiring April 5, 2004. The lease included purchase options
exercisable on April 5, 2002 and April 5, 2004 at an amount equal
to the present value, calculated in 2003, of the remaining income
stream.
[16] Commcorp borrowed $27,770,896.80 from
Barclays to be secured by an assignment of specific amounts of
rent owing under the lease to Eagle and a portion of the amounts
payable upon termination of the lease which corresponded to the
outstanding balance of the loan at the relevant time. Recourse
under the loan was limited to the assigned payments.
[17] Eagle and MLL entered into a sublease
and license agreement whereby Eagle subleased and licensed the
software to MLL on substantially the same terms as the software
had been leased and licensed by 1004583 to Eagle. One difference
was that MLL was obliged on closing to prepay to Eagle the total
rent payable under the sublease and the first option price.
[18] MLL and BHP entered into a
sub-sub-lease and license agreement under which MLL sub-subleased
and licensed the software to BHP on substantially the same terms
as the lease and sublease, except that MLL was obliged at the
request of BHP to accept prepayment of $31,933,061 in
satisfaction of BHP's obligation under the lease.
[19] 1004583 entered into an agreement with
Commcorp to sell the software to Commcorp on April 5, 1994 for
$33,091,255.
[20] 1004583, Commcorp, Eagle, MLL and BHP
and Whitaker Nominees Limited, a New Zealand corporation
("Whitaker") entered into an escrow agreement under
which BHP was to deliver to Whitaker a copy of the source code,
object code and software documentation for the software, a copy
of the software and the original valuation. The deposit of this
material with Whitaker, as escrow agent, was acknowledged to be
delivery of the software under the three software sale and
assignment agreements. (i.e. BHP to MLL, MLL to 1004583 and
1004583 to Commcorp)
[21] Commcorp, Barclays and Eagle entered
into the Eagle Support agreement under which Eagle agreed to pay
to Barclays $3,996,708. Barclays also agreed that wherever a
"specified payment" became payable under the lease
Barclays would pay or cause Eagle to pay out of the proceeds of
the bond which Barclays agreed would be purchased with the
$3,996,709. The "specified payments" were the scheduled
rents and other amounts (including termination amounts) payable
under the lease except to the extent that they were not assigned
as security under the loan.
[22] Barclays wrote to Commcorp agreeing
that it would ensure that Eagle would meet its obligations under
the agreements to which it was a party.
[23] MLL, Eagle, Commcorp and 1004583
entered into a coordination agreement under which the obligations
under the lease were divided into "Recourse covenants"
which MLL agreed to perform and "Limited recourse
covenants" which Eagle agreed to perform. Eagle gave MLL a
power of attorney to exercise all of the rights of Eagle under
the lease. As part of that agreement Commcorp executed a
Non-Disturbance undertaking under which Commcorp agreed to be
bound by the terms of the sub-sublease even if there was a
default by Eagle or MLL under the lease or the sublease so long
as BHP was not in default under the sub-sub lease.
[24] MBL guaranteed the obligations of MLL
under various agreements.
[25] The moneys flowed as contemplated by
the various agreements all from and to accounts of the parties at
Barclays and the transactions closed on April 5, 1993. Paragraphs
11 and 12 of the PSAF summarize what happened that day:
11. On April 5, 1993 (see
attached Exhibit A):
(a) BHP assigned and
transferred the Software to MLL pursuant to the BHP NZS Software
Sale and Assignment Agreement and MLL executed a Promissory Note
in favour of BHP in the amount of $33,091,255.00 in satisfaction
of the purchase price payable by MLL for the Software,
(b) MLL assigned and
transferred the Software to 1004583 pursuant to the MLL Software
Sale and Assignment Agreement and 1004583 executed a Promissory
Note in favour of MLL in the amount of $33,091,255.00 in
satisfaction of the purchase price of the Software,
(c) 100483
(sic) leased and licensed the Software to Eagle pursuant
to the Lease,
(d) 1004583 assigned
and transferred the Software to Commcorp pursuant to the Commcorp
Software Sale and Assignment Agreement,
(e) Eagle subleased
and sublicenced the Software to MLL pursuant to the Sub-Lease and
Licence Agreement.
The documents included in Tabs 26-50 were also executed in
connection with the above.
12. On April 5, 1993, the
flow of funds was as follows:
(a) Barclays
advanced $27,770,896.80 to Commcorp pursuant to the Loan;
(b) Barclays
transferred the amounts referred to in the Directions, in
accordance with the Directions, namely:
(i)
$33,091,255 from Commcorp to 1004583;
(ii) $1,100,000 from
Commcorp to MLL;
(iii) $33,091,255 from
1004583 to MLL; and
(iv) $33,091,255 from MLL
to BHP;
(c) BHP paid the
amount of $31,933,061 to MLL, as prepayment of the rents and
purchase option under the Sub-Sub Lease and Licence
Agreement;
(d) MLL paid the
amount of $31,767,604 to Eagle, as a prepayment of the rents and
purchase option under the Sub Lease and Licence Agreement;
(e) Eagle paid
Barclays the amount of $3,996,708, in accordance with the Eagle
Support Agreement;
(f) Barclays
purchased the Bond for $3,996,709 in accordance with the Eagle
Support Agreement, which Bond matured on March 15, 2002 for the
amount of $8.13 million;
(g) The amount of
$27,770,897 ($31,767,604-$3,996,708) remained within Eagle and
provided funds to make the lease payments to secure and discharge
the Loan.
Attached as Exhibit B is a diagram of the cash payments that
were made on April 5, 1993. The amount of $1,100,000 referred to
above was paid by Commcorp to MLL as fees in connection with the
transaction.
[26] I need not summarize the other formal
steps taken in connection with the transactions including legal
opinions. These matters were fully and competently dealt with by
major law firms throughout the world. The transactions and the
documents that underlay them represent genuine, legally binding
and enforceable obligations. They were what they purported to
be.
[27] A few of the other facts as set out in
the PSAF warrant being reproduced because they were considered
significant by one or other of the parties and were referred to
in argument:
18. The purchase price of
the Software was funded by Commcorp using the $27,770,897 to be
borrowed from Barclays and $6,420,358 from internal sources. This
amount includes $1,100,000 which Commcorp paid to MLL as fees in
respect of the transaction. Eagle paid $3.996M to Barclay. The
balance of the prepayment rested with Eagle.
19. Before entering into
the deal, Commcorp knew that BHP would pay an amount to MLL and
MLL would pay an amount to Eagle which was sufficient to cover
the obligations of Commcorp to Barclays over the life of the
loan.
20. There was no
indication to Commcorp that either BHP or Macquarie needed
financing. This was not a deal whereby BHP or Macquarie obtained
financing from Commcorp.
21. The net cash flow to
Commcorp over the term of the deal regardless of whether the
first purchase option is exercised is $1,711,907 (as set out
Schedules I and II at Tab 72). The payments under the Lease and
the principal interest payments required under the Loan are set
out in Schedules at Tab 72. This schedule also includes details
of the tax deductions anticipated in respect of CCA and interest
and the income inclusions for tax purposes.
22. In computing its
income for the 1993 taxation year (ending December 31, 1993),
Commcorp included the amount of $34,191,255 in its cost of
depreciable property in Class 12 of Schedule II to the Income Tax
Regulations and deducted the amount of $17,313,740 as capital
cost allowance under paragraph 20(1)(a) of the Income Tax Act in
computing its income for that year, which amount was disallowed
as a deduction by the Minister of National Revenue in reassessing
the Appellant.
23. The arrangement in
issue in this case was described to the Board of Directors of
Commcorp (Tab 73) as a "tax predicated computer software
lease that will provide Commcorp with $15MM of tax shelter in
1993 and an additional $15MM in 1994".
24. It was understood by
Commcorp before entering into the deal that Eagle would sub-lease
and licence the software to MLL and MLL would sub-sub-lease and
licence the software to BHP. It was also understood that BHP and
Macquarie would prepay their obligations under the Sub-Sub-Lease
and Licence Agreement and the Sub-Lease and Licence Agreement,
respectively.
25. The Appellant entered
into two other deals in the following year based on the same
model, involving prepayments to the leasee which was a subsidiary
of the lending bank. They are referred to in Minutes of the Board
of Directors Meeting of September 21, 1994 (Tab 74) as "tax
advantaged software leases booked by the Corporation".
26. Commcorp's primary
purpose in entering into the arrangement in issue in this case
was to obtain the capital cost allowance arising from the
acquisition of the Software in that the capital cost allowance
was deductible in the first two years and the rents were
receivable over the term of the lease.
[28] I am attaching as Exhibits I and II,
the Schedules at tab 72 of the PSAF because they illustrate the
real economics of the transaction. Obviously the predominant
economic motivation from Commcorp's standpoint lay in the tax
write off. This is not disputed. Presumably there were tax
advantages in New Zealand to BHP, but I do not know exactly what
they were. They are not germane to this case. The schedules
however demonstrate something else - the essential circularity of
the transaction. $33,000,000 was inserted at one end and it ended
up back where it started. The only thing that did not move was
the software. It stayed where it started - with BHP. No one -
Commcorp, the bank, Eagle, BHP or anyone else in the chain - bore
the slightest risk. The $33,000,00 was based on a valuation by
MCC which had been selected by MBL who had an interest in seeing
the deal go through because of the $1,100,000 commission that MLL
was to receive. The $33,000,000 valuation bore little relation to
the actual cost of developing the software which was more like
$11,000,000.
[29] The simple fact is any number would
have worked because the money all came back to the starting
point. I shall deal with this point more fully when I discuss the
fmv of the software.
[30] The third reason that the schedules are
important is that they illustrate the difficulty of determining
the tax consequences to deny the tax benefit "as is
reasonable in the circumstances". Here the Minister has
decided that this is an avoidance transaction not saved by
subsection 245(4). The tax benefit is the deduction of CCA
on the software. Therefore, so the reasoning goes, the deduction
of the CCA should be denied but all of the other consequences
over the term of the transaction from 1995 to 2004 are left
untouched. Admittedly, the GAAR is something of a blunt
instrument, but while it may not be a scalpel neither is it a
sledgehammer.
[31] The large income and deduction
anticipated over the years from 1995 to 2004 are undoubtedly
predicated on a cost of $33,000,000 with its consequent tax write
off. The GAAR seems to have been enacted to enable the Minister
to combat overly aggressive tax avoidance schemes whose fiscal
purposes far overshadowed their commercial purposes. It should
not be used as an instrument to punish people for engaging in tax
avoidance schemes that the Minister does not like by taking away
the fiscally beneficial aspect of an avoidance transaction but
leaving intact the detrimental aspects.
[32] I turn first to the question of the fmv
of the software. Two valuations were submitted in evidence by the
appellant. The first valuation was prepared by Mr. Geoffrey H.
Cooper (the "MACC valuation") before the lease was
entered into. The second valuation was made by Mr. Peter Hatges
of KPMG. The Crown did not have an opportunity of examining the
software before trial because it was not available. When it
became available I adjourned the trial to permit the Crown's
experts to examine it, but apparently the state it was in after
10 years, with all of the changes that continued use over that
period entailed, made any meaningful determination of value as of
April 1993 impossible. The Crown therefore relied on reports
prepared by Mr. Howard E. Johnson of Campbell Valuation Partners
Limited in which he comments on the appellant's expert
reports.
[33] Before discussing the reports, let us
look at just what we are trying to do here. Commcorp bought the
software in a non-arm's length transaction and therefore the
price under section 69 is deemed to be the fmv. The primary
purpose of the purchase was to obtain a tax write off. Therefore
one would not expect a party such as Commcorp or anyone else in
the chain such as MLL, 1004583 or Eagle or, indeed, BHP, to be
concerned about striking a deal that bore any relationship to its
inherent commerciality because there was no commercial risk
involved to anyone. We are not dealing with a parcel of land or
shares in a company. The property to be valued is a unique,
special purpose software package developed by BHP for its own
purposes in running its steel mill in New Zealand. Its true value
to BHP in the conduct of its business is unknown. It could vary
within a range of indeterminate magnitude depending on the
criteria used. The conventional definition of fmv is too well
known to require repetition but it involves postulating a
hypothetical vendor and purchaser who are at arm's length,
knowledgeable and canny and who would like to make a deal but are
not desperate to do so and deciding what sort of a bargain these
hypothetical negotiators would strike. As Viscount Simon said in
Gold Court Selection Trust Limited v. Humphrey (Inspector of
Taxes), [1948] A.C. 459 at 473.
... If the asset is difficult to value but is none the
less of a money value, the best valuation possible must be made.
Valuation is an art, not an exact science. Mathematical certainty
is not demanded, nor indeed is it possible. It is for the
commissioners to express in the money value attributed by them to
the asset their estimate, and this is a conclusion of fact to be
drawn from the evidence before them.
[34] The difficulty with attempting to
determine the fmv of the software here is that there is no
evidence that software that is specifically designed and
developed to run a particular steel mill in New Zealand is likely
to be bought and sold on the open market. I find it difficult to
believe that a steel company is some other part of the world
would pay anywhere near $33,000,000 for software designed and
developed for BHP's steel mill in New Zealand. The uniqueness
of the software is illustrated by section 3.0 of the MACC report
which reads:
3.0 The
Computing Environment in BHP NZS
BHP NZS is New Zealand's largest steel producer employing
some 1,900 staff. The company operates an integrated steel mill,
comprising iron making plant, steelmaking plant, treatment plant,
continuous casting machines, hot strip mill, cold reduction mill,
tube and pipe mill and coil coating lines. The company's
approach to steel making is different to most in that it uses
iron bearing sand as a raw material rather than iron ore. The
company exports the majority of its output (approx. 66%) and
hence needs to be competitive in the international market place.
It does this through being a niche supplier and differentiating
itself from major producers by being flexible as to the quantity
of a product that can be produced and providing responsive client
service. The ability to change schedules and produce small order
quantities economically is critical to its
competitiveness.
BHP NZS is relatively progressive in its implementation of
computerised processing management systems and computer based
business systems. The company has prepared a Computer Services
Business Plan which outlines how CSD will support the operations
of the company from the conversion of the raw material through to
the despatch of the finished product. The plan identifies the
following 6 levels of computerisation.
Level 5
-
Corporate System
Level 4
-
Business Support
Level 3
-
Process Management
Level 2
-
Process Control
Level 1
-
Real Time Control
Level 0
-
Sensors and Activators
The plan identifies and maps the extent of integration between
each level. The knowledge of and adherence to this plan appeared
to be well understood and accepted throughout the total
organisation. In fact CSD was seen to be a critical element in
the overall achievements of the organisations's strategic and
annual plans.
While the business and computer information systems had been
addressed in the past, the current focus is on computer
integrated manufacturing (C.I.M.) which is the horizontal and
vertical integration of systems, thereby allowing each level of
management access to key data through which decisions,
optimisation and continuous improvement can be made.
Full implementation of all C.I.M. systems is planned to lead
to further cost reductions, e.g. through inventory reduction and
cheaper process costs, because of the increased co-ordination
that will be possible between individual plants.
The approach to application development revolves around heavy
user involvement in all development activities - in fact for
system development projects the delineation between user and CSD
staff is hard to determine, so integrated is the team. While
there are formally some 55 staff in CSD there are at least 150
staff company wide involved in computer application development
and support roles.
The company has adopted an "open system" and
distributed processing approach to computer application
development using a variety of languages (LINC, ORACLE, FORTRAN)
running on multiple hardware platforms (UNISYS, IBM, DEC, SUN). A
site wide network based on fibre optics links all processors and
facilitates data integration.
[35] I have not reproduced the lengthy
section 4.0 (Functional and Technical Evaluation of the Software
except for one short passage:
Based on our review the following broad conclusions were
formulated regarding the appropriateness and maintainability of
the Software:
· While three
different programming languages were used for the application
development, each language is operationally efficient and well
supported. The decision as to which programming language to use
was based [on] the hardware platform, nature of processing and
likely future technological directions.
· There was a
heavy emphasis on user involvement and user responsibility
acceptance in most stages of software development. This
relationship between CSD and users was one of close
co-operation.
·
Considerable effort was devoted to alignment of the application
requirements with the business needs. This was a substantial
factor in ensuring the delivery of effective, supportive and
functional applications.
· The
Software code appears to be efficiently written and structured to
ensure that efficiencies were gained in the overall application
development and maintenance tasks.
· An
"open" system approach has facilitated the integration
of applications whereby data can be transmitted between
applications regardless of technical platform.
· Extensive
user functions have been incorporated into the Software to
support the business operations.
· The
Software has been developed using modular programming techniques
which aid the system modification of enhancement effect.
· The
Software is well documented from a technical and user
perspective.
· While not
formally documented, industry standards and methodologies current
at the time of development appear to have been used in the
development process.
· The
Software appears to be well maintained with new functionality
resulting from user interaction and requests for enhancements and
modifications.
· The
Software has been developed according to a formal strategic plan
which identifies the discrete applications and the extent of
(vertical and horizontal) integral/relationship with other
applications.
[36] These passages as well as other
passages in the report illustrate the excellence of the software
in its application to the business of BHP. I accept this
conclusion but what it demonstrates is that however valuable it
may be to BHP in running its business it could not be readily
adapted to anyone else's business. Its very value to BHP is
in inverse relationship to its value to another mill and hence
would adversely affect what the software would fetch on the open
market. Put differently, any company that wanted to install a
computer system to run its steel mill would find it easier and
cheaper to develop its own system than to pay $33,000,000 or any
other amount for BHP's system and then try to adapt it. The
more precisely tailored it is to BHP's business and hence the
greater its commercial value to BHP, the less useful or valuable
it will be to another user.
[37] Faced with the difficulty of finding an
open market for the software the MACC report has adopted a
valuation method that takes into account replacement cost and
historical effort. The following appears in the report:
For the purposes hereof, our conclusions as to the value of
the Software at any time represents the price, expressed in terms
of money, obtainable for the Software in an open and unrestricted
market between informed and prudent parties acting at arm's
length and under no compulsion to transact.
This type of wording with slight variations appears in
virtually all valuation reports. It is the traditional definition
of fmv used by valuators. I do not take exception to it as a
definition. It is not however what the MACC report is
determining. No open and unrestricted market for software of this
type has been identified. That is why replacement cost had to be
resorted to as the only available method that would yield a
value.
[38] In Aikman v. R., [2000] 2 C.T.C.
2211, affirmed [2002] 2 C.T.C. 147 (F.C.A.) I expressed serious
reservations about the use of replacement cost to determine fmv
and considered a number of cases where the courts had to
determine the fmv of property for which there was no open market.
The property there was a disassembled prototype of a
lighter-than-air aircraft called the Cyclo-Crane. In that case,
however, I had the evidence of a recent arm's length
purchase. I do not view the series of purchases for $33,000,000
as affording evidence of fmv.
[39] In this case we have the experts for
both parties using replacement cost as an acceptable method of
determining fmv and, as Viscount Simon said in the Gold
Coast case, one has to do the best one can. Moreover the
replacement cost of a property that is developed for use in
carrying on a business may be some evidence of fmv in the absence
of an open market whereas the replacement cost of a museum
artifact with no commercial function is clearly not a reliable
test.
[40] Since all the experts agree that
replacement cost is an acceptable test and since I have nothing
else to go on, I shall deal with the evidence of value on that
basis.
[41] The MACC report states:
We have formed an opinion as to the value of the Software
using the following valuation methodologies:
· Replacement
cost - the value calculated to be the current cost of replacing
the Software in terms of design, development together with user
training and documentation.
· Historical
effort - the effort incurred by BHP NZS in design, development
and implementation of the Software.
In determining the value of the Software a two staged approach
was adopted, which involved:
· Estimating
the value of the Software based on quantitative methods related
to replacement cost.
· Adding a
factor to replacement cost based on the perceived intellectual
knowledge, special skill required to design and develop the
Software, its importance to the business operations of the group,
and its contribution to earnings.
[42] In his testimony Mr. Cooper stated that
the words "and implementation" should be deleted. I
agree. The cost of implementation forms no part of replacement
cost. However the rest of the evidence of Mr. Cooper does not
support the contention that he did not include the cost of
implementation.
[43] In determining the replacement cost of
the software Mr. Cooper used several methods:
(i) Lines of code. This involves
counting the number of lines of operational code (or program
statements) and then applying an average number of lines
completed per day to arrive at a total number of person days for
this application.
(ii) Development time. This
involves counting the total time required to perform all
activities and tasks associated with the development of the
software. Mr. Cooper refers to a book by Capers Jones
"Applied Software Measurement" where the observation is
made that corporate trading systems accidentally omit 30 percent
to 70 per cent of effort in the development of software products.
Accordingly he made upward adjustments to take this factor into
account.
(iii) Function point analysis.
Function points are the weighted sums of five or six different
factors such as algorithms, inputs, outputs and so forth and
determining how many can be completed per person month.
(iv) Backfire conversion. This is
simply a method of verifying the function point analysis by
working backwards to the source code.
(v) Estimation formulae. This involves
applying formulae to the number of lines of code produced.
I shall
not discuss this method further because Mr. Cooper did not
consider it reliable and rejected it. I can see why. It resulted
in a replacement cost of $58.7 million (NZ), over $10 million
higher than the next highest method.
[44] The result of Mr. Cooper's analysis
using the various methods is as follows:
Lines
of code
- $47,000,000 (NZ)
Development
time
- $41,100,00 (NZ)
Backfire
conversion
- $52,000,000 (NZ)
[45] Mr. Cooper, after setting out these
numbers, makes a number of comments on the variances. One such
observation is the following:
There are variances in value between methods applied across
the applications. This demonstrates the volatility of factors
used in the calculations.
He concludes that the current replacement cost of the software
is NZ$47,000,000. To this he adds another NZ$3,000,000 because of
the competitive advantage that the use of the software gives
it.
[46] Frankly, one does not need the
assistance of an expert to see the fallacy of this reasoning. If
you are determining replacement cost you do so by the best method
you can. You do not add to the replacement cost that you have
determined some arbitrary percentage based on an unrelated factor
having to do with how good the product is.
[47] Before I deal with the rebuttal report
prepared by the respondent's expert witness there is one
piece of evidence that sticks out like a sore thumb. On
March 16, 1993 a lawyer acting on behalf of BHP sent a
letter to the Commissioner of Inland Revenue of New Zealand
setting out the development costs for each application. The
figures contained in that letter, as well as those in the MACC
report, are as follows: (they were reproduced in
Mr. Johnson's report):
Application
|
|
Tax Letter
|
|
MACC Report
|
|
|
|
|
|
|
|
Sales Order Administration ('MARKET')
|
NZ$
|
1,230,000
|
NZ$
|
6,710,000
|
|
Product Design System
|
|
725,000
|
|
6,380,000
|
|
Master Production Scheduling ('MPS')
|
|
670,000
|
|
3,910,000
|
|
Galvanising ('GOSPR')
|
|
820,000
|
|
3,960,000
|
|
Supply System
|
|
1,200,000
|
|
4,400,000
|
|
Financial System Database ('FINSYS')
|
|
890,000
|
|
4,600,000
|
|
Cost Management System ('CMS')
|
|
610,000
|
|
2,665,000
|
|
Rolling Mills Production ('ROLLPC')
|
|
1,370,000
|
|
5,600,000
|
|
Primary Plant Level 3 ('PPL3')
|
|
3,000,000
|
|
8,165,000
|
|
CPD System
|
|
675,000
|
|
3,800,000
|
|
|
|
|
|
|
|
Total
|
NZ$
|
11,190,000
|
NZ$
|
50,000,000
(rounded)
|
|
Conversion Rate to Cdn.$ at the Valuation Date
|
|
0.6618
|
|
0.6618
|
|
|
|
|
|
|
|
Total - Cdn.$
|
Cdn.$
|
7,405.823
|
Cdn.$
|
33,091.255
|
[48] In the MACC report reference was made
to the Capers Jones book Applied Software Measurement which
states that most companies tracking systems do not record between
30 and 70 percent of the real effort that goes into software.
These percentages are unsubstantiated and may be arbitrary
estimates. Nonetheless, let us accept them and see where they
take us.
[49] If we add the lower and higher
percentages for slippage to the figures given to the
New Zealand tax authorities of NZ$11,190,000 we arrive at
figures of NZ$14,587,000 and NZ$19,023,000 respectively or
Cdn.$9,627,204 and Cdn.$12,589,421.
[50] I do not intend this calculation to be
a stand-alone valuation but it demonstrates how far out of line
the MACC report is. Moreover it is consistent with the figures in
Mr. Johnson's report.
[51] Mr. Johnson's criticisms of the
MACC report are specific and detailed. His conclusions are
summarized in a letter to respondent's counsel.
In summary, based on our review and analysis, in our view, the
MACC Report's conclusion of NZ$50 million (approximately
Cdn.$33 million) for the fair market value of the Software at the
Valuation Date likely is significantly overstated. As explained
more fully in Appendix A to this letter, the reasons for our view
are:
· a letter to the
Commissioner of the Inland Revenue Department of New Zealand
dated March 16, 1993 wherein BHP New Zealand Steel
('BHP NZS') purports the development cost of the
Software to be NZ$11,190,000 (approximately Cdn.$7.4 million);
and
· specific issues
in the MACC Report, which suggest that its value conclusion is
overstated. These include the:
√ inclusion of implementation and
training time as a component of replacement cost, which costs do
not form part of replacement cost for the Software program itself
from the standpoint of Commcorp Financial Services Inc.
('Commcorp'). Such costs may represent between 20% and
35% of the total time expended,
√ daily labour charge rate of NZ$700
adopted by MACC, which may be overstated by 20% to 30%,
√ double-counting of overhead costs
relating to management and administrative time, which may
overstate the adjustment for 'involvement of non-CSD
staff' by 5% to 10%,
√ double-counting of software
development time in respect of 'general purpose code' for
certain applications, which may overstate the value of the
Software by approximately NZ$2 million (approximately
Cdn.$1.3 million), and
√ application of a 'business
factor' premium, which may overstate the fair market value of
the Software to Commcorp by NZ$3 million (approximately Cdn.$2.0
million).
Adjusting for the apparent errors and inconsistencies in the
MACC Report based on our analysis, the fair market value of the
Software at the Valuation Date falls in the range of
approximately NZ$18.5 million to NZ$27.3 million, or
approximately Cdn.$12 million to Cdn.$18 million.
Finally, notwithstanding the adjustments that may be
appropriate as outlined above, the MACC Report's conclusion
that the residual value of the Software 9 years following the
Valuation Date is 35% of the initial fair market value (being
NZ$17.5 million, or approximately Cdn.$11.6 million) likely
is significantly overstated as well.
[52] He supports these conclusions in
appendices to the letter. I will not reproduce them here except
for the calculation below. I found Mr. Johnson an impressive
witness and I accept his conclusions. His recalculation of the
fmv of the software is set out in Schedule 2 as follows:
Schedule 2
Recalculation of Fair Market Value of the
Software
|
|
|
Low
|
|
High
|
|
|
|
|
|
|
|
Fair market value per MACC Report
|
NZ$
|
50,000,000
|
NZ$
|
50,000,000
|
|
|
|
|
|
|
|
Business factor premium
|
|
(3,000,000)
|
|
(3,000,000)
|
|
General purpose code
|
|
(1,880,000)
|
|
(2,115,000)
|
|
|
|
|
|
|
|
Sub-total
|
|
45,120,000
|
|
44,885,000
|
|
|
|
|
|
|
|
Management and administrative time
|
|
10%
|
|
5%
|
|
|
|
(4,512,000)
|
|
(2,244,250)
|
|
|
|
|
|
|
|
Sub-total
|
|
40,608,000
|
|
42,640,750
|
|
|
|
|
|
|
|
Average daily labour charge rate
|
|
30%
|
|
20%
|
|
|
|
(12,182,400)
|
|
(8,528,150)
|
|
|
|
|
|
|
|
Sub-total
|
|
28,425,600
|
|
34,112,600
|
|
|
|
|
|
|
|
Implementation and training costs
|
|
35%
|
|
20%
|
|
|
|
(9,948,960)
|
|
(6,822,520)
|
|
|
|
|
|
|
|
Adjusted fair market value
|
NZ$
|
18,476,640
|
NZ$
|
27,290,080
|
|
|
|
|
|
|
|
Conversion Rate to Cdn.$ at the Valuation Date
|
|
0.6618
|
|
0.6618
|
|
|
|
|
|
|
|
Total - Cdn.$ (rounded)
|
Cdn.$
|
12,000,000
|
Cdn.$
|
18,000,000
|
[53] The MACC report calculates the
residual value of the software at the end of nine years to be
35 percent of its fmv in 1993. The implausibility of this
assertion is illustrated in Mr. Johnson's report where
he says:
Residual Value Determination
The MACC Report estimates the residual value of the Software
at the end of 9 years to be 35% of its 'Assumed Value'
(i.e. NZ$17.5 million).
When determining the residual value of the Software, the MACC
Report assumes (at Section 8.0) that the Software will not be
maintained (or modified by Commcorp). Conversely, in
Section 10.0, the MACC Report states that "much of the
Software is "state of the art" hence to maintain its
competitive advantage and value the Software needs to be
continually modified and enhanced to cater for changing
requirements and business needs". Accordingly, it seems
unrealistic that the Software would retain a significant portion
of its original value (35%) 9 years after the Valuation Date,
with no maintenance or enhancements, when such things are
considered critical by MACC.
The residual value of 35% after 9 years implies an average
compound 'physical depreciation' rate of approximately
11% per annum over that period which, in the absence of
appropriate maintenance and enhancements, further serves to
illustrate the implausibility of MACC's residual value
assumption. By way of comparison, at Section 11.0, the MACC
Report states that the residual value of the Software at the end
of its remaining life (of 12.75 to 12.8 years from the Valuation
Date as estimated by MACC) will be approximately 5% of its value
at the Valuation Date (in constant dollars). Accordingly, MACC
has assumed that the fair market value of the Software will
decline at an average annual compound rate of approximately 11%
from 1993 through 2002, but then at an average annual compound
rate of approximately 40% from 2002 through the end of 2005.
[54] I agree. The idea that software
developed in 1993 to run a steel mill would after nine years
retain 35 percent of its value strikes me as well beyond the
realm of possibility. The fact that the Crown was unable, in
2003, to determine the fmv of the software because of the
extensive changes in it over 10 years illustrates how useless it
would be without constant upgrading. It is common knowledge that
software becomes obsolete very quickly and the evidence here
confirms this fact.
[55] Even if we applied the unrealistic
percentage of 35 percent to Mr. Johnson's estimate
of between $12,000,000 and $18,000,000 (Cdn.) we still end up
with $4,200,000 to $6,300,000. This strikes me as high. I do not
think that realistically one can justify a figure in excess of 15
percent of the fmv in 1993 and this is probably generous.
[56] The Crown's own expert gives us a
fmv in 1993 of between $12 and $18 million (Cdn.). I am not bound
to accept this or any other expert opinion and I think that the
actual costs of developing the software as supplied to the
New Zealand tax authorities are as reliable an indication of
the fmv of this property as one is likely to find, at least as a
starting point. Therefore, I think the figure of $18,000,000 is
too high.
[57] So far as the slippage factor suggested
by Capers Jones of between 30 and 70 percent is concerned
the 30 percent figure may be realistic whereas the 70 percent
factor applied to the BHP figure puts it roughly into the Johnson
range of $12,000,000. I think it is fair to give the appellants
the benefit of the higher figure which results in a fmv of
$13,100,000. I arrive at this figure by adding 70 percent to
$7,405,823, the Canadian dollar equivalent of NZ$11,190,000, to
arrive at $12,589,899 (Cdn) and rounding it up to Cdn.$13,100,000
which is the figure stated in the Reply to the Notice of Appeal.
This is within the range suggested by Mr. Johnson. Thus we have a
substantial convergence of these calculations - the figure in the
Crown's reply, the range suggested by the Crown's witness
and the figure arrived at by taking BHP's own recorded costs
and adding the high end of the Capers Jones slippage factor.
[58] I can deal briefly with the KPMG
report. The rent payable under lease of the software by 1004583
to Eagle is based upon the $33,000,000 MACC valuation. The KPMG
report does not value the software. It values the lease based on
the rentals which are themselves based on a valuation that in my
view is about $20,000,000 too high. If you start from a value of
$33,000,000, base your rents on that figure knowing that the
money is going to come back to where it started it is not
surprising if you end up with a value for the lease that is equal
to $33,000,000.
[59] If they had started with a figure of
$100,000,000 and based the rentals on that amount they could have
determined the discounted cash flow to be $100,000,000 and
therefore the value of the software with the lease to be that
amount. Mr. Johnson commented on the KPMG report as
follows:
All monetary amounts set out in this letter are expressed in
Canadian dollars.
In our view, KPMG Report's conclusion that the fair market
value of the Software at the Valuation Date was approximately
$34 million is unsubstantiated. The principal reasons for
our view are that:
• KPMG does not
specifically address the value of the Software, but rather the
lease agreement in respect of the Software. KPMG assumes that its
calculation of the present value of the Software lease payments
can be taken as the fair market value of the Software. Based on
its scope of review, it is evident that KPMG did not perform any
meaningful assessment of the Software itself that would have
enabled it to reasonably estimate the fair market value of the
Software independent of the Software lease;
• the approach
adopted by KPMG represents a circular calculation. The lease
payments were established based on the Software purchase price of
approximately $33 million, which in turn was linked to the report
prepared by MACC Partners Australia Limited ('MACC')
dated April 5, 1993 (the 'MACC Report'). Therefore, it is
not surprising that KPMG's calculation of the present value
of the lease payments approximates the purchase price amount. If
this were not the case, it would represent an arbitrage profit
opportunity for either the lessor or the lessee. Therefore,
KPMG's approach serves only to verify the mathematical
accuracy of the lease payments. It does not address whether the
fair market value of the Software on which those lease payments
were established is reasonable. As explained in our letter dated
April 11, 2003 containing our comments on the MACC Report, in our
view MACC's estimate of the fair market value of the Software
of approximately $33 million likely is materially overstated. As
a result, KPMG's attempt to ascribe the present value of the
lease payments to the fair market value of the Software serves to
reinforce the errors made by MACC in its determination of the
fair market value of the Software; and
• the discount rate
adopted by KPMG of 8.64% represents a near risk-free rate of
return. While such a rate might be appropriate when valuing the
Software lease, given the secure nature of the lease payments, it
does not represent a reasonable rate of return by which to
determine the fair market value of the Software itself.
Commercial rates of return required in the development and sale
of complex proprietary software programs are significantly higher
than risk-free rates of return given the inherent risks
involved in such activities. Had KPMG adopted a commercial rate
of return reflective of the risks of the Software itself, its
calculation of the fair market value of the Software would have
been significantly less than $34 million.
[60] I accept these conclusions. It is
in my view artificial to base a valuation on the discounted cash
flow under a lease where the rentals are themselves based upon an
excessive valuation and in which it makes no difference what the
rentals are because they substantially come back in one form or
another to where they started.
[61] During argument counsel for the
appellant used an analogy of two identical buildings, side by
side, where one is completely leased on long term leases to
triple A tenants such as a provincial or federal government
department and the other on short leases to hippies and beatniks.
He says the former building would fetch a much better price than
the latter. No doubt he is right but the analogy does not stand
up. In the case of the two buildings the leases were negotiated
in the open market. Here the lease was simply one link in a
closed circuit.
[62] The KPMG report confirms the
mathematical accuracy of the arrangement but it does not prove
anything about fmv.
[63] The rest of these reasons will
proceed on the basis that the fmv of the software as of April 5,
1993 was $13,100,000. The result that flows from this under
section 69 is that Commcorp, who acquired the software from
1004583 in a non-arm's length transaction, acquired it from
1004583 at a cost of $13,100,000. I make no finding on the
question of the $1,100,000 paid as a fee to MLL. It was included
by Commcorp in the $34,191,255 which it claimed was the capital
cost of this software. No separate argument was made on this
item. Since the parties will be drafting the formal judgment in
any event they can deal with this point or, failing agreement, I
can be spoken to.
[64] This leaves then the question of
reasonableness under section 67 and GAAR under section 245.
[65] Section 67 may in some cases
perform the same function in an arm's length situation as
section 69 does in a non-arm's length one. Generally it is
unreasonable to pay in excess of fmv for property and this is
true whether the payor is at arm's length with the vendor or
not. Section 69 provides an automatic mechanism to reduce an
excessive price to fmv where the parties are not at arm's
length. In an arm's length relationship section 67 does not
automatically apply. Here, however, it is subsumed in the
application of section 69. Put differently, section 67 can,
at least in this case, add nothing to what section 69 has already
done.
[66] What, then, about GAAR? The
application of GAAR involves a number of steps. GAAR is a weapon
of last resort to be invoked only where a transaction or series
of transactions perceived as having as their purpose the
avoidance of tax work, i.e. achieve their intended fiscal result.
If the transaction does not work apart from GAAR in any event
there is no need to invoke GAAR. If it does work the next step is
to determine whether there is an avoidance transaction resulting
in a tax benefit that should be denied. Before coming to the
determination of the method of denying the tax benefit "as
is reasonable in the circumstances" it must be determined
whether it may be considered that the transaction would result
directly or indirectly in a misuse of the provisions of the Act
or an abuse having regard to the provisions of the Act other than
section 245, read as a whole. If it is decided that subsection
(4) does not exclude the operation of subsection (2) the
consequences must be determined under subsection (2). In
applying subsection (2), subsection (5) presents a wide but
not exhaustive list of things that may be done.
[67] Subsection (7) requires that the
tax consequences must be determined only through a notice of
assessment, reassessment, additional assessment or
determination.
[68] On an appeal from an assessment
in which GAAR is applied, the Court is entitled to consider every
one of the elements described above. If it considers that GAAR
applies it is entitled to decide whether the Minister's
determination of the tax consequences is appropriate and, if it
concludes that it is not, the Court may substitute its own
determination. Section 245 gives the Minister no discretionary
powers either in deciding that GAAR applies or in deciding the
appropriate remedial action to be taken. Therefore the
court's powers on an appeal from an assessment made under
section 245 are at least as far reaching as the
Minister's.
[69] Here the appellant admits that
the transaction is an avoidance transaction as defined in
subsection (3) but contends that there is no misuse or abuse as
contemplated by subsection (4) and accordingly subsection (2)
does not apply.
[70] The Minister has denied the
entire claim for CCA. This strikes me as unreasonable and an
overreaction. Section 245 is not a penal section. It is not
contended by the respondent that the transactions were shams,
that the software was not acquired or that it was not acquired
for the purpose of gaining or producing income. These arguments
were originally pleaded as assumptions but were dropped at trial.
Had such arguments been advanced and accepted there would have
been no need to invoke section 67 or 69. The result of accepting
any of these arguments would have been a denial of the entire CCA
claim. GAAR would not even have come into the picture. What I
find rather odd is for the Minister to have "assumed"
facts that would completely destroy the claim for CCA and then
"assume" that GAAR applies. The GAAR assumption is
logically inconsistent with the other assumptions pleaded. They
cannot stand together. This does not seem to have troubled the
Minister who shares, as I assume he does, Emerson's view of
inconsistency.
[71] In a case such as this one must
first consider whether other sections of the Act are
effective to eliminate or attenuate the beneficial tax result
sought by the taxpayer. This would include specific rules such as
section 55, the at-risk rules, the leasing property rules,
section 67, or section 69, to mention only a few. If what remains
after the application of the specific rules is still a result to
which the Minister believes that GAAR should be applied then it
must be considered after the specific sections have been
considered and, if possible, applied. GAAR does not subsume or
encompass the other sections of the Income Tax Act, nor is
it a substitute for them.
[72] If one proceeds from the twofold
premise that the software was acquired for the purpose of gaining
or producing income (and the acceptance of this proposition is
implicit in the Crown's abandonment of the opposite
assumption) and that its cost was $13,100,000 then what room is
left for GAAR? The transaction even ceases to be an avoidance
transaction and the abuse that results from claiming CCA on an
artificially high capital cost is completely eliminated by
section 69.
[73] If I had concluded that GAAR had
applied I would have determined that the appropriate means of
denying the tax benefit was to base the CCA on the fmv of the
software and not to deny it entirely. It is no abuse of
subsection 20(1) of theIncome Tax Act or the
Regulations to claim CCA on property at the favourable
rates provided for that property. The abuse lies in claiming CCA
on an artificially inflated price. That, however, is not an abuse
or misuse of the provisions of the Act that requires GAAR.
It is an abuse that is readily counteracted by
section 69.
[74] If one applies GAAR and bases the
claim for CCA on the fmv of the software or if one applies
section 69 and bases the claim for CCA on the fmv precisely the
same result is achieved. If the same result can be achieved
without section 245 as with it obviously, as a provision of
last resort, section 245 need not be resorted to and it has
therefore no application. There is no need to invoke a general
anti-avoidance provision to do what a specific provision can do
simply and efficiently.
[75] The appeal is allowed and the
reassessment is referred back to the Minister of National Revenue
for reconsideration and reassessment on the basis that the
software was acquired by Commcorp from 1004583 at its fair market
value of Cnd.$13,100,000.
[76] The parties are directed to
prepare a draft judgment incorporating the conclusions stated in
these reasons as well as any other matters in this litigation
that they have settled.
[77] Failing agreement on costs the
parties should communicate with the Court to determine a suitable
date for making representations.
Signed at Ottawa, Canada this 5th day of September, 2003.
A.C.J.
Email this Content