globalization and civil war: an empirical analysis · globalization and civil war: an empirical...
TRANSCRIPT
Globalization and civil war:
An empirical analysis
Roberto Ezcurra and Beatriz Manotas
Department of Economics
Universidad Publica de Navarra
January 7, 2013
Abstract
This paper investigates the empirical relationship between globalization and within-country conflict in a sample of 138 countries over the period 1970-2009. To thatend we use a measure of globalization that distinguishes the social and politicaldimensions of integration from the economic dimension, which allows us to adopta broader perspective than in most of existing studies and to examine the effectof these three distinct aspects of globalization on civil violence. The results of thepaper show that the degree of integration with the rest of the world contributessignificantly to increasing the incidence of civil wars. In particular, the dimensionsof globalization most robustly related with internal conflict are social and politicalintegration. These findings are not affected by the inclusion in the analysis of addi-tional explanatory variables, changes in the definition and sources of data on civilwars, and the employment of alternative estimation methods. Our results contrastdirectly with those arguments that defend that globalization has the beneficial effectof deterring within-country armed conflicts.
Keywords: Globalization, civil war.
JEL classification: F02, F60.
1 Introduction
The consequences of globalization are nowadays the subject of an active public de-
bate in different forums (Stiglitz, 2002; Bhagwati, 2004). The interest surrounding
this issue is clearly related to the increasing relevance of the process of globalization
currently underway. This does not imply that globalization is a new phenomenon, as
its origins go back, at least, to the 19th century (Findlay and O’Rourke, 2007). Nev-
ertheless, during the last few decades the world has experienced unprecedented levels
of integration, surpassing the peak reached before the First World War. This process
is characterized by the opening of national borders to a variety of flows including
people, goods and services, capital, information and ideas (Clark, 2000). Although it
is difficult to agree on a precise definition, there is a wide consensus that globalization
tends to erode the relevance of national borders, generating complex relations among
different actors at multi-continental distance (Norris, 2000). These increasing mutual
interactions have important consequences on many relevant facets of contemporary
societies, including economic, social, cultural and political aspects. Accordingly, un-
derstanding the effects of globalization is essential to address the numerous challenges
posed by this process, and be able to identify who wins and who loses, not only within
each country but also across countries.
Against this background, the last years have seen the publication of numerous
studies on the impact of globalization on economic growth (Frankel and Romer, 1999;
Dreher, 2006), income inequality and poverty (Dollar and Kraay, 2004; Milanovic,
2005), labour market (Dreher and Gaston, 2007; Tomohara and Takii, 2011), envi-
ronmental quality (Antweiler et al., 2001; Frankel and Rose, 2005), or democracy and
human rights (Rudra, 2005; Dreher et al., 2012). Likewise, there are various con-
1
tributions that examine the potential link between globalization and within-country
conflict, using different indicators of trade openness and foreign direct investment to
measure the relevance of globalization (Hegre et al., 2001; Fearon and Laitin, 2003
; Bussmann and Schneider, 2007; Martin et al., 2008). From a policy perspective,
the relationship between these variables and civil war is definitively important, as it
provides information on the role played in this context by economic integration. Nev-
ertheless, the degree of trade openness and foreign direct investment are not useful
to capture the incidence of other dimensions of globalization identified in the polit-
ical economy literature, such as social integration and political integration (Prakash
and Hart, 1999; Keohane and Nye, 2000). This is potentially important, given that
it is not evident that the various dimensions of globalization affect internal conflict
in the same way. Bearing this in mind, and in a quest for empirically well-founded
stylized facts, this paper aims to provide a comprehensive analysis of the relationship
between globalization and the incidence of civil armed conflict. To that end, we adopt
a broader perspective than in most of existing studies on this topic and investigate in
a systematic way the consequences on civil war of the economic, social and political
dimensions of globalization.
To the best of our knowledge, only Olzak (2011) has considered so far in this
context the multidimensional nature of the process of globalization. Nevertheless, our
study is different from this prior work in three major aspects. First, the measure used
to quantify the relevance of globalization within the various countries differs in the
two papers. Olzak (2011) employs in her main analyses a composite indicator known
as “Globalindex” (Raab et al., 2008) and we use the KOF index of globalization
constructed by Dreher (2006) and updated by Dreher et al. (2008). Both measures
are based on principal component analysis and are highly correlated. However, there
2
are some differences between them. Thus, the Globalindex distinguishes between
economic, sociotechnical, cultural and political integration, whereas the KOF index
combines the sociotechnical and cultural indicators into a single component of social
globalization. This conceptual distinction may be relevant if, as in the case of Olzak
(2011), one is interested in testing the effect on internal conflict of these two particular
aspects of globalization. Nevertheless, we have decided to use in our study the KOF
index because it is available for more countries than the Globalindex. Accordingly,
the employment of the KOF index allows us to use a greater sample size than in
Olzak (2011) and to include in the analysis an important number of countries located
in some of the more conflictive regions of the world during the last decades, as is
the case for example of Sub-Saharan Africa (Michalopoulos and Papaioannou, 2011).
Second, there are relevant methodological differences between our study and Olzak
(2011). In particular, the empirical analysis carried out by Olzak (2011) does not
take into account that there are important reasons that suggest that endogeneity
may influence the analysis of the link between integration and within-country conflict
(Martin et al., 2008). As outlined below, the presence of endogeneity is particularly
relevant from an econometric perspective and may affect seriously the results of the
analysis. In view of this potential problem, we adopt in our study an instrumental
variables approach to investigate the causal link between globalization and civil war.
Third, the two papers focus their attention on different aspects of internal violence.
In particular, the dependent variable used by Olzak (2011) is the number of fatalities
from conflict, which is a measure of the intensity of civil war. By contrast, we are
interested in the effect of globalization on the incidence of civil war. This allows us
to relate our paper to the abundant literature on the determinants of the incidence
of within-country conflicts (Miguel et al. 2004; Montalvo and Reynal-Querol, 2005;
3
Esteban et al., 2012a,b).
The remainder of the paper is organized as follows. After this introduction, section
2 reviews several of the theoretical arguments proposed in the literature to justify the
possible connection between globalization and internal conflict. Section 3 describes
the measure used in our study to quantify the incidence of globalization in the various
countries and also provides evidence on the evolution and the process of integration
during the last decades. Section 4 presents the empirical undertaken in the paper to
investigate the link between globalization and civil war. The robustness of our findings
is examined in section 5. The final section offers the main conclusions from our work.
2 The relationship between globalization and internal
conflict
From a theoretical perspective there are several arguments to believe that globalization
and within-country conflict may be related. Nevertheless, this is a complex relation-
ship, as attempting to explain how globalization affects conflict implies to take into
consideration multiple factors and mechanisms. Specifically, it is important to note
that economic, social and political integration can have different effects on conflict
(Keohane and Nye, 2000).
Most of the existing literature has focussed exclusively on the link between inter-
national trade and civil war. Thus, according to Martin et al. (2008) there are two
mechanisms relating trade and the opportunity cost of internal conflict, which work
in opposite directions. The first of these mechanisms is the deterrence effect. This
effect is based on the idea that the opportunity cost of conflict is positively associated
with the degree of trade openness of the country in question, as the economic benefits
4
generated by international trade can be threatened by the existence of within-country
armed violence. According to this effect, trade openness reduces the potential risks of
civil war. Nevertheless, Martin et al. (2008) also recall that international trade can
be a substitute to internal trade during civil war episodes, thus acting as an insurance
and reducing the opportunity cost of conflict. This insurance mechanism also implies
the weaken of the degree of economic interdependence of the various regions and eth-
nic groups within a country, which increases the feasibility of internal conflict (Martin
et al., 2004). Therefore, the final impact of international trade on the incidence of
civil war depends in the last instance on the magnitude of both effects, which may be
related to the degree of intensity of conflict. In particular, the deterrence effect should
be more relevant in high intensity conflicts, whereas the insurance effect should be
less important in these type of conflicts (Martin et al., 2008).
Moreover, various authors have emphasized the relevance of economic integration
in promoting growth and economic development (Frankel and Romer, 1999; Alcala and
Ciccone, 2004; Dreher, 2006), which is particularly relevant for the establishment of the
rule of law and political stability (La Porta et al., 1999). Given that the opportunity
cost of conflict increases with the advances in the process of economic development
(Collier and Hoefler, 2002; Fearon and Laitin, 2003), this argument suggests that
economic globalization may contribute indirectly to reducing the risk of civil war.
In any case, although economic globalization is possibly benefiting the countries
involved in the process in aggregate terms, is also generating winners and losers within
these countries. In fact, the opening of national economies to world markets has led
to greater inequality in numerous countries (Stiglitz, 2012). According to the tra-
ditional view, economic inequality is perceived as a major driver of social conflict.
Thus, for example, Sen (1973) points out that “the relationship between inequality
5
and rebellion is indeed a close one”. Yet, intuitive and natural as it might seem, the
link between income inequality and conflict has not received so far a conclusive and
definitive empirical support (Esteban et al., 2012 a,b). Anyway, other dimensions of
inequality are potentially important in this context. For example, economic globaliza-
tion also contributes to increase spatial inequality (i.e. inequality across the various
regions of a country) (Ezcurra and RodrA-guez-Pose, 2012). This is particularly rel-
evant, since a high level of spatial inequality may lead to internal conflicts about the
territorial distribution of resources and to rise the risks of secession, undermining so-
cial and political stability. Furthermore, the positive and negative effects of economic
globalization are often unevenly distributed across the members of different ethnic
groups.
Accordingly, the degree of economic integration with the rest of the world in-
fluences on ethnic inequality (i.e. inequality across ethnic groups), favouring some
ethnic groups over others (Olzak, 2011). The implications of economic globalization
on ethnic inequality may be especially important in lower income countries, where
the most benefited are generally those ethnic groups that hold a political dominant
position, whereas other groups tend to be excluded and experience few benefits from
the process of integration (Chua, 2003). In order to keep their privileged situation
and limit the degree of mobilization of disadvantaged groups, the dominant ethnic
group usually adopts practices including the deterioration of civil and political rights
of minority groups. This setting leads to an intensification of social unrest based on
ethnic cleavages (Østby, 2008; Wimmer et al., 2009), which is consistent with the
increasing relevance of ethnic violent conflicts during the last decades (Chua, 2003).
The social dimension of globalization can also affect conflict. Thus, the flows of
information and ideas that characterize social integration boost internal movements
6
based on claims for self-determination and expanded minority rights (Soysal 1994;
Frank and McEneaney 1999; Schofer and Fourcade-Gourinchas 2001; Tsutsui 2004;
Tsutsui and Wotipka 2004). Social globalization helps to reduce the cultural distance
between countries, thus providing an ideological platform and an international audi-
ence predisposed to support these claims (Olzak, 2011). In this setting, the minority
groups have greater capacity to mobilize against repressive regimes that deny them
their rights, which in turn raises the risks of armed violent conflict. Moreover, the
advances in this dimension of globalization give rise to an increase of migratory flows
across national borders (Goldberg and Pavnick, 2007). These migratory flows often
lead to a negative reaction of native citizens and the aggravation of existing ethnic
tensions.
Social globalization also generates greater international pressure on repressive
regimes, as a result of the increasing information available via the Internet and other
global communication media (Dreher et al., 2008). In this context, the existence of a
violent armed conflict within a country affects negatively the probability of receiving
foreign investment and international aid. Indeed, this effect is particularly important
in those countries highly dependent of tourism, as the economic gains generated by
tourism are put at risk due to the negative publicity of internal violence. This argu-
ment seems to suggest that this aspect of social globalization increases the opportunity
cost of civil war, thus reducing the risk of conflict. It should be recalled, however, that
additionally the advance of the new technologies of information facilitates the capacity
of mobilization of insurgents, as can be observed in the recent wave of demonstrations,
protest and civil wars that has shaken the Arab world.
Finally, political globalization may also be connected with the incidence of within-
country conflict through different mechanisms. An important aspect of this dimension
7
of globalization has to do with the increasing relevance of international economic and
political organizations. The membership in these organizations provides numerous ad-
vantages of different nature that member countries do not want to endanger adopting
repressive polices against the rights of specific minority groups (Dreher et al., 2012).
For example, in the case of the European Union, the member states must sign the Eu-
ropean Convention for the Protection of Human Rights and Fundamental Freedoms.
Likewise, several UN declarations order that action be taken against those countries
violating the rights of ethnic minorities (Koenig, 2008), although examples like the
Darfur conflict suggest the scant relevance in practice of international pressure in this
respect. Anyway, this seems to suggest that political globalization should have the
beneficial consequence of deterring internal armed conflict.
Furthermore, international organizations such as the WTO, the FMI or regional
trade unions often defend the strategic interests of power blocks like the United States
or the European Union (Dreher, 2012). Accordingly, the decisions adopted by these
organizations tend usually to be based on asymmetric trade and financial relations,
which can affect the internal situation and the economic performance of low and
middle income countries (Stiglitz, 2006). This may have implications on the level of
dispersion of the income distribution, the degree of ethnic inequality or the magnitude
of spatial disparities within these countries. As outlined above, all these factors are
especially important in explaining the potential for social unrest and civil conflict.
In any case, empirical research is key to illustrating the potential link between
globalization and conflict. During the last years various studies have investigated this
relationship empirically, paying particular attention to the impact of international
trade on civil war (Barbieri and Reuveny, 2005; Bussmann and de Soya, 2005; Buss-
mann and Schneider, 2007; Martin et al., 2008). The analysis of the link between
8
trade openness and civil war is doubtless useful for attempting to examine the effect
of economic globalization on internal conflict, but it does not provide any information
on the role played in this context by social and political globalization. Although the
different aspects of globalization are often positively correlated, this omission is po-
tentially important, as the various arguments discussed above show that social and
political globalization may have a direct effect on the the incidence of conflict. Accord-
ingly, the impact on conflict of economic integration observed in the literature may be
affected by the omission from the analysis of social and political globalization (Dreher,
2006). Bearing this in mind, in this paper we follow the strategy adopted in a recent
paper by Olzak (2011) and use an extensive concept of globalization, which allows us
to examine comprehensively the overall effect on civil war of economic, social, and
political integration.
3 The process of globalization: Data and dynamics
Our empirical analysis requires comparable and reliable information on the incidence
of globalization in the various countries. Nevertheless, this is not an easy task because,
as discussed above, globalization is a multidimensional process and cannot be captured
by a single variable (Clark, 2000; Keohane and Nye, 2000). Bearing this in mind, the
measure of globalization that we use is the KOF index of globalization constructed
by Dreher (2006) and updated by Dreher et al. (2008). This is a composite index
widely employed in recent literature to examine different aspects of the consequences
of globalization (e.g. Dreher, 2006; Dreher and Gaston, 2007, 2008; Bergh and Nilsson,
2010; Rao and Vadlamannati, 2011).1
1A comprehensive list of papers based on the KOF index of globalization can be found athttp://globalization.kof.ethz.ch/.
9
The KOF index is based on a set of 23 variables associated with different di-
mensions of globalization. These variables are used to obtain three indexes on the
incidence of economic, social and political integration, by means of principal compo-
nent analysis (see Dreher et al. (2008) for further details). In turn, the information
provided by these three indexes is employed to calculate an overall index of global-
ization. Specifically, the index of economic integration is a weighted average of two
subindexes measuring respectively actual economic flows and existing restrictions on
trade and capital. The index of social integration is a weighted average of three
subindexes capturing respectively the importance of personal contacts, information
flows and cultural proximity. The degree of political integration is proxied by the
number of embassies in a country, the membership in international organizations, the
participation in UN Security Council missions, and the ratification of international
treaties. Finally, the overall index of globalization is obtained as a weighted average
of the three indexes of economic, social and political integration. Table ?? includes
further details on the different components of the KOF index, as well as the weights
attached to each individual variable to derive the various indexes. In turn, Table ??
displays the Spearman’s rank correlation coefficients between the overall measure of
globalization and the three indexes of economic, social and political integration. As
expected, all the correlation coefficients are positive and statistically significant at the
1% level. Their magnitude, however, reveals the existence of discrepancies between
the orderings generated by the various dimensions of globalization identified by the
KOF index. This shows that the distinction between economic, social and political
globalization is empirically relevant and is not exclusively a conceptual issue.
[INSERT TABLE 2 AROUND HERE]
10
According to the KOF index, in 2009 the world’s most globalized country was Bel-
gium (score of 92.7), as a result of the high degree of economic and political integration
of this country with the rest of the world. Belgium was followed by other European
countries, such as Ireland (92.0), Netherlands (90.9) and Austria (90.6). In the other
end of the scale we find East Timor, which was the world’s least globalized country in
2009 (score of 23.4). Other countries ranking low on the overall index include Equa-
torial Guinea (25.9), Laos (26.4) or Solomon Islands (27.0). A first observation from
this ranking is that the incidence of globalization appears to be positively associated
with the level of economic development of the various countries. This impression is
confirmed by the scatter plot displayed in Figure 1, which indicates that high income
countries tend in general to register greater levels of integration with the rest of the
world than low and middle income countries.
[INSERT FIGURE 1 AROUND HERE]
In order to complete this information, we now study the dynamics of the process
of globalization throughout the study period. To do so, instead of resorting to a set
of descriptive statistics, we examine the entire distribution of the KOF index. In
particular, we apply a non-parametric approach to estimate the external shape of the
distribution, using a Gaussian kernel function and the optimal smoothing parame-
ter proposed by Silverman (1986). The employment of non-parametric techniques is
particularly useful in the context that concerns us, given the lack of generality and
flexibility associated with parametric methods. Our estimates are shown in Figure
??. As can be observed, the initial situation has not remained stable over time. Thus,
there has been a clear shift of the density functions towards the right between 1970 and
11
2009, which indicates that the incidence of the process of globalization has increased
throughout the study period, particularly during the last two decades (???). Further-
more, Figure ?? reveals that the distance between the two ends of the distribution
has decreased. This implies the narrowing of the gap between the most and the least
globalized countries. Additionally, the density located around the single mode has
grown between 1970 and 2009. These changes suggest that the diffusion of globaliza-
tion all over the world has contributed to reducing existing differences in the degree of
integration of the various countries (Villaverde and Maza, 2011). In order to confirm
the relevance of this process of convergence, we calculate the standard deviation of
the KOF index over time. The value of this statistic is found to have decreased by
25% between 1970 and 2009.
[INSERT FIGURE ?? AROUND HERE]
4 Is there an empirical link between globalization and
civil war?
In this section we investigate empirically the relationship between globalization and
within-country conflict. To that end, we consider the estimation of a probit model
for the incidence of civil wars (Reynal-Querol, 2004). This model can be expressed as
follows:
Cit = α+ βGit + γ′Xit + εit (1)
12
where C is a binary variable that takes a value of one if civil conflict occurred in country
i during year t and zero otherwise, G is the KOF index of globalization described above,
X denotes a set of variables that control for additional factors that are assumed to
have an influence on internal conflict, and ε is the corresponding disturbance term.
The coefficient of interest throughout the paper is β, which measures the effect of
globalization on the incidence of within-country conflict.
The primary data on conflict used in the paper are drawn from the UCDP/PRIO
dataset. This dataset is the result of a collaborative project between the Department
of Peace and Conflict Research at Uppsala University and the Centre for the Study
of Civil War at the International Peace Research Institute located in Oslo.2 The
UCDP/PRIO data, which are described in detail by Gleditsch et al. (2002), have been
frequently used in recent years by numerous researchers and policy makers (e.g. Collier
et al., 2003; Miguel et al., 2004; Esteban et al., 2012 a,b). Given the nature of our
study, we focus our attention in intermediate and high-intensity conflicts. Thus, the
basic dependent variable of model (3) takes into account whether the internal conflict
has exceeded 1,000 battle-related deaths throughout its course. More specifically, a
conflict is coded as zero as long as it has not resulted over time in more than 1,000
battle-related casualties. Once a conflict reaches this threshold is coded as one. This
definition of civil war is consistent with the approach adopted among many others by
Doyle and Sambanis (2000), Fearon and Laitin (2003) or Montalvo and Reynal-Querol
(2005). Figure 3 shows the number of civil conflicts that meet the above criterion
between 1970 and 2009. The graph shows clearly that the global trend in within-
country conflict has not been uniform throughout the study period. The maximum
number of civil wars was reached at the beginning of the 1990s and was the result of
2For further details see http://www.pcr.uu.se/research/ucdp/
13
the steady and gradual accumulation of conflicts since the mid 1970s. Coinciding with
the end of the Cold War, the incidence of civil conflicts decreased slightly. This seems
to suggest that the increase in the incidence of internal violence in some countries
associated with the Soviet collapse was offset by improved management strategies by
states and international organizations (Gurr, 2000). In any case, Figure 3 reveals that
in 2009 there are still 23 ongoing civil wars all over the world, which implies that
about one in eight countries is affected by internal conflict.3
[INSERT FIGURE 3 AROUND HERE]
The control variables included in vector X have been selected on the basis of ex-
isting studies on the explanatory factors of civil war (e.g. Doyle and Sambanis, 2000;
Collier and Hoefler, 2002; Fearon and Laitin, 2003; Montalvo and Reynal-Querol,
2005; Collier et al., 2009; Esteban et al., 2012a,b). Considering the findings of this
literature, we take as our baseline specification of model (3) the following set of con-
trols: GDP per capita, population, percentage of mountainous terrain, non-contiguity
of country territory, ethnic fractionalization and democracy.4 Although the detailed
analysis of the links between these variables and civil war is beyond the scope of this
paper, it is worth pausing for a moment to recall briefly that GDP per capita can
be interpreted as a proxy for “a state’s overall financial, administrative, police and
military capabilities” (Fearon and Laitin, 2003, p.80). This suggests that rebels can
expect a greater probability of success in low income countries. Moreover, a higher
level of GDP per capita implies a greater opportunity cost of engaging in a civil war
3It is interesting to note that the basic pattern in Figure 3 is observed in a broad range of other datasets on civil wars (e.g. Fearon and Laitin, 2003; Esteban et al., 2012b).
4See the Appendix for further details on the definitions and sources of these variables.
14
(Collier and Hoefler, 2002). The size of population is also important in this context
because the number of potential rebels that can be recruited by the insurgents is
greater in large countries, whereas the government of these countries must face more
difficulties to exercise its authority and keep the control at the local level (Montalvo
and Reynal-Querol, 2005). Additionally, as is usual in the literature, the threshold
used to define the dependent variable of model is not normalized by the population
of the country in question, which tends to bias civil wars in favour of large countries
(Esteban et al., 2012a). The inclusion of the population control in the list of regressors
allows one to take care of this problem.
Furthermore, geographical factors may be related to the incidence of internal con-
flict. Rough and mountainous terrain can be used by rebel groups to hide from gov-
ernment forces. Likewise, the existence of a territorial base separated geographically
from the country’s centre should favour insurgency and civil war (Fearon and Laitin,
2003). In turn, ethnic cleavages are commonly perceived as an important cause of
internal conflict. This has to do with the belief that ethnically diverse societies often
register a greater degree of violence, which may lead in the final instance to civil war
(Horowitz, 1985; Esteban et al., 2012a,b). Bearing this in mind, we include in our
baseline model a measure of fractionalization commonly used in the literature to de-
scribe the ethnic diversity of the society. Finally, democracy may also be related to
the presence of internal conflict. Democratic states are generally characterized by less
repression of minority groups and by the observance and respect of civil and political
rights of their citizens (Fearon and Laitin, 2003). Although the relationship between
democracy and conflict is complex (Collier and Rohner, 2012), this argument seems
to suggest that democracy may reduce the risk of civil war.
15
Before estimating the model it is important to note that the existence of an inter-
nal conflict may also affect the incidence of globalization in the country in question
(Martin et al., 2008). Moreover, there may be some measurement errors in the val-
ues of the globalization index, particularly in the case of low income countries where
the probability of conflict is also greater. The implications of these two problems
are potentially important from and econometric perspective (Yatchew and Griliches,
1985; Wooldridge, 2010), but they could be solved if we had a suitable instrument
for globalization. Such an instrument must not be correlated with the error term in
model (3), but must be an important factor in accounting for the variation in the
incidence of globalization that we observe in our sample. Finding an instrument that
fulfils these two conditions is not an easy task in our context given the nature of the
KOF index. Nevertheless, we can use as instrument for globalization the degree of
remoteness of the various countries, which has been widely employed in the recent
literature as one of the determinants of the level of openness of national borders to
international trade (Baier and Bergstrand, 2004). In fact, Martin et al. (2008) employ
this variable as instrument for trade openness in their analysis of the link between civil
war and international trade. Intuitively, our choice of this instrument is based on the
idea that remote countries tend to register low levels of integration with the rest of
the world. According to the literature, we calculate the remoteness variable according
to the following expression:
Rit = − ln
n∑
j 6=i
GDPjt
dij
(2)
where dij is the bilateral distance between country i and country j.
16
At this point we should investigate to what extent this instrument is correlated
with the globalization index. To that end we present in Table 3 the results of the first
stage regressions of the form:
Git = δ + ζRit + φ′Xit + υit (3)
[INSERT TABLE 3 AROUND HERE]
As can be observed, in all the regressions the instrument has a positive and sta-
tistically significant effect on overall globalization and its three dimensions identified
by the KOF index: economic, social and political integration. These relationships are
illustrated in Figures 4-7 with the partial regression plots of the remoteness indica-
tor versus the various measures of globalization conditional on the full set of control
variables described above. Our findings confirm that those countries with low (high)
values of the remoteness variable are characterized generally by registering high (low)
levels of globalization. Table 3 also shows that the F-statistics for the excluded instru-
ment are in all cases well above the threshold of 10 suggested by Staiger and Stock
(1997) when there is a single endogenous regressor. The relevance of the instrument
is confirmed additionally by the partial R-squared, which measures the correlation
between the different indexes of globalization and the instrument after partialling out
the effect of the remaining controls.
[INSERT FIGURE 4 AROUND HERE]
17
[INSERT FIGURE 5 AROUND HERE]
[INSERT FIGURE 6 AROUND HERE]
[INSERT FIGURE 7 AROUND HERE]
The information provides by the first stage regressions in Table 3 indicates that
the remoteness variable is significantly associated with the incidence of globalization
in the various countries. To be a valid instrument, however, the remoteness variable
should not affect civil war, beyond its impact through the degree of integration. This
condition cannot be tested formally in the absence of other instruments. Nevertheless,
it seems reasonable to assume that R does not exert a direct effect on the incidence
of internal conflict, which suggests that remoteness is a plausible instrument in this
context.
Table 4 shows the results obtained when a pooled instrumental variable probit
is used to estimate model (3) by maximum likelihood with robust standard errors
clustered at the country level. The first column of the table shows that the coefficient of
the overall index of globalization is positive and statistically significant. Accordingly,
the degree of integration with the rest of world contributes to increasing the incidence
of civil wars, which raises worrying implications on the consequences of the process
of globalization currently underway. More specifically, this result questions directly
the validity of those arguments that defend that the advances of globalization can
help to promote stability and peace, reducing the risks of internal conflicts all over
18
the world (Barbieri and Reuveny, 2005; Blanton and Apocada, 2007). The different
controls included in our baseline specification of model (3) show that this is not a
spurious correlation resulting from the omission of relevant variables in this context.
In particular, in view of the information provided by Figure 1, it is important to note
that the overall index of globalization has a statistically significant effect on within-
country conflict war even when we control for the level of GDP per capita. This
indicates that the degree of international integration makes an important contribution
in explaining the incidence of civil wars and is not simply capturing the effect of the
level of economic development.
[INSERT TABLE 4 AROUND HERE]
So far we have investigated the overall impact of globalization on civil wars. In
order to complete our results, we now use the information provided by the KOF index
to examine the role played in this setting by economic, social and political integration.
This is particularly interesting in this context, given that it is not clear a priori that
these three dimensions of globalization affect civil wars in the same way (Brown, 2000;
Dreher, 2006). Bearing this in mind, model (3) is estimated again using the indexes of
economic, social and political integration in turn as regressors, instead of the overall
indicator employed so far. Columns 2-4 of Table 4 present the findings obtained
when the three dimensions of globalization identified by Dreher (2006) are analysed
individually. The results are similar in all cases. Regardless of the specific index
employed in each case, the coefficients of the different measures of globalization are
positive and statistically significant in all the specifications considered. Having said
that, it should be noted that the positive association between economic globalization
19
and internal conflict observed in the second column of Table 4 is consistent with the
empirical evidence provided by Martin et al. (2008) and Olzak (2011), although it
contrast with the negative correlation between trade openness and the risk of civil war
reported by Bussmann and De Soya (2005), Barbieri and Reveney (2005), or Bussmann
and Schneider (2007). Nevertheless, at this point it is important to recall that only
Martin et al. (2008) has taken into account to date the issue of reverse causality
between globalization and conflict. Anyway, this result should be treated with some
caution, since the coefficient of the index of economic globalization is statistically
significant only at the 10% level. Additionally, the third and fourth columns of Table
4 show that the incidence of civil war in a given country is positively related to its level
of social and political globalization, which is consistent with several of the arguments
discussed in section 2. These findings are an important contribution of the paper,
taking into account the scant attention paid so far by the literature to the potential
link between these two dimensions of globalization and the incidence of civil wars.
With respect to the control variables included in model (3), Table 4 reveals that the
results are consistent with the findings of the existing literature on the determinants
of internal armed conflicts (Collier and Hoeffler, 1998, 2004; Fearon and Laitin, 2003;
Miguel et al. 2004; Montalvo and Reynal-Querol, 2005, Martin et al. 2008; Esteban
et al. 2012a,b). Thus, our estimates show that the coefficient of GDP per capita is
negative and statistically significant, indicating that the incidence of civil war is greater
in those countries with low levels of economic development. In turn, internal violence
is more likely in larger countries. Likewise, the presence of rough and mountainous
terrain is also related to higher rates of civil conflict. Furthermore, the index of
fractionalization exerts a significant effect on the dependent variable, which indicates
that ethnic diversity is a relevant factor in explaining the existence of internal violence.
20
The coefficients of the remaining controls have in all cases the usual signs in the
literature, but none of them is statistically significant consistently across the various
regressions included in Table 4. As pointed out by Martin et al. (2008), this may have
to do with the employment in our analysis of clustered standard errors.
5 Robustness checks
The analysis carried out so far suggests the existence of a positive impact of globaliza-
tion on the incidence of civil war. In particular, our estimates seem to indicate that
the three main dimensions of globalization (economic, social and political integration)
are positively associated with the presence of civil armed conflicts. In this section we
investigate the robustness of these findings.
Alternative measures of conflict
As mentioned above, the dependent variable in model (3) is a binary variable that
reports all conflicts with more than 1,000 battle-related deaths over its course. Our
findings, however, may be affected by the choice of this specific threshold of deaths.
For this reason, as a first robustness check, we examine to what extent the previous
results depend on the definition of civil war used to construct the dependent variable in
model (3). To this end, we now employ an alternative indicator based on UCDP/PRIO
data that corresponds to conflicts with 25 or more battle-related deaths in a given year
(prio25 ), which allows us to include in the analysis low-intensity conflicts (Miguel et
al., 2004; Montalvo and Reynal-Querol, 2005; Esteban et al., 2012a).
[INSERT TABLE 5 AROUND HERE]
21
Table 5 reports the results obtained when prio25 is used as dependent variable in
model (3). As can be checked, this change has little effect on the main findings of the
paper. More specifically, the first column of the table shows that the coefficient of
the overall index of globalization continues to be positive and statistically significant.
Similar results are obtained in the regressions with the indexes of of social and political
globalization (third and fourth columns of Table 5). The only appreciable change in
comparison with the estimates in Table 3 has to do with the economic dimension
of integration. Thus, the second column of Table 5 still indicates the presence of
a positive association between the index of economic globalization and civil conflict.
Nevertheless, in this case the corresponding coefficient is not statistically significant at
the usual levels, which raises doubts on the robustness of the link between the degree
of economic integration with the rest of the world and civil war.
At this point it should be recalled that the information to construct the two de-
pendent variables employed so far in Tables 3 and 5 were drawn in both cases from
the UCDP/PRIO dataset. In view of this, one may wonder if our results could be
affected by the use of this dataset. In order to investigate this issue, we resort to
the data employed by Doyle and Sambanis (2000) and Fearon and Laitin (2003), who
constructed two alternative lists of conflicts based on the information provided by the
Correlates of War (COW) project and other sources.5 Doyle and Sambanis (2000)
define civil war as a conflict that: ”(a) it caused more than 1,000 deaths; (b) it chal-
lenged the sovereignty of an international recognized state; (c) it occurred within the
recognized boundary of that state; (d) it involves the state as a principal combatant;
(e) it included rebels with the ability to mount organized armed opposition to the
state; and (f) the parties were concerned with the prospects of living together in the
5See Sambanis (2002) for further details about the COW civil war data.
22
same political unit after the end of a war”. In turn, for Fearon and Laitin (2003), a
violent conflict should meet the following criteria to be coded as a civil war: (a) it
should involve the fighting between agents of (or claimants to) a state and organized,
non-state groups who sought either to take control of a government, take power in
a region, or use violence to change government policies; (b) the conflict killed or has
killed at least 1,000 over its course, with a yearly aver- age of at least 100 deaths; (c)
at least 100 were killed on both sides (including civilians at- tacked by rebels).
[INSERT TABLE 6 AROUND HERE]
[INSERT TABLE 7 AROUND HERE]
Tables 6 and 7 show respectively the impact of globalization on civil war using
the data from Doyle and Sambanis (2000) and Fearon and Laitin (2003). Although
the employment of these alternative datasets on civil war obliges us to reduce the
sample size, it is important to stress that our major findings remain unaffected. Our
estimates show that the different indexes of globalization exert in all cases a positive
and statistically significant effect on within-country conflict, regardless of the specific
dataset used in each case. This confirms the evidence provided by Table 4.
Additional controls and regional dummies
As an additional robustness check, we now investigate the possibility that our results
are driven by an omitted variable. We address this issue by controlling for different
covariates that could plausibly be correlated with both civil war and globalization,
and checking whether the inclusion of these covariates affects our estimates.
23
According to this strategy, we include in our baseline specification a measure of
the degree of natural resource abundance based on Esteban et al. (2012a). Numerous
studies have highlighted the link between natural resource abundance and violent
civil conflict (Collier and Hoefler, 1998, 2004; Ross, 2006; Brunnschweiler and Bulte,
2009) Resource-rich countries are often characterized by land expropriation, inadquate
job opportunities and labour migration, which may breed social unrest in different
sectors of the society (Rosser, 2006). Natural resource abundance may also increment
the potential gains of these officials who are in charge on the exploitation of such
resources, which may give rise to more corruption and poor governance (Ades and
Di Tella, 1999). In addition, natural resources may provide an important source of
funding for rebel forces, although the presence of resource rents may also increase the
probability of foreign intervention.
As is usual in the literature, our baseline specification includes an index of fraction-
alization to capture the degree of ethnic heterogeneity in the various countries. Nev-
ertheless, the empirical evidence suggests that there is less ethnic violence in highly
homogeneous and highly heterogeneous societies, whereas the intensity of internal
conflict is greater in those societies where a large ethnic minority must face an ethnic
majority. In view of this, various authors argue that an index of polarization captures
the risk of potential ethnic conflict better than traditional indexes of fractionalization
(Montalvo and Reynal-Querol, 2005; Esteban et al., 2012a,b). For this reason, we
include in the list of regressors of model (3) a measure of ethnic polarization proposed
by Esteban et al. (2012a).
We also control for the economic growth rate in earlier years, which is a variable
commonly used in the literature to proxy for the income foregone by enlisting as
a rebel (Collier and Hoefler, 2002; Miguel et al. 2004). The lower is the growth
24
rate, the lower is the opportunity cost of engaging in a civil war. Furthermore, we
introduce in the baseline specification a dummy for former French colonies. This
may be important in this context, since France has shown more willing than other
colonial powers to intervene militarily to preserve the political order in its former
territories. Accordingly, actual French military interventions or simply the threat of
such interventions may work as a disincentive to rebellion, thus reducing the risk of
an internal armed conflict (Collier and Hoefler, 2002).
[INSERT TABLE 8 AROUND HERE]
Table 8 presents the results obtained when model is estimated again including
these additional controls. As can be seen, none of these covariates is consistently
significant in the various regressions and their inclusion in our baseline specification
has little effect on the main results of the paper. In particular, Table 8 indicates
that the additional controls considered do not affect the estimates of the impact of
globalization on within-country conflict. The different indexes of globalization hold
positive and statistically significant in all cases, corroborating the robustness of our
findings.
We now check whether the results are sensitive to the inclusion of regional dum-
mies. To that end, we add to the list of regressors of model (3) three dummies for
countries in the most conflictive regions of the world during the study period: Sub-
Saharan Africa, Latin America and Asia. This allows us to test whether some omitted
factor make these regions more prone to internal conflict than predicted by our base-
line model. As reported in Table 9, the coefficients of the regional dummies are not
statistically significant in any of the regressions considered. This suggests that Sub-
25
Saharan Africa, Latin America and Asia are not subject to an additional risk factor
as these regions do not differ significantly in their incidence of conflict from that pre-
dicted in the model. Focusing on the main aim of the paper, it is important to note
that the coefficients of the various globalization indexes continue to be positive and
statistically significant in all cases, which is in line with our previous findings.
[INSERT TABLE 9 AROUND HERE]
Alternative estimation methods
Are the results of the paper robust to the way used to estimate model (3)? So far
our analysis of the empirical relationship between globalization and civil war has been
based on the information provided by a maximum likelihood estimator applied in the
context of an instrumental variable approach. We now investigate to what extent this
choice affects the main findings of the paper. To do so we resort to the minimum chi-
squared estimator proposed by Newey (1987) to estimate a probit model when there
are continuous endogenous regressors. Table 10 shows the results obtained when
this alternative method is used to estimate our baseline model. As can be seen, our
main findings still hold, confirming the observed relationship between the degree of
integration with the rest of the world and within-country conflict.
[INSERT TABLE 10 AROUND HERE]
26
6 Conclusions
As is well-known, civil wars account for an enormous share of deaths and hardship
in the world today. In addition to the direct impact on battle-related deaths, within-
country conflicts give rise to an important number of indirect deaths due to disease
and malnutrition, as well as the forced displacements of numerous people (Lacina and
Gleditsch, 2005; Esteban et al., 2012b). Therefore, the analysis of the explanatory
factors of internal armed conflicts is particularly relevant. Against this background,
in this paper we have investigated the link between the process of globalization and
civil war using data on 138 countries over the period 1970-2009. Unlike most of
existing studies on this issue, the paper employs an extensive notion of globalization
including its three main dimensions: economic integration, social integration and
political integration.
The results obtained from an instrumental variable approach show a positive as-
sociation between globalization and the incidence of civil war. Accordingly, the ad-
vances in the process of globalization currently underway contribute to increasing
significantly the risk of internal armed conflict. This finding contrasts directly with
those arguments that defend that globalization has the beneficial effect of deterring
within-country conflicts. The observed link does not depend on the specific dimen-
sion of globalization considered in the analysis, although the aspects of integration
more robustly correlated with civil war are social and political globalization. This is
particularly interesting, given the scant attention received so far in the literature by
these two dimensions of globalization. The results of the paper are not affected by the
inclusion in the analysis of additional explanatory variables, changes in the definition
and sources of data on civil wars, and the employment of an alternative estimation
27
method.
The analysis carried out raises some potentially interesting implications. Thus, our
findings indicate that not only internal factors of countries are relevant to explain the
existence of armed violent conflicts within national borders, which should be taken
in consideration by those who are in charge of designing initiatives and measures
aimed at reducing the incidence of such conflicts. Furthermore, numerous studies
have found that the degree of integration with the rest of the world plays a key role in
fostering growth and economic development (Frankel and Romer, 1999; Dreher, 2006).
Nevertheless, our results suggest that, in addition to this direct impact, globalization
also has a negative influence on economic development through its effect on civil
war. In any case, further research is required to identify and analyse in detail the
various causal mechanisms which explain in the final instance the complex link between
globalization and internal conflict. Only by pursuing this strand will we be able to
attain a more complete understanding about how globalization affects civil war.
Acknowledgements
The authors would like to thank Javier Hualde and Sara Martınez de Morentın for
very helpful comments and suggestions. Roberto Ezcurra gratefully acknowledges the
financial support of the Spanish Ministry of Economy and Competitiveness (Project
ECO2011-29314-C02-01).
28
References
Ades, A. and Di Tella, R. (1999): Rents, competition, and corruption, American
Economic Review 89, 982-993.
Alcala, F. and Ciccone, A. (2004): Trade and Productivity, Quarterly Journal of
Economics, 613-646.
Antweiler, W., Copeland B. R. and Taylor M. S. (2001): Is free trade good for the
environment?, American Economic Review 91, 877-908.
Baier, S. L. and Bergstrand, J. H. (2004): Economic Determinants of Free Trade
Agreements, Journal of International Economics 64, 29-63.
Barbieri, K. and Reuveny, R. (2005): Economic globalization and civil war, The
Journal of Politics 67, 1228-1247.
Bergh, A. and Nilsson, T. (2010): Good for living? On the relation between global-
ization and life expectancy, World Development 38, 1191-1203.
Bhagwati, J. (2004): In Defense of Globalization. Oxford University Press, New
York.
Blanton, R.G. and Apocada, C. (2007): Economic globalization and violent civil con-
flict: Is openness a pathway to peace?, The Social Science Journal 44, 599-619.
Brown, L. D., Khagram, S., Moore, M. H. and Frumkin, P. (2000): Globalization,
NGOs, and multisectoral relations, in Governance in a Globalizing World. Eds J.
S. Nye, J. D. Donahue. Brookings Institution Press, Washington, D.C., 271-296.
Brunnschweiler C. N. and Bulte E. H. (2009). Natural Resource and Violent Conflict:
Resource Abundance, Dependence and the Onset of Civil Wars, Oxford Economic
29
Papers 61, 651-674.
Bussmann, M. and De Soysa, I. (2005): How Taxing is Trade? Globalization, State
Capacity and Civil War, mimeo, University of Konstanz.
Bussmann, M. and Schneider, G. (2007): When globalization discontent turns violent:
foreign economic liberalization and internal war, International Studies Quarterly
51, 79-97.
Chua, A. (2003): World On Fire. How Exporting Free Market Democracy Breeds
Ethnic Hatred and Global Instability. New York: Anchor Books.
Clark, W. C. (2000): Environmental globalization, in Governance in a Globalizing
World. Eds J. S. Nye and J. D. Donahue. Brookings Institution Press, Washing-
ton, D.C., 86-108.
Collier, P. and Hoeffler, A. (1998): On Economic Causes of Civil War, Oxford Eco-
nomic Papers 50, 563-573.
Collier, P. and Hoeffler, A. (2002): On the Incidence of Civil War in Africa, Journal
of Conflict Resolution 46, 13-28.
Collier, P. and Hoeffler, A. (2004): Greed and Grievance in Civil War, Oxford Eco-
nomics Papers 56, 563-595.
Collier, P., Hoeffler, A. and Rohner, D. (2009): Beyond Greed and Grievance: Feasi-
bility and Civil War, Oxford Economics Papers 61, 1-27.
Collier, P. and Rohner, D. (forthcoming): Democracy, development and conflict, The
Journal of the European Economic Association.
Dollar, D. and Kraay, A. (2004): Trade, growth and poverty, Economic Journal 114,
F22-F49.
30
Doyle, M.W. and Sambanis, N. (2000): International Peacebuilding: A Theoretical
and Quantitative Analysis, American Political Science Review 94(4), 779-801.
Dreher, A. (2006): Does globalization affect growth? Evidence from a new index of
globalization, Applied Economics 38, 1091-1110.
Dreher, A. and Gaston, N. (2007): Has globalization really had no effect on unions?,
Kyklos 60, 165-186.
Dreher, A. and Gaston, N. (2008): Has globalization increased inequality?, Review of
International Economics 16, 516-536.
Dreher, A., Gaston, N. and Martens, P. (2008): Measuring Globalization: Gauging its
Consequences. Springer, New York.
Dreher, A. and Gassebner, M. (2012): Do IMF and World Bank programs induce
government crises? An empirical analysis, International Organization 66, 2: 329-
358.
Dreher, A., Gassebner, M. and Siemers, L. H. (2012): Globalization, economic freedom
and human rights, Journal of Conflict Resolution 56, 3: 509-539.
Esteban, J. Mayoral, L. and Ray, D. (2012): Ethnicity and Conflict: An Empirical
Study, American Economic Review102, 1302-1342.
Esteban, J. Mayoral, L. and Ray, D. (2012): Theory and facts, Science 336, 858.
Fearon, J. and D. Laitin (2003): Ethnicity, Insurgency, and Civil War, American
Political Science Review 97, 75-90.
Findlay, R. and O´Rourke, K. H. (2007): Power and Plenty: Trade, War and the
World Economy in the Second Millenium. Princeton University Press, Princeton,
NJ.
31
Frankel, J. A. and Romer, D. (1999): Does trade cause growth?, American Economic
Review 89, 379-399.
Frankel, J.A. and Rose, A. K. (2005): Is trade good or bad for the environment?
Sorting out the causality, Review of Economics and Statistics 87, 85-91.
Gleditsch, N-P., P. Wallensteen, M. Eriksson, M. Sollenber, and H. Strand (2002):
Armed Conflict 1946-2001: A New Dataset, Journal of Peace Research 39, 615-
637.
Goldberg, P.K. and Pavcnik, N.(2007): Distributional effects of globalization in de-
veloping countries, Journal of Economic Literature 45, 39-82.
Hegre, H., Gissinger, R. and Gleditsch, N.P. (2003): Globalization and internal con-
flict, in Globalization and armed conflict. Eds G. Schneider, K. Barbieri and N.
P. Gleditsch. Lanham, MD. Rowman and Littlefield, 251-275.
Horowitz, D.L. (1985): Ethnic Groups in Conflict. Berkeley, CA: University of Cali-
fornia Press.
Keohane, R. O. and Nye, J. S. (2000): Introduction, in Governance in a Globalizing
World. Eds J. S. Nye, and J. D. Donahue. Brookings Institution Press, Washing-
ton D.C., 1-44.
Koenig, M. (2008): Institutional change in the world polity: International human
rights and the construction of collective identities, International Sociology 23,
95-114.
Lacina, B. and Gledisch, N. P. (2005): Monitoring Trends in Global Combat: A New
Dataset of Battle Deaths, European Journal of Population 21, 145-166.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A. and Vishny, R. (1999): The quality of
32
government, Journal of Law, Economics and Organization 15, 222-279.
Martin, P., Mayer, T. and Thoenig, M. (2008): Civil War and International Trade,
Journal of the European Economic Association Papers and Proceedings 6, (-3):
541-550.
Miguel, E., Satyanath, S. and E. Sergenti (2004): Economic Shocks and Civil Conflict:
An Instrumental Variables Approach, Journal of Political Economy 112, 725-753.
Montalvo, J. G. and M. Reynal-Querol (2005): Ethnic Polarization, Potential Conflict
and Civil War, American Economic Review 95, 796-816.
Milanovic, B. (2005): Can we discern the effect of globalization on income distri-
bution?. Evidence from household surveys, World Bank Economic Review 19,
21-44.
Newey, W. K., West, K.D. (1987): A Simple, Positive Semi-definite, Heteroskedasticity
and Autocorrelation Consistent Covariance Matrix, Econometrica 55 (3), 703-708.
Norris, P. (2000): Global governance and cosmopolitan citizens, in Governance in a
Globalizing World. Eds J. S. Nye and J. D. Donahue. Brookings Institution Press,
Washington, D.C., 155-177.
Olzak, S. (2011): Does Globalization Breed Ethnic Discontent?, Journal of Conflict
Resolution 55, 3-32.
Østby, G. (2008): Horizontal Inequalities, Political Environment and Civil Conflict:
Evidence From 55 Developing Countries, Horizontal Inequalities and Conflict:
Understanding Group Violence In Multiethnic Societies, 136-157.
Prakash, A. and Hart, J. (1999): Globalization and governance: an introduction, in
Globalization and Governance, Eds A. Prakash, J. Hart. Routledge, London, 1-24.
33
Raab, M., Ruland, M. Schonberger, B., Blossfeld, H.P., Hofacker, D. and Buchholz,
S. (2008): Globalindex: A sociological approach to globalization measurement,
International Sociology 23, 599-634.
Rao, B B. and Vadlamannati, K C. (2011): Globalization and growth in the low
income African countries with the extreme bounds analysis, Economic Modelling
28, 795-805.
Reynal-Querol, M. (2005): Does democracy preempt civil wars?, European Journal of
Political Economy 21, 445-465.
Ross, M. (2006): A Closer Look at Oil, Diamonds and Civil War, Annual Review of
Political Science 9, 265-300.
Rosser, A. (2006): Escaping the Resource Curse, New Political Economy 11 (4), 557-
570.
Rudra, N. (2005): Globalization and the strengthening of democracy in the developing
world, American Journal of Political Science 49, 704-730.
Schofer, E. and Fourcade-Gourinchas, M. (2001): The structural contexts of civic en-
gagement: Voluntary association membership in comparative perspective, Amer-
ican Sociological Review 66, 806-28.
Sen, A. (1973): On Ignorance and Equal Distribution, American Economic Review
63, 1022-1024.
Shorrocks, A.F. (1980): The class of additively decomposable inequality measures,
Econometrica 48, 613-625.
Siverman, B. W. (1986): Monographs on Statistics and Applied Probability. London:
Chapman and Hall.
34
Soysal, J. (1994): Limits of citizenship: Migrants and postnational membership in
Europe. Chicago: University of Chicago Press.
Staiger, D. and Stock, J.H. (1997): Instrumental Variables Regression with Weak
Instruments. Econometrica 65, 557-586.
Stiglitz, J. E. (2002): Globalization and Its Discontents. W. W. Norton, New York.
Stiglitz, J. E. (2006): Making globalization work. W. W. Norton, New York.
Stiglitz, J. E. (2012): The Price of Inequality: How Today’s Divided Society Endangers
Our Future. W. W. Norton, New York.
Tomohara, A. and Takii, S. (2011): Does globalization benefit developing countries?.
Effects of FDI on local wages, Journal of Policy Modeling 33, 511-521.
Tsutsui, K. (2004): Global civil society and ethnic social movements in the contem-
porary world, Sociological Forum19, 63-88.
Tsutsui, K. and Wotipka, C. M. (2004): Global civil society and the international
human rights movement: Citizen participation in human rights international non-
governmental organizations, Social Forces 83, 587-620.
Villaverde, J. and Maza, A. (2011): Globalisation, Growth and Convergence, The
World Economy, Wiley Blackwell 34(6), 952-971, 06.
35
Wimmer, A., Cederman, L.E. and Min, B. (2009): Ethnic politics and armed conflict:
A configurational analysis of new global data set. American Sociological Review
74, 316-337.
Wooldridge, J. M. (2010): Econometric Analysis of Cross Section and Panel Data,
MIT Press Books, The MIT Press, edition 2, volume 1, number 0262232588,
September.
Yatchew, A., and Z. Griliches. (1985): Specification error in probit models. The
Review of Economics and Statistics 67, 134-139.
36
Data appendix: Definitions and sources
Dependent variables:
The different dependent variables used are described in detail in the main text.
Measures of globalization:
Overall index of globalization: Index constructed with principal components anal-
ysis comprising twenty- three variables measuring globalization. The index is a
weighted average of the indexes of economic, social and political globalization.
See Table 1 for further details. The index is expressed in natural logs. Source:
Dreher (2006) and Dreher et al. (2008).
Economic globalization: Index based on various measures of actual economic
flows and restrictions. See Table 1 for further details. The index is expressed in
natural logs. Source: Dreher (2006) and Dreher et al. (2008).
Social globalization: Index based on various measures on personal contacts, infor-
mation flows, and cultural proximity. See Table 1 for further details. The index
is expressed in natural logs. Source: Dreher (2006) and Dreher et al. (2008).
Political globalization: Index based on various measures capturing the degree of
political integration. See Table 1 for further details. The index is expressed in
natural logs. Source: Dreher (2006) and Dreher et al. (2008).
37
Control variables:
GDP per capita: Natural log of real GDP per capita expressed in constant 2005
international dollars. Source: Penn World Tables 7.0.
Population: Natural log of total population (thousands of people). Source: Penn
World Tables 7.0.
Mountainous terrain: Percentage of mountainous terrain. The variable is ex-
pressed in natural logs as log(1 +mount). Source: Fearon and Laitin (2003).
Noncontiguous states: Dummy variable that takes the value one for those coun-
tries with territory holding at least 10,000 people and separated from the land
area containing the capital city either by land or by 100 kilometers of water, zero
otherwise. Source: Fearon and Laitin (2003).
Fractionalization: Index of ethnic fractionalization capturing the probability that
two individuals randomly drawn from the population belong to different ethnic
group. The index is calculated as F =∑m
i=1 ni(1−ni), where ni is the population
share of group i and m is the number of groups. Source: Fearon and Laitin
(2003).
Democracy: Institutionalized democracy. Democracy ranges from zero (low)
to ten (high) (Polity IV Project). Using this information a dummy variable is
constructed to identify those countries where the democracy score is higher or
equal to four (Montalvo and Reynal-Querol, 2005). Source: Esteban et al. 2012a.
Natural resources: Dummy variable that takes the value one if the country is
“rich in oil” and zero otherwise. A country is “rich in oil” if the average value
of its oil production in a period is greater than 100 US dollars in 2,000 constant
38
dollars. Source: Esteban et al. (2012a) and US Geological Survey Mineral.
Polarization: Index of ethnic polarization. The index is calculated as P =∑mi=1
∑mj=1 n
2injkij , where k = 1 − sij and sij the degree of similarity between
two languages i and j, given by the ratio of the number of common branches to
the maximum possible number N fifteen for the entire tree.
Economic growth: Annual growth rate of real GDP per capita expressed in
constant 2005 international dollars. Source: Own calculations based on the
information provided by the Penn World Tables 7.0.
French colony: Dummy variable that the value one if the country in question is
a former French colony. Source: Norris, 200?.
39
Tables and figures
Table 1: Components of the KOF index of globalization.
Indices and Variables Weights
Economic Globalization [36%]
Actual flows [50%]Trade (percent of GDP) (21%)Foreign direct investment, stocks (percent of GDP) (28%)Portfolio investment (percent of GDP) (24%)Income payments to foreign nationals (percent of GDP) (27%)
Restrictions [50%]Hidden import barriers (24%)Mean tariff rate (27%)Taxes on international trade (percent of current revenue) (26%)Capital account testrictions (23%)
Social Globalization [37%]
Data on personal contacts [34%]Telephone traffic (25%)Transfers (% of GDP) (4%)International tourism (26%)Foreign population (percent of total population) (21%)International letters (per capita) (25%)
Data on information flows [35%]Internet users (per 1000 people) (33%)Television (per 1000 people) (36%)Trade in newspapers (percent of GDP) (32%)
Data on cultural proximity [31%]Number of McDonald’s restaurants (per capita) (44%)Number of Ikea (per capita) (45%)Trade in books (% of GDP) (11%)
Political globalization [26%]
Embassies in country (25%)Membership in international organizations (28%)Participation in UN Security Council Missions (22%)International treaties (25%)
Source: http://globalization.kof.ethz.ch/
40
Tab
le2:
Sp
earm
an’s
ran
kco
rrel
atio
nco
effici
ents
bet
wee
nth
eva
riou
sd
imen
sion
sof
glob
ali
zati
on
.
Ove
rall
ind
exE
con
omic
Soci
al
Pol
itic
al
ofgl
obal
izat
ion
glob
aliz
atio
ngl
ob
ali
zati
on
global
izati
on
Ove
rall
ind
exof
glob
ali
zati
on1.
000
Eco
nom
icglo
bal
izati
on0.
853
1.00
0S
oci
al
glo
bal
izati
on0.
923
0.77
91.
000
Pol
itic
al
glo
bal
izati
on0.
596
0.24
00.
382
1.000
Note
s:D
ata
for
179
countr
ies
in2005.
All
the
corr
elati
on
are
stati
stic
ally
signifi
cant
at
the
1%
level
.
41
Table 3: Globalization and remoteness. First stage regressions.
Dependent variable Overall index Economic Social Politicalof globalization globalization globalization globalization
(1) (2) (3) (4)
Remoteness -0.172*** -0.183*** -0.171*** -0.180***(0.021) (0.030) (0.027) (0.034)
GDP per capita (t-1) 0.197*** 0.190*** 0.332*** 0.104***(0.016) (0.025) (0.024) (0.017)
Population(t-1) -0.001 -0.058*** -0.068*** 0.117***(0.010) (0.020) (0.015) (0.014)
Mountainous terrain -0.026** -0.030 -0.018 -0.018(0.011) (0.019) (0.017) (0.013)
Noncontiguos state 0.040 0.108* -0.013 -0.012(0.031) (0.060) (0.070) (0.048)
Fractionalization 0.083 0.192 -0.064 0.057(0.072) (0.127) (0.104) (0.090)
Democracy 0.131*** 0.111*** 0.181*** 0.124***(0.023) (0.039) (0.038) (0.036)
Constant -0.582* -0.199 -1.346*** -0.806(0.345) (0.524) (0.447) (0.503)
F-test 64.09*** 36.47*** 40.88*** 27.71***Partial R-squared 0.181 0.094 0.085 0.114Countries 137 132 137 138Observations 4779 4600 4779 4818
Notes: Robust standard errors adjusted for clustering in parentheses. * Significant at 10%level, ** significant at 5% level, *** significant at 1% level.
42
Table 4: The relationship between globalization and civil war. Instrumental variableprobit.
(1) (2) (3) (4)
Overall index of globalization 1.597***(0.619)
Economic globalization 1.104*(0.582)
Social globalization 1.414***(0.522)
Political globalization 1.413**(0.580)
GDP per capita (t-1) -0.447** -0.364** -0.582*** -0.278**(0.175) (0.168) (0.210) (0.127)
Population(t-1) 0.171*** 0.230*** 0.248*** -0.010(0.063) (0.076) (0.067) (0.094)
Mountainous terrain 0.252*** 0.249*** 0.225*** 0.244***(0.074) (0.078) (0.071) (0.072)
Noncontiguous state 0.621** 0.602* 0.624** 0.674**(0.292) (0.310) (0.299) (0.292)
Fractionalization 1.304*** 1.160** 1.354*** 1.382***(0.425) (0.480) (0.394) (0.427)
Democracy -0.482** -0.312 -0.490** -0.469**(0.214) (0.198) (0.209) (0.206)
Constant -6.003*** -5.456*** -4.335*** -5.373***(1.408) (1.626) (1.010) (1.215)
Wald exogeneity test 10.69*** 3.737*** 8.754*** 8.033***Countries 137 132 137 138Observations 4779 4600 4779 4818
Notes: The dependent variable is a binary variable that takes a value of one for conflictswith 1,000 or more battle-related deaths over time and zero otherwise. Robust standarderrors adjusted for clustering in parentheses. * Significant at 10% level, ** significant at 5%level, *** significant at 1% level.
43
Table 5: Robustness analysis. Alternative measures of conflict (I). Instrumental variableprobit.
Dependent variable prio25
(1) (2) (3) (4)
Overall index of globalization 1.284**(0.631)
Economic Globalization 0.894(0.592)
Social Globalization 1.139**(0.572)
Political Globalization 1.140*(0.594)
Control variables Yes Yes Yes YesWald exogeneity test 9.920*** 3.596* 6.822*** 6.702***Countries 137 132 137 138Observations 4779 4600 4779 4818
Notes: The dependent variable is a binary variable that takes a value of one for conflictswith 25 or more battle-related deaths in a given year and zero otherwise. All the regressionsinclude the full set of control variables described in the text. Robust standard errors adjustedfor clustering in parentheses. * Significant at 10% level, ** significant at 5% level, ***significant at 1% level.
44
Table 6: Robustness analysis. Alternative measures of conflict (II): Doyle and Sambanis(2002) definition of civil war. Instrumental variable probit.
Dependent variable Doyle and Sambanis (2002)
(1) (2) (3) (4)
Overall index of globalization 2.819***(0.841)
Economic globalization 2.071***(0.655)
Social globalization 2.454***(0.529)
Political globalization 2.227***(0.698)
Control variables Yes Yes Yes YesWald exogeneity test 8.633*** 5.058** 7.549*** 7.428***Countries 135 130 135 136Observations 3411 3282 3411 3440
Notes: All the regressions include the full set of control variables described in the text.Robust standard errors adjusted for clustering in parentheses. * Significant at 10% level, **significant at 5% level, *** significant at 1% level.
45
Table 7: Robustness analysis. Alternative measures of conflict (III): Fearon and Laitin(2003) definition of civil war. Instrumental variable probit.
Dependent variable Fearon and Laitin (2003)
(1) (2) (3) (4)
Overall index of globalization 3.028***(0.725)
Economic globalization 2.294***(0.479)
Social globalization 2.585***(0.459)
Political globalization 2.408***(0.607)
Control variables Yes Yes Yes YesWald exogeneity test 11.17*** 9.450*** 8.714*** 7.908***Countries 136 131 136 137Observations 3436 3307 3436 3465
Notes: All the regressions include the full set of control variables described in the text.Robust standard errors adjusted for clustering in parentheses. * Significant at 10% level, **significant at 5% level, *** significant at 1% level.
46
Table 8: Robustness analysis. Additional controls. Instrumental variable probit.
(1) (2) (3) (4)
Overall index of globalization 1.658***(0.621)
Economic globalization 1.193**(0.572)
Social globalization 1.441***(0.516)
Political globalization 1.586***(0.612)
Natural resources 0.323 0.133 0.460* 0.292(0.264) (0.258) (0.275) (0.263)
Polarization 3.115 2.460 3.430* 3.030(2.061) (2.162) (2.024) (1.953)
Economic growth (t-2) -0.195 -1.115*** -0.165 -0.015(0.305) (0.417) (0.279) (0.251)
French colony -0.215 -0.148 -0.101 -0.504**(0.273) (0.299) (0.302) (0.252)
Control variables Yes Yes Yes YesWald exogeneity test 12.08*** 4.504** 9.816*** 8.709***Countries 136 131 136 137Observations 4528 4359 4528 4565
Notes: The dependent variable is a binary variable that takes a value of one for conflicts with1000 or more battle-related deaths over time and zero otherwise. All the regressions includethe full set of control variables described in the text. Robust standard errors adjusted forclustering in parentheses. * Significant at 10% level, ** significant at 5% level, *** significantat 1% level.
47
Table 9: Robustness analysis. Regional dummies. Instrumental variable probit.
(1) (2) (3) (4)
Overall index of globalization 1.629**(0.747)
Economic globalization 1.105*(0.651)
Social globalization 1.602**(0.629)
Political globalization 1.338**(0.645)
Sub-Saharan Africa -0.445 -0.398 -0.251 -0.429(0.481) (0.473) (0.500) (0.456)
Latin America 0.134 0.134 0.260 -0.025(0.422) (0.433) (0.405) (0.384)
Asia -0.008 -0.181 0.151 0.136(0.367) (0.359) (0.379) (0.351)
Control variables Yes Yes Yes YesWald exogeneity test 7.767*** 2.809*** 7.063*** 6.486**Countries 132 128 132 133Observations 4592 4452 4592 4631
Notes: The dependent variable is a binary variable that takes a value of one for conflicts with1000 or more battle-related deaths over time and zero otherwise. All the regressions includethe full set of control variables described in the text. Robust standard errors adjusted forclustering in parentheses. * Significant at 10% level, ** significant at 5% level, *** significantat 1% level.
48
Table 10: Robustness analysis: Alternative estimation methods. Minimum chi-squaredestimator (Newey, 1987).
(1) (2) (3) (4)
Overall index of globalization 1.880***(0.326)
Economic globalization 1.220***(0.300)
Social globalization 1.790***(0.334)
Political globalization 1.631***(0.298)
GDP per capita (t-1) -0.526*** -0.402*** -0.736*** -0.321***(0.076) (0.069) (0.121) (0.046)
Population (t-1) 0.201*** 0.255*** 0.314*** -0.012(0.022) (0.027) (0.032) (0.041)
Mountainous terrain 0.296*** 0.275*** 0.285*** 0.282***(0.024) (0.024) (0.024) (0.022)
Noncontiguous state 0.732*** 0.665*** 0.790*** 0.778***(0.079) (0.079) (0.084) (0.080)
Fractionalization 1.535*** 1.282*** 1.714*** 1.594***(0.132) (0.133) (0.144) (0.130)
Democracy -0.568*** -0.345*** -0.620*** -0.541***(0.084) (0.079) (0.098) (0.079)
Wald exogeneity test 83.17*** 26.74*** 64.83*** 50.97***Countries 137 132 137 138Observations 4779 4600 4779 4818
Notes: The dependent variable is a binary variable that takes a value of one for conflictswith 1,000 or more battle-related deaths over time and zero otherwise. * Significant at 10%level, ** significant at 5% level, *** significant at 1% level.
49
Figure 1: Globalization and GDP per capita.
50
Figure 2: Density functions of the overall index of globalization.
51
Figure 3: Number of civil wars from 1970 to 2009.
52
Figure 4: Partial regression plot: Overall index of globalization and remoteness.
53
Figure 5: Partial regression plot: Economic globalization and remoteness.
54
Figure 6: Partial regression plot: Social globalization and remoteness.
55
Figure 7: Partial regression plot: Political globalization and remoteness.
56