In recent years, international organizations (IOs) have become
increasingly pervasive features of the global landscape. Both the number
of such organizations and the range of issue-areas they cover have grown
rapidly. The implications of this development have been studied
extensively and hotly debated in the field of international relations.
Whereas some researchers believe that IOs have little effect on state
behavior, many observers argue that the proliferation of these
institutions will facilitate interstate cooperation and help to resolve
the interstate conflicts that do arise.1
In contrast, relatively little research
has been conducted on the factors that prompt states to enter IOs. This
gap in the literature is both surprising and important. While many
countries have rushed to join IOs, others participate in few of these
organizations. What determines the propensity of states to join IOs?
We argue that changes in a state's regime type are crucially
important in this regard. Countries undergoing a democratic transition are
especially likely to enter IOs because leaders have difficulty credibly
committing to sustain liberal reforms and the consolidation of democracy.
Chief executives often have an incentive to solidify their position during
democratic transitions by rolling back political liberalization. Entering
an IO can help leaders in transitional states credibly commit to carry out
democratic reforms and can reduce the prospect of reversions to
authoritarianism, especially if the organization is composed primarily of
democratic members.
To test this claim, we examine whether regime change has influenced IO
membership from 1965 to 2000. Consistent with our argument, we find strong
evidence that democratic transitions prompt states to enter these
organizations. Furthermore, states in the throes of democratization tend
to join IOs composed of democratic members; and the likelihood that a
democratizing state will subsequently backslide in an autocratic direction
is reduced if it enters a relatively democratic organization.
Consequently, political liberalization and IO membership seem to go hand
in hand.
Why Do States Enter International
Organizations?
IOs are “associations established by governments or their
representatives that are sufficiently institutionalized to require regular
meetings, rules governing decision- making, a permanent staff, and a
headquarters.”2
Since
membership in such an organization may limit the policy discretion of
national leaders, why would decision makers choose to join one? Most of
the existing research on this topic emphasizes that states enter these
arrangements to help solve coordination problems, to prompt third parties
to change particular policies, or to help enforce bargains.
3
Moreover, most of this work has evolved in the fields of international
political economy and security studies. Traditional work in the field of
international organization still describes the role of IOs as pursuing
“common or converging national interests of the member
states.”
4 The primary purpose of
IOs, for the vast majority of this literature, is to resolve cross-border
issues that cannot otherwise be addressed domestically.
5 Our contention, however, is that a key impetus to IO membership
emanates from the domestic political arena, particularly from transitions
toward democracy. This is not to imply that international factors are
unimportant influences on IO membership, as we discuss at greater length
below. Nonetheless, states frequently join international institutions with
an eye toward domestic politics, an issue that has received very little
attention to date.
The scholarship that does exist on the domestic sources of IO
membership focuses primarily on regional organizations.6
Moreover, little of this
research addresses the effect of regime type; virtually none of it
addresses the influence of regime change. Some studies have analyzed the
relationship between regime type and membership in political-military
alliances, but the results have not produced any consensus on the nature
or the strength of this relationship.
7 Other research has
focused on the links between regime type and organizations that monitor
and regulate foreign trade. Mansfield, Milner, and Rosendorff, for
example, found that democratic states are especially likely to enter
preferential trading arrangements (PTAs).
8 However, alliances and PTAs make up only a small
portion of IOs; indeed, many alliances (for example, nonaggression pacts)
would not qualify as an IO at all.
The most direct analysis of regime type and IO membership was
undertaken by Jacobson and his colleagues.9
They found that democracy promotes IO membership,
but that democratization reduces the rate at which states enter
international institutions. However, the strength of democracy's
effect varied depending on how their statistical model was specified and
their research was based on only two years of data.
10 Russett and Oneal concluded that democracy has fostered IO
involvement during the period since World War I, although the magnitude of
democracy's effect is quite small.
11 Moreover, as they pointed out, their analysis of IO
membership was rather cursory.
Democratization and IO Membership
The aim of this article is to provide a more systematic analysis of
whether democratization influences the propensity of states to join IOs.
The past three decades have been marked by a wave of democratization:
dozens of countries throughout the world have undergone transitions to
democracy.12
This development has stimulated a large body of
research, but little of it addresses the foreign policy implications of
such transitions. Our central argument is that democratization is a potent
impetus to IO membership. A key problem for democratizing countries is the
difficulty that leaders face in credibly committing to sustain domestic
reforms. Heads of state may have reason to solidify their hold on power
during the transition by canceling elections, suspending reforms, reneging
on promises, or even cracking down on potential antiregime forces. Some of
these tactics may stem from a desire to consolidate political power at the
expense of reform, but it is also possible that the lack of a perceived
commitment to liberalization itself hampers reform. Entering IOs can help
leaders in transitional states to make a credible commitment to sustain
democratic reforms.
The credibility problems faced by democratizing states emanate from
the uncertainty that accompanies a democratic transition. In some cases,
leaders in nascent democracies limit reform, consolidate their personal
power, or attempt to distort and weaken emerging democratic institutions,
limiting the consolidation of democracy.13
In
others, leaders are sincere in their efforts to forge and consolidate
democracy. The problem is that neither domestic nor international
audiences can be certain about which type of leader they are facing. Such
uncertainty arises from multiple sources.
First, some transitional governments initiate reforms that they have
no intention of completing, especially if incomplete reforms yield rents
for key constituencies or create institutions that solidify the
leader's hold on power.14
Since transitional governments
are new, observers have difficulty determining whether such a government
will sincerely attempt to complete the reforms that it launches. More
generally, transitional regimes face reputational problems, including the
prospect that they lack restraint and cannot be trusted to honor
commitments.
15 In
contrast to established regimes, transitional governments do not have an
established track record of honoring or violating policy commitments,
prompting considerable uncertainty about their reliability. Compounding
this problem is that during many transitions, groups that rise to power
alter policies and institutions to suit their purposes.
16 This,
in turn, undermines the credibility of any commitment they make not to
further shift policies and institutions in the posttransition period.
Second, transitional regimes may have time-inconsistent preferences. A
regime's optimal policy ex ante may differ from its
evaluations of proper policy ex post.17
Governments
that come to power with the aim of promoting and sustaining political
liberalization may find that incentives exist to reverse reforms. Although
this problem is not limited to new democracies, the instability of such
countries and the high rate of turnover in their leadership exacerbate the
probability of
ex post policy reversals.
18 Leaders in nascent democracies would benefit from
the ability to tie their hands by creating a mechanism that increases the
cost of
ex post policy changes.
The inability of democratizing states to make credible commitments can
generate various problems. Elites often distrust one another in the
transitional period and fear that the new regime will not serve their
interests.19
More generally, if the
population does not believe that announced political reform efforts are
sincere, it is unlikely to support the new regime. This lack of support
from society can jeapordize democratic consolidation.
20 Some
groups may actively oppose the regime, even turning to violent measures or
allying with other disaffected groups in society. The lack of societal
support can also lead the government to take actions that undermine
democracy, for example, cracking down on opponents or dismissing the
legislature. More generally, as Whitehead points out, “if each
political sector concludes that the democratic commitment of the other is
lukewarm, this will reduce the motivation of all, and so perpetuate the
condition of fragility.”
21 Furthermore, the inability of transitional states to make credible
commitments can cause economic harm. Firms and financial institutions may
be reluctant to invest in or aid democratizing countries that are unable
to demonstrate that they will follow through on political liberalization
and enact sound economic policy.22
In fact, if economic actors believe that reform is
ephemeral, they are likely to take actions that distort markets and damage
a country's economic performance.
23 Leaders who wish to liberalize can benefit
from a mechanism that ties their hands, provides information about their
policy goals, and distinguishes them from leaders who want to use the
rhetoric of reform to accrue power without genuine liberalization.
Membership in IOs can help the leader of a democratizing country
credibly commit to reform efforts by establishing a mechanism that
increases the cost of deviating from these efforts and backsliding. This
mechanism stems from information provided by the organization about
members' actions, conditions imposed by the organization for new
members, and the reputational impact of violating an IO's rules.
Accepting conditionality heightens the credibility of a democratizing
regime's commitment to reform, because monitoring and enforcement are
handled by a third party with the ability to publicly sound an alarm in
the event that reform efforts falter.24
Conditions and conditionality refer to any terms of joining
an organization, including the terms of economic arrangements that a state
may join.
In addition, the costs associated with membership (fulfilling the
initial conditions as well as the traditional costs of membership) lend
credibility to the regime's commitment to the IO by conveying to
domestic and international audiences that its accession is not
“cheap talk.” The IO-imposed conditions also raise the costs
of limiting reform because any reversal can jeopardize the benefits
stemming from membership in the organization. Equally, membership can
create expectations about the behavior of a democratizing regime
(regardless of whether the institution sets conditions for membership),
generating audience costs for the government if these expectations are not
met.25
In this vein, Moravcsik argues that the origins
of the European Convention on Human Rights “lie in self-interested
efforts by newly established (or reestablished) democracies to employ
international commitments to consolidate democracy—‘locking
in’ the domestic political status quo against their nondemocratic
opponents.”
26 Similarly, observers of the Council of Europe have
suggested that Council membership for postcommunist countries can be
viewed “as a way of positively ‘locking’ a country into
an intergovernmental democratic network, with its binding international
conventions and treaties, so as to protect it more effectively from its
own antidemocratic enemies within.”
27 Indeed,
the then-Secretary General of the Council, Daniel Tarchys, suggested that
although Central and Eastern European states did not have spotless records
of democracy when admitted to the Council, the commitments accompanying
Council membership would expand and solidify democracy in those
states.
28 In addition, various IOs require participants to adhere to practices
that are likely to ease tensions among competing groups within a
democratizing country. For example, the European Union (EU) stipulates
that members must respect property rights. As Whitehead argues, this
requirement has “offered critical external guarantees to the
business and propertied classes of southern Europe.”29
This was especially true in Spain and Portugal, where
economic elites initially had been quite hostile to democracy.
30 For Spanish elites,
who were a potential obstacle to democracy, the stipulations of the
European Economic Community “provided guarantees and reassurances to
those who faced the postauthoritarian future with apprehension.”
31 In Central Europe, the commitments made under the auspices
of the EU (mostly through Association Agreements) have been equally
important. Pridham notes that even that in the absence of full EU
membership, “the general recognition given to Eastern and Central
European governments as European partners by the EU and its member states
… is both a means of confidence building for new democratic elites
but is also likely to promote the credibility and legitimacy of democratic
institutions in these countries, assuming that national publics are
broadly favorable to EU entry prospects.”
32 Pridham 2001, 76. Indeed, while
there is some variation in public support for EU membership across
countries and over time, such support tends to be quite strong. See Henderson 2000, 239; and Haerpfer
2002. In our empirical tests, we consider association status as
membership, but do not count observer status.
In some cases, joining an IO can also help a democratizing state send
a credible signal to international audiences about its commitment to
follow through on political reform. For example, Poland, Hungary, and the
Czech Republic formed the Central European Free Trade Area (CEFTA) in 1991
at least partly to demonstrate to the EU that they were committed to both
political and economic liberalization.33
The EU had demanded that these countries conduct
sweeping political and economic reforms before it would consider their
applications for membership. Promoting both kinds of liberalization became
a cornerstone of CEFTA.
Reneging on conditions set out by an IO can be costly for a
democratizing states, serving as a deterrent to influential groups within
government that have reason to derail liberal reforms and thus increasing
the credibility of commitments to sustain such programs. Violating the
terms of membership is likely to lead to a suspension of specific benefits
and even risks expulsion from the organization. In fact, IOs do punish
violators. For example, the European Community (EC) suspended
Greece's associate membership in 1967 after the military came to
power.34
The Organization of American States compelled
the leader of a coup in Guatemala to step down after an interruption of
democratic rule in 1993.
35 More recently, the EU “strictly enforced” the conditions
of its Association Agreement with Slovakia when President Vladimir
Meciar's behavior toward Slovakia's Hungarian minority and his
political opponents did not meet EU standards.36
Not only
did Brussels bombard Slovakia with
demarches, the EU also
formally removed Slovakia from its first list of applicant countries.
Since Meciar was replaced in October 1998, “EU and NATO entry is the
major unifying policy commitment” of the new Dzurinda
government.
37 More
generally, the EU and especially the Council of Europe has actively
enforced its terms of membership, with the Council now issuing public
reports on compliance with human rights and democracy commitments at
regular intervals.
38 Of course, there are likely to be some groups in a democratizing
society that have no interest in liberalization. Some of them may have the
ability to jeopardize the transition. The military, for example, can pose
such a threat. It is not unusual for the military's position in
society to weaken during a democratic transition.39
Fortunately, the costs imposed by IOs for reneging
on reform apply to these groups as well, discouraging them from attempting
to topple a nascent democratic regime. Such a situation materialized in
1996, when Mercosur members convinced would-be perpetrators of a military
coup in Paraguay that the costs the organization planned to impose if the
coup was undertaken would be too high for the junta to bear.
40 This dynamic was an explicit consideration as Greece
moved to join the EC. As Greek Foreign Minister Constantine Mitsotakis
mentioned in the months before accession:
Naturally, we do not expect our nine partners in the
Community to become the guardians of Greek democracy. By joining a broader
group of like-minded Western democracies, however, our own democratic
institutions will be reinforced, through constant contact and interchange,
but mainly because from now on Greece will share the destiny of its
Community partners…. They [prospective dictators] are
bound to know that the abolition of democracy entails immediate ostracism
from the Community. This could have grave internal and external
consequences. So, in this respect, the EC is a safe haven.41
Even if the conditionality policy of the IO is unclear or there is a
possibility of nonenforcement by the organization itself, reneging on
international agreements imposes reputational and domestic audience costs
on the regime. Concluding an international agreement places a transitional
state's reputation on the line. Any reversal, backsliding, or
abrogation of its treaty obligations can damage the state's
reputation, even if these actions do not elicit sanctions from the
organization. Accession itself can be a form of “international
recognition of a country's democratic credentials.”42
The domestic audience is likely to be attuned to these issues
because association with a highly democratic IO is an early chance to
break with the vestiges of an authoritarian past.
43
For Spain, both EC and NATO membership supplied such an opportunity to end
its isolation under the Franco regime.
44 As the
Financial Times reported, “The impact
of entry for Spain is mainly psychological, but is by no means a
negligible one. Achieving membership was the political equivalent of a
doctor's certificate—a sign of acceptance of recognition of
Spain as a ‘normal’ country.”
45 Financial Times, 1 December 1986, S1.
Similarly, in Central and Eastern Europe, joining IOs has been viewed
as an important signal to mass publics and a key means to lock in
democratic institutions. As one Romanian scholar opined, “I do not,
in all fairness, know whether Romania's joining Europe is the only
formula for a good future for the Romanians…. But it is my
strongest belief that Europe is the only strong incentive, for both the
political class and the people, to further the democratization of the
country.”46
Thus one strategy for leaders in nascent democracies who want to
consolidate democracy is to tie their own hands while sending a costly
signal to international and domestic observers that they are serious about
political reform. One way to accomplish this goal is by entering an IO. Of
course, membership in such organizations is far more likely to achieve
this end if the other participants are relatively democratic. A
transitional democracy joining the Warsaw Pact, for example, would
undoubtedly find that the associated credibility of its commitment to
reform is far weaker than a similar state that enters the EU. Nonetheless,
we begin with a more general analysis of whether democratization
influences IO membership. We then address whether the regime type of
existing IO members influences the propensity of democratizing states to
join these organizations.
In testing our argument, it will be important to ensure that any
observed effect of democratization on IO membership does not stem from a
more general tendency for democracies—whether stable or
transitional—to enter international institutions. There are various
reasons why democracies of all sorts might be drawn to such organizations.
Some observers have argued that established democracies tend to join
IOs—especially those populated by other democracies—because
doing so helps to reinforce and strengthen their democratic
institutions.47
Others have advanced a functionalist
explanation, emphasizing that governments join IOs to provide their
constituents with benefits that cannot be realized without international
collaboration. In order to retain office, leaders in democracies have to
be more attentive to constituent demands than their nondemocratic
counterparts, giving democratic leaders a particular impetus to enter
IOs.
48 Still another reason why democratic leaders tend to join IOs is that
voters have difficulty distinguishing between events that adversely affect
the country and that are beyond the leader's control and adverse
consequences arising from the leader's poor performance in office. As
a result, voters may remove a democratic head of state from office because
they believe the leader has done a bad job when in fact this is not the
case. Joining an IO can help chief executives to guard against this
possibility. Such institutions often are able to furnish reliable
information about the behavior of member-states. Countries that violate
their commitments to an IO will trip an alarm sounded by other members or
the organization itself. By publicizing the actions of democratic leaders,
IOs help them to avoid being turned out of office because voters
mistakenly believe the leaders have performed poorly. In nondemocracies,
by contrast, electoral dynamics are far less important, giving leaders
much less incentive to join IOs.49
In the following analysis, we will consider the effects of both
democracy and democratization on the rate at which states enter IOs in
order to ensure that democratization's impact on IO membership is not
just an outgrowth of democracy. Our argument is not that democratization
is a more important influence on such membership than democracy. Whether
that is the case is an empirical matter and we will provide some of the
first evidence bearing on it. Rather, our argument is that regardless of
whether democracies tend to join international institutions, democratic
transitions are an independent impetus to IO membership. We now turn to a
test of this argument.
A Statistical Model of IO Membership
To analyze the effects of transitions to democracy on the frequency
with which states enter IOs, we estimate the following model:
Our dependent variable, Δ#io, is the change in the
number of IOs to which each state, i, is a party from year
t to year t + 1. We code this variable using the new
Intergovernmental Organizations version 2.1 data from the Correlates of
War (COW) Project.50
This data set contains yearly information on IO membership
for all members of the interstate system from 1965 to 2000.
51 Data on IO membership exist before 1965; but
from 1815 to 1965, the data are coded in five-year increments, which is
not suitable for our analysis.
We create an annual count of all
IOs in which state
i is a member in year
t and again in
year
t + 1.
52 This includes IOs in
which state i is an associate member. We exclude IOs where the
state is an observer.
The dependent variable is the difference
between these yearly counts. There are various cases in the data set where
an organization disbands. Under these circumstances, members generally
have no choice about whether to exit the IO. Consequently, we do not code
states as leaving organizations that fold, although we revisit this issue
below.
To test our hypothesis, we include an independent variable indicating
whether state i experienced a democratic transition. To construct
this variable, we use a 21-point index of regime type developed by Gurr
and his colleagues that we refer to as regime type.53
This index—which
is derived using an updated version of the Polity IV data set—ranges
from 10 for the most democratic states to −10 for the most
autocratic ones.
54 We assess
the regime type of state
i in year
t − 5 and then
again in year
t. Following Jaggers and Gurr, we define countries
where
regime type > 6 as democracies, those where
regime
type < −6 as autocracies, and all remaining countries as
incoherent or “anocratic” regimes.
55
democratization equals 1 if state
i changes from a
nondemocratic polity (either autocracy or anocracy) to a democracy or from
an autocracy to an anocracy between
t − 5 and
t, 0
otherwise. Our expectation is that the coefficient of
democratization will be positive, because states undergoing a
democratic transition have particular reason to join IOs.
To ensure that the effect of democratization does not reflect a more
general tendency for regime change of any sort to influence IO membership,
we include autocratization. This variable equals 1 if state
i changes from either a democracy or an anocracy to an autocracy
or from a democracy to an anocracy between t − 5 and
t, 0 otherwise. Equally, to ensure that the influence of
democratization does not stem from a more general tendency for democracies
to enter IOs at a more rapid clip than other states, we include regime
type, which is Gurr's 21-point index of state i's
regime type in year t. If, in contrast to our argument, the
effect of democratization is merely an outgrowth of such a tendency, then
the coefficient estimate of regime type will be positive and
statistically significant and the coefficient estimate of
democratization will not be statistically significant. Our
expectation, however, is that the latter estimate will be positive and
significant even after accounting for the influence of regime
type.
In order to adequately analyze the effect of democratization on
changes in IO membership, it is crucial that we control for other factors
that also may prompt states to enter or exit international institutions.
Five of these factors are political. First, major power is a
dichotomous variable coded 1 if state i is considered a great
power in year t by the COW Project.56
Singer and Small 1994. During the
period we analyze, the major powers were China, France, Great Britain,
Russia/Soviet Union, and the United States.
Major powers are
likely to create and join IOs at a higher rate than weaker states, using
these institutions to consolidate their influence and maintain
international order.
57 Second,
independence is the number of years
that state
i has been an independent nation-state, based on the
dates of independence furnished by the COW Project.
58 At least one study has concluded that the
length of time since statehood shapes a country's propensity to enter
IOs.
59 There is also limited evidence that the
time since states gained independence is correlated with transitions to
democracy, suggesting that the variable should be included in our
model.
60 Third, former communist is a dummy variable that equals 1 if
state i was ruled by a communist government during some portion
of the period from 1965 to 2000 and it is no longer ruled by this type of
government in year t. This variable equals 0 otherwise.61
The communist countries in our sample are
Albania (until 1991), Bulgaria (until 1989), Cambodia (1976–90),
China, Cuba, Czechoslovakia (until 1989), East Germany (until 1989),
Hungary (until 1989), Laos (1975–91), Mongolia (until 1990), North
Korea, Poland (until 1989), Romania (until 1988), the Soviet Union (until
1991), Yugoslavia (until 1991), and Vietnam.
We analyze
former communist because a large number of Eastern European
countries and former Soviet republics rushed to gain access to Western IOs
after the collapse of the Berlin Wall. Many of them have also undergone
democratic transitions, and it is important to ensure that any observed
effect of democratization on IO membership is not being driven by this set
of countries alone. Fourth,
dispute is the number of Militarized
Interstate Disputes (MIDs) involving state
i that are ongoing in
year
t. We rely on the MID 3.0 data to code this variable.
62 Functionalist and neoliberal arguments suggest that
past conflict is likely to stimulate the formation of supranational
institutions, as states and societal groups attempt to avert future
conflict.
63 On the
other hand, participants in an IO may be reluctant to grant membership to
a state involved in various interstate disputes, because its belligerence
could adversely affect the organization. Further, states embroiled in
militarized disputes may focus their foreign policy on resolving these
conflicts rather than gaining entry into IOs. Fifth,
hegemony
measures the relative size of the largest state in the international
system. It is calculated by dividing the gross domestic product (GDP) of
the largest state in the system (for the entirety of this analysis, the
United States) by global GDP in year
t, using data compiled in
the Penn World Table.
64 This measure
of hegemony, although somewhat crude, is closely related to that used in
many previous studies.
65 Earlier research
has found that hegemony influences the formation of IOs as well as
patterns of international conflict and cooperation; however, the strength
of these relationships remains controversial.
66 In addition to political factors, we analyze three key aspects of each
state's economy. Past research has linked economic wealth and size to
both IO membership rates and democratic transitions, so it is important to
control for these factors.67
Consequently, we include
development, which is country
i's per capita GDP in year
t, and its
gdp
in
t.
openness is country
i's total
foreign trade (imports plus exports) divided by its GDP in year
t. There is some evidence that increasing openness increases the
propensity for states to join IOs.
68 Furthermore, because the flow of overseas commerce
is affected by the regime types of the trading partners, it is important
to ensure that any relationship between democratization and IOs is not an
outgrowth of international trade.
69 We also include a trend (year) in the model to ensure that
any observed relationship between regime type and IO membership does not
stem from the spread of both democracy and international institutions over
time.70
To
control for possible “ceiling” or “floor” effects
in IO membership, we introduce #
io, which is the total number
of IOs that state
i belongs to in year
t. As a state
participates in a growing number of IOs, the marginal benefit of joining
an additional one may decline. Further, the number of existing IOs that a
state does not belong to shrinks as it joins more organizations, reducing
the number it could join in the future. As such, #
io may be
inversely related to the change in IO membership. Alternatively, this
relationship may be direct. States that participate in a large number IOs
may be “joiners,” predisposed to enter many international
institutions, whereas those that belong to few IOs may have a general
aversion to joining such organizations. Regardless, we need to account for
this factor because it is likely to be related to regime change.
71 Equally, because many IOs tend to be regional
in nature, patterns of IO membership may be similar within geographic
regions.
72 We therefore
include regional indicator variables in model (1).
73
Finally, ε is a stochastic error term.
Table 1 presents descriptive statistics for
each of the variables used in this model (as well as some variables
analyzed later). To estimate the model, we use ordinary least squares.
Tests of statistical significance are based on panel-corrected standard
errors, which account for heteroskedastic disturbances and contemporaneous
correlation across each panel.74
We do not adjust for
autocorrelation in the data, because we found little evidence of an
autoregressive data generating process.
The Results
Table 2 presents the estimates of model (1).
As shown in the first column, democratization promotes IO membership.75
It is important to recognize that our findings
are not at odds with recent research indicating that democratization
promotes war and that participation in IOs reduces the prospect of
conflict. See Russett and Oneal 2001; Mansfield and Snyder 2005. First, that research
emphasizes the conflict-promoting effects of a particular aspect of
democratization, namely, transitions that stall before the establishment
of a coherent democracy in countries where political institutions are
weak. Second, the existing literature focuses on whether the likelihood of
conflict between states depends on their participation in IOs, rather than
on their initial entry into IOs. Furthermore, there is considerable
disagreement about whether IO membership actually inhibits conflict. Some
studies have found that IOs increase conflict under certain circumstances;
others have found that whether IOs dampen hostilities depends on the type
of IO being analyzed. See Kinsella and Russett
2002; Boehmer, Gartzke, and Nordstrom
2004; Pevehouse and Russett
forthcoming.
The estimate of
democratization is
positive, statistically significant, and large. States that experienced a
democratic transition over the past five years increase their number of IO
memberships by 20 percent relative to states that did not undergo a regime
change.
76 We express all marginal effects
calculations as the predicted change in IO memberships. To make these
calculations, we set all continuous variables to their mean value and all
discrete variables to their modal value. All of the regional indicator
variables are set to zero.
Moreover, the effects of
democratization do not reflect a more general tendency for regime change
to promote IO membership. Democratizing countries join nearly 40 percent
more IOs per year than autocratizing states, a difference that is highly
significant (χ
2 = 8.81,
p < .01). In fact,
autocratizing states are less likely to enter IOs than stable regimes. The
coefficient estimate of
autocratization is negative and
statistically significant. These findings strongly support our argument.
They also cast doubt on the results of Jacobson and his colleagues, who
found that democratization actually led states to decrease their
involvement in IOs.
77
The effects of democratization on changes in IO membership,
1965–2000
In addition to democratizing countries, democracies have a marked
tendency to join IOs. The estimate of regime type is positive and
statistically significant, indicating that more democratic states join IOs
at a more rapid rate than less democratic countries. In fact, increasing
regime type from what is often considered to be the threshold for
an autocracy (−7) to the threshold for a democracy (+7) yields a
nearly 20 percent rise in the predicted number of IOs that a given state
enters in a given year.78
Turning to the remaining variables, the coefficient estimates of
major power and former communist are positive and
statistically significant. Whether IOs are used by great powers to promote
their own agendas, preserve their favored rules, increase burden sharing,
or foster international cooperation, these states accede to international
institutions at a rate more than 40 percent higher in any given year than
their weaker counterparts. States that had been ruled by communist
governments enter IOs at a rate about 80 percent higher than other states,
once these governments fall from power.79
We
also included a dummy variable indicating whether the Cold War had ended.
The estimate of this variable was not statistically significant and
including it did not appreciably change the remaining estimates.
Further, the coefficient estimates of
dispute,
hegemony,
and
year are negative and statistically significant. States
involved in more MIDs display a lower propensity to join international
institutions, possibly because their diplomatic efforts are focused on
resolving these conflicts or because existing IO members are reluctant to
add belligerent states to their rolls. In addition, eroding hegemony
stimulates entry into IOs, a finding that accords with various critiques
of hegemonic stability theory.
80 There is also a tendency for states that participate in more IOs to
experience an increase in IO membership. The estimate of #io
is positive and statistically significant, indicating that some states
have a deep and ongoing interest in participating in IOs whereas other
states do not. Equally, the region-specific variables are jointly
significant (χ2 = 8.67, p < .01), reflecting
the tendency for many IOs to be regional in nature and for patterns of IO
membership to be similar within geographic regions. However, there is no
evidence that the number of years a state has been sovereign, the extent
of its economic development, its GDP, or its degree of commercial openness
influence changes in IO membership.
Assessing the Stability and Validity of the
Results
Having generated a set of initial results, we now turn to some
supplementary tests of our argument. First, we have implicitly assumed
that the supply of IOs does not vary across countries; that is, that all
countries have equal access to membership in such organizations. But many
IOs are not universal, most obviously those that are limited to members in
the same geographic region.81
A number of
IOs seem to emphasize democracy as a condition of membership. As such, it
could be that democratic and democratizing countries enter more IOs
because they have access to more organizations than other states. However,
this is unlikely to be much of a problem for our analysis. First, few IOs
make any explicit reference to the regime type of members. Second, even
those that do make such reference sometimes sidestep the issue in
practice. For example, although the NATO preamble contains references to
democracy as a underlying principle, one of its founding members was a
European dictatorship (Portugal) and military coups in member states
resulted in neither major changes within NATO nor pressure to end
authoritarian rule (Greece and Turkey). The North American Free Trade
Agreement (NAFTA) is another example of imprecise conditionality—the
implicit U.S. criterion for NAFTA expansion seems to include democracy,
although there are no formal written conditions. Whitehead 1993. Third, while some IOs may have
stressed democracy as a condition for membership, others (most notably
those that were established by the Soviet Union to manage the Communist
bloc) were restricted to autocracies. Hence, there is little reason to
worry that our results are driven by a tendency for democracies to have
greater access to IOs than other countries.
To address this
issue, we code the number of organizations to which a state has access by
assuming that each state has an option to join each IO that operates in
its region of the world as well as every universal organization (for
example, the United Nations and the International Labor Organization). We
then replace #
io with
saturation, the ratio of the
number of IOs in which state
i does participate in year
t to the number of IOs in which this state could participate.
As shown in the second column of Table 2,
the estimate of this variable is positive and statistically significant,
indicating that states that already belong to a large number of the IOs
they have access to are inherently “joiners” and tend to enter
even more organizations. States that participate in few IOs have an
aversion to these organizations and continue to eschew membership. Most
important for our purposes, however, is that regardless of whether we
replace #io with saturation, the estimated effect of
democratization (as well as the other independent variables in the model)
is virtually identical. In the remainder of this article, we include
saturation rather than #io in our models of IO
membership, although this decision has no bearing on any of the following
results.
Second, we estimate the model after eliminating development,
gdp, and openness. As shown in Table
1, there are roughly 25 percent more data for the remaining
variables in our analysis than for these three variables. Furthermore,
there is an overwhelming tendency for the countries that are missing data
for these three variables, but that have complete information for the
remaining variables in our model, to be autocracies. This tendency could
introduce a source of bias in our earlier results. Equally, none of these
factors has a statistically significant influence on IO membership. The
results reported in the third column of Table 2
clearly indicate that removing these variables from the model has
relatively little effect on the estimated coefficient of
democratization, although the estimate of
autocratization is no longer statistically significant. In the
interests of analyzing the broadest and most representative sample
possible and because these economic variables have little influence on
changes in IO membership, we exclude them from the following analysis.
However, it is important to recognize that this change in the model's
specification has little bearing on the results presented below.
Third, we mentioned earlier that in coding the dependent variable, we
do not consider members of an IO that disbands to be exiting the
organization because its participants usually have no choice but to leave.
Nonetheless, to ensure that this coding decision does not unduly influence
our results, we include in Δ#io cases where an
organization ceases to exist and then reestimate the model. As shown in
the fourth column of Table 2, there is no
evidence that the influence of democratization or any of the control
variables except saturation and a few of the regional indicators
depend on how we code members of disbanded IOs.
Fourth, it is useful to more fully evaluate the effects of stable and
transitional regimes on IO membership. In our analysis of regime change,
the reference category is a stable regime. Hence the positive coefficients
of democratization in Table 2 indicate
that democratizing countries are more likely to join IOs than countries
that do not undergo any sort of regime change. It is possible, however,
that different types of stable regimes vary in their propensity to enter
such organizations. We therefore replace regime type with one
indicator variable (stable democracy) that is coded 1 if a state
remains a democracy from year t − 5 to year t, and
another variable (stable autocracy) that is coded 1 if a state
remains autocratic during this interval. The reference category in this
analysis is a stable anocracy.
As reported in the last column of Table 2,
the coefficient estimate of democratization continues to be
positive and statistically significant. A state making a transition to
democracy experiences a higher rate of entry into IOs than any other
regime type considered here: such states experience more than a 50 percent
rise in IO membership relative to stable autocracies and autocratizing
countries, a greater than 30 percent increase in IO membership relative to
stable anocracies, and more than a 10 percent rise in IO membership
compared to stable democracies.
These results and our earlier findings provide strong evidence that
democratization promotes IO membership, even after accounting for the
effects of democracy. As another way to address this issue, we include in
our model a variable indicating the number of consecutive years that a
state has been a democracy as of year t. Regardless of whether we
use regime type or stable democracy and stable
autocracy to measure a state's regime type and regardless of
whether we include democratization and autocratization
in the model, the estimate of this variable is negative and statistically
significant. Consequently, democratic states are most likely to enter IOs
immediately after they become democratic. They become less likely to enter
IOs as democracy matures.
Fifth, it is possible that our results stem from the fact that both
stable and transitional democracies have common interests, which in turn
predisposes them to join many of the same IOs.82
Because most of the IOs in our
sample are regional rather than universal, we address this possibility in
a preliminary way by introducing a measure (tau-beta) of the similarity of
alliance portfolios between each state,
i, and the most powerful
state in
i's geographical region, based on procedures
developed by Bueno de Mesquita and by Bennett and Stam.
83 The coefficient
estimate of this measure, however, is nowhere close to being statistically
significant and its inclusion has no bearing on the sign, size, or
statistical significance of any remaining variable in our model.
Sixth, it is important to ensure that our results do not reflect any
reverse causality that could arise if changes in IO membership influence
democratization. Recent research, for example, indicates that regional
organizations composed of democratic countries have stimulated both
democratization and the consolidation of democracy.84
To address
this issue, we measure Δ#
io from year
t −
1 to year
t and
democratization from year
t to
year
t + 5. Then we estimate a logistic regression of
democratization on Δ#
io, a set of
region-specific variables, and a spline function of the number of years
since state
i last experienced a democratic transition to account
for any temporal dependence in the data.
85 The results provide no evidence that the change in
the number of IO memberships has a statistically significant effect on
democratization.
86 These results
are not inconsistent with Pevehouse's findings, because there is a
sizeable difference in the sample of IOs used in his study and in ours.
Pevehouse 2005. While we examine all IOs,
Pevehouse examines only regional organizations. Moreover, his statistical
association was between the level of democracy within an IO and
democratization, whereas we examine the change in the number of
IOs.
As such, our results do not seem to be threatened by a
simultaneity bias.
87 We also analyze whether
IO membership influences regime type by measuring Δ#io
from year t−1 to year t and regime type
in year t. Then we regress regime type on
Δ#io and a set of region-specific variables using
feasible generalized least squares, a technique that involves using
ordinary least squares to estimate the model and then purging the errors
of serial correlation. Here we assume that the serial correlation is first
order (that is, AR[1]) and base tests of statistical
significance on panel-corrected standard errors that account for any
heteroskedasticity and contemporaneous correlation of the errors across
countries. On this technique, see Beck and Katz
1995. The results furnish no evidence that changes in IO membership
influence a state's regime type.
Finally, although our argument focuses primarily on the domestic
political conditions that prompt states to join IOs, the value of
Δ#io is determined by the frequency with which states
enter and exit these organizations. It is useful to analyze separately the
influences on joining and leaving IOs. To this end, we estimate two
models. The dependent variable in the first model is the number of IOs
that state i joined from year t to year t + 1;
the dependent variable in the second is the number of IOs that state
i left during this interval. The independent variables in each
model are those included in our earlier analysis, specifically model (1.3)
in Table 2. Because the error terms in these two
models are likely to be related, it is more efficient to estimate the
models jointly rather than separately.88
Because
each of the dependent variables is a count, we rely on seemingly unrelated
binomial regressions to conduct this analysis.
89 Various studies employ count models in which the underlying
process generating the dependent variable is assumed to have a Poisson
distribution. A key feature of this distribution is the assumption that
the probability of an event occurring in a given interval of time is
independent of previous events in the interval and that the rate at which
events take place in this interval is constant and does not depend on
previous events. We find statistically significant evidence that this
assumption is violated for both the joining and the leaving model,
indicating that a negative binomial specification should be used. On this
issue, see King 1989; Greene
1993.
As reported in Table 3, the sign of each
coefficient in the IO “joining model” is the same as the
corresponding coefficient in model (1.3); and the estimates that are
statistically significant when we analyze Δ#io remain
significant when we focus on the frequency with which states join IOs. It
is particularly noteworthy that countries undergoing a democratic
transition join IOs significantly more rapidly than either stable regimes
or autocratizing states, and that there is no marked difference between
stable and autocratizing countries in this regard. Moreover, while the
size of the coefficients in Tables 2 and 3 cannot be directed compared since the techniques
used to derive these results are quite different, the effects of
democratization on IO accession are quite sizeable. Democratizing states
enter over 20 percent more IOs than stable regimes and over 25 percent
more than autocratizating states.
Seemingly unrelated count models of joining and leaving IOs,
1965–2000
In contrast, democratization usually does not prompt states to leave
IOs. In the “leaving model,” the coefficient estimate of
democratization is negative, but it is not statistically
significant. Thus democratizing states generally do not renounce
membership in the organizations that its autocratic predecessor
participated in. They do, however, display a pronounced tendency to join
the rolls of additional organizations. Interestingly, the only factor
included in our model that has a strong influence the rate at which states
exit IOs is regime type. The estimate of this variable is
negative and statistically significant, indicating that autocratic states
are more likely to leave IOs than their democratic counterparts. Indeed,
this tendency is quite pronounced: increasing the mean value of regime
type by one standard deviation generates about a 25 percent jump in
the predicted number of IOs a given state exits.
The Effect of IO Members' Regime
Type
The preceding results strongly support our argument. As we noted
earlier, however, democratizing states are likely to be drawn with the
greatest force to IOs in which the bulk of the members are democratic. The
commitments made by a democratizing state to sustain and deepen political
reforms are likely to gain more credibility if it joins an IO composed of
democracies—countries with a far greater interest in political
reform than other states—than if it enters an IO with a different
make-up. Equally, from a signaling perspective, it does little good for a
new democracy to join an IO with a sizeable number of autocratic members.
Finally, more uniformly democratic IOs are more likely to enforce the
conditions of membership. The transparency of democracies lessens the
likelihood that any one state in the organization will shirk its
enforcement of the IO's rules.90
On the
greater transparency of democratic systems than others, see Fearon 1994.
More generally, democracies are
less likely to renege on commitments than other countries.
91
Knowing this, leaders attempting to tie their hands will tap IOs made up
primarily of democracies.
Indeed, past work has shown that only regional organizations composed
of highly democratic countries are especially effective at promoting
democracy and assisting in its consolidation.92
One
observable implication of our theory is that democratizing countries
should attempt to join those IOs whose membership is more democratic,
because these organizations will be more likely to undertake the functions
that we have emphasized.
To test this proposition, we make three changes to model (1.3) in
Table 2. First, we analyze a different dependent
variable (dem level io), the average regime type within the IOs
joined by state i from year t to year t +
1.93
The average level of democracy in each
IO is computed exclusive of state i.
To compute this
variable, we derive the average regime type score (which, recall, ranges
from −10 to 10) of the members of each IO that state
i
joins and then take the mean of these averages. Second, we restrict the
sample to countries that join at least one IO between
t and
t + 1. Consequently, we add a new variable,
instrument[join], which is the predicted probability of
state
i entering an IO in year
t + 1 based on model
(1.3). We include this variable to account for any selection bias caused
by the fact that many states do not join any IOs between years
t
and
t + 1 and the possibility that the very factors prompting
states to enter democratic IOs could be the same as those influencing
whether they enter any IOs.
94 This is
therefore a Heckman-style selection model. See Heckman
1979.
Third, we omit
year because there is little
evidence of a secular trend in the dependent variable. We also omit
saturation since there is no reason to control for preexisting IO
membership in the current analysis.
95 Whether or not we include development, gdp,
and openness makes no difference in the following
results.
Excluding these variables is also useful because in
order to identify the instrumental variable
(
instrument[join]), it is necessary to exclude some of
the variables that determine its value from the model being estimated. We
include the remaining control variables in our earlier analysis in an
effort to proceed cautiously and avoid misspecifying the model. However,
it is not clear what effect most of these variables are likely to have on
the average level of democracy among members of a IO and we do not
emphasize those estimates in the following discussion.
The first column of Table 4 presents the
results of this analysis. As expected, the coefficient estimate of
democratization is positive, statistically significant, and
large. Democratizing states therefore seem to pay close attention to the
regime characteristics of IO members, acceding to organizations with
democracy scores that are, on average, about 60 percent higher (on the
21-point scale that ranges from −10 to 10) than those joined by
stable regimes and nearly 30 percent higher than those joined by
autocratizing countries. New democracies tend to seek membership in highly
democratic IOs, supporting our argument that such organizations are best
positioned to help states cement transitions to democracy.
The effects of democratization on the average regime type in the IOs a
state joins, 1965–2000
In the second column of Table 4, we replace
regime type with two indicator variables, each of which is
measured from year t − 5 to year t, to distinguish
the influence of stable democracies (stable democracy), stable
autocracies (stable autocracy), and stable anocracies. Doing so
yields results that are much the same as before, although there are some
noticeable differences. The estimate of democratization continues
to be positive and statistically significant. States undergoing a
democratic transition become involved in IOs where the average democracy
score is about half a point higher than the IOs joined by stable
anocracies, two points higher than the IOs joined by stable autocracies,
and more than one point higher than the IOs joined by autocratizing
countries. Only stable democracies enter international institutions that
are, on average, more democratic than the ones that democratizing states
join. This difference, however, is quantitatively small and statistically
insignificant (χ2 = 1.81, p = .18).
IO Membership and Backsliding by
Democratizing Countries
Central to our argument is that leaders in democratizing countries
have difficulty making credible commitments to sustain political reforms
and the consolidation of democracy. Entering an IO composed largely of
democratic states can help leaders in such countries to address this
problem by creating a mechanism that increases the cost of deviating from
reforms and backsliding in an autocratic direction. Thus another
implication of our argument is that democratizing countries participating
in IOs made up of democracies should be less likely to subsequently
experience a reversion to autocracy than democratizing countries that
participate in less democratic IOs.
Although a comprehensive test of this hypothesis is beyond the scope
of our analysis, we can address it in a preliminary way. In our sample,
there are 380 episodes where a country experienced a democratic transition
over some five-year interval from year t − 5 to year
t; forty-seven of these episodes were then followed by an
autocratic transition from year t to year t + 5, whereas
333 were not.96
Note that all of the
following results are essentially unchanged if autocratization is measured
from year t + 1 to year t + 6 rather than from year
t to year t + 5.
Focusing on all IOs that these
states participated in between
t − 5 and
t, the
average regime score of the members of these organizations (
dem level
io) is −0.91 for the democratizing states that experienced an
autocratic transition from year
t to year
t + 5,
compared to .76 for the remaining states, a difference that is highly
significant (
t = 6.58,
p < .001). An analysis of the
average regime score for the IOs that these states joined between
t − 5 and
t yields almost identical results: an
average of −0.89 for countries that made an autocratic transition
from
t to
t + 5 and an average of 0.90 for states that
did not. Again, this difference is highly statistically significant.
Shifting our focus from the average regime score of all IOs a
democratizing state participates in and joins to the average score of the
most democratic IO the state belongs to or enters has no bearing on our
findings. The average regime score of the most democratic IO that a
democratizing state participates in is about 1.3 points higher if it did
not subsequently undergo an autocratic transition than if it did
experience such a transition. This score is roughly 2 points higher if we
restrict attention to the IOs that democratizing states joined between
years t − 5 and t. Both differences are
statistically significant at the 0.01 level.
To further address this issue, we code each IO as democratic if the
average regime score of its members is 7 or greater on Gurr's
21-point index (from −10 to 10). On average, democratizing states
that do not subsequently experience an autocratic transition are members
of 1.65 democratic IOs. The corresponding figure for democratizing states
that do autocratize is 0.60. Equally, democratizing states that do not
autocratize in the following five-year period join an average of 0.59
democratic IOs, compared to an average of 0.09 for states that do
autocratize. To assess the robustness of these results and because
defining a democratic IO as one where the average regime score is 7 or
greater is somewhat arbitrary, we reset this threshold to 6 and greater
and then again to 5 and greater. As this threshold value declines, there
is even a more pronounced tendency for democratizing states that do not
backslide in an autocratic direction to participate in and join a larger
number of democratic IOs than states that do undergo an autocratic
transition. Furthermore, all of these differences continue to be
statistically significant.
Clearly, these tests are preliminary and do not control for other
factors that may influence the sustainability of democratization and IO
membership. Nonetheless, the findings are consistent with a key
implication of our argument: participating in and joining democratic IOs
reduces the prospect that democratizing states will backslide in an
autocratic direction.
Conclusions
We have argued that democratization is an important impetus to IO
membership. States undergoing democratic transitions have a strong
incentive to join IOs, because doing so sends a credible signal to
domestic and international audiences that political reform efforts are
sincere. Entering an IO can help leaders in transitional states credibly
commit to carry out reforms since these institutions convey information,
help ameliorate time-inconsistency problems, and improve the reputation of
new member states. Membership can also discourage regime opponents from
threatening emerging regimes by imposing potentially high costs on
countries that renege on IO commitments. Each of these mechanisms can
assist in the process of deepening democracy, giving leaders in nascent
democracies strong incentives to join IOs.
Our statistical findings confirm these hypotheses. Using a new data
set on IO membership covering the period from 1965 to 2000, we find strong
evidence that democratizing states join IOs more frequently than other
countries. Moreover, democratizing states do not attempt to join just any
organizations. Instead, they tend to enter ones composed of relatively
democratic members—forming clubs of democracies.
In addition to democratizing states, democracies display a tendency to
join IOs. Democracies are somewhat less likely to join all IOs and
marginally more likely to join IOs with a highly democratic membership
than democratizing countries. However, democracy and democratization have
independent influences on the propensity of states to join IOs. Even after
accounting for the effects of democracy, democratization has a
statistically and substantively significant influence on the decision to
enter international institutions. Finally, there is evidence that joining
IOs composed of relatively democratic members can help democratizers
reduce the prospect of a reversion to authoritarianism, a finding
consistent with past research.97
Most of the existing literature on IO formation focuses on the impact
of international political forces. Our findings confirm the importance of
some of these forces. Eroding hegemony and the absence of
political-military conflict, for example, prompt to states to enter
international institutions. Equally, major powers join more IOs than
weaker states. In addition to these international variables, however, our
findings underscore the importance of domestic politics in shaping the
decision to enter an IO, particularly the role of domestic institutions.
Although studies of international institutions have placed relatively
little emphasis on the effects of domestic politics, the results of our
analysis strongly suggest that the recent wave of democratization is
likely to expand the size and number of IOs. Moreover, this expansion is
likely to bode well for the long-term survival of democracy.
