Bilateralism or Multilateralism? The Political Economy of Avoiding International Double Taxation

Abstract

Why do states cooperate bilaterally or multilaterally? This article addresses the issue using the example of international double tax avoidance. It is argued that double tax avoidance exhibits the strategic structure of a coordination game with a distributive conflict. The distribution of tax revenues depends on the asymmetry of investment flows between treaty partners. Since investment flows are defined dyadically, bilateral bargaining can best accommodate countries’ concerns for the distribution of tax revenues and other economic benefits connected to the tax base. Moreover, because there are no serious externality problems with bilateral agreement, this solution is also viable. At the same time, there is a need for a multilateral organization to disseminate information and shared practices in the form of a model convention that provides a focal point for bilateral negotiations. This solution minimizes transaction costs. Since agreements are self-enforcing in coordination games there is no need for third-party enforcement. Instead, the Mutual Agreement Procedure (MAP) is a device to address problems of incomplete contracting.

Full text

European Journal of
International Relations
16(4) 589–614
© The Author(s) 2010
Reprints and permissions: sagepub.
co.uk/journalsPermissions.nav
DOI: 10.1177/1354066109346891
ejt.sagepub.com
Article
Corresponding author:
Thomas Rixen, Social Science Research Center Berlin (WZB), Germany.
Email: rixen@wzb.eu
Bilateralism
or multilateralism? The
political economy of avoiding
international double taxation
Thomas Rixen
Social Science Research Center Berlin (WZB), Germany
Abstract
Why do states cooperate bilaterally or multilaterally? This article addresses the issue
using the example of international double tax avoidance. It is argued that double tax
avoidance exhibits the strategic structure of a coordination game with a distributive
conflict. The distribution of tax revenues depends on the asymmetry of investment flows
between treaty partners. Since investment flows are defined dyadically, bilateral bargaining
can best accommodate countries’ concerns for the distribution of tax revenues and
other economic benefits connected to the tax base. Moreover, because there are no
serious externality problems with bilateral agreement, this solution is also viable. At
the same time, there is a need for a multilateral organization to disseminate information
and shared practices in the form of a model convention that provides a focal point
for bilateral negotiations. This solution minimizes transaction costs. Since agreements
are self-enforcing in coordination games there is no need for third-party enforcement.
Instead, the Mutual Agreement Procedure (MAP) is a device to address problems of
incomplete contracting.
Keywords
bilateralism, coordination, distribution, multilateralism, rational institutional design, taxation
Introduction
Economic activities are increasingly globalized, but the power to tax remains bound to the
nation-state. One of the problems resulting from this incongruity is double taxation, which
stems from an overlap of jurisdiction to tax between a residence state — the country where
a taxpayer lives — and a source state — the country where the taxpayer’s income was
generated. If both countries exert to the full their power to tax, then the tax burden for
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

590 European Journal of International Relations 16(4)
international investments is higher than for national investments causing an inefficient
allocation of capital. In order to prevent this, governments engage in efforts to avoid dou-
ble taxation.1 The institutional form of international double tax avoidance exhibits several
features that are in need of explanation.
The most remarkable feature is the fact that double tax agreements (DTAs) are pre-
dominantly bilateral (Vann, 1991; Whalley, 2001: 17–18).2 This bilateralism contrasts
with many other international regimes in the economic sphere. Most prominent among
these is the GATT/WTO — a multilateral regime with a commitment to achieving pro-
gressive, coordinated trade liberalization in simultaneous negotiations. In a classic model
Mundell (1957) shows that the free flow of goods and the resultant common prices for
them lead to factor prices being equalized across countries. Likewise, free factor flows
and common factor prices lead to equal goods prices. What can be achieved with goods
flows can also be achieved with factor flows. As Whalley (2001: 17–18) has pointed out,
Mundell’s factor–goods equivalence could be taken to suggest that the institutional form
of the tax and trade regimes (which deal with factor and goods flows, respectively)
should be similar or the same. So the puzzle is: why is the regime of international double
tax avoidance not multilateral?
Posing the question in this way is somewhat imprecise, because there actually are
some multilateral elements of tax cooperation. The OECD, as a multilateral organization,
deals with double tax avoidance and other issues of international taxation. It publishes a
so-called model convention (MC) that is negotiated and agreed upon by its member
countries. The OECD MC is non-binding. In practice, however, nearly all of the bilateral
tax treaties are based upon this instrument (Vann, 1991: 99). Thus, more precisely stated,
the question is why there is no binding multilateral agreement. What is the MC’s function
in international tax cooperation? How can one account for the coexistence of binding
bilateral treaties and this multilateral document that serves as a template for them?
When turning to theories of international cooperation for potential answers, one
finds that the institutional choice between bilateralism and multilateralism has hardly
been addressed. For one, cooperation theory has an inherent analytical bias towards
multilateralism and generally disregards the antipode to this institutional form, namely,
bilateralism (Odell, 2000: 13).3 Even work that explicitly focuses on the institutional
form of cooperation fails to contrast multilateralism with bilateralism (cf. Ruggie,
1993). The disregard of bilateralism in international relations (IR) theory cannot be
explained with the empirical irrelevance of this category — the United Nations treaty
database collected 5130 bilateral treaties adopted from 1990 to 1999 (United Nations,
2003).4 Second, and more important for the research question at hand, research was also
largely blind towards the specific advantages of an institutional form (Caporaso, 1993:
62). When do states cooperate bilaterally? What are the relative benefits and costs of
bilateral compared to multilateral cooperation? Cooperation theory not only disregards
bilateralism, it also does not systematically investigate the choice between bilateralism
and multilateralism.
This leads to an incomplete understanding of cooperation, including multilateral
cooperation. As I demonstrate, states consciously choose between the two institutional
forms and often design institutions with a mix of both bilateral and multilateral elements.
A straightforward classification of a regime as bilateral or multilateral is not always pos-
sible. In many regimes, cooperation exhibits bilateral and multilateral elements which
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 591
interact in systematic ways. Therefore, if we are to gain an appropriate understanding of
cooperation, it is necessary to develop theories of the choice of bilateralism and multilat-
eralism, and the ways in which these forms can interact or coexist.
My account of international tax governance demonstrates a theoretically informed
way of thinking about this choice. I construct an explanation of the tax regime’s par-
ticular institutional mix of bilateralism and multilateralism by employing some well-
established mechanisms that have been shown to have explanatory power for the design
of international institutions (Koremenos et al., 2001). Thus, the present article makes two
contributions. First, it opens up the field of international tax cooperation to IR research.
Second, it contributes to theories of rational institutional design by addressing the choice
between bilateralism and multilateralism.
My explanation can be summarized as follows: I differentiate between the bargaining,
agreement and enforcement phases of cooperation and show that double tax avoidance
is a coordination game with a distributive conflict. The distribution of tax revenues
depends on the symmetry or asymmetry of investment flows between treaty partners.
As I show in a regression analysis, the terms of bilateral tax treaties vary systematically
with treaty partners’ investment positions. This is evidence that bilateral bargaining
can best accommodate countries’ concerns for the distribution of tax revenues and
other economic benefits connected to the tax base. At the same time, a multilateral
organization to disseminate information and shared practices in the form of an MC can
minimize the transaction costs of bilateral bargaining. As there is no free rider problem
in double tax avoidance, however, agreement can remain bilateral. Since equilibria are
self-enforcing in coordination games, the institutions of double tax avoidance do not
have to be equipped with enforcement capabilities. Instead, the mutual agreement
procedure (MAP) is a device for dealing with problems of incomplete contracting.
To develop this argument, I first present the basic stylized facts of the institutional
design of double tax avoidance by differentiating between the bargaining, agreement and
enforcement phases of international cooperation. Second, I derive a game-theoretic
model that can explain the various design features, and subject it to an empirical test.
Third, I briefly discuss a potentially competing explanation. Finally I summarize the
approach and findings of this article, and draw some conclusions for future research
strategies and theory development.
The institutions of international taxation
The history of double tax avoidance
The history of the double tax regime goes back to the beginning of the last century
when a few continental European states signed bilateral double tax treaties mostly with
their neighbours. The issue became more prominent in the 1920s when the League of
Nations appointed economists to address the problem of double taxation, and convened
several conferences of technical experts and government officials (League of Nations,
1923, 1927). The objective during the ‘League years’ was to draft a multilateral treaty.
While governments persistently rejected this, they were nonetheless supportive of
developing a model convention that could be employed as a template for bilateral
negotiations. They insisted on keeping the model convention non-binding, because that
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

592 European Journal of International Relations 16(4)
would allow the necessary flexibility to make nationally differing tax systems compat-
ible with one another (Picciotto, 1992: 38). The work of the League resulted in a series
of model conventions that were published in 1928, 1935, 1943 and 1946.
In the 1950s and 1960s the OECD replaced the League of Nations (and briefly the
United Nations) as the main multilateral policy forum for discussions of international tax
issues. Countries’ positions remained unchanged. They expressed their opposition to a
multilateral treaty, but were supportive of further developing and adapting the model
convention. The OECD published its first MC and commentary in 1963, followed by a
revised version in 1977. In 1991, the OECD decided to publish the MC in loose-leaf
format, in order to be able to better adapt it to changes in the economic environment.
Since then the MC has been updated continuously. After the conclusion of the 1977 MC —
and with further liberalization of capital markets and increasing tax ratios in industrialized
countries — the number of treaties increased rapidly to over 1500 in 1998. Even though
countries are not obliged to use the MC in their bilateral negotiations, almost all of the
more than 3000 tax treaties that are in force today follow it (Owens and Bennett, 2008;
Rixen, 2008, 108–115). Bilateral treaty-making basically consists of the treaty partners
agreeing on the MC and adapting some provisions to their needs.
This development was not without conflicts. The most important one is that between
the residence and source principle of taxation. In general, developed countries are in
favour of the residence principle, whereas developing countries prefer the source prin-
ciple (see below). One embodiment of this conflict is the existence of different model
conventions. Since the OECD MC accords greater weight to residence taxation, the UN
developed a competitor model that leans towards source taxation. However, apart from
these divergences in the distribution of taxing rights, the UN MC, published in 1980
and in a modernized version in 2000, reproduces many of the articles and relies on the
same mechanics and principles as the OECD MC. Altogether, the influence of the UN
MC is limited but visible. Many developed countries subsequently granted more taxa-
tion at source in treaties with developing or transition economies (Kosters, 2004: 4).
Over time, the arrangement of a non-binding multilateral model convention as a tem-
plate for bilateral treaty negotiations has become firmly institutionalized. The Commission
on Fiscal Affairs (CFA) of the OECD — a body of government officials and tax experts,
namely, the same persons negotiating bilateral treaties for their countries — meets on a
regular basis (Messere, 1993). The CFA is the global forum for countries to cooperate in
matters of taxation (Radaelli, 1998). Non-OECD member countries participate in these
negotiations. In an ongoing process the CFA strives to modernize and adapt the MC.
Often technical innovations stemming from bilateral treaties are integrated into the
model; other innovations are developed within the CFA. In this process common under-
standings of bilateral tax treaty-making and interpretation are developed and published
in the commentary accompanying the MC, which enjoys considerable authority with
courts, lawyers and other tax practitioners (Arnold and McIntyre, 1995: 98–100).
The ‘mechanics’ of double tax avoidance: Residence vs source
Double taxation results from an overlap of jurisdiction to tax between a residence state,
where the recipient of income lives, and a source state, where the income was generated.
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 593
If both exert their power to tax to the full extent, the total burden on trans-border eco-
nomic activities is prohibitively high. In order to obtain the benefits of liberalization,
governments have a common interest in avoiding such double taxation. In principle,
there would be two solutions to this problem. Countries could delegate the power to tax
international income to an international authority (conjoint taxation) or they could agree
on some rule to share the jurisdiction to tax between them. Leaving aside the European
Union, the first option has never seriously been contemplated and is generally believed
to be utopian (Tanzi, 1999). Consequently, the problem of double taxation has been dealt
with along the lines of the second option. Framed in this way the basic question to be
answered is: which country has the right to tax the income, and which country must
restrict its tax claims?
No general consensus on a best principle could be achieved. Instead, the solution
embodied in the various MCs — whose fundamentals up to this day have remained
unchanged since the 1920s — represents the outcome of bargains in which conflicting
tax claims have been traded off against each other on a case-by-case basis (Graetz,
2001).5 Jurisdiction to tax is assigned to either the source country or the residence
country for different kinds of income. These rules, codified in Articles 6 to 22 of the
OECD model, ‘perform the function of dividing items of income between countries.
[They are] a set of arbitrary rules that were carefully crafted to support a specific com-
promise’ (Brauner, 2003: 278–279). The residence country is obliged to provide relief
from double taxation in cases of full or limited source taxation. This can be done either
by allowing a credit for the tax paid at source on the tax due in the home country or by
exempting the income taxed at source from home tax altogether (Article 23 of the
OECD MC). Importantly, however, basically all countries already provide for relief of
double taxation unilaterally. Their national tax codes contain exemption or credit pro-
visions for taxes paid in the source country. In a few cases, the national rules only
foresee partial tax relief (deduction).6
In order to implement this compromise solution between residence and source taxa-
tion, the MC refers to a series of legal constructs that establish a nexus between the
transnational tax base and a country. The constructs represent plausible assumptions
(and make them legally tractable) about the correspondence between transactions
across borders and the territorial base of the underlying economic activity that is the
target of national taxation (Bird and Wilkie, 2000: 91–93). The important point about
these constructs is that they are chosen in such a way as to interfere as little as possible
with national tax laws. Bilateral treaties do not contain comprehensive rules for taxa-
tion, but in essence achieve nothing more (and nothing less) than disentangling the
transnational tax base and assigning it to different jurisdictions. Once jurisdiction to tax
is established, the country is basically free to use its own domestic tax law on the
respective income. The bilateral treaties merely help to ‘coordinate divergent national
tax laws’ (Li, 2003: 33); they regulate the interface of autonomous national tax systems.
In this way governments retain almost unlimited legal independence over the taxation
of ‘their’ share of the transnational tax base. The tax treaty regime is built on sovereignty-
preserving cooperation.
Apart from their main function of avoiding double taxation by disentangling the
transnational tax base, DTAs also contain provisions on the implementation of these
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

594 European Journal of International Relations 16(4)
rules. Article 25 lays out the mutual agreement procedure (MAP). It serves to resolve any
conflicts that arise between treaty partners in the application of the DTA. Often, the
procedure is initiated when taxpayers claim to have been double-taxed or otherwise
disadvantaged by either state. In this case, they can ask their home country authorities to
enter into a mutual agreement procedure with the other contracting party. Governments
are not required to necessarily come to an agreement in this procedure; in cases where
they do not, double taxation persists. However, in the overwhelming majority of cases
an agreement is reached (Sasseville, 1999). Since the MAP is a ‘political rather than a
judicial procedure’ (Züger, 2001: 15), there have been many proposals to replace it with
binding arbitration. However, governments have generally refused that. Finally, Article
26 of the OECD MC foresees information exchange. This enables administrative ease in
collecting taxes.
In summary, international double tax avoidance takes place on three interrelated
levels: the unilateral, bilateral and multilateral levels. For one, all countries relieve
double taxation in their national tax laws, that is, on the unilateral level. They give up
entirely or partially their right to tax foreign source income in order to avoid interfering
with other tax systems. Second, in bilateral negotiations, countries conclude double tax
treaties which deal with the cooperative avoidance of double taxation and the division
of taxing rights. Third, international tax policy takes place on the multilateral level.
Technical experts, national tax administrators and scientific advisors cooperate in inter-
national organizations to develop model conventions, disseminate information on treaty
practices, standardize the bilateral treaties and monitor the treaty network. The resultant
MCs are legally non-binding, but very influential in practice.
Sorting the empirical evidence: Bargaining, agreement and enforcement
How can we make sense of this institutional pattern? I suggest subsuming the empirical
evidence under the established distinction between bargaining and enforcement (cf. e.g.
Fearon, 1998; Koremenos et al., 2001). Sometimes, bargaining is referred to as an ex ante
problem of cooperation, whereas enforcement concerns ex post problems (Williamson,
1985). I add ‘agreement’ to this differentiation, which is the dividing line between bar-
gaining and enforcement.
Bargaining is about negotiating the terms of an agreement. Agreement is the stage of
binding decision. It captures the formal conclusion of the bargaining period. Enforcement
refers to the ex post stage of the cooperation process in which countries have to ensure
that all treaty partners comply with the agreement. Figure 1 depicts the three stages of
cooperation on a timeline.
enforcementbargaining
t0t2
agreement (t1)
Figure 1. Three stages of the cooperation process
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 595
In each of the stages, countries can choose between bilateralism and multilateralism.
Thus, cooperation is not dichotomous per se. Taking into account the whole process,
cooperation can be characterized by a mix of bilateral and multilateral elements. The
following picture of double tax avoidance emerges: bargaining is bilateral and multi-
lateral. On the multilateral level, governments bargain about the non-binding MC. On
the bilateral level, they bargain about binding double tax treaties and agree on the
actual concessions they grant each other. Multilateral bargaining has an influence on
bilateral bargaining, because the MC is the starting point of bilateral bargains. Also, the
commentary to the MC is sometimes changed in response to problems that were
encountered in bilateral negotiations. Multilateral bargaining is complementary to
bilateral bargaining. Agreement is bilateral in double tax avoidance. Only the bilateral
treaties contain provisions that are binding on the countries. The final stage of the
cooperation process is enforcement. With respect to this, we can note that there is no
external enforcement mechanism in double tax avoidance. Rather, disputes about the
application of agreements are resolved through the bilateral MAP, which is more a
diplomatic than a judicial mechanism. Table 1 summarizes these features.
Explaining the institutional choice
The strategic structure: Double tax avoidance as a coordination game
From the fact that they have engaged in conscious efforts to this effect, we can conclude
that governments have a common interest in avoiding double taxation.7 Further, they also
have an individual incentive to avoid double taxation. This is evidenced by the fact that
all countries are willing to grant double tax relief unilaterally as part of their domestic
laws, irrespective of whether other governments reciprocate in the same way. The problem
of avoiding double taxation can thus be framed as a coordination game.
This derivation of the strategic structure from governments’ observed preferences
is in line with two common theoretical assumptions in political economy about their
preferences. First, if one assumes that a government’s goal is to maximize national wel-
fare, then the strategic interaction is represented by a coordination game (Dagan, 2000).
Second, if we assume that the government pursued its own egoistic goal of budget maxi-
mization (potentially at the expense of national welfare), the same conclusion could be
drawn. While unilateral tax relief can lead to decreasing tax revenues, which may have
negative consequences for public spending, this negative effect impacts the whole popu-
lation of a country. Following Olson (1965), such a big group faces a substantial problem
of collective action and cannot exert effective political influence. In contrast, the business
Table 1. Empirical observations on the three stages of cooperation
Stage of Cooperation Empirical Observation
Bargaining 1) Bilateral on (binding) tax treaties
2) Multilateral on (non-binding) MC
(Binding) Agreement Bilateral on DTAs
Enforcement No external enforcement, only bilateral MAP
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

596 European Journal of International Relations 16(4)
lobby is a small group that can manage to make their interests heard. There is no other
well-organized domestic interest group that would oppose granting full unilateral tax
relief in the form of credit or exemption to foreign investment. Thus, apart from the fact
that providing full double tax relief may be adequate for the maximization of national
income, there is an additional domestic political argument that makes such a strategy
attractive for a government (cf. also Bird and Mintz, 2003: 439).
Why conclude tax treaties at all?
If, however, governments have an individual incentive to avoid double taxation the ques-
tion arises as to why they should bother to negotiate and conclude DTAs at all. What are
they bargaining about? The answer is that the residence country still has an incentive to
conclude a treaty in order to limit the source country’s right to tax. There are two related
reasons for this. First, a limit on source taxes reduces the tax burden of the residence
country’s investors abroad. Second, if the source tax were lowered, countries using the
credit (or deduction) method to avoid double taxation could collect the residual taxes on
the foreign income. In other words, while the residence country is willing to grant unilateral
tax relief no matter how much tax the source country collects, it would be even better off
if, at the same time, source taxation were limited. In such a situation, the level of foreign
investment flows would be the same because the overall tax level on the investment
remained constant; but the residence country could have a larger share of the tax revenue
and its resident investors would face better tax treatment abroad. In other words, under
the unilateral relief interaction, the residence country only achieves its second-best
outcome. It could improve upon this outcome if taxation at the source were limited. This
can be achieved through tax treaties (Dagan, 2000: 982–983; Davies, 2004: 779).
This leaves the question of why the source country should be willing to enter into such
a treaty that limits its right to tax. The answer is that countries are generally residence
and source countries at the same time. Residents of one state invest in the other state
and vice versa. Consequently, investment as well as the resulting income flows in both
directions. Thus, the interest in limiting other countries’ taxation at source should hold
irrespective of relieving double taxation unilaterally as a residence country. Governments
can only achieve a limit on source taxation in other countries if they come to a binding
cooperative agreement with them. Tax treaties are a cooperative mechanism for the
reciprocal lowering of source taxation.
There are further benefits to be gained from treaty formation. For one, tax treaties
lower the administrative costs of taxation, for example, through information exchange.
Another advantage is increased legal certainty. International investors do not have to
rely on potentially conflicting national rules, but can claim protection under an inter-
national agreement. The conclusion of tax treaties has a signalling function to investors
that goes beyond that of favourable unilateral policies (Dagan, 2000: 986). Again, by
their very nature, these benefits can only be obtained through cooperation with other
governments.
So far, it has been established that countries are willing to provide tax relief unilater-
ally, but that they may have an additional, mutual interest in concluding tax treaties. This
reasoning does not, however, make a case for a particular institutional form of such a
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 597
cooperative agreement. In order to explain the institutional form, the strategic structure
has to be analysed in more detail.
Bilateral bargaining accommodates distributive concerns: The model
There is a distributive conflict built into double tax avoidance, which stems from the fact
that investment flows between countries are often not symmetric. While countries are gen-
erally residence and source countries at the same time, they are so to different degrees. A
country that is a net capital importer favours more extended source taxation; it has ‘source
interests’. A net capital exporter is in favour of residence taxation; it has ‘residence inter-
ests’ (Kingson, 1981: 1158; Rigby, 1991: 409–410). Governments often disagree about the
extent of limitations on source taxation depending on whether they have residence or
source interests. In a nutshell, the distributive conflict is about who gets how much of the
tax revenue and what the tax burden is for a country’s resident investors abroad.
Overall, the structure of the double tax avoidance game is that of a coordination game
with a distributive conflict (‘battle of the sexes’): adopting unilateral relief is always
preferred to not relieving double taxation, but the distributive consequences of that strategy
are more or less favourable to the country depending on whether they are net capital
exporters or importers. Now, being a net exporter or net importer is a relational attribute
that can vary with respect to different countries. Country A could have source interests in
relation to country B, if A is a net capital importer from B. At the same time, A might have
residence interests in relation to country C, exporting capital to C. In relation to country
D, there might not be any distributive conflict, if capital flows between A and D are
symmetric. Hence, the nature and intensity of the distributive conflict depend on dyadic
characteristics. Therefore, as I argue in the following, bilateral bargaining can accom-
modate countries’ conflicting distributive interests.
Figure 2 depicts the bargaining situation for the case of symmetric and asymmetric
capital flows between potential treaty partners. The pay-offs of country R are shown on
the x-axis, those of country S on the y-axis. The diagrams show three different welfare
levels represented by the diagonal lines. By moving from no tax relief to a situation of
unilateral tax relief and then to coordinated relief under a treaty, both countries gain and
can thus reach a higher welfare line. In the case of symmetric capital flows, the unilateral
relief interaction results in an equal distribution of the benefits. In the case of asymmetric
capital flows the net capital importer, country S in Figure 2, gets a bigger share of the
benefits (compare point A on the left-hand side to point A on the right). The difference in
benefits stems from the difference in national income that the countries receive from the
foreign investment. The source country can exert some taxation at source without reducing
the inflow of capital from the other country. Since double tax relief is in place, source
withholding taxes do not drive away foreign investment.8 The residence country therefore
has an incentive to lower the withholding taxes at source, because they directly diminish
its national income either in the form of private income (in exemption countries) or tax
revenue (in credit countries).
The starting point of treaty negotiations (point A in both graphs) lies in the middle of
the welfare line for the unilateral relief interaction of symmetric countries, whereas in the
case of asymmetric countries, it is tilted towards the net capital importer. Accordingly,
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

598 European Journal of International Relations 16(4)
the bargaining space, which is the room for Pareto improving negotiation outcomes,
demarcated by the dotted lines, is different in both cases. The extent of source taxation in
the case of symmetric countries should not be controversial. They both benefit equally
from a reduction of source taxes. Thus, the expectation is that source taxes are lower in
the case of symmetric countries. In contrast, there should be conflict over this question
in the case of asymmetric countries with net capital exporters in favour of low and net
capital importers in favour of high source taxes. Since the bargaining space is tilted
towards the net importer’s interests, the outcome of this bargain should be higher source
taxes than in the case of symmetric countries. This is indicated in the diagram by the fine
dotted lines that facilitate comparison of the outcomes of treaty negotiations between
symmetric and asymmetric countries (point B in both graphs).
If the preferred treaty rate depends on the symmetry or asymmetry of capital flows,
this could be an argument for bilateral treaties, since capital flows vary among different
country pairs. The qualitative evidence on the conflict between developing and devel-
oped countries presented above supports this. One of the reasons governments refused to
conclude a multilateral treaty was the distributive conflict over the allocation of the tax
base between countries with residence and source interests.
Quantitative evidence
In addition the model can also be subjected to a quantitative test. The central provisions
of the typical tax treaty concerning the extent of source taxation are the withholding tax
rates on passive investment income. An observable implication of the model is that these
treaty rates should be higher the more asymmetric the bilateral investment flows are. In
addition, the outcome of negotiations should also depend on the relative bargaining
power of countries. A more powerful country should be able to press for its preferred
treaty rate more successfully.
To test these propositions, I use a dataset of 80 DTAs that were signed between
1985 and 2003, and set up a linear regression model. The treaties were selected from
Figure 2. Tax treaty bargaining between symmetric and asymmetric countries
Hypothetical outcome of
uncoordinated
interaction if R were net
capital importer
C =
Treaty outcomeB =
Equilibrium outcome −
uncoordinated
interaction
A =
Symmetric Countries
A
A
B
B
C
R R
Uncoordinated,
unilateral relief
interaction
Welfare level
without
double tax
avoidance
Coordinated
treaty relief
interaction
SS
Asymmetric Countries
(R is net capital exporter)
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 599
a set of all DTAs currently in force for which foreign direct investment (FDI) data
were publicly available from one source. The treaties and the year of signature are
listed in the Appendix. The negotiated withholding tax rate for each country pair, as
the dependent variable, is taken from Tax Analysts (2004). A treaty contains four kinds
of withholding taxes: taxes on dividends, taxes on dividends between associated enter-
prises, taxes on interest and taxes on royalties (Articles 10 to 12 of the OECD MC);
thus the number of observations on the dependent variable is 320. Since I am using all
four kinds of treaty withholding rates simultaneously, the standard errors are adjusted
for clustering. The coefficients can be interpreted as the relation between the respec-
tive independent variable and the average level of withholding taxes rather than a
specific tax.
The dyadic investment position, that is, the asymmetry of bilateral FDI stocks, is the
first independent variable of interest. It is taken from OECD (2004) and measured as the
difference between the ‘outward FDI stock’ (the stock that the first country holds in the
second) and the ‘inward FDI stock’ (the stock that the second country holds in the first).9
I use the investment position of the year in which the respective treaty was signed. All
country pairs have been arranged so that the net capital exporter is in first position. Thus,
the investment position is always a positive number. The expectation is that the more asym-
metric FDI stocks are (i.e. the higher the investment position), the higher the negotiated
withholding rate. The coefficient should be positive.
As a proxy for bargaining power, the other independent variable, I use the concept
and the data of the Correlates of War project which constructs its ‘capability index’ as a
mix of military expenditure and personnel, energy consumption, iron and steel production,
and total and urban population (COW, 2001).10 I measure the bargaining power of the
first country, that is, the capital exporter, as the relative share of the sum of the bargaining
power of both treaty partners in the year of treaty signature. Since the capital exporter
should favour lower over higher withholding rates, the variable should have a negative
effect on the treaty rate. Since the outcome of negotiations should depend on the invest-
ment position and the bargaining power at the time of treaty conclusion, the data for
all variables are those for the year in which the treaty was signed.
Additionally, I control for the wealth of treaty partners, measured as the sum of per
capita GDP in the year of treaty signature. The source of these data is the Penn World
Tables.11 It is conceivable that wealthier countries are less dependent on trying to tax the
foreigner and thus the expectation is that the coefficient has a negative sign. In an analysis
similar to mine, it was found that this variable influenced the negotiated tax rate (Chisik
and Davies, 2004: 1136).
Column 2 of Table 2 shows the results of this first regression model. Under this
specification, the overall explanatory power of the model is low (R2 of 0.064). A possible
reason for this could be that a particular withholding rate can best be explained by
reference to the income on which it is levied. In order to test for this possibility, I
include dummy variables for each kind of withholding rate. The dummy variable for, let
us say, dividend taxes is 1 if the respective dependent variable is the dividend tax rate,
and 0 if it is any of the other tax rates. Thus, the coefficients report the average tax rate
for the respective type of income. The result of the second regression model is depicted
in column 3 of Table 2. The inclusion of the control variables increases the overall
explanatory power of the model: 44.9 percent of the overall variance can be explained.
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

600 European Journal of International Relations 16(4)
The results confirm the idea that negotiated withholding rates vary systematically with
respect to the investment position. The coefficient has the expected sign and is significant
at the 0.1 percent level.12 It suggests that an increase in the asymmetry of capital flows by
one million dollars leads to an increase in the withholding tax of 0.0000553 percentage
points. While this number may appear small, it can be shown to be highly plausible. The
estimation suggests that the average withholding rate in a treaty between the United States
and Japan, where the asymmetry of investment stock in the year of signature was $102,458
million, should be 5.5 percent higher than that between Australia and New Zealand (asym-
metry of $2,430 million). In other words, since there are sizeable differences in countries’
capital positions, the resultant differences in tax rates are also sizeable.
The expectations concerning the influence of bargaining power do not hold. The
coefficient has a positive sign and is insignificant. This result contradicts anecdotal
evidence in the literature that more powerful nations try to pressure less powerful countries
to agree on the tax rate as the former see fit (e.g. McIntyre, 1993: 318).
The dummy variables on the different tax rates are all highly significant. The coef-
ficient of the constant, which is 18.41, can be interpreted as the effect of the dividend tax
dummy on the withholding rate. The effect of the associated dividend tax dummy is
10.085 (–8.325 + 18.41), that of the interest tax dummy 10.572 (–7.838 + 18.41) and that
of the royalty tax dummy is 10.335 (–8.075 + 18.41). Together with the increase in R2 this
suggests that the tax dummies can explain the general level of the respective rates. This
may be interpreted as evidence that there are focal points of generally accepted rates for
different kinds of income and the variance around these rates can be explained by the
asymmetry in the stocks of FDI.
Table 2. Determinants of negotiated withholding taxes — OLS regressions
Model 1 Model 2
(Constant) 12.35*** 18.41***
(7.44) (11.52)
Investment position (Asymmetry in FDI stock) 0.0000553*** 0.0000553***
(3.63) (3.62)
Bargaining power (Share of Capability Index) 0.95 0.95
(0.85) (0.84)
Sum of per capita GDP -0.000125*** -0.000125***
(-3.59) (-3.57)
Dividend tax dummy N.A.
Associated dividend tax dummy -8.325***
(-17.05)
Interest tax dummy -7.838***
(-11.35)
Royalty tax dummy -8.075***
(-12.67)
Number of observations (N) 320 320
R2 0.064 0.449
t-values in parentheses (Robust Standard Errors)
***significant at 0.1 percent level
**significant at 1 percent level
*significant at 5 percent level
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 601
Interestingly, the estimated effect of the dividend tax dummy and the interest tax
dummy are quite close to the suggestions of the OECD MC (15 and 10 percent, respec-
tively). However, the coefficients of the associated dividend tax dummy and the royalty
tax dummy diverge considerably from the suggestions of the MC (5 and 0 percent, respec-
tively). This implies that, for the latter two kinds of income, the OECD MC is not as well
accepted as a focal point as it is for interest and dividend income. This interpretation is
further substantiated by an analysis of the commentary to the OECD MC. Governments
have registered far more reservations and observations on the suggested rates for associ-
ated dividends and royalties than for dividend and interest payments (OECD, 2005:
Commentary, Reservations and Observations on Articles 10, 11 and 12).
Finally, the negotiated rate is indeed lower if treaty partners are wealthy. The coeffi-
cient of sum of per capita GDP is negative and highly significant.
The results are in line with those of Chisik and Davies (2004), and Rixen and
Schwarz (2009). Overall, this provides support for the notion that the politics of double
tax avoidance are driven by the distributive conflict between residence and source
states. The dyadic nature of countries’ investment positions militates in favour of bilat-
eral bargaining in which the terms of the tax treaty can be designed to accommodate
each country pair’s particular investment position. Under multilateral bargaining, gov-
ernments would find it difficult or impossible to agree on one precise sharing rule that
serves their revenue interests in relation to all of the others. In other words, multilateral
bargaining would be very expensive in terms of transaction costs. But, if countries pre-
fer bilateral bargains, this then raises the question of why governments engage in com-
plementary multilateral cooperation on the development of the MC.
The interaction of bilateralism and multilateralism: Constructed focal points
In order to answer this question, one has to differentiate between the technical side of the
problem of double tax avoidance and the problem of the distribution of benefits. Both
issues have to be resolved through bargaining. On the technical side, the legal constructs
to implement the avoidance of double taxation have to be decided upon. This is repre-
sented by a pure coordination game. It may not matter so much which concepts are
chosen; the important point is that any standard is agreed upon. Since all governments
were in favour of sovereignty–preserving solutions, there was no serious bargaining prob-
lem involved. In contrast, the distributive conflict can be expected to be very tough.
Precisely because there is no serious enforcement problem in double tax avoidance, the
stakes in the bargaining process are high and governments have an incentive to ‘hold out’
in the hope of securing a more favourable agreement (Fearon, 1998: 270–271). Of course,
holding out incurs costs for both countries, so that there is a mutual interest to moder-
ate the intensity of bargaining in order to minimize transaction costs. The instrument to
achieve this is the multilateral MC, as I will show in this section.
Bargains very often find their solution in a so-called focal point, which is the point of
convergent expectations of actors with an overriding interest to agree on a coordinated out-
come. Focal points are defined as social conventions that are followed ‘automatically’
because they have become self-evident. They are more obvious, conspicuous and prominent
points of agreement than other possible solutions. A focal point can have different sources. It
may come from history, social norms and culture, or simply represent a status quo. If actors
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

602 European Journal of International Relations 16(4)
are in a bargaining situation where such a focal point exists, bargaining should consume less
time and effort because the solution gravitates towards the focal point (Schelling, 1980:
57–80). As shown above, there is no self-evident solution to the problem of double tax avoid-
ance. Neither the academic nor the political debate about the proper allocation of jurisdiction
to tax to the residence or source country has ever been settled. In other words, a focal point is
not available. The creation of institutions ‘can fill this void. By embodying, selecting, and
publicizing particular paths on which all actors are able to coordinate, institutions may pro-
vide a constructed focal point’ (Garrett and Weingast, 1993: 176; emphasis in original). In the
case at hand, governments engage in the intentional creation of a focal point in the form of
model tax conventions that limit the range of possible solutions.
Of course, when negotiations at the League of Nations began, the initial goal was not
to construct a focal point for bilateral bargains, but to come to a binding multilateral DTA.
However, this proved to be elusive, so governments contented themselves with coming to
an agreement on a model convention for bilateral bargains. The attempt to find a focal
solution was subject to bargaining problems similar to those faced in negotiations about
the binding agreement itself. Importantly, it proved to be comparably easy to come to
agreement on the technical side of the problem, that is, to agree on legal constructs that are
capable of achieving the coordination of different national tax laws. While there were
important technical problems to be resolved that required some ingenuity from tax experts,
all actors agreed that the technical solution should be sovereignty preserving. However,
the distributive conflict was more difficult to solve; the difficulties encountered in trying
to forge agreement on the MC within international organizations and the fact that there are
two competing models in terms of the distribution of benefits are evidence of this.
The fact that the OECD MC is non-binding can explain why it was nonetheless
possible to forge an agreement. Since governments know ex ante that they are allowed to
deviate from the convention in their bilateral agreements, they are more willing to
subscribe to a model, even if it may not entirely conform to their distributive preferences.
The flexibility inherent in soft law is one of the main reasons that it is chosen by govern-
ments. Rather than having to ‘accommodate divergent national circumstances within a
single text’, it leaves ‘flexibility in implementation’ (Abbott and Snidal, 2000: 445). The
multilateral MC pre-structures the bilateral bargains, but it does not predetermine them
entirely. It is merely the starting point for bilateral negotiations, in which a binding
solution to accommodate parties’ distributive interests is achieved.
Importantly, the non-binding nature of the MC does not matter for its effectiveness
with respect to the technical side of double tax avoidance. Since all countries have the
desire to be coordinated, any workable solution that is found is accepted and there is no
reason to deviate from this solution in bilateral bargains. In this sense, the OECD MC
provides standards, which countries voluntarily adopt in their bilateral bargains. As is
the case in most standard-setting regimes, while making the rules binding on countries
would not meet with resistance, there is no need to do so, because they wish to follow
them in any case. The rules are self-enforcing. At the same time, since there are indeed
various technical problems involved in developing these standards, there is a demand
for pooling the expertise and information, and making it available to other governments.
A multilateral institution that is specialized in collecting, creating and disseminating
information can fulfill this task (Snidal, 1985: 928).
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 603
That distributive problems and not technical issues are at the heart of difficulties in
achieving agreement on a focal point is evident in the historical development. As shown,
agreement on the technical side of the problem emerged early under the League of
Nations. However, the League years ended with two technically identical conventions,
the Mexico and London models, but each with different distributive implications — one
emphasizing the source, the other the residence, principle. Agreement on one model
convention was elusive because of the heterogeneous group of countries in the League
of Nations. During the OECD years, governments managed to agree on one model. This
success was facilitated by the fact that the OECD is made up of a relatively small group
of countries with relatively symmetric capital flows between them. Therefore, the
distributive conflict between these countries is weaker. Accordingly, in combination with
the non-binding character of the model, agreement was easier. Importantly though, the
technical solutions that were developed in the previous period were not challenged, but
merely further developed and refined.
The adoption of a model by such a small and exclusive group of countries also has
consequences for the countries remaining outside the agreement. Since OECD countries,
which are mostly capital exporters, were the ‘first at the table’, they could implement
a system that favoured residence countries. The best that the ‘last at the table’ — the
developing countries — could do was to follow the OECD, even though the rules were
less favourable to them (Horner, 2001: 183–184). The first-mover advantage, a feature
of coordination games, accounts for the fact that, in general, the double tax rules favour
developed countries.
The disadvantaged countries have tried to change this situation and attempted to
counterbalance the OECD MC through the UN MC. While the UN did not attempt to
challenge the technical solutions, its MC aims to be a corrective to the distributive solution
that emerged within the OECD. The OECD MC has quite clearly not achieved universal
acceptance as a focal point with respect to the issue of the distribution of benefits. The
empirical evidence concerning the influence of the tax dummy variables on the level
of different withholding rates supports this view. The suggested rates for royalties and
associated dividends are less well accepted than those for dividends and interest.
But even the coexistence of two distributively divergent model conventions facilitates
bilateral bargaining. The fact that the disagreement about the distribution of benefits is
embodied in multilateral model conventions sponsored by well-respected international
organizations legitimizes the distributive conflict. The discussions of these problems in
multilateral forums allow treaty negotiators to anticipate the areas where conflict can be
expected. Multilateral bargaining about focal solutions rationalizes the distributive
conflict and thus enables its quicker resolution in the bilateral setting.
The enforcement phase and the mutual agreement procedure
As argued above, there are no enforcement problems with DTAs. One of the most impor-
tant sources of such problems is strongly mitigated in double tax avoidance: monitoring
is not problematic in the case of tax treaties. Any violation of a treaty can be easily
detected because there is a natural third party to the agreement — namely, the taxpayer.
If one of the treaty partners violates a DTA, the taxpayer will notice this violation and
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

604 European Journal of International Relations 16(4)
notify the competent authorities in his/her home country. The countries then enter into
MAP negotiations to reach an agreement over the treaty violation.
The absence of a serious enforcement problem does not, however, mean that there are
no conflicts in applying the provisions of the treaty. Rather than resulting from con-
scious and deliberate efforts to cheat on treaty partners, disputes often stem from prob-
lems in interpreting the agreement correctly. Given that domestic and international tax
rules are complicated, avoiding double taxation can be understood as a complex transac-
tion. Not all future contingencies can be dealt with at the time of concluding the agree-
ment because they are not known; or, even if they can be anticipated, the transaction
costs of agreeing on contractual provisions for them would be too high. Tax treaties are
thus necessarily incomplete contracts that involve indeterminacy and often have to
be amended to accommodate new circumstances ex post. Contract theory suggests that,
in such circumstances, treaty partners should not try to agree on detailed rules ex ante,
but instead should keep the treaty more general and agree on a procedure to be followed
should a dispute arise over the application of the provisions. The procedure chosen in
tax treaties is the MAP. Its major function is that of a flexible mechanism of ex post
treaty negotiations. It enables ‘ad hoc and ex post agreement’ between governments
through which divergent treaty interpretations can be brought into line and ‘temporary
or unforeseen problems’ be addressed (Aoyama, 2004: 653).
Given that DTAs are incomplete contracts it is understandable that governments did
not choose an external enforcement mechanism, and generally resisted the introduction
of binding arbitration. They wish to determine the terms of agreement by themselves,
rather than grant the power of treaty interpretation and ex post amendment to a third
party (Green, 1998: 129–137). The absence of a major enforcement problem makes such
a solution viable. The fact that the MAP enables the adaptation of existing tax treaties to
new circumstances may also explain why, on average, treaties between OECD countries
are changed only every 14 years (Sasseville, 1999: 56). Given the flexibility of the MAP,
formal treaty renegotiations are not considered necessary.13
Contrasting cases: Bilateral agreement and the absence of free rider problems
So far, it has been argued that bilateral bargaining is well suited to accommodate govern-
ments’ distributive concerns. In addition, there is no major enforcement problem that
would require third-party enforcement. But this does not explain why binding agreement
is bilateral in tax treaty-making. As we have seen, there is an important role for comple-
mentary multilateral bargaining in order to provide focal points for bilateral bargains. This
raises the possibility that countries could also come to a binding multilateral agreement. In
fact, it is easy to draft a multilateral tax treaty that contains the same provisions as the
OECD MC (Lang, 1997). Governments could, in principle, agree on a multilateral tax
treaty that would leave them distributional flexibility. The distributively sensitive aspects
could still be determined in bilateral bargains and subsequently all countries involved
could agree on one multilateral document that included a series of bilaterally varying
provisions.14 Why, then, is agreement not multilateral in double tax avoidance?
The answer is that there is no need for such a binding multilateral framework in
double tax avoidance because there is no free rider problem. This point can be worked
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 605
out by contrasting the cases of the international trade and tax regime (the following argu-
ment is developed in more detail in Rixen and Rohlfing, 2007). In international trade
liberalization, which has for a long time been achieved through bilateral agreement,
countries have to balance the interests of ‘their’ import-competing and exporting indus-
tries. They generally do not engage in unilateral trade liberalization and can best
achieve this balance through bilateral agreements. However, once they have struck a
deal, the balance achieved in relation to one country may be upset by a subsequent trade
agreement between their treaty partner and a third country. If that agreement is more
favourable to the third country, the exporters of the first country may suffer. In order
to prevent this, governments introduced Most Favoured Nation (MFN) clauses into their
bilateral agreements, so that their exporters would always be given the best treatment
that their treaty partners grant to any country. Thus, MFN treatment, which became man-
datory under the GATT, is intended to ensure that the domestic balance between import
competitors and exporters is not disturbed. However, once a series of bilateral agree-
ments with MFN treatment is in place, a positive externality for other countries is pro-
duced. Governments have an incentive to lean back and wait for other countries to
conclude agreements which the former are then able to access without granting any con-
cessions themselves. In this situation, multilateral agreement is an institutional safe-
guard by which such free riding can be prevented. After all concessions have been
exchanged, each member country can consider the bargains in conjunction. If one
country believes that another intends to free ride, the former can withhold some of the
concessions and insist on additional concessions by the potential free rider.
In international taxation the situation is different. A third-country effect of this kind
does not exist. Initially, it is conceivable that a government would like to ensure that no
third country gets a better deal from one of its treaty partners than it could itself secure.
One motive might be concerns about competitiveness; a government may want to make
sure that its own investors get at least the same concessions as investors from other coun-
tries. Accordingly, that government could insert an MFN clause into its treaties. The fact
that MFN clauses are sometimes found in tax treaties indicates that such considerations
may play a role in tax treaty-making. However, MFN treatment is an exception rather
than the rule in tax treaties. In fact, the quantitative evidence presented above suggests
that the kind of third-country effects that would make MFN treatment desirable cannot
be very strong. Otherwise, the correlation between withholding tax rates and bilateral
investment positions would not be as strong, because an MFN clause clearly upsets the
distributional balance of reciprocal concession-making in bilateral bargains. And indeed,
it can be shown that the way MFN clauses are used in tax treaties has to do with the
desire to balance bilateral, reciprocal deals.15
The decisive difference between taxation and trade is that, in the trade regime, MFN
treatment has become mandatory for solving an enforcement problem: if there is no
mandatory MFN treatment, the fear is that an unravelling of all bilateral bargains would
occur due to the third-country effect. This fear is clearly not present in the DTA regime.
In trade, the domestic balance between import competitors and exporters that has been
disturbed by the third-country effect provides an incentive to defect from the prior
treaty. In international taxation, since a comparable politically salient domestic conflict
does not exist, a country would not defect from prior agreements, even if one made the
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

606 European Journal of International Relations 16(4)
questionable assumption of a strong third-country effect. This explains why there is no
mandatory MFN treatment in double tax avoidance and why it is rarely used in practice.
Consequently, the possibility to free ride on the concessions other countries have made
is not a relevant factor. Whereas bilateral agreements in international trade create exter-
nalities that countries wish to internalize by means of binding multilateral agreement,
externalities of this kind do not exist in double tax avoidance.16 Governments do not
come to a binding multilateral agreement because there is no need for it.
The fact that the tax regime has a strong bilateral element and a switch to a multi-
lateral tax treaty did not occur challenges the claim that ‘solving coordination prob-
lems is institutionally neither complex nor particularly demanding, and it was the
domain in which multilateralism … flourished in the nineteenth century’ (Ruggie,
1993: 22). My argument turns Ruggie’s logic around. Precisely because the underlying
strategic structure does not necessarily require multilateral agreement, binding agree-
ment could remain bilateral. The multilateral institutions of double tax avoidance
‘only’ have a facilitating role for resolving bargaining problems — a task that they
accomplish successfully, as the continuing growth and substantial degree of homoge-
nization of the treaty network shows.
A competing explanation: Bilateralism preserves sovereignty
I do not have the space for an in-depth analysis of competing explanations here, but will
briefly consider the most common one. The bilateral nature of DTAs is often ‘explained’
by a broad and unspecified reference to the desire of countries to maintain their tax
sovereignty (cf. e.g. Abbott and Snidal, 2000: 441). While this claim is hardly ever
spelled out in detail, it could be understood as a competing explanation.
First of all, it is true that governments want to preserve their tax sovereignty. The
institutional set-up is constructed in a sovereignty-preserving way, so that double taxa-
tion can be avoided by interface regulation. Governments cherish the flexibility of this
set-up, which allows them to design their domestic tax laws independently of other coun-
tries. However, this fact by itself is not a sufficient condition for the bilateral nature of
tax treaties. For one, it would be conceivable to conclude a multilateral tax treaty
based on the very same legal constructs and technical solutions currently used in bilat-
eral tax treaties. Neither does multilateralism as such restrict sovereignty, nor is the bilat-
eral form necessarily more sovereignty preserving. Rather, the substantive provisions
contained in a treaty determine whether or not it preserves national sovereignty. For exam-
ple, bilateral tax treaties that were not based on separate accounting but would instead
use unitary taxation with formula apportionment would require the definition of a com-
mon tax base. This would restrict a single government’s tax sovereignty more than a
multilateral treaty based on the arm’s length standard.17
Second, as shown elsewhere (Rixen, forthcoming), governments are, if only grudg-
ingly, willing to compromise some of their national tax sovereignty in the fight against tax
evasion and avoidance. This is a reaction to the functional requirements of this problem,
which are fundamentally different from those of double tax avoidance. The difference in the
reaction to the two problems shows that the sovereignty-preserving and bilateral approach
in the realm of double tax avoidance hinges on the underlying strategic structure. While the
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 607
desire to preserve national sovereignty may be one of the reasons why governments find
this particular set-up attractive, the fact that such a solution was viable cannot be explained
by reference to this desire. Consequently, the notion that countries’ desire to preserve their
tax sovereignty can explain bilateralism is incomplete.
Conclusion
In this article I have constructed an explanation of the design features of international
tax cooperation. Bilateral bargaining is preferred over multilateral bargaining because
the asymmetry of investment flows can be better accommodated in bilateral bargains.
Despite this preference for bilateral bargains, governments have an interest in develop-
ing model conventions and a multilateral forum for discussion, serving as a constructed
focal point. Concerning the ex post phase of cooperation, it was shown that there is no
need for external enforcement mechanisms. The mutual agreement procedure is best
understood as a device for dealing with the problems of incomplete contracting.
Concerns for third-country benefits, that is, externalities of bilateral agreements, are
not relevant, so that there is no free rider problem and thus no need for multilateral
agreement.
In the case of double tax avoidance, the problems of cooperation lie mainly within
the sphere of bargaining. Cooperation is made difficult not by the fact that enforcement
is problematic but by struggling over the terms of the agreement. Since the question of
‘who gets what’ is difficult to resolve, most of the governance design elements — for
example, constructed focal points and the MAP — concern the facilitation of successful
bargaining between countries. This finding lends support to the argument that consid-
ering bargaining and distribution is as important as enforcement and should receive
more attention by IR scholars (Fearon, 1998: 297–299; Koremenos et al., 2001: 765;
Simmons and Martin, 2002: 204). However, as shown, this does not mean that the
enforcement phase can be ignored. On the contrary, the absence of enforcement prob-
lems amplifies the intensity of bargaining problems. This absence also accounts for the
fact that multilateral agreement is not necessary. Thus, it is crucial to understand both
the bargaining and enforcement phases, and their interaction, in order to make sense of
institutional design (cf. for a similar argument, Drezner, 2000; Barkin, 2004).
These findings exemplify, first, that a straightforward and dichotomous classification
of international cooperation into bilateral or multilateral cooperation is unhelpful. Rather,
as the comparison with international trade suggests, it is quite likely that different policy
fields exhibit different mixes of bilateralism and multilateralism at different stages of the
cooperation process.
Second, these findings suggest that the choice between a bilateral and multilateral form
of cooperation is contingent on a particular configuration and interaction of distribution
and enforcement problems. An important conclusion to be drawn from my findings for
future efforts at theory development is thus to pay closer attention to matters of bargaining
and distribution, and to give greater consideration to how these interact with enforcement
problems. Additional issue areas that exhibit different configurations of distribution and
enforcement problems need to be analysed to generate generalized knowledge on the
choice between, or the coexistence of, bilateralism and multilateralism.
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

608 European Journal of International Relations 16(4)
Acknowledgements
I received helpful comments from Martin Binder, Matthias Ecker-Erhardt, Steffen Ganghof,
Philipp Genschel, Benny Geys, Achim Helmedach, Markus Jachtenfuchs, Peter Mayer, Ingo
Rohlfing, Peter Schwarz, Susanne Uhl, Lora Viola, Dieter Wolf, Michael Zürn and two anonymous
reviewers. Mary Kelley-Bibra, Manuel Domes and Vanina Nikolova provided research assistance.
Special thanks to all of them. Part of the research for this article was conducted when I was a mem-
ber of the Collaborative Research Center 597 ‘Transformations of the State’ at Jacobs University
Bremen, Germany. Financial support by the German Research Foundation (DFG) is gratefully
acknowledged.
Appendix: List of tax treaties included in the
regression analysis
Exporter Importer Signed
Australia New Zealand 1995
Austria Singapore 2001
Austria Slovenia 1997
Canada Czech Republic 2001
Canada Mexico 1991
Czech Republic Slovakia 2002
Denmark Canada 1997
Denmark Germany 1995
Denmark Poland 2001
Denmark Portugal 2000
Denmark Singapore 2000
Denmark United States 1999
Finland Austria 2000
Finland Netherlands 1995
France Austria 1993
France Italy 1989
France Mexico 1991
France Spain 1995
France United States 1994
Germany Austria 2000
Germany Canada 2001
Germany India 1995
Germany Indonesia 1990
Germany Italy 1989
Germany Korea (R.O.K.) 2000
Germany Mexico 1993
Germany Norway 1991
Germany Poland 2003
Germany Russia 1996
Germany Turkey 1985
Germany United States 1989
Italy Denmark 1999
Italy Turkey 1990
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 609
Appendix: (Continued)
Exporter Importer Signed
Japan Canada 1986
Japan France 1995
Japan Korea (R.O.K.) 1998
Japan Malaysia 1999
Japan Norway 1992
Japan Turkey 1993
Japan United States 2003
Netherlands Canada 1986
Netherlands Denmark 1996
Netherlands Indonesia 2002
Netherlands Italy 1990
Netherlands Poland 2002
Netherlands Portugal 1999
Netherlands United States 1992
Norway Austria 1995
Norway Canada 2002
Norway Ireland 2000
Norway Netherlands 1990
Norway United Kingdom 2000
Portugal Brazil 2000
Russia Austria 2000
Russia Finland 1996
Russia France 1996
South Africa Australia 1999
South Africa Austria 1996
Sweden Canada 1996
Sweden France 1990
Sweden Germany 1992
Sweden United States 1994
Switzerland Canada 1997
Switzerland United States 1996
United Kingdom Australia 2003
United Kingdom India 1993
United Kingdom Italy 1988
United Kingdom Korea (R.O.K.) 1996
United Kingdom Malaysia 1996
United Kingdom Singapore 1997
United Kingdom South Africa 2002
United Kingdom United States 2001
United States Austria 1996
United States Indonesia 1988
United States Ireland 1997
United States Italy 1984
United States Mexico 1992
United States South Africa 1997
United States Spain 1990
United States Venezuela 1999
Source: Tax Analysts (2004).
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

610 European Journal of International Relations 16(4)
Notes
1. The other tax problem resulting from the incongruity between an international economy and
national tax sovereignty is so-called double non-taxation. The related phenomena of tax eva-
sion, tax avoidance, tax competition and tax havens are all instances of double non-taxation.
I address these important issues elsewhere (Rixen, 2008: 117–151, 181–203; forthcoming),
but set them aside for this article.
2. There is one minilateral treaty among the Scandinavian countries that was signed in 1983
(Mattsson, 2000).
3. A particularly ironic critique in this respect can be directed towards early game-theoretic
regime research that used two-person games to analyse multilateral cooperation. Even though
the bilateral form of the games would suggest otherwise, one hardly finds an explicit discus-
sion of bilateralism in this work. Exceptions are Oye (1985) and Snidal (1985).
4. This figure is only the lower bound of all bilateral agreements because states are not obliged
to deposit their treaties with the UN.
5. The conflict between residence and source states on the avoidance of double taxation is
not only interest driven. There is also persistent disagreement in the academic literature on
whether the residence or source principle is preferable when measured against the normative
ideals of equality and efficiency (for an overview, see Rixen, 2008: 57–63).
6. Under the deduction method, taxes paid abroad are subtracted from the tax base on which
home taxation is based. Thus, double taxation is not fully relieved. All OECD countries
provide unilateral tax relief and in most cases they use the credit or exemption method, that
is, provide full relief (see Yoo, 2003).
7. For a more extensive derivation of the strategic structure by both theoretical and empirical
reasoning, see Rixen (2008: 31–54, 156–161).
8. The most obvious case in which imposing source taxation does not drive away foreign invest-
ment is that of the residence country granting a foreign tax credit. In that case, taxation at
source merely results in a revenue transfer from the residence country’s to the source country’s
treasury. However, even under different constellations the imposition of source withholding
taxes does not drive away foreign investment (Davies, 2004: 785–787).
9. There are two kinds of FDI data. There is the FDI flow in a given year, measuring new
investment flowing into and out of a country, and the FDI stock accumulated over time in
a country. This latter measure should be relevant for tax treaty negotiations, since it is the
income generated from the stock of foreign investment that is to be subjected to taxation.
On the basis of this, the preceding discussion of the effect of ‘capital flows’ might appear
incorrect. However, it can be justified because the stock measure represents the long-term
capital flows between two countries more accurately than the annual measures of capital
flows, which fluctuate significantly. In addition, using the stock data instead of the flow
data can ameliorate potential endogeneity problems — if one thought that the flow of FDI
depended on the negotiated withholding rate rather than the other way around.
10. In regressions not reported here, I have experimented with other possible specifications of
bargaining power. Using the relative share of GDP or the relative share of military expenditure
yields similar results as those reported below.
11. Available from the Center for International Comparisons (CIC), University of Pennsylvania,
http://pwt.econ.upenn.edu/ (last accessed 9 February 2009).
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 611
12. Note that the dataset is actually biased against my prediction. Due to problems of FDI data
availability, the set contains mostly OECD countries. The capital flows between these countries
is generally more symmetric than that between developed and developing countries.
13. Recently, the OECD added an additional paragraph to Article 25 of the MC, which foresees the
submission to arbitration if the affected taxpayer so requests, and if the competent authorities
do not reach agreement within two years (OECD, 2007). Rather than indicating that there may
after all be a serious enforcement problem, this has simply been done to speed up the MAP.
The fact that few cases have moved to the stage of arbitration is evidence that complementary
arbitration is indeed successful in achieving this goal (see Rixen, 2008: 173–175).
14. In fact, the one instance of multilateralism (or rather minilateralism) — the Scandinavian tax
treaty — is of this kind (Vann, 1991: 151).
15. The treaty between Norway and Australia may serve as an example. Norway was not able
to push through its desired low withholding rates. In order to at least ‘win’ something in
the negotiations, it managed to introduce an MFN clause that foresees the renegotiation
of the treaty should Australia grant lower withholding taxes in the future (Lennard, 2005:
99–100). In most tax treaties, MFN treatment is not granted automatically, but only consists
of a commitment to renegotiate.
16. This reasoning leaves aside negative externalities resulting from issues of tax evasion, avoid-
ance or competition (‘double non-taxation’). While it is certainly true that these externalities
may pressure countries to find multilateral solutions, the abstraction from these externalities
is justified by the empirical record. The problem of double non-taxation is addressed by con-
structing a multilateral support structure, which is institutionally distinct from the bilateral
tax treaty network. While it would be desirable to deal with the two problems of double
taxation and double non-taxation in an integrated fashion, for example, under the umbrella
of an ‘international tax institution structured like the GATT’ (Vann, 1991: 100), it is unlikely
that the institutional path of bilateral tax treaty-making can easily be left. Nevertheless, the
increasing importance of tax competition, avoidance and evasion has led to gradual changes
in the DTA regime (Rixen, forthcoming).
17. On the difference between the arm’s length standard and unitary taxation with formula appor-
tionment, see e.g. Li (2003: 509–620).
References
Abbott KW, Snidal D (2000) Hard and soft law in international governance. International
Organization 54(3): 421–456.
Aoyama K (2004) Mutual agreement procedure: Addressing new challenges in transfer pricing.
Tax Notes International 35(August 16): 651–664.
Arnold BJ, McIntyre MJ (1995) International Tax Primer. Den Haag: Kluwer Law International.
Barkin JS (2004) Time horizons and multilateral enforcement in international cooperation.
International Studies Quarterly 48(2): 363–382.
Bird RM, Wilkie JS (2000) Source- vs. residence-based taxation in the European Union: The
wrong question? In: Cnossen S (ed.) Taxing Capital Income in the European Union. Issues
and Options for Reform. New York: Oxford University Press, 78–109.
Bird RM, Mintz JM (2003) Sharing the international tax base in a changing world. In: Cnossen S, Sinn
HW (eds) Public Finance and Public Policy in a New Century. Cambridge: MIT Press, 405–446.
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

612 European Journal of International Relations 16(4)
Brauner Y (2003) An international tax regime in crystallization. Tax Law Review 56: 259–328.
Caporaso JA (1993) International relations theory and multilateralism: The search for foundations.
In: Ruggie JG (ed.) Multilateralism Matters: The Theory and Praxis of an Institutional Form.
New York: Columbia University Press, 51–90.
Chisik R, Davies RB (2004) Asymmetric FDI and tax-treaty-bargaining: Theory and evidence.
Journal of Public Economics 88: 1119–1148.
COW (2001) National Material Capabilities from 1816-2001 (v3.02). Available at: http://
correlatesofwar.org/
Dagan T (2000) The tax treaties myth. Journal of International Law & Politics 32: 939–996.
Davies RB (2004) Tax treaties and foreign direct investment: Potential vs. performance.
International Tax and Public Finance 11(6): 775–802.
Drezner DW (2000) Bargaining, enforcement, and multilateral sanctions: When is cooperation
counterproductive? International Organization 54(1): 73–102.
Fearon JD (1998) Bargaining, enforcement, and international cooperation. International
Organization 52(2): 269–305.
Garrett G, Weingast BR (1993) Ideas, interests, and institutions: Constructing the European com-
munity’s internal market. In: Goldstein J, Keohane RO (eds) Ideas and Foreign Policy. Beliefs,
Institutions, and Political Change. Ithaca: Cornell University Press, 173–206.
Graetz MJ (2001) Taxing international income: Inadequate principles, outdated concepts and
unsatisfactory policies. Tax Law Review 54: 261–336.
Green RA (1998) Antilegalistic approaches to resolving disputes between governments: A compari-
son of the international tax and trade regimes. Yale Journal of International Law 23(1): 79–139.
Horner FM (2001) Do we need an international tax organization. Tax Notes International
24(October 8): 179–187.
Kingson CI (1981) The coherence of international taxation. Columbia Law Review 81(6):
1151–1289.
Koremenos B, Lipson C, and Snidal D (2001) The rational design of international institutions.
International Organization 55(4): 761–799.
Kosters B (2004) The United Nations model tax convention and its recent developments. Asia-
Pacific Tax Bulletin 5(1/2): 4–11.
Lang M (1997) The concept of a multilateral tax treaty. In: Lang M, Loukota H, Rädler AJ, et al. (eds)
Multilateral Tax Treaties. New Developments in International Tax Law. Wien: Linde, 187–196.
League of Nations (1923) Report on Double Taxation. Submitted to the Financial Committee
by Professors Bruins, Einaudi, Seligman and Sir Josiah Stamp. Geneva: League of Nations,
E.F.S.73.F.19.
League of Nations (1927) Double Taxation and Tax Evasion. Report Presented by the Committee
of Technical Experts on Double Taxation and Tax Evasion. Geneva: League of Nations, C. 216.
M 85. 1927 II.
Lennard M (2005) The GATT 1994 and direct taxes: Some national treatment and related issues.
In: Lang M, Herdin J, and Hofbauer I (eds) The WTO and Direct Taxation. The Hague: Kluwer
Law International, 73–101.
Li J (2003) International Taxation in the Age of Electronic Commerce: A Comparative Study.
Toronto: Canadian Tax Foundation.
Mattsson N (2000) Multilateral tax treaties – A model for the future? Intertax 28(8–9): 301–308.
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

Rixen 613
McIntyre MJ (1993) Guidelines for taxing international capital flows: The legal perspective.
National Tax Journal 46(3): 315–321.
Messere K (1993) The precursors and successors of the new OECD model tax convention on
income and capital. European Taxation 33(8): 246–251.
Mundell RA (1957) International trade and factor mobility. American Economic Review 47(3):
321–356.
Odell JS (2000) Negotiating the World Economy. Cornell: Cornell University Press.
OECD (2004) International Direct Investment Statistics. Available at: www.sourceoecd.org
OECD (2005) Model Tax Convention on Income and on Capital. Electronic Version. Paris:
OECD.
OECD (2007) Improving the Resolution of Tax Treaty Disputes. Available at: www.oecd.org/
dataoecd/17/59/38055311.pdf
Olson M (1965) The Logic of Collective Action. Public Goods and the Theory of Groups.
Cambridge: Harvard University Press.
Owens J, Bennett M (2008) OECD model tax convention: Why it works. OECD Observer.
Available at: www.oecdobserver.org/m/fullstory.php/aid/2756
Oye KA (1985) Explaining cooperation under anarchy – Hypotheses and strategies. World Politics
38(1): 1–24.
Picciotto S (1992) International Business Taxation. A Study in the Internationalization of Business
Regulation. New York: Quorum.
Radaelli CM (1998) Game theory and institutional entrepreneurship: Transfer pricing and the
search for coordination in international tax policy. Policy Studies Journal 26(4): 603–619.
Rigby M (1991) A critique of double tax treaties as a jurisdictional coordination mechanism.
Australian Tax Forum 8(3): 303–427.
Rixen T, Rohlfing I (2007) The institutional choice of bilateralism and multilateralism in interna-
tional trade and taxation. International Negotiation 12(3): 389–414.
Rixen T (2008) The Political Economy of International Tax Governance. Basingstoke: Palgrave/
Macmillan.
Rixen T (forthcoming) From double tax avoidance to tax competition: Explaining the institutional
trajectory of international tax governance. Review of International Politicial Economy.
Rixen T, Schwarz P (2009) Bargaining over the avoidance of double taxation: Evidence from
German tax treaties. FinanzArchiv: Public Finance Analysis 65(4): 442–471.
Ruggie JG (1993) Multilateralism – The anatomy of an institution. In: Ruggie JG (ed.) Multi-
lateralism Matters: The Theory and Praxis of an Institutional Form. New York: Columbia
University Press, 3–47.
Sasseville J (1999) The future of the OECD model tax convention. In: Gassner W, Hemetsberger-Koller
H, Lang M, et al. (eds) Die Zukunft des Internationalen Steuerrechts. Vienna: Linde, 41–58.
Schelling TC (1980) The Strategy of Conflict. Cambridge: Harvard University Press.
Simmons BA, Martin LL (2002) International organizations and institutions. In: Carlsnaes W,
Risse T, and Simmons B (eds) Handbook of International Relations. London: Sage, 192–211.
Snidal D (1985) Coordination versus prisoner’s dilemma: Implications for international coopera-
tion and regimes. American Political Science Review 79: 923–942.
Tanzi V (1999) Is there a need for a world tax organization? In: Razin A, Sadka E (eds) The
Economics of Globalization. Policy Perspectives from Public Economics. Cambridge:
Cambridge University Press, 173–186.
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from

614 European Journal of International Relations 16(4)
Tax Analysts (2004) Worldwide Tax Treaties in Force. Available at: www.taxanalysts.com.
Supplements available at: web.lexis-nexis.com
United Nations (2003) United Nations Treaty Series. Available at: http://untreaty.un.org
Vann RJ (1991) A model tax treaty for the Asian-Pacific region? (Part I and II). Bulletin for
International Fiscal Documentation 45(3 and 4): 99–111 and 151–163.
Whalley J (2001) Puzzles over international taxation of cross border flows of capital income.
National Bureau of Economic Research Working Paper 8662.
Williamson OE (1985) The Economic Institutions of Capitalism. New York: Free Press.
Yoo K-Y (2003) Corporate taxation of foreign direct investment income 1991–2001. OECD
Economics Department Working Papers No. 365.
Züger M (2001) Arbitration under Tax Treaties. Improving Legal Protection in International Tax
Law. Amsterdam: IBFD Publications.
Biographical note
Thomas Rixen, political scientist and economist, is a research fellow at the Social
Science Research Center Berlin (WZB) in the research unit ‘Transnational Conflicts and
International Institutions’. His work focuses on international and comparative political
economy and the theory of institutions.
at Wissenschaftszentrum Berlin on December 8, 2010ejt.sagepub.comDownloaded from