Академический Документы
Профессиональный Документы
Культура Документы
Transfer-Pricing Rules?*
Abstract
This paper uses a strategic tax compliance model to examine taxpayer reporting and tax
authority audit strategies in an international setting with two tax authorities. The setting
features both information asymmetry between the taxpayer and the tax authorities and
inconsistent tax transfer-pricing rules. The latter creates the possibility of each country trying
to tax the same income. We study the effect of the probability of transfer-price rule inconsis-
tency on the strategies and payoffs of the taxpayer and the tax authorities. We find that an
increase in the probability of transfer-price rule inconsistency induces more aggressive
auditing by governments. It therefore deters taxpayers from shifting income to the country
with the lower tax rate in situations in which the transfer-pricing rules are consistent, and
can either increase or decrease the income reported to the low-tax-rate country in cases in
which the transfer-pricing rules are inconsistent. We find that an increase in transfer-price
rule inconsistency could either increase or decrease the taxpayer's exp)ected tax liability and
could either increase or decrease the deadweight loss from auditing. Our results call into
question the conventional wisdom that the prospect of double taxation due to transfer-price
rule inconsistency increases a firm's expected tax liability and governments' expected audit
costs.
Keywords Double taxation; Tax compliance; Tax law inconsistency; Transfer pricing
Accepted by Ken Klassen. An earlier version of this paper was presented at the 2(X)4 Contemporary
Accounting Research Conference, generously supported by the Canadian Institute of Chartered
Accountants and Ordre des complables agrees du Quebec. We thank Jon Davis, John Hand.
Bart Kamp, Ken Klassen, Suzanne Paquette. Shelley Rhoades-Catanach, workshop participants at
Tilburg Universiiy, the 2004 American Accouniing Association convention, and the 2004 Contem-
porary Accounting Research Conference, and two anonymous reviewers for helpful comments.
Contemporary Accounting Research Vol. 23 No. 1 (Spring 2006) pp. 103-31 © CAAA
104 Contemporary Accounting Research
Condense
Les atJteurs se penchent sur la question de la double imposition d'une society multinationale
en eiaborant et en analysant un modele d'obiigations fiscales ititemationales. Le benefice
realise par une societe multinationale peut faire Tobjet d'une double imposition si plusieurs
iidministrations fiscales affimient leur droil de lever un impot sur ce benefice. Le cas se produit
notamment lorsque ies gouvernements appliquent des regies sur les prix de transfert qui
sont differente.s en ce qui a trait a la repartilion du benefice entre les pays.
Pour Illustrer les questions etudiees dans leur modele, les auteurs proposent un exemple
representatif de I'enjeu monetaire eteve des litiges en matiere de prix de transfert, puisque
cet exemple met en cause des actifs incorporels appartenant a une societe multinationale
rentable. Une societe mere exer9ant ses activites dans un pays possede une liliale qui exerce
les siennes dans un autre pays. La filiale fabrique et vend un produit qui genere un proHt de
30 dollars. La tiliale utilise un bien incorporel apparlenant a la societe mere dans la fabrication
et (ou) la vente de soti produit. Elle doit verser une redevance a la societe mere afin de remu-
nerer cette demiere pour Futilisation dudit bien. Si cette redevance s'eieve a 10 dollars, la
societe mere paiera Timpot sur 10 dollars de benefice et la filiale paiera Timpot sur 20 dollars
de benefice ; si la redevance s'eieve a 20 dollars, la societe mere paiera Timpot sur 20 dollars de
benefice et la filiale paiera I'impot sur tO dollars de benefice. Les societ^s peuvent done
choisir le montant de la redevance. mais il leur faut declarer le benefice imposable corres-
pondant a ce choix. Ainsi s'assure-t-on que le benefice imposable total dont Ia societe fait
6tal dans les deux pays est egal au profit resultant de I'operation.
Reste a savoir si Taccord relatif a la redevance sera accepte par les deux pays, aux fins
fiscales. Les operations entre apparentes preoccupent tout particulierement les administra-
tions fiscales parce qu'elles peuvent etre utilisees pour transferer des benefices vers ie pays
ayant Ie taux d'imposition le plus faible. Dans le modele des auteurs. Ies deux pays adherent
au « principe de pleine concurrence » en ce qui a trait a la determination des prix de transfert
par les apparentes. En vertu de ce principe, le montant de la redevance que verse la filiale a
la societe mere doit correspondre h la somme que paierait une partie non apparentee concluant
une operation comparable dans des circonstances analogues. L'application de ce principe
peut s'operer selon differentes methodes, notamment la methode du prix de marche com-
parable, la methode des profits comparables et la methode du partage des profits. En outre,
le mode d'application de ces methodes depend, dans une certaine mesure, des operations
entre parties non apparentees qui sont jugees comparables aux operations entre apparentes.
Si les deux pays, compte tenu de ces diverses methodes. derivent la meme redevance dans des
conditions nonnales de concurrence, les regies sur les prix de transfert sont jugees coherentes.
Si les deux pays derivent des redevances differentes, les regies sur les prix de transfert sont
jugees incoherentes. Les auteurs font remarquer que, puisque les deux pays adherent au
principe de pleine concurrence, il y a double imposition du fait de incoherence du mode
d'appiication des regies sur les prix de transfert, dans la pratique.
Le modele des auteurs etablit la distinction entre trois etats : Ies regies sur les prix de
transfert sont coherentes et favorisent le pays ayant ie taux d'imposition eleve ; les regies
sur les prix de transfert sont coherentes et favorisent le pays ayant le faible taux
d'imposition ; ou les regies sur les prix de transfert sont incoherentes. La societe observe
privement la situation et declare le benefice faisant l'objet du litige soit dans le pays qui
applique le faible taux d'imposition, soit dans le pays qui applique le taux d'imposition
eleve. Les pays connaissent uniquement la distribution de probabilites des differents etats,
Les auteurs definissent les six equilibres auxquels peut donner lieu le modele. En general,
les equilibres I et V correspondent au « comportement declaratif audacieux » selon lequel la
societe procede au partage du benefice en declarant parfois le benefice dans le pays ayant le
faible taux d'imposition alors que ledit benefice est indeniablement imposable dans le pays
ayant le taux d'imposition eleve (c'est-a-dire que les regies sur les prix de transfert sont
coherentes et favorables au pays ayant le taux d'imposition eleve). Les equilibres IV et VI
correspondent au « comportement de controle audacieux », selon lequel le pays ayant le faible
taux d'imposition et le pays ayant le taux d'imposition eleve procedent tous deux a un con-
trole, dans leur effort pour tirer des recettes de la double imposition. Les equilibres II et IU
sont des cas intermediaires dans lesquels la societe ne se livre pas au transfert des benefices
et le pays ayant le faible taux d'imposition ne controle aucune declaration.
Apres avoir defini ces equitibres, les auteurs examinent les repercussions de changements
dans la probabilite d'incoherence des regies relatives aux prix de transfert sur la strategie
declarative de la societe et les strategies de controle du pays ayant un faible taux d'imposition
et du pays ayant un taux d'imposition eleve. Les auteurs constatent qu'une augmentation de
la probabilite d'incoherence des regies sur tes prix de transfert pousse les administrations
fiscales a adopter un comportement de controle plus audacieux qui, en retour, incite les
societes a se livrer moins volontiers au transfert de benefices lorsque les regies sur les prix
de transfert sont coherentes. Cette augmentation peut egalement soit hausser soit reduire la
probabilite que le contribuable declare des benefices au pays ayant le faible taux d'imposition
lorsque les regies sur tes prix de transferi sont incoherentes.
Les auteurs se demandent ensuite comment un changement dans rincoherence des
regies sur les prix de transfert influe sur les obligations fiscales prevues de la soci6te. Ils
constatent qu'une diminution de I'incoherence peut soit augmenter soit reduire ces obligations
fiscales. En effet, une probabiiite plus elevee d'incoherence des regies sur les prix de transfert
beneficiera a la soci^te si la probabilite qu'un pays impose un benefice deja impose par
I'autre pays est suffisamment faible, Cette situation est hautement probable dans les cas ou
l'ecan entre les taux d'imposition des deux pays est important.
Enrtn, les auteurs se demandent si les couts du controle augmenteront ou diminueront
si la probabilite d'incoherence des regies sur les prix de transfert diminue. Encore une fois,
les deux situations sont possibles. Meme si I'un ou I'autre pays adopte un comportement de
controle moins audacieux devant une declaration dtSfavorable, les couts pr6vus du controle
peuvent soit augmenter soit diminuer, etant donne qu'une diminution de I'incoherence des
regies sur les prix de transfert peut pousser la societe a un comportement declaratif plus ou
moins audacieux, Dans les cas ou une augmentation de I'incoherence des regies sur les prix
de transfert augmente les recettes fiscales prevues et diminue les couts de controle prevus, il
est avantageux pour les gouvernements de maintenir un minimum d'incoherence des regies
sur les prix de transfert.
Ces resultats aboutissent a deux conclusions etonnantes. Premierement. une augmentation
de rincoherence des regies sur les prix de transfert peut diminuer les obligations fiscales
prevues de la societe, de sorte que I'incoherence peut etre avantageuse pour les societes.
Deuxi^mement, une augmentation de Tincoherence des regies sur les prix de transfert peut
soit augmenter soit diminuer le poids des pertes associ6es aux coflts de controle. Les auteurs
definissent les cas dans lesquels la diminution de rincoherence des regies sur les prix de
transfert peut a la fois reduire les recettes fiscales des gouvemements et augmenter les couts
de controle prevus. Dans ces situations, les gouvemements n'ont pas interet a coordonner
trop rigoureusement leurs regies sur les prix de transfert et leurs pratiques en cette matiere.
Les resuitats obtenus par les auteurs reniettent en question I'idee precon^ue selon laquelle la
perspective d'une double imposition attribuable a Tincoherence des regies sur les prix de
transfert est favorable aux gouvemements. defavorable aux contribuables et socialement
inefficace. Une politique gouvemementaie visant a augmenter rincoherence des regies sur
les prix de transfert n'est ni clairement avantageuse ni indiscutablement desavantageuse
pHJur les societds ou les gouvemements.
1. Introduction
In this paper we examine the issue of double taxation of a multinational enterprise
by developing and analyzing a model of international tax compliance. Income
earned by a multinational enterprise can be subject to double taxation if multiple
tax authorities assert the right to tax the income. In particular, double taxation
arises to the extent that governments use different transfer-pricing rules to allocate
income between countries. In the 1997, 1999, and 2001 Ernst & Young Transfer
Pricing Global Surveys, over 80 percent of multinationals identified "double tax
relief" as an important intemational tax issue because transfer-pricing adjustments
typically result in double taxation (Ackerman and Hobster 2001).
Mechanisms such as mutual agreement procedures (MAPs) exist to make it
possible for a taxpayer to get relief from double taxation. However, in practice the
MAP system often fails to provide such relief because governments are not required
to resolve the conflict in a manner that eliminates double taxation (Mortier 2002).
The failure of the MAP system to provide relief from double taxation has led to a
demand for alternative means of resolving disputes, such as advance pricing agree-
ments (Ring 2000) and arbitration (Morgan 2003).
The standard tax compliance model treats the interaction between the taxpayer
and the tax authority as a game between a wealth-maximizing taxpayer and a tax
authority trying to maximize government revenues net of audit costs (Graetz, Rein-
ganum, and Wilde 1986; Reinganum and Wilde 1986; Beck and Jung 1989; Sansing
1993; Rhoades 1997, 1999; Mills and Sansing 2000; Feltham and Paquette 2002).
In the standard model, the taxpayer privately observes its income and submits a
report to the tax authority. The model in this paper considers a multinational whose
worldwide income is common knowledge, but the division of income between the
countries is open to dispute. It therefore adds two features to the standard tax com-
pliance model. First, since the taxpayer's income is potentially subject to tax in
two different countries with different tax rates, the firm has an incentive to shift
income to the country with the lower tax rate through its choice of transfer price.
Second, we allow for the possibility of what we refer to as transfer-price rule
inconsistency. Inconsistency arises when both countries claim to adhere to the
same transfer-pricing principles, but apply these principles in different ways.
These differences can result in both tax authorities trying to tax the same income
(double taxation). The possibility of transfer-price rule inconsistency in turn affects
the reporting and auditing strategies of tbe taxpayer and the two tax authorities.
To illustrate tbe issues that we examine in the model, we offer the following
example that is representative of high-dollar transfer-pricing disputes in that it
involves intangible assets owned by a very profitable multinational enterprise (Sul-
livan 2004). A parent corporation operating in one country owns a subsidiary
operating in another country. The subsidiary produces and sells a product generat-
ing a profit of $30. The subsidiary uses intangible property owned by the parent in
its production and/or sales process. The subsidiary must pay a royalty to the parent
to compensate the parent for the subsidiary's use of the intangible property. If the
subsidiary pays a royalty of $10, then the parent pays tax on $ 10 of income and the
subsidiary pays tax on $20 of income; if the subsidiary pays a royalty of $20, then
the parent pays tax on $20 of income and the subsidiary pays tax on $10 of
income. We emphasize that the royalty payment is an actual transaction between
two distinct legal entities (a parent and its subsidiary), and that both report the roy-
alty truthfully on their respective tax retums.' In other words, we allow the firms to
choose the royalty, but require them to report taxable income in a way that is con-
sistent with that choice. This ensures that the aggregate taxable income that the
firm reports to the two countries is equal to the profit from the transaction.
The issue is whether the royalty arrangement will be accepted by the countries
for tax purposes. Transactions between related parties are of particular concem to
tax authorities because they can be used to shift income to the jurisdiction with the
lower tax rate. Both countries adhere to the "arm's-length standard" when deter-
mining transfer prices used by related parties in our model. Under the arm's-length
standard, the amount of the royalty that the subsidiary pays to the parent should be
the amount that an unrelated party engaging in a comparable transaction under
comparable circumstances would pay. There are various methods that can be used
when applying this principle in practice, including the comparable uncontrolled
transaction method, the comparable profits method, and the profit-split method.
Furthermore, how these methods are applied in practice depends in part on which
transactions between unrelated parties are considered to be comparable to the
related-party transaction. If, when faced with these various methods, the two coun-
tries would derive the same arm's-length royalty, we consider the transfer-pricing
rules to be consistent. If the two countries would derive different royalties, we con-
sider the transfer-pricing rules to be inconsistent. We emphasize that because both
countries claim to adhere to the arm's-length standard, double taxation arises due
to the inconsistency of how the transfer-pricing rules will be applied in practice.
An extreme example of the potential for transfer-price rule inconsistency is pro-
vided by Kenya. Section 18(3) of the Kenyan Income Tax Act gives the Kenya
Revenue Authority (KRA) the power to restate transactions in accordance with the
arm's-length standard, but the KRA provides no guidelines regarding which trans-
fer-pricing methods are acceptable (Harris 2004).
Inconsistent application of the arm's-length standard can cause the aggregate
taxable income in the two countries to exceed the worldwide economic income
from the transaction, hui cannot cause aggregate taxable income to be less than
worldwide economic income. This asymmetry in outcomes reflects the asymmetric
position of the taxpayer and the tax authorities. The tax authorities can argue that
the form of the transaction (in this case, the royalty paid by the subsidiary) can be
ignored for tax purposes if it diverges from the substance of the transaction as
determined under the arm's-length standard. The taxpayer, in contrast, having chosen
the form of the transaction, must accept the tax consequences of that choice (see
Commissioner v. National Alfalfa Dehydrating).'^ A consequence of this asym-
metry is that aggregate taxable income (after audit) from the transaction can
exceed worldwide economic income from the transaction, but cannot be less.
Our analysis proceeds in two stages. In the first stage, we take the probability
of transfer-price rule inconsistency as exogenously fixed, and derive the equilibriutn
reporting strategy of the firm and audit strategies of the two countries. We find that
an increase in the probability of transfer-price rule inconsistency induces the tax
authotity to adopt a more aggressive audit strategy. This increase in aggressiveness
in turn induces the firm to engage in less income shifting when the transfer-pricing
rules are consistent. A more aggressive audit strategy can also either increase or
decrease the probability that the taxpayer will report income to the low-tax-rate
country when the transfer-pricing rules are inconsistent.
In the second stage, we view the probability of transfer-price rule inconsistency
as being influenced by both governments. Governments can decrease transfer-price
rule inconsistency by increasing the extent to which they coordinate their transfer-
pricing rules and practices (through regulations, treaties, or informal agreements).
First, we ask how a change in transfer-ptice rule inconsistency affects the firm's
expected tax liability. We find that a decrease in inconsistency could either increase
or decrease the firm's expected tax liability. The firm benefits from a higher probabil-
ity of transfer-price rule inconsistency if the probability that a country successfully
taxes income that is already taxed by the other country is sufficiently low. This out-
come is most likely to occur in settings in which there is a large difference in tax
rates between the two countries.
Second, we ask whether audit costs will increase or decrease if the probability
of transfer-price rule inconsistency is decreased. Again, either outcome is possible.
Audit costs can either increase or decrease because a decrease in transfer-price rule
inconsistency can either induce more or less aggressive tax reporting by the firm.
In cases in which an increase in transfer-price rule inconsistency would increase
expected tax revenues and decrease expected audit costs, a certain minimum level
of transfer-price rule inconsistency is beneficial to governments.'*
We emphasize that our analysis of the effects of a change in the probability of
transfer-price rule inconsistency is partial equilibrium in the sense that we do not
consider the possibility that a change in the probability could induce a change in either
the firm's worldwide income or the amount involved in the transfer-pricing dispute.
Eor example, we do not consider the possibility that an increase in transfer-price rule
inconsistency could induce the firm to transact with an unrelated party instead of
with a subsidiary, thereby avoiding a transfer-pricing dispute in the first place. Our
results apply to settings in which such responses are of second-order importance.
2. Model
This section describes the game, the players' strategies, and the possible payoffs.
applied in an inconsistent manner hy the two governments, we assume that the tax-
payer does not pay a penalty when the state is _vg because although the government
has prevailed, the taxpayer's interpretation ofthe arm's-length standard was not an
unreasonable one. Double taxation occurs because the other country would not
refund the taxes already paid, because both countries claim the right to tax the
income. While the probability of douhle taxation is affected by both p and £, the
two parameters are conceptually distinct. The parameter p is the probability that
the governments apply the arm's-length standard in ways that result in both coun-
tries claiming the right to tax the income under dispute, assuming that the report is
audited. The parameters is the probability that the government would prevail in liti-
gation against the firm.
Auditing is costly. The audit cost incurred by country H (L) is kf^ (ki). We
assume that i^ <ETi,k^ <£'^H^ a"*^ ^H < [^//^ +?r)]/2, which ensure that the cost
of auditing is sufficiently low that the threat to audit is credible. The game tree is
presented in Figure 1. The payoffs to the players given the state, report, and audit
decisions are summarized in Table 1.
^
F reports JCH
r
L and H decide to
T"
Payoffs are
accept or audit realized
F privately
observes state
of nature F reports J:^ L and H decide to Payoffs are
accept or audit realized
L_
Inconsistent Multinational Tax Transfer-Pricing Rules 111
Strategies
Each player will sometimes have a dominant strategy and other times will not. We
characterize the strategies employed by each player below.
Strategies for F
• F has a dominant strategy of reporting x^ when it observes _v^.
• When F observes _y//, it reports x^f with probability a and .v^ with probability
1 -a.
• When F observes yg, it reports X// with probability p and x^ with probability
Strategies for H
• H has a dominant strategy of never auditing a report of X/y.
• H audits reports of jc^ with probability y.
Strategies for L
• L has a dominant strategy of never auditing a report of J^.
• L audits reports of X;^ with probability S.
3. Equilibria
In this section we characterize the six equilibria that can arise in the model. We
first preview our results with an example. Figure 2 illustrates the regions within
which each equilibrium occurs over the range of values for p and f. The boundary
at e = 0.125 in Figure 2 reflects the assumption thatei// > kf^. Generally speaking,
equilibria I and V are "aggressive reporting" equilibria that arise when eitherp or£
TABLE 1
Payoffs
F's payoff L's payoff H's payoff
State yf^
xi, H audits -T£_ Ti -kff
Xi, H does not audit -T^ T/_ 0
x^. L audits —Ti{ 1 + T ) TI( I + T ) ~ ^ I 0
Xff., L does not audit -iff 0 Tf]
State yff
x/^,H audits -T//(l +n) 0 %(1 +K)-kff
jc^, H does not audit -Tf^ Ti, 0
Xff., L audits -T// —^ Xff
.T/y, L does not audit —Xff 0 T//
State Vfl
.J:^^, H audits ~'^L~^'^H Tt ^H~^H
Xf^, H does not audit -T[^ ti 0
Xff, L audits -Xff — exi^ ex^ — k^ T^
Xfj, L does not audit -Tu 0 T//
~x k
Figure 2 Equilibria regions when Xff = 40%, T^ = 20%. K = 30%, kf^ - it^ = 0.05, and
"1 r
0.2 0,4 0.6 0.8
The value p* is the value of/? for which H would be indifferent between auditing
and not auditing reports of x^^ if the firm would never shift income to L when there
is no inconsistency, and would always report .\[^ in case of inconsistency. The value
/;* (/?*) is the value of/J for which neither country is willing to adopt a pure strategy
of never auditing if (T,f - t[)/Tf^ <e<jr+ [{Xf/ - Xi)lxij \ (resp. £ >ff+ (T// - W/TH)-
The assumptions we have made ensure that 0 </?*</?*< I and 0 < /?* < /J2 ^ I.
Equilibrium I arises when p is lower than p* and the tax rate differential and/or
the penalty rate is sufficiently large.
The proof and the values of a* and / are presented in the appendix. In equi-
librium I, there is no pure strategy equilibrium with respect to the strategies a and
7. The relatively low value of p implies that if F reports J:^ whenever V// occurs
(a =^ 1). H should never audit ( y - 0); but if H never audits, F should always shift its
income to L (a - 0). On the other hand, the cost of an audit for H is sufficiently
low that, if F always shifts its income to L under V// (a = 0), then H should always
audit (y = 1). Therefore, the only equilibrium is for F to choose a mixed reporting
strategy, a*, so as to keep H indifferent between auditing and not auditing reports
of .V/^, and for H to choose a mixed audit strategy, y", so as to keep F indifferent
between reports when it observes yff. The resulting audit frequency by H is low
enough that it is optimal for F Io always report .v^^ under inconsistency (j3 ^ 0).
Consequently, L should never audit {S ^ 0). The low probability of inconsistency
causes H's audit strategy to be driven by its desire to deter F from engaging in
income shifting when v// occurs.
In general, equilibrium I corresponds to settings where the level of coordina-
tion between governments regarding the application of their transfer-pricing rules
is very high because p < p*. For example, we would expect equilibrium I to arise
in transactions involving countries that both adhere to the Organisation for Economic
Co-operation and Development's (OECD's) transfer-pricing guidelines and have a
well-functioning competent authority process that reduces the probability that the
two countries will try to impose different transfer prices on the same transaction.
Equilibrium II arises when both the probability, p, of transfer-price rule incon-
sistency is higher than p* and the tax rate differential is large relative to£.
The proof is in the appendix. In equilibrium II, all players choose pure strat-
egies. Tbe probability, £, of H prevailing in its effort to tax income that is also
taxed by L is sufficiently small that reporting x/, when the transfer-pricing rules are
inconsistent is optimal even given a certain audit. F prefers to face an expected tax
liability of r^ + eXff from reporting .v£ to a tax liability of at least T// from reporting
J://, because the probability of H prevailing in litigation is low relative to the differ-
ence in tax rates, Xff - r^. Consequently, F optimally always reports AJ^ under
inconsistency (jS = 0). and L has no incentive to audit (5 = 0) because F only
reports Xf^ when the state is yf^. The probability, p, of transfer-price rule inconsis-
tency is high enough compared with H's audit costs that, because F always reports
xi under inconsistency, it is optimal for H to audit all reports of .v^ (y= 1), regardless
of F's reporting strategy under >>;/. Because H audits all reports of x^, F reports x//
when yff occurs (« ^ 1). The combination of the high probability of inconsistency
and the low probability that H would prevail in litigation implies that the firm should
always report .r^ under inconsistency and H should always audit a report of .v^.
Equilibrium II corresponds to settings in which there is low coordination
between governments regarding the application of their transfer-pricing rules and
there is a large difference between the tax rates imposed by the two countries. The
smaller the tax rate differential, the less likely it is that the condition £<(Zff- XjJIXfj
will be satisfied. A transaction between related entities located in an OECD country
and a "tax haven" country would likely correspond to equilibrium II.
Equilibrium III arises when the tax rate differential is low, the penalty rate is
high, and the probability of transfer-price rule inconsistency is between the two
boundary values />* and p'^.
The proof and the values of^* and/* are presented in the appendix. In equilib-
rium III, the fact that £ is bigger than the tax rate differential implies that it cannot
be an equilibrium for H to always audit. If H would audit all reports of x^, F should
optimally respond by never shifting income to L under yff (a = 1). Moreover, F's
optimal response under inconsistency to a certain audit by H would be to keep L
from always auditing. But because p is not very high (p < p*), the frequency with
which F needs to report ,v^ under inconsistency to keep L from always auditing is
low enough that, combined with a - 1, always auditing is not optimal for H. As a
The proof and the values of )8* and S* are presented in the appendix. In equi-
librium IV, p and £ are large enough to induce either H or L to audit in the hope of
achieving double taxation. The high probability of inconsistency implies that it is
not optimal for F to choose its reporting strategy under inconsistency, j8, in order to
keep H indifferent, because p is now high enough that L would then have an incen-
tive to audit all reports of J:^. F therefore chooses a mixed strategy so as to keep L
indifferent between auditing and not auditing repOTts of Xff, and L chooses an audit
strategy so as to keep F indifferent between reports when it observes yg. Given
these strategies, F has a pure strategy of reporting Xff when it observes y^f and H
has a pure strategy of auditing all reports of .v^. In this case, the very high probability
of transfer-price rule inconsistency {p > /J*) induces H to audit all reports of Xf^. L
in turn audits some reports of Xff in order to deter F from always reporting Xff when
it observes yg.
Equilibrium IV corresponds to settings in which tax rates are similar but there
are significant disagreements between the two countries over transfer-pricing
methods. Transactions between the United States and other OECD countries that
involve intangible assets often fall into this category because in such transactions
the Internal Revenue Service often favors the comparable profits method, which is
not allowed under the OECD guidelines.
Equilibrium V arises when the tax rate differential is low, the penalty rate is
low, and p is low.
The proof and the values of a* and y* are presented in the appendix. In equi-
librium V, e is sufficiently high and the penalty rate, 7i, is sufficiently low that the
equilibrium level of auditing by H deters F from reporting to L when the transfer-
pricing rule is inconsistent, but does not deter income shifting. In addition, p is suffi-
ciently low that L does not have an incentive to audit reports of Xff. The equilibrium
features F choosing a mixed strategy when yff occurs to keep H indifferent
between auditing and not auditing reports of x^, while H adopts a mixed auditing
strategy to keep F indifferent between reports when y^ occurs. As in equilibrium I,
equilibrium V corresponds to transactions between related entities in OECD coun-
tries. However, equilibrium V holds in cases where the tax rates are similar and the
penalty rate is low.
Finally, equilibrium VI arises when the tax rate differential is low, the penalty
rate is low, and/j is intermediate.
The proof and the equilibrium values of a*,/?*, j ' * , and 8* are in the appendix.
In equilibrium VI, £ and p are sufficiently high that H and L have an incentive to
audit some reports. However, ;r is sufficiently low that F is willing to employ
mixed strategies both when yff and yg occur. The equilibrium features F adopting a
mixed strategy both when it observes V// and yg, and both H and L adopting mixed
audit strategies. As was the case in equilibrium III, equilibrium VI corresponds to
settings in which there is an intermediate level of coordination between countries,
but in equilibrium VI the tax rates of the two countries are similar and the penalty
rate is low.
The strategies of the players are summarized in Table 2.
that an increase in p weakly increases the probability that a report will be audited,
weakly decreases the probability that F will engage in income shifting, and has
countervailing effects on F's reporting strategy when the transfer-pricing rules are
inconsistent. The analyses in this and the next section assume that the change in p
affects neither the worldwide income (z) nor the amount of income under dispute
(normalized to one).
There are three cases to consider. First, we consider the case in which (T// - T^)/r;/
> £, which we denote the high tax rate differential case. Second, we consider the
effects in the case of low tax rate ditterential and high penalty — that is, (i/y - Ti^ )IT^
<E<K + {x^-Xi )lTyf. Third, we consider the case of a low tax rate differential and
low penalty — that is. £ > ;i: -i- (T// - Xi)lTff.
'
TABLE 2
Equilibrium strategies
Strategy Equil. 1 Equil, II Equil. Ill Equil. IV Equil. V Equil. VI
a a* I I 1 a* a*
/3 0 0 j3* ^* 1 ^'
7 f I f \ f Y*
S 0 0 0 5* 0 6*
When the critical level is crossed {p > p*), the probability of transfer-price
rule inconsistency is sufficiently high that if F always reports A:^ under inconsis-
tency, it is optimal for H to audit all reports of A:^, even when F always reports x//
when it observes _v//. Because the tax rate differential is high, F has a pure strategy
of reporting A^ when it observes yg; L in tum has a pure strategy of never auditing
a report of A/y. F's strategy of always reporting .v/^ under inconsistency implies that,
when the critical level p* is crossed, the probability of inconsistency is high
enough that auditing for the sole purpose of achieving double taxation is beneficial
for H. There is a pure strategy equilibrium in which a = I. p = O,y=], and 5 = 0
(equilibrium 11). A further increase in/? does not affect these strategies.
The effects of changes in p on the players' strategies when e < (T// - T^)/T//
are summarized in Table 3. An increase in p when the tax rate differential is high
decreases income shifting when yf^ occurs (a increases) and increases the fre-
quency with which H audits low reports (/ increases).
TABLE 3
Effeclsofpon strategies when e < (r/y-T^)/r//(high tax rate differentia!)
a increases 1 N/A
fi N/A 0 0 0
Y ^ j^^'^^ 1 1
r^(l +;r)-r^
S 0 0 0 N/A
be an equilibrium for L to always audit. If L always audits, F will report Xi^ under
inconsistency, regardless of H's strategy, because T^ + ysTf^ < ^// + ^^L f*"" ^"y
choice of 7. But if F always reports j ^ , L should never audit. In equilibrium, F
therefore decreases P to make L indifferent between auditing and not auditing
reports of % ; L chooses a mixed strategy (S > 0) to make F indifferent between
reporting Xf^ and jr^ when _Vg occurs; and H audits all low reports (equilibrium IV).
As p increases, F continues to decrease (i to keep L indifferent between auditing
and not auditing reports of JT//.
The effect of changes in p on players' strategies when (r^^ - '^L)I'^H ^ ^ '^ ?r +
(T// - T^)/T// are summarized in Table 4. An increase in the probability of transfer-
price rule inconsistency when the tax rate differential is low and the penalty rate is
high affects F's reporting strategy both when the transfer-pricing rules are incon-
sistent and when they are consistent. An increase in/j increases the probability that
H (L) audits a report ofx^ (jc/y) (7 and 5 increase). F responds in tum by decreasing
the probability of income shifting when it observes y^ {a increases). An increase
in/j has ambiguous effects on/3*, F's reporting strategy when the transfer-pricing
ruies are inconsistent.
UJ
VI
VI 3-
*— UJ
UJ
UJ
practices. Once the transaction occurs, p is simply a fixed parameter. In this section,
we consider the effects of changes in p on the firm's expected tax liahility and on
the deadweight loss associated with auditing.
TABLE 5
Effects ofp on strategies when f > ;r + (% - r^^)/!// (low tax rate differential, low penalty)
a — increases 1 N/A
^ 1 decreases decreases 0
1 1
1 + ;r) - T/
is higher, an increase in p will be more likely to decrease the firm's expected tax
liability when the tax rates of the two countries are far apart. Iff: is sufficiently
high, however, an increase inp will increase F's expected tax liability.
6. Conclusions
In this paper we extend the standard tax compliance model to an intemational setting
with two tax authorities. The focus of our research is on the possihility of transfer-
Appendix
Proof of Proposition I
We show that
t^(\+7t)i\p) + 2{per^k^) ,^
( l ) [ { l ) ^ ] '^ '•^
+ (1 -
(d) 5 ^ 0 is optimal for L because not auditing Xff yields a payoff of zero and
auditing Xff yields a payoff of -kf^ < 0. •
Proof of Proposition 2
We show that
Proof of Proposition 3
We show that
-k,A\ + p)
(d) 5 - 0 is optimal for L when it observes Xff because not auditing Xff yields a
payoff of zero and auditing v/y yields a payoff of
because
Proof of Proposition 4
We show that
-Ti,(l-e)
implies ^ < 1.
(c) 7 = 1 is optimal when H observes jc^ because not auditing yields a payoff of
zero and auditing yields a payoff of ^
because
(d) L is willing to randomize when it observes Xff because not auditing yields
a payoff of zero and auditing yields a payoff of
Proof of Proposition 5
We show that
5= 0
(d) 5 = 0 is optimal for L because the expected payoff from auditing a report
of Xff would be
because/7< /;*.
Proof of Proposition 6
We show that
a =
+7:-
* Xu-X,
f=—.
Proof of Proposition 7
(a) F's expected tax liability, denoted Lj, is (1 -
Substituting 7* ^ |(T^ -r^)/[r//(l +K) - T ^ , ] ) from Proposition 1 and differenti-
ating with respect to p yields
dp 2[T^(
which is greater than (less than) zero if e > (<) fT/y(l +;r)
(b) F's expected tax liability, denotedL2. is (1 -p)[(T// +t^)/2J +/?(T/,
Differentiating with respect top yields
dp 2
2 ^ 2 •
dp 2 ^•
+ piti +
which is greater than (less than) zero if e > (<) ix^ - Ti)/2Tf^. Because £ > (Tff
T/y in Proposition 4, F's tax liability within equilibrium IV is increasing in p.
(c3) F's expected tax liability in equilibrium VI, denoted L5, is
dp 2\x,,( ^
Proof of Proposition 8
(a) In equilibrium 1, the expected deadweight loss, denotedy4], is
ri+P,(i-p)(i-«")i,>,-
L 2 2 y '^^
Substituting the values fora* a n d / ' from Proposition I and differentiating with
respect lo p yields
dp 2[Xff(\ +7r)-
^A^ ku
dp 2
Substituting the values foTp* and y* from Proposition 3 and differentiating with
respect top yields
dp 2(£Xff-kff)
dp V 2
Substituting the value for^* from Proposition 4 and differentiating with respect to
p yields
Substituting the values for a* and^* and differentiating with respect to p yields
Endnotes
1. We assume that the tax authorities can costlessly verify that the taxpayer has reported
the transaction in a consistent fashion on both tax retums and penalize the taxpayer if it
fails to do so. In the U.S. context, the Internal Revenue Service (IRS) can obtain the
foreign country tax retum through an information document request (IDR), one of
several intemational tax enforcement methods available to the IRS (Sharp and
Sheppard 2003).
2. If one country measures taxable income on the basis of separate accounting and arm's-
length transfer prices, while the other uses formulary apportionment, aggregate taxable
income could be less than worldwide economic income. Our paper investigates settings
in which both countries measure income under separate accounting using the ami's-
length standard, so aggregate taxable income cannot be less than worldwide economic
income in our model.
3. This result is similar to the case of formulary apportionment, in which govemments
will tend to choose different apportionment formulas instead of coordinating on a
single formula (Anand and Sansing 2000).
4. The assumption of equal penalty rates in the two countries is without loss of generality
because it is never in F's interest to report Xf] when yi^ occurs.
References
Ackerman, R.. and J. Hobster. 2001. Transfer pricing practices, perspectives, and trends in
22 countries. Tax Notes International 24 (12): 1151 - 8 .
Anand, B., and R. Sansing. 2(XK). The weighting game: Formula apportionment as an
instrument of public policy. National Tax Journal 53 (2): 183-99.
Beck, P., and W. Jung. 1989. Taxpayers' reporting decisions and auditing under information
asymmetry. The Accounting Reviei\' 64 (3): 468-87.
Commissioner v. National Alfalfa Dehydrating. 1974. 417 U.S. 134.
Emst& Young. 1997, 1999, 2001. £™5/(fe Young transfer pricing global survey. Uevj York:
Emst & Young.
Feltham, G., and S. Paquette. 2002. The mterrelationship between estimated tax payments
and taxpayer compliance. Journal ofthe American Taxation Association 24
(Supplement): 27-45.
Graetz, M., J. Reinganum, and L, Wilde. 1986. The tax compliance game: Toward an
interactive theory of law enforcement. Journal of Law. Economics, and Organization
2(1): 1-32.
Harris, A. 2004. Multinationals in Kenya face transfer pricing uncertainty. Tax Notes
Internationan5 {\0): 891-2.
Mills, L., and R. Sansing. 2000. Strategic tax and financial reporting decisions: Theory and
evidence. Contemporary Accounting Research 17 (1): 85-106.
Morgan. J. 2003. Arbitration clauses in international tax treaties could benefit developing
states. Tax Notes International 31 (7): 681 - 9 1 .
Mortier, R 2002. International tax arbitration: Toward bener taxpayer protection. Tax Notes
International 27 (14): 53-8.
Reinganum, J.. and L. Wilde. 1986. Equilibrium verification and reporting policies in a
model of tax compliance. International Economic Review 27 (3): 739-60.
Rhoades, S. 1997. Costly false detection errors and taxpayer rights legislation: Implications
for tax compliance, audit policy, and revenue collections. Journal ofthe American
Taxation Association 64 (Supplement): 27-47.
Rhoades, S. 1999. The impact of multiple component reporting on tax compliance and audit
strategies. The Accounting Review 74 (1): 63-85.
Ring, D. 2000. On the frontier of procedural innovation: Advance pricing agreements and
the struggle to allocate income for cross border taxation. Michigan Journal of
International Law 2\ (2): 143-234.
Sansing, R. 1993. Information acquisition in a tax compliance game. The Accounting
Review 6S {4): ^14-M.
Sharp, W., and H. Sheppard. 2003. Privilege, work-product doctrine, and other discovery
defenses in U.S. IRS's international tax, enforcement. Tax Notes International 32 (4):
377-96.
Sullivan. M. 2004. With billions at stake, Glaxo puts U.S. APA program on trial. Tax Notes
International 34 (5): 456-63.