Professional Documents
Culture Documents
R
in Accounting
Vol. 9, No. 1 (2014) 1–57
c 2014 R. Sansing
DOI: 10.1561/1400000037
Richard Sansing
Tuck School of Business at Dartmouth
Hanover, USA
and
CentER, Tilburg University
The Netherlands
richard.c.sansing@tuck.dartmouth.edu
Contents
1 Introduction 2
2 Historical Perspective 7
5 Optimization Models 21
5.1 Transfer price discretion . . . . . . . . . . . . . . . . . . . 21
5.2 Two sets of books . . . . . . . . . . . . . . . . . . . . . . 24
5.3 Endogeneity of comparable uncontrolled prices . . . . . . . 29
5.4 Transfer pricing and investment incentives . . . . . . . . . 31
ii
iii
8 Formulary Apportionment 45
9 Future Research 48
9.1 Methods of shifting income from intangible assets . . . . . 48
9.2 Allocating the returns to organizational capital . . . . . . . 49
10 Conclusions 50
Acknowlegements 52
References 53
Abstract
2
3
the sine qua non of the MNE. About 40% of U.S. international trade
occurs between a U.S. firm and a related party in a foreign country
[Clausing, 2003]. The Organization for Economic Co-operation and
Development (OECD) estimates that more than 60% of world trade
takes place between related parties in MNEs [Wittendorf, 2010, p. 5].
For some pairs of countries, 75% of cross-boarder trade occurs within
MNEs [Brem, 2004]. The increase in intrafirm trade means that transfer
pricing has become much more important over time.
If the MNE could pick any price in an unconstrained fashion, it
could shift all of its taxable income to the country with the lower
income tax rate.1 Therefore, tax authorities have coordinated on the
arm’s length standard for determining whether the resulting allocation
of taxable income is valid for tax purposes. The arm’s length standard
involves a thought experiment, in which one asks what allocation would
have arisen had the parties been unrelated, each striving to maximize
its own income. This approach raises serious conceptual difficulties,
as the economic theory of the firm indicates that two related entities
within a vertically integrated firm will interact quite differently than
will independent firms.
The research on international transfer pricing is found in journals
that reflect very different research traditions. Leading papers can be
found in economics journals, accounting journals, law reviews, and
tax practitioner journals. This monograph puts more weight on the
accounting and tax practitioner literatures, while striving to incorpo-
rate the most important contributions from the economics and legal
research literatures. Eden [1998] and Wittendorf [2010] provide more
comprehensive surveys of the economics and legal research literatures,
respectively. Transfer pricing is just one part of how MNEs are taxed;
Blouin [2011] provides a review of the larger literature on the taxation
of MNEs.
It is important to understand and appreciate the history of a field
of study in order to put current disputes into perspective. Section 2
1
Transfer prices are not the only way to shift income from a high-tax rate country
to a low-tax rate country. For example, an MNE can reduce taxable income in a
high-tax country by having high-tax rate affiliates do all of the borrowing or making
all of the tax-favored R&D investments.
4 Introduction
7
8 Historical Perspective
p. 92]. After World War II, the Organization for European Co-operation
(OECC), which became the OECD in 1961, continued to embrace this
approach.
In the U.S., the arm’s length standard was included in §45-1(b) of
the 1935 U.S. Treasury Regulations. §45 was renumbered as §482 in
1954. Treasury Regulations interpreting §482 were finalized in 1968.
The regulations provided a hierarchy of methods, in which the compa-
rable uncontrolled price method had the highest priority, followed by
resale price, cost plus, and other methods in that order. The compa-
rable uncontrolled price (CUP) method uses the price of a compara-
ble good or service in a transaction between unrelated taxpayers as a
benchmark for evaluating whether the transfer price satisfies the arm’s
length standard. The resale price method uses the gross profit expressed
as a percentage of the sales price of an unrelated seller of goods as
a benchmark. An arm’s length transfer price results in the related
party seller of goods having the same gross profit percentage as the
unrelated seller. The cost plus method uses the gross profit expressed
as a percentage of cost of an unrelated manufacturer of goods as a
benchmark. An arm’s length transfer price results in the related party
manufacturer having the same gross profit percentage as the unrelated
manufacturer.
The most contentious issues that arose after the 1968 regulations
were issued dealt with intangible assets. U.S. firms could transfer intan-
gible assets to foreign subsidiaries in low-tax jurisdictions as tax-free
contributions to capital under IRC §351. Because the foreign subsidiary
is then the owner of the intangible asset, the taxable income from the
intangible asset would be allocated to the foreign subsidiary rather
than the U.S. parent, even if the U.S. parent had deducted research
and development costs while developing the asset.
Congress responded in 1976 by denying tax-free treatment to
the transfer of intangible assets to foreign subsidiaries. In addition,
Congress amended IRC §482 in 1986 by mandating that income from
transferred or licensed intangible assets must be commensurate with
the income (CWI) attributable to the intangible asset. The Treasury
Department then published a White Paper, A Study of Intercompany
9
The difference between the two arises when a profit level indicator
includes, but is not limited to, the profit from the controlled transac-
tion. The OECD guidelines would appear to not allow the use of such
a profit indicator; the Treasury Regulations would.
In addition, the OECD guidelines express a preference for the tradi-
tional transaction methods (CUP method, resale price, and cost plus)
over the TNMM method or profit-split methods, with CUP being
favored among the traditional transaction methods. Paragraph 2.3 of
11
Cheng [1995] and Levy and Wright [1995] compare the U.S. regu-
lations to the OECD guidelines. Although the Final Regulations are
more similar to the OECD guidelines than were the Proposed Regula-
tions, important differences remain. The most important differences are
between the comparable profit method and the transactional net mar-
gin method, the possibility of periodic adjustments to implement the
commensurate-with-income standard, and the OECD’s preferences for
particular methods as opposed to the best method rule in the Treasury
Regulations.
3
Evidence of Income Shifting
12
3.2. Intangible assets 13
during these years. Firms shifted income to the U.S. in 1987, a year in
which the U.S. corporate income tax rate decreased.
Bartelsman and Beetsma [2003] examine income shifting by
manufacturing firms within 16 OECD countries from 1979 to 1997.
They find evidence of income shifting among OECD countries that is
both statistically and economically significant. The authors estimate
that at least 65% of the additional tax revenue that a government
would expect to receive due to a unilateral tax rate increase is lost due
to income shifting.
Clausing [2009] examines firm responses to tax rate differences for
U.S. MNEs from 1982 to 2004. The study looks at both account-
ing responses, such as using transfer prices to shift income, and real
responses, such increasing investment or employment in foreign coun-
tries. A one percentage point decrease in the foreign tax rate is asso-
ciated with a 0.5 percentage point increase in the foreign affiliate’s
return on sales. A one percentage point decrease in the foreign tax rate
is associated with a 1.6% increase in foreign employment, 2.9% increase
in foreign sales, and a 4.8% increase in foreign assets.
Huizinga and Laeven [2008] examine parent companies and sub-
sidiaries of European MNEs using 1999 data. They find evidence of
income shifting by European MNEs away from high-tax countries,
notably Germany, and toward European countries with lower tax rates.
low-tax countries and out of the U.S., and shifting income out of high-
tax countries and into the U.S. As in Harris [1993], firms with high levels
of intangible assets and debt financing were more likely to engage in
income shifting.
Grubert and Slemrod [1998] use 1987 data to investigate the joint
decision to invest in Puerto Rico and shift income from the U.S. to
Puerto Rico. They find that a large fraction of U.S. investment into
Puerto Rico is attributable to the ability to engage in income shifting.
Firms with high levels of intangible assets are more likely to invest in
Puerto Rico, consistent with the income-shifting hypothesis.
Grubert [2003] extends this research to examine the relations among
intangible assets, production location, and income shifting in 60 loca-
tions that account for virtually all investment by U.S. MNEs. The study
finds that about half of the difference in profitability between low-tax
rate and high-tax rate countries is attributable to income shifting asso-
ciated with R&D investments. Furthermore, MNEs with high levels of
R&D are more likely to invest in countries with very high and very low
statutory tax rates, consistent with the strong income shifting benefits
associated with intangible assets.
Some studies examine data at the product level rather than the firm
level to investigate tax-motivated income shifting. Swenson [2001]
examines U.S. imports from Canada, France, Germany, Japan, and
the U.K. from 1981 to 1988 using product-level data. The study finds a
statistically significant albeit economically small relation between tax
rate changes and price changes; a 5% decline in the foreign tax rate is
associated with a 0.024% price increase.
Using international trade price data from 1997 to 1999, Clausing
[2003] finds that intrafirm trade with low-tax countries is associated
with lower export prices and higher import prices, both of which have
the effect of shifting taxable income to the lower tax rate country. A
one percentage point lower tax rate is associated with export prices
3.3. Product level data 15
that are 0.94% lower and import prices that are 0.64% higher, both
relative to non-intrafirm transactions.
Using Census Bureau and Customs Bureau data, Bernard et al.
[2006] find that the arm’s length price of undifferentiated exported
goods is 8.8% higher when sold to an unrelated buyer compared to a
foreign related party. The difference is 66.7% for differentiated goods.
The difference is negatively associated with the income tax rate in the
foreign country; a one percentage point decrease in the foreign tax rate
is associated with a price increase of about 0.6%.
4
Transfer Pricing Fundamentals
16
4.2. Economic profits or accounting profits? 17
the cost of capital associated with producing q units is r(q), then the
correct representation of (4.1) should be:
(1 − τ )[pq − m(q)] − r(q). (4.3)
The fact that the normal return to equity capital is an economic cost
but not an accounting cost is important when analyzing transfer pric-
ing methods based on “costs.” In a competitive equilibrium, abnormal
returns are zero, and thus expression (4.1) is equal to zero. But when
applying the cost-plus method, “cost” simply refers to accounting costs;
the markup on accounting costs provides the normal return to capital.
Treas. Reg. §1.482-3(d)(3)(ii)(A) makes this clear:
A producer’s gross profit provides compensation for the per-
formance of the production functions related to the product
or products under review, including an operating profit for
the producer’s investment of capital and assumption of risk.
The difference between economic costs and accounting costs is easily
forgotten, however. Berry [1989] discusses the economic underpinnings
of the various transfer pricing methods. He discusses a hypothetical
competitive button-making industry:
At arm’s length, no manufacturer would long produce these
buttons for returns less than those available from using
the injection-molding facilities for some other purpose, and
since the buttons are not (by assumption) patented or
trademarked, no manufacturer will long be able to sell such
buttons at an usual profit without attracting the entry of a
competitor. The manufacturing price, therefore, that would
be expected to prevail at arm’s length can be estimated
by adding the profit margin generally realized by indepen-
dent manufacturers of similar molded plastic products to
the actual costs of the captive manufacturer.
Consider this claim in light of expression (4.3). Expression (4.3)
will equal zero for all manufacturers in a competitive equilibrium.
20 Transfer Pricing Fundamentals
21
22 Optimization Models
country imposes a tariff, the choice depends joint on the tariff rate and
the income tax rate.
Samuelson [1982] follows the general approach in Horst [1971], but
notes that if either the output market in the selling country is imper-
fectly competitive or marginal production costs are not constant, the
firm’s choices of production quantities will affect the ultimate selling
price, the marginal cost of production, or both. Because the assumption
that the tax authorities will accept any transfer price within that range
implies that the transfer price will be one of those two endpoints, the
transfer price becomes endogenous to the firm’s choices. Horst acknowl-
edges this possibility, but dismisses it as small, asserting: “In words,
the firm can ignore the way an increase in sales or production would
affect the upper and lower limits on the export transfer price, π.”
Because the firm always uses either the upper or lower bound of the
range of transfer prices, the choice of output quantity affects not only
the firm’s pretax income but also the transfer price used to allocate
taxable income between the two countries. As a result, the firm will
not produce so as to equate marginal revenue and marginal cost, as it
does when the range of transfer prices is exogenous to the firm’s choices.
The comparative statics, showing the effects of a tax rate change on
output quantity, are different in the exogenous and endogenous cases
when the firm prefers a high transfer price.
Diewert [1985] adopts a slightly different approach to the modeling
of transfer price rules. In that study, the firm can choose its transfer
price from any element in the interval [w1 , w2 ]. As in Horst [1971], the
firm always chooses one of the endpoints as its transfer price for tax pur-
poses. As in Horst, the endpoints are exogenous. However, the interval
itself could be completely disconnected from the underlying economics
of the model, whereas Horst’s range was from a low of marginal pro-
duction cost to a high of output price.
Diewert’s study focuses on the efficiency consequences of transfer
prices. He first characterizes the production decisions that maximize
firm income in a setting without taxes. The key feature in the model is
the existence of an intermediate good that a division operating in one
country produces and transfers to a division operating in a different
5.1. Transfer price discretion 23
Many studies in this literature examine the dual roles played by transfer
prices. One role is determining the allocation of taxable income within
an MNE. The other role is a mechanism by which divisions within
a vertically integrated firm with decentralized decision rights coordi-
nate their actions. These studies are of two types. One type constrains
the firm to using one transfer price for both external tax and internal
coordination purposes. The other type permits the firm to use differ-
ent transfer prices for external tax and internal coordination purposes.
I first review the leading papers in which firms are constrained to use
the same price for both purposes. I then review the leading papers in
which the prices are decoupled. Finally, I discuss the arguments for
both approaches.
5.2. Two sets of books 25
At one extreme is the approach that assumes that the price used
for external tax and internal coordination purposes must be identical.
This approach is adopted in Narayanan and Smith [2000]. In their set-
ting, the head office located in one country manufactures a product
and transfers it to a sales office in another country. The head office has
private information regarding production costs, whereas the sales office
has private information regarding demand for the product. Communi-
cation regarding private information is exogenously precluded. In one
setting, the firm is a monopolist; in a second setting, the firm engages
in Bertrand duopolistic competition with differentiated products.
The firm can choose whether to act in a centralized fashion or a
decentralized fashion. In a centralized firm, the head office chooses the
output price, the transfer price, and produces to meet demand. The
transfer price can be any value greater than the manufacturing cost and
less than the output price, as in Horst [1971]. The sales office makes no
decisions. Centralization implies that output price is chosen based on
expected demand, uninformed by the private information of the sales
office. In a decentralized firm, the head office chooses the transfer price,
but delegates the choice of output price to the sales office. The sales
office is evaluated on its own profits, and chooses the output price after
observing the transfer price and its private signal regarding demand
for the product.
When the firm is a monopolist and the tax rate in the manufactur-
ing country is weakly greater than the tax rate in the selling country,
operating in a decentralized fashion dominates operating in a central-
ized fashion and the transfer price is equal to the manufacturing cost. In
this case, the transfer price both signals marginal cost to the sales office
and allocates all taxable income to the country with the lower tax rate.
When the tax rate in the selling country is strictly greater than the tax
rate in the manufacturing country, the optimal transfer price for the
decentralized firm exceeds marginal cost, reflecting a tradeoff between
increasing pretax profits and trying to shift income to the manufactur-
ing country. In contrast, the centralized firm uses a transfer price equal
to the selling price, shifting all income to the manufacturing country.
It is optimal for the firm to centralize in that case if the uncertainty
26 Optimization Models
The preceding analysis assumes that the firm has no ability to devi-
ate from the transfer price implied by either the comparable profit
method or the comparable uncontrolled price method. If instead the
firm has some ex post discretion, it will use that discretion to shift
income to the lower-tax country. This in turn can either exacerbate or
mitigate investment distortions relative to the benchmark case with no
tax rate differentials.
De Waegenaere et al. [2012] study the effects of transfer prices on
investments in tangible and intangible assets. In their model, a large
number of firms make investments in research and development to
engage in a patent race. One of the firms wins the race, and subse-
quently makes a tangible investment that uses the patented technol-
ogy. Making the tangible investment in a low-tax rate foreign country
makes it possible for the MNE to shift income attributable to an intan-
gible asset developed in the high-tax rate domestic country unless the
domestic country can perfectly enforce the arm’s length allocation of
income implied by CPM. The greater the ability to deviate from the
arm’s length standard, the greater the incentive the firm has to locate
tangible investment in the foreign country.
The present value of taxes collected by the domestic country on
intangible investments depends on whether one looks at the winner
of the patent race or all firms that entered the race. Taxes collected
by the domestic country from the winner are positive. However, if one
considers all the firms that entered the race, the present value of the
domestic country’s tax collections is zero if the domestic country can
perfectly enforce the arm’s standard, and negative if the winner can
shift income attributable to the patent to the low-tax country.
In addition, deviating from the arm’s length standard increases
investment in research and development. This could increase or
decrease efficiency, because the privately optimal level of research and
development investments in a patent race could be either higher or
lower than the socially optimal level.
6
Transaction Cost Economics
34
6.2. Intangible assets 35
activities are carried out within a single firm, the production and pric-
ing strategy that maximizes firm income equates marginal revenue and
marginal cost. Now suppose instead that the manufacturer sells to an
unrelated retailer. Then the manufacturer will choose the wholesale
price that equates the manufacturer’s marginal revenue and marginal
cost, and the retailer will in turn choose the retail price that equates
the retailer’s marginal revenue and marginal cost, a process known
as double marginalization [Tirole, 1990, p. 174]. Double marginaliza-
tion yields a lower output and higher price relative to case in which a
vertically integrated firm performs both the manufacturing and retail-
ing functions. There are alternative arrangements that avoid double
marginalization (e.g., the manufacturer charges the retailer a fixed fran-
chise fee), but they have drawbacks of their own, such as inefficient risk
sharing. This is a simple illustration of how the effort to put “a con-
trolled taxpayer on a tax parity with an uncontrolled taxpayer” can
break down; the two affiliated entities are in a fundamentally different
economic relationship than are two independent firms. Establishing tax
parity between a controlled taxpayer and a corresponding independent
firm becomes impossible because the aggregate taxable incomes of the
two pairs of firms will likely differ, perhaps substantially.
Harris and Sansing [1998] explore this issue. In their study, a ver-
tically integrated group can avoid the double marginalization problem
that arises when both the manufacturing firm and selling firm can
exercise market power. However, they also consider the case in which a
local independent selling firm has private information regarding market
demand, and thus can base its output strategy on the realization of the
stochastic demand parameter. In contrast, the MNE can only base its
decision on expected demand.
The authors first investigate the distributional effects of using the
price that the independent manufacturer charges the independent seller
to allocate taxable income between members of the vertically integrated
group when tax rates are equal. They find that the manufacturer in the
vertically integrated group is allocated a higher gross margin per unit
than what its independent counterpart earns. In the special case of a lin-
ear demand curve, 100% of the profits are allocated to the manufacturer
within the vertically integrated group. This illustrates why using an
6.3. Returns to internalization 37
Most transfer pricing models assume that the transfer price used, how-
ever it is derived, will be the same for calculating the income of each
related party. In practice, however, double taxation due to different
countries using different transfer prices is common. Although govern-
ments are supposed to use competent authority procedures when cases
of double taxation arise, there is no requirement that the governments
arrive at a compromise that prevents double taxation. Survey evidence
indicates that 47% of parent firms have had transfer price adjustments
that result in double taxation since 2009 [Ernst and Young, 2013].
According to the 2007 Ernst & Young survey [Ernst and Young, 2007,
p. 3]:
42
7.2. Advance pricing agreements 43
45
46 Formulary Apportionment
It is also not clear that other countries will follow the U.S. lead.
Anand and Sansing [2000] show that U.S. states with high levels of
natural resource income have not increased the weight on the sales
factor in their apportionment formulas, even though other states have.
Finally, the OECD [2010] is strongly opposed to formulary appor-
tionment. Paragraph 1.25 of the OECD Guidelines states:
In this section, I discuss two promising but yet unexplored avenues for
future research. Both relate to the transaction cost economics perspec-
tives discussed in Section 6 of this monograph.
48
9.2. Allocating the returns to organizational capital 49
50
51
52
References
53
54 References
L. Samuelson. The multinational firm with arm’s length transfer price limits.
Journal of International Economics, 13:365–374, 1982.
A. Sandmo. Investment incentives and the corporate income tax. Journal of
Political Economy, 82:287–302, 1974.
R. Sansing. Relationship-specific investments and the transfer pricing para-
dox. Review of Accounting Studies, 4:119–134, 1999.
G. Schjelderup and A. Weichenrieder. Trade, multinationals, and transfer
pricing regulations. Canadian Journal of Economics, 32:817–834, 1999.
M. Smith. Ex ante and ex post discretion over arm’s length transfer prices.
The Accounting Review, 77:161–184, 2002a.
M. Smith. Tax and incentive trade-offs in multinational transfer pricing.
Journal of Accounting, Auditing and Finance, 17:209–236, 2002b.
J. Sokatch. Transfer-pricing with software allows for effective circumvention
of subpart F income: Google’s “sandwich” costs taxpayers millions. Inter-
national Lawyer, 45:725–747, 2011.
I. Springsteel. Separate but unequal. CFO Magazine, 15:89–91, August 1999.
D. Swenson. Tax reforms and evidence of transfer pricing. National Tax
Journal, 64:7–25, 2001.
J. Tirole. The Theory of Industrial Organization. MIT Press, Cambridge,
1990.
A. Tomohara. Inefficiencies of bilateral advanced pricing agreements (BAPA)
in taxing multinational companies. National Tax Journal, 57:863–873, 2004.
U.S. Department of the Treasury. A Study of Intercorporate Pricing (the
White Paper). Government Printing Office, Washington, 1988.
O. Williamson. Markets and Hierarchies: Analysis and Antitrust Implications.
Free Press, New York, 1975.
O. Williamson. The Economic Institutions of Capitalism. Free Press, New
York, 1985.
P. Wilson. The role of taxes in location and sourcing decisions. In A. Giovan-
nini, G. Hubbard, and J. Slemrod, NBER, editors, Studies in International
Taxation, pages 195–234. University of Chicago Press, Chicago, 1993.
J. Wittendorf. Transfer Pricing and the Arm’s Length Principle in Interna-
tional Tax Law. Kluwer Law International, Netherlands, 2010.
J. Zimmerman. Accounting for Decision Making and Control. McGraw-Hill
Irwin, New York, 8th edition, 2014.