Abstract
The proliferation and scope expansion of regional organizations (ROs) is one of the most prominent features in contemporary international politics. In particular, many regional economic organizations (REOs) have expanded into the security realm. This trend has often resulted in an overlap with regional security organizations (RSOs) already in place. This study sheds light on this phenomenon by identifying the conditions under which REOs trespass into the security policy domain despite the fact that preexisting RSOs already fulfill security functions. We argue that the presence of strategic rivalries is an important driver of the creation and depth of organizational overlap through scope expansion. Specifically, RSOs that include significant interstate rivalries propel a subgroup of like-minded states to advance and deepen security cooperation through their existing REOs. Using an original data set of security cooperation within economic and security ROs and a quantitative analysis, we find substantial support for this argument.
Keywords
The proliferation and scope expansion of regional organizations (RO) is one of the most prominent features of the contemporary international environment. Many ROs that were initially designed to promote economic cooperation have moved into the security sphere, thus drifting from their purported core mandate (Haftel and Hofmann 2017). Doubly surprising, perhaps, these regional economic organizations (REOs) have entered into the security domain even in instances in which regional security organizations (RSOs), which include some of the same member-states, were already in place. In so doing, such REOs purportedly encroach on the turf of their security-oriented counterparts, with potentially significant consequences for their organizational performance and regional peace and stability (Hofmann forthcoming). This phenomenon is illustrated by REOs such as West Africa’s Economic Community of West African States (ECOWAS) and its overlap with the security-oriented African Union (AU), the Union of South American States (UNASUR) that moved into the Organization of American States (OAS) security “territory,” or the Gulf Cooperation Council’s (GCC) overlapping security agenda with the League of Arab States (LAS).
To be sure, not all REOs that have an overlapping membership with RSOs address security matters. And among those that do, some engage in such cooperation more deeply than others. For example, most member-states of ECOWAS, the Common Market for Eastern and Southern Africa (COMESA), and the Arab Maghreb Union (AMU) are also members of the AU, but they espouse deep, shallow, and no security cooperation, respectively. The former is heavily involved in peace operations through the ECOWAS Monitoring Group (ECOMOG). On the other hand, COMESA only provides for regular meetings among security officials and a conflict early warning system (CEWS), and the AMU has no security components at all. In light of this variation, this article takes up two related questions: (1) why do REOs trespass into the security policy domain despite the fact that preexisting RSOs already fulfill security functions? and (2) given such overlap, what accounts for the depth of REO security cooperation?
To answer these questions, we introduce insights from research on strategic rivalry (Bremer 1992; Vasquez 1993; Klein, Goertz, and Diehl 2006; Colaresi, Rasler, and Thompson 2008) into the study of (overlapping) international organizations (IOs; Alter and Raustiala 2018; Jupille, Mattli, and Snidal 2013; Hofmann forthcoming). We argue that strategic rivalries within RSOs trigger a subgroup of member-states, which either include only one of the rivals or neither of them, to look for alternatives. Frustrated with rivalrous tensions that create uncertainty within the RSO that can potentially hamper the conduct of multilateral security policies, this subgroup looks for other forums where trustful relations are more readily available. In instances of partial membership overlap between an RSO and an REO, the latter, indeed, offers such a constructive alternative. We thus expect, all else being equal, REOs to trespass into the security realm when an overlapping RSO is mired by strategic rivalries and the overlapping REO is not. These conditions are especially conducive to deep, rather than shallow, security cooperation within the REO’s institutional framework.
A focus on strategic rivalries can, for example, account for the reality that regional powerful actors such as Nigeria, which faced rivals such as Morocco, Algeria, Gaddafi’s Libya and South Africa, instead of choosing to invest its military in the OAU/AU, pushed for security cooperation within ECOWAS. On the other hand, given the rivalry between Algeria and Morocco, these two countries did not opt to militarize the AMU, in which both are leading members. In the Americas, this theoretical lens highlights UNASUR as a political project of several like-minded left-leaning South American governments that wanted to create a security institution independent of a US-induced uncertain security agenda. Recently, many UNASUR member-states have experienced a political turn to the right, and consequently, some of them no longer perceive the United States as a rival but rather as an ally. They have therefore refocused their attention to the OAS while suspending their membership or withdrawing from UNASUR.
We employ quantitative methods to evaluate this theoretical argument. We first develop a metric that gauges the degree of security cooperation within REOs and RSOs and present a new data set of numerous ROs around the world on this institutional feature. We then map overlapping REOs according to their value on strategic rivalry and the corresponding value of their overlapping RSOs. This exercise offers face-value support for our theoretical framework. Next, we use multivariate regression techniques to test the effect of existing RSO performance and the severity of RSO internal rivalry on the extent of security cooperation within REOs. This analysis further corroborates our key argument: under conditions of overlapping membership, strategic rivalries in the RSO are associated with greater REO security cooperation.
These findings offer important lessons to research on international cooperation. First, by linking strategic rivalry to regional (economic) IOs, they reinforce the need to do away with the artificial and now dated assumption that the economic and security spheres are mutually detached and independent of each other. Second, they demonstrate that the assumption that ROs operate in an institutionalized vacuum is untenable and that one ought to recognize the ubiquity of organizational complexity and, in turn, investigate its causes and consequences (Alter and Raustiala 2018; Hofmann 2011; Panke and Stapel 2018a). Relatedly, they further advance recent research on institutional choice (Jupille, Mattli, and Snidal 2013; Lenz and Viola 2017), by specifying additional conditions under which institutional selection and change are likely to take place. We underscore the possibility that states will pursue institutional change despite the existence of a status quo institutional alternative. Thus, not every newly created overlapping regional security structure needs to become a new focal IO. Given the significant impact of ROs on regional peace and stability (Haftel 2007, 2012), these conclusions have important implications for our understanding of intra- and interstate crisis management and conflict resolution.
This article proceeds as follows. The next section develops the theoretical framework and derives the main hypothesis with respect to the consequences of strategic rivalry for the emergence of overlapping security substructures within REOs. The third section presents the research design with particular emphasis on the definition, measurement, and coding of the dependent and independent variables. It also takes a first look at the relationships between these variables. The fourth section presents the results of the statistical analysis. The final section concludes.
Theorizing Strategic Rivalries and Security Cooperation within REOs
States sometimes create organizational structures that overlap with existing organizations. They do so despite the fact that encroaching on others’ organizational mandate can be a costly affair for member-states: new expertise needs to be institutionalized (Littoz-Monnet 2017), potential fallouts among member-states can arise (Hofmann 2009), new (potentially conflicting) rules and obligations need to be agreed upon (Abbott et al. 2015, 7; Panke and Stapel 2018b), additional policy coordination needs to be scheduled because of “different timetables” (Raustiala and Victor 2004, 280), and member-states might even lose more national autonomy (Arel-Bundock 2017).
We nonetheless observe this phenomenon, in particular with REOs that create security structures in the shadow of existing RSOs. We extend existing theory to emphasize that states that experience rivalries firsthand sometimes jeopardize potential cooperative gains in performing RSOs, chart into uncertain institutional territory, and increase transaction costs between ROs in order to expand the mandate of an REO. We elaborate on the claim that rivalry dynamics within the existing RSO is a major driver of REO expansion into the security domain, thereby creating overlapping security capabilities and responsibilities.
Why are states willing to duplicate their organizational efforts and commitments? Why do they sometimes create additional autonomous organizational structures instead of reforming existing ones? We build on international security and regime complexity scholarship to develop an argument rooted in strategic rivalry dynamics.
Strategic rivalry is a competitive relationship that exists between political leaders, but it can also transcend single leaderships and preoccupy states over an extended period of time (Colaresi and Thompson 2002, 275). Mistrust nurtures and sustains strategic rivalries; mistrust can be rooted in historical experience and perceived threats (Colaresi and Thompson 2002, 263), territorial or status claims or other political, ideological or regime-related tensions (Bremer 1992; Mitchell and Thies 2011; Dreyer 2012). Many rivalries occur at the regional level, where states compete over how to organize the region or over regional hegemony (Colaresi, Rasler, and Thompson 2008, 77-81; Thompson 1995, 205). Rivals primarily observe each other’s capabilities and intentions. Extant research shows that rivals are conflict-prone and that they might resort to violent disputes, crisis, or even war if they cannot resolve their political disagreements peacefully (Dreyer 2010, 779), whereby territorial claims are the most conflictual and enduring (Vasquez 1993). The concept has been widely applied in conflict studies but has hardly made inroads into the study of IOs (not to mention economic ones).
How do rivalries impact ROs? We show how competitive relationships give voice to dissatisfaction and create uncertainty not only between adversaries but also among their partners within their ROs. Rivalrous relations can push some states to transform parallel regional arrangements to become either overlapping or nested. To the extent that strategic rivalries within institutionalized multilateral cooperation agreements exist, they can produce two kinds of processes. Looking, first, at rival—nonrival dynamics, we argue that nonrivalrous members can become dissatisfied with the ongoing political battles in their RO and potentially even worried and uncertain about their RO’s performance in the future. For nonrivalrous member-states, dancing around political tensions can be a time-consuming or simply unpleasant diplomatic exercise. While they have committed to and invested in the RO materially and normatively, which in turn reduces the likelihood that they will pull out from the organization completely, they might look for institutional alternatives that are not costly to build up. Other ROs that are active in multilateral policy-making can be such low-investment springboards.
Second, rival-to-rival dynamics can push rivals to use other ROs to project and realize their preferences. Rivals within an RO can push each other to look for institutional alternatives where their preferences or status are not challenged. For example, within asymmetric rivalries, the more powerful rival(s) might dominate the RO’s operations, sidelining rival member-states and hence fueling the potential for further political discord (Lepgold 1998). No matter symmetric or asymmetric rivalries, only those rivals who are members of parallel regional cooperation agreements can use those as reliable platforms. This occurs in particular if these parallel arrangements have no rivalries within them. Insofar as minor rivalries exist in the parallel RO, scope expansion is still possible but will most likely be shallow.
Rivalries need not necessarily impact actual RO cooperation patterns, performance, and efficiency. Many ROs do not only encourage rivalrous member-states to exchange necessary information, sometimes through informal channels, but also to formulate and implement multilateral policies with bureaucratic oversight (Jones 2007; Greig and Diehl 2006). Also, given the various sources of rivalries, some rooted in the countries’ past, they do not a priori have to inform different policy goals among rivals. Instead, rivals might share an interest in stabilizing their region and hence engage in multilateral cooperation. For example, rivals might cooperate to attract donors and outside investment (Dietrich 2016).
Unsurprisingly, perhaps, RSOs—organizations with a mission to advance peace and stability—can include states with a history of rivalry that nonetheless engage in institutionalized security-providing cooperation (Snyder 1997; Schroeder 1976, 256–257). RSOs provide a forum where states can discuss their security, exchange information, train their militaries together, and coordinate peace operations, if needed. These activities require scarce, expensive, and interoperable resources, as well as a strategic commitment. States that engage in multilateral crisis management, either among their membership or abroad, agree to participate in institutional decision-making bodies, invest in military and civil–military expertise, negotiate procedures to share national (confidential) information, and mediate between conflicting parties. They may also formulate a common security strategy, earmark national military forces to multilateral ones, and coordinate military exercises.
Strategic rivals have joined one and the same RSO, for example, in the AU, OAS, and or the LAS. Distrust and dissatisfaction among certain member-states did not necessarily prevent the RSO from remaining operative and conducting peace operations or mediation. In one example from the Americas, the OAS involvement in Haiti has been ongoing while the US and many Latin American countries did not see eye-to-eye on security matters (Herz 2011). In Africa, rivals such as Algeria and Nigeria cooperated in the AU’s AMISOM operation, and lingering rivals such as Kenya and Uganda provided troops for this multilateral force (Williams 2018). In the Middle East, various rivals that are members in LAS agreed to suspend Syria’s membership in this RO in 2011. They also unanimously supported United Nations (UN) resolution 1973, which arguably prevented a Russian and Chinese veto in the UN Security Council (UNSC), paving the way for the UN-sanctioned North Atlantic Treaty Organization (NATO)-led intervention in Libya.
At the same time, mutual mistrust creates political tensions within the RSO. The sharing of classified information is a good case in point—states that do not trust each other and expect to engage in political or military conflict in the future will be hesitant to share information that has strategic value. This leaves rivals wondering about their respective intentions and nonrivals frustrated with the possible emerging political tensions that permeate the hallways in the IOs’ bodies. So why are member-states dissatisfied with internal rivalries not leaving the RSO? The incentive to exit an RSO is weak because for rivals and nonrivals alike, membership has arguably taught them some restraint (Ikenberry 2000; Pevehouse and Russett 2006). It also fulfills a variety of other purposes (Haftendorn, Keohane, and Wallander 1999), such as providing a framework to face a common threat (internal or external) to the region (Krebs 1999). For rivals, in addition, staying within the RSO allows a check on rival(s)’ institutional and strategic movements and security policies.
One potential solution to this predicament could be to reform the RSO in order to increase its flexibility and functionality. Efforts to modify an RSO and to prevent some security cooperation from shifting to an existing REO are doomed to fail, however. Given that rivalries are a problem that lies within IO membership and that RSOs are resistant to qualified majority voting (Hooghe and Marks 2014), the RSO’s institutional design is unlikely to be adjusted to accommodate this particular membership constellation. Any institutional reform that could either neutralize rivals’ formal decision-making powers or give one rival more power over the other is likely to be blocked, as exemplified by the enduring yet futile reform discussions within the UNSC. In addition, emphasizing rivalrous cleavages and putting them on the RSO agenda as a reason for contemplating reform can magnify these existing rivalries or make rivalries the dominant cleavage within the RSO, overshadowing others. Setting up a completely new RSO with just a subgroup of states is also unlikely as this can be a very costly endeavor, especially in such a policy domain as international security, which heavily depends on expensive and scarce resources and assets.
As rivalries hardly lend themselves to be bargained over or talked away so that RSO reforms could become an option, having institutional alternatives that are rivalry-free become desirable to rivals and nonrivals alike. 1 They allow states to discuss regional security projects and give voice to their dissatisfaction without the “sword of Damocles” that are rivalries hanging over their heads. Creating additional institutional structures is conducive to the demonstration of disagreement, exhaustion, or competitiveness, while at the same time trying to settle political tensions and disagreements diplomatically (Johnston 2001; Snidal 1985).
RSO member-states seek institutional alternatives that are not riddled with rivalries and possibly not too costly to erect. A facilitating factor that provides them with an infrastructure to do so are preexisting institutional structures such as an REO whose scope can be expanded toward a new policy domain (Haftel and Hofmann 2017). REOs possess general assets that can be transferred to additional functions so that states only have to invest in domain-specific assets and capabilities (Wallander 2000). For example, while the “OAS has been the most influential and hands-on IO in South America for decades” (Allen 2010, 45), it also has been riddled with tensions, as mentioned above. Some Latin American members even declared the United States to be a threat to their national security (Weiffen, Wehner, and Nolte 2013, 382). Consequently, several left-leaning Latin American states under the leadership of Brazil’s Luiz Inácio Lula da Silva started expanding the scope of UNASUR into the security realm (Sanahuja 2012). This occurred despite their shared OAS membership and continued engagement in this IO. The case of UNASUR is a good case in point, as it illustrates that these moves can be reversible as well: with many right-wing governments taking office in Latin America, rivalrous tensions with the United States have subsided and UNASUR is gradually imploding (Paraguassu 2018).
On a shallow level, REO scope expansion can provide one RSO rival with an institutional alternative in which it can organize security policy multilaterally without losing sight of its rival in the RSO. On a deeper level, the absence of rivalries in REOs is an important factor to explain why they are willing and able to expand the organization’s scope into an already occupied policy domain. REOs that are rivalry-free include member-states that share common cooperative interests, might seek regional status, and understand the REO to be an expression of shared ideas and identity of a regional political community (Cho and Park 2014; Börzel 2016). This is conducive to scope expansion, in particular as the importance of cohesion increases as organizations deal with more contentious issues (Jupille, Mattli, and Snidal 2013, 47). 2
Rivalries within an RSO can even externally reinforce REO cohesion and ultimately trust among its membership. The existing RSO, to which at least some REO member-states belong, can serve as “the other”—often bigger and imposing RO—based on which REO member-states want to distinguish themselves (Mercer 1995). RSO rivalry dynamics can strengthen the political bonds between REO members. This increases the chances that REO member-states agree to mirror at least some of the RSOs institutional structures, despite the uncertainty of whether the new security structures will perform well. For these reasons, it becomes more likely that REOs will invest deeply in security structures. This logic leads to the following key hypothesis:
As a further elaboration of this hypothesis, it is useful to contrast it with three other possible combinations of RO rivalry under conditions of overlap, depicted in Figure 1. We maintain that the conditions captured by Hypothesis 1, located in Quadrant II, should result not only in a high probability of an REO expanding into the security domain, but deeply so. Moreover, we argue that such expansion is more likely to occur under these conditions, compared to the other three scenarios.

Theoretical expectations regarding rivalry, overlap, and regional economic organization security cooperation.
Starting with the situation in which both the RSO and the REO are nonrivalrous, located in Quadrant I, REO member-states should be satisfied politically with the RSO and feel less pressure to expand security cooperation in the REO. However, it is still possible that REO members will create security structures. Here, RSO rivalry is not the trigger for REO scope expansion, but rather internal REO dynamics that build on institutional capacity (Haftel and Hofmann 2017) or ideological congruence. With respect to the latter, REO member-states wanting to foster a political union might push for scope expansion, even if this entails the creation of a more complex organizational constellation for them in the form of organizational overlap (Hofmann 2013). 3
Moving to Quadrant III, we should observe no REO scope expansion if there is no rivalry in the RSO but rivalry in the REO, for the very reasons underpinning our hypothesis (turned on its head). After all, the existence of rivalry dynamics within an REO makes it very unlikely for REO member-states to agree to push for a possibly costly institutional endeavor. Interestingly, though, there is no instance of this combination in our sample (see Figure 2 in the next section). That is, we were not able to identify a case in which the REO is rivalrous but the RSO is not.

REO/RSO rivalry dynamics and REO security depth.
Finally, should both the REO and the RSO experience rivalry dynamics, depicted in Quadrant IV, expanding the scope of the REO into the activities of the RSO is also still theoretically possible, but very unlikely. Only in circumstances in which the rivalry within the RSO is so debilitating or targets all RSO member-states that are also part of the REO, should we envisage that the RSO rivalry trumps the REO rivalry. In these instances, REO member-states might leave their political competition aside and create additional institutional structures. However, given internal REO rivalries, it is unlikely that these security structures will be meaningful.
In summary, we submit that under conditions of RSO-REO membership overlap, the latter is most likely to embrace security cooperation when it does not suffer from internal rivalries but the RSO does. We turn to an empirical investigation of this hypothesis next.
Research Design
This study examines the sources of security substructures within REOs and their relationship to preexisting RSOs. The empirical analysis is therefore conducted at the regional level, defined by organizational membership. Given that ROs often evolve and change over several years (especially with respect to the implementation of agreements), annual observations might artificially inflate the number of observations without adding much variation to the data. We therefore code the various ROs in five-year intervals from 1982 to 2012. All other variables are also constructed in five-year increments so as to fit this setup. 4 For similar reasons, all independent variables are lagged one five-year period. This setup is compatible with our theoretical expectations, which envision a sequential process that leads from rivalry, and other explanatory factors, to organizational scope expansion. For instance, values on the dependent variable for the 2008 to 2012 period correspond to values on independent variables for the 2003 to 2007 period.
The values of the main dependent variable are ordinal (see below). They are therefore estimated with an ordered probit model with robust standard errors clustered by REO. To account for temporal dynamics, all models include time dummies. To assess the robustness of the results, we transformed the main dependent variable into a binary one. We test this variable with random-effects and fixed-effects logit models. The rest of this section briefly discusses the dependent variable, elaborates on the independent variables, and outlines the control variables. 5
Dependent Variable
This study accounts for the depth of security cooperation embedded in REOs in instances in which RSOs overlap with the REO’s membership. The variable of security cooperation is constructed with reference to five types of security organs and activities, grouped in two categories: shallow and deep. 6
Beginning with the former, superficial levels of security cooperation entail relatively modest institutionalization. In these cases, security substructures aspire to assure member-states against aggression and conflict as well as facilitate the exchange of information and perspectives between relevant stakeholders. 7 Such bodies may be instrumental in building mutual trust and confidence among the members and foster the coordination of security policies within the region and with respect to external actors (Lundgren 2017).
We consider three specific institutions in this context: (1) security commission—a body that facilitates the coordination of security and defense policies. It often involves officials from the ministry of defense or military personnel and mostly tackles technical issues, (2) ministerial council—an institutional setup that brings together top-level government officials, that is, either ministers of defense, security, or foreign affairs or heads of states. Generally, this body engages in decision making and deals with controversial political issues, and (3) CEWS—an arrangement designed to collect information on domestic and international conflicts, and in so doing to prevent escalation of exiting strains. It usually involves a network of offices in different parts of the region that analyze and report tensions, incidents, and other security problems.
Deeper levels of security cooperation envisage the execution of common security policies in the event that violent conflict erupts. Such cooperation requires greater delegation of authority to the RO and an agreement regarding the coordination of security strategy and operational procedures. This is vital for an effective performance of the organization, when called to action. This category contains two security substructures: (1) designed military exercise—an agreement on a framework of regular joint military exercises of either national units or regional forces and (2) designed military operation—an agreement to establish a multinational military force for the purpose of military cooperation and peace operations. This may involve the establishment of a central command or other headquarter structures that engage the implementation of such activities.
With these indicators and categories, we produce the ordinal variable REO Security Depth. It scores two for deep security cooperation, that is, if either military exercises or operations are present; one for shallow security cooperation, that is, if an REO has either a commission, or a council, or a CEWS; and zero if the organization has none of the five indicators and thus no security cooperation. To assess the effect of the covariates on REO security cooperation as such, we construct an alternative variable, labeled REO Security Cooperation. It is coded one if the REO engages in either low or high levels of security cooperation and zero otherwise.
As already mentioned, the unit of analysis is the REO, which is one type of IO and should therefore have a continuous institutional framework, a formal structure, and at least three member-states (Pevehouse, Nordstrom, and Warnke 2004). It is “regional” in the sense that membership in these organizations is restricted to geographically proximate states and “economic” to the extent that the promotion of economic policy cooperation among the organization’s members is one of their original and primary (but not necessarily exclusive) goals (Mansfield and Milner 1999; Haftel 2013). Here, consistent with our theoretical framework, most REOs started with an economic mandate and then expanded into security cooperation. Thus, there is a clear increase in the value on REO Security Depth as an organization ages. The mean value on this variable for REOs in their first ten years of existence is about 0.26. This value rises to 0.47 for REOs in existence from twenty to thirty years and to 0.76 for those that exist forty years or more.
Surveying most, if not all, existing international economic agreements and organizations, we identified twenty-eight REOs that correspond to these criteria and for which sufficient information on institutional design is available, listed in Table 1. The sample is very comprehensive, spans all continents, and includes the majority of states worldwide. To code the five indicators pertaining to security cooperation, we surveyed all the available agreements and protocols and recorded the articles that deal with security matters. These texts provide a very good sense of the kind of security arrangements and institutions. In several instances, the actual texts were not available, and in others, important details were hammered out in declarations, memorandums, and decisions rather than in formal treaties. In such cases, we surveyed relevant primary documents and secondary sources, which included the REOs’ own websites and scholarly articles and reports, thereby depicting a complete picture of planned security cooperation.
REOs and Overlapping RSOs,1982 to 2012.a
aFull names of all regional organizations referred to in this table and elsewhere in this article are listed in the Online Appendix.
REO and RSO Overlap: Establishing the Baseline
Our point of departure for this analysis is that REOs expand into the security domain even in instances in which overlapping RSOs already exist. Before turning to our main independent variables, we take the initial step of demonstrating (rather than assuming) this reality. This is required because simple functional logic suggests that REOs that operate in a region where meaningful security organizations already perform well will be less likely to expand into the security realm, compared to REOs that work in a region with no such RSOs. For example, scholars who observed the transformation of NATO after the end of the Cold War have argued that “if the marginal costs of maintaining an existing institution outweigh the considerable costs of creating an entirely new set of norms, rules, and procedures, states will choose to sustain existing arrangements rather than abandon them” (Wallander 2000, 706).
Given this logic, one might expect that only when an RSO does not perform adequately—for example, is not willing or capable to send out troops to address conflicts—we should expect that member-states who want their organization to fulfill its mandate, look for other institutional solutions, such as expanding the scope of an existing REO into security matters (Jupille, Mattli, and Snidal 2013, 45-47; Koremenos, Lipson, and Snidal 2001, 762; Urpelainen and Van de Graaf 2015). In contrast, our theoretical framework indicates that creating security structures within an REO is possible even when it overlaps with a functional and meaningful RSO.
Empirically establishing this baseline requires us to identify “meaningful” RSOs and determine whether they overlap with a given REO. With respect to the first task, we begin with the design of security cooperation and employ the same coding scheme described for the dependent variable. We consider only those ROs that score two on REO Security Depth as “meaningful RSOs.” Here, we include those REOs that embrace deep security substructures as well as RSOs that do not address economic issues directly, such as NATO and LAS. We identify nine RSOs in total. To the extent that the majority of a given REO are also members of at least one RSO, the variable RSO Overlap scores one, and zero otherwise. Table 1 lists all the REOs included in our sample, the RSOs with which they overlap (if at all), and the years for which overlap was present.
Arguably, looking at institutional design alone is insufficient. To the extent that some ROs have ambitious plans that remain on paper (Gray 2018; Gray and Slapin 2012; Haftel 2013), other organizations in the region may fill the void and take up pressing security challenges. As a first cut into this issue, we distinguish between high-performing RSOs and those that are less active. The former have conducted at least two military operations or employed at least 4,500 personnel in a five-year period. 8 Thus, High-Performing RSO Overlap is a categorical variable that scores one for high-performing RSOs that overlap with a given REO and zero otherwise. We identify several high-performing RSOs that overlap with REOs in our sample, listed in Table 1. With this in mind, we now turn to our main explanatory variables.
Key Independent Variables
REO’s internal rivalry
Using CRT’s list of rivalries, we identified all the strategic rivalries within the REO. We then distinguished between major and minor rivalries. The former involves the regional hegemon or, if no clear hegemon can be identified, the more powerful state or states. REOs that have one or more major rivalries or at least two minor rivalries during a given five-year period score one on the variable REO Rivalry and zero otherwise. The exclusion of REOs with one minor rivalry from this definition ensures that we capture security dynamics that are likely to have an impact on regional political processes. 9 As the various examples mentioned in previous sections illustrate, many REOs are rivalrous, while others are not. In addition, some REOs have turned from rivalrous to nonrivalrous, or vice versa. For example, ECOWAS and CACM were rivalrous until the early 1990s but then became free of major rivalries. This timing corresponds to ECOWAS’s establishment of ECOMOG. On the other hand, APTA turned from nonrivalrous to rivalrous in the early 2000s, once China (a strategic rival of India) joined this organization. We thus have a meaningful regional and temporal variation on this variable.
Rivalry in an overlapping RSO—our main hypothesis shifts the focus to characteristics of the overlapping RSO in relations to the REO. In line with the theoretical framework, we expect RSOs with strategic rivals to find it more difficult to accommodate all member-states political outlooks. Thus, RSOs that contain strategic rivalries are expected to push a subgroup of like-minded states to advance security cooperation within their existing REOs. We capture these dynamics by, first, determining whether the overlapping RSO is rivalrous. We use the same criteria employed with respect to REO Rivalry for the coding of this variable, labeled RSO Rivalry. Here, however, we are presented with an interesting dilemma: NATO has only one “minor” rivalry, the one between Greece and Turkey, which would render this RSO nonrivalrous. Research suggests, however, that this may be a special case in which a minor rivalry has a significant impact on an RSO (Krebs 1999). We therefore report the results with NATO coded as either rivalrous or nonrivalrous. 10
Next, because we expect the strongest effect to take place when the REO is rivalry-free (which is not always the case, as pointed out above), we calculate the difference between RSO Rivalry and REO rivalry. This variable, labeled Rivalry Gap, takes the value of 1 if the RSO is rivalrous and the REO is not; it takes the value of zero either if both ROs are rivalrous or both are nonrivalrous. 11 Models estimating the effect this variable include only REOs that overlap with RSOs and exclude the rest. In addition, given that we use REO rivalry and RSO Rivalry to compute Rivalry Gap, we exclude the former two variables from models including the latter. As we show later, substituting Rivalry Gap with REO Rivalry and RSO Rivalry does not affect our results.
As a first empirical cut into our primary hypothesis, Figure 2 locates all REOs that engage in security cooperation and overlap with an RSO in the quadrants presented in Figure 1. The acronym of the REO is bolded and the acronym of the corresponding overlapping RSO is in brackets. The reality depicted in Figure 2 nicely comports with our theoretical expectations. Four of the five REOs with high levels of security cooperation listed in Quadrant II are not rivalrous and overlap with a rivalrous RSO. The odd REO out of this category, the EU, falls in Quadrant I because NATO is coded as nonrivalrous. As mentioned above, however, the EU is probably the most cohesive RO that exists and the EU-NATO dynamics are actually quite consistent with the theoretical framework. The creation of the EU’s Common Security and Defense Policy (CSDP) is a prime example of congruent national governments’ preferences (Hofmann 2013). In the end of the 1990s, all major governments in Europe were ideologically aligned to push for the European project to become more political, including security organs, and created the CSDP in spite of their NATO membership and US objections.
Looking next at REOs with low levels of security cooperation, four of the seven REOs that overlap with an RSO fall in Quadrant II as well, as expected. 12 Two rivalrous REOs, CAN and COMESA, are located in Quadrant IV, suggesting that low levels of security cooperation are still possible in rivalrous REOs, but only if the RSO is also rivalrous. There is no instance of security cooperation in a rivalrous REO that overlaps with a nonrivalrous RSO in our sample. To be sure, the relationships depicted in Figure 2, while providing an initial support for the hypotheses, are vulnerable to the risk of spuriousness. We therefore subject them to a more rigorous statistical test that accounts for several alternative explanations, discussed in the next section.
Alternative Explanations
Various other factors might account for security cooperation within REOs and REO-RSO overlap. 13 To begin with, regions that suffer from militarized conflicts might require more robust security institutions to respond to such crises, compared to more peaceful regions. On the other hand, consistent with our emphasis on the detrimental effect of rivalry, armed conflict might hamper security cooperation. We account for these opposing arguments with the level of regional violence, measured with the number of armed conflicts as reported in the Uppsala Conflict Data Program’s Armed Conflict Dataset (Gleditsch et al. 2002; Themnér and Wallensteen 2012). The variable Conflict is a sum of all violent conflicts within member-states and between member-states of a given REO in a given five-year period.
Another commonly cited explanation for regional cooperation is the presence of a regional hegemon. Such an undisputed leader may have the capabilities and will to underwrite security cooperation. As we argued above, however, the willingness of the hegemon to engage in security cooperation, as well as the willingness of other members to trust the hegemon, is likely to depend on the political relationships among the REO’s member-states. We measure this factor, labeled Hegemony with the concentration ratio
The flipside of rivalry is shared political interests. When member-states see international issues eye-to-eye, they may be more likely to cooperate on security matters, compared to those that have more divergent interests (Hawkins et al. 2006; Hofmann 2013). We account for this logic with Ideal Point STD, which is the standard deviation of the members’ ideal point distance, based on UN General Assembly (UNGA) voting patterns (Bailey, Strezhnev, and Voeten 2017). Given the theoretical expectations, higher values on this variable should be associated with lower security cooperation.
Next, we control for Delegation, which refers to the empowerment of IOs to fill-in relevant agreements, provide expert information, select or prioritize tabled proposals, propose policy initiatives, and make and enforce decisions (Hooghe et al. 2017). This variable takes into account the transfer of agenda-setting or decision-making authority to four corporate bodies—a council, a secretariat, an assembly, and consultative bodies—with respect to six competencies. It is noteworthy that Delegation takes into account only the organization’s general institutional structure and disregards the more specialized bodies and arrangements that tackle security matters.
Previous studies suggest that the propensity to combine economic and security cooperation is more pronounced in the developing world (perhaps with the exception of the EU). We account for this factor with a binary variable labeled South–South REO. It scores one for REOs that include only developing countries and zero otherwise. Finally, it appears that more members are associated with greater security cooperation, perhaps because larger ROs may be able to amass greater capacity required for security cooperation. We thus include Members, which is a count of the REO’s member-states.
Results
Tables 2 and 3 present the results of the statistical analysis. Models 1 and 2 in Table 2 report two baseline models that focus on the effect of RSO overlap and internal REO rivalry on REO Security Depth. The first model includes the variable RSO Overlap, which is substituted with High-Performing RSO Overlap in the second model. Models 3 and 4 shift the focus to interorganizational dynamics, our main theoretical interest, and thus include Rivalry GAP. NATO is coded as a nonrivalrous RSO in the third model and as a rivalrous RSO in the fourth one. Table 3 presents four models that assess the robustness of the results. Model 5 in this table substitutes Rivalry Gap with REO Rivalry and RSO Rivalry. Models 6 to 8 substitute REO Security Depth with the binary dependent variable—Regional Security Cooperation. Table 4 reports substantive effects for statistically significant variables, based on the results reported in Table 2.
Ordered Probit Estimates of the Sources of REO Security Depth, 1982 to 2012.
Note: Standard errors are clustered and robust. All models include time fixed effects. Figures in parentheses are z statistics. REO = regional economic organization; RSO = regional security organization; NATO = North Atlantic Treaty Organization.
*p < .1.
**p < .05.
***p < .01 (two-tailed).
Additional Models of the Sources of Security Substructures within REOs, 1982 to 2012.
Note: Standard errors are clustered and robust. All models include time fixed effects. Figures in parentheses are z statistics. REO Security Depth is the dependent variable in model 5. REO Security Cooperation is the dependent variable in models 6 to 8. REO = regional economic organization; RSO = regional security organization.
*p < .1.
**p < .05.
***p < .01 (two-tailed).
Predicted Probability of REO Security Depth Due to Changes in Statistically Significant Independent Variables.
Note: Effects are generated using Spost (Long and Freese 2005). Variables except the variable of interest are held at mean values. For continuous variables, high and low values are one standard deviation above and below the mean, respectively. For binary variables, high and low values equal one and zero, respectively. The results are based on models 1 and 3 (for rivalry gap and ideal point STD) in Table 2. Because Spost cannot accommodate time fixed effects, a Year variable replaces the time fixed effects (statistical and substantive results are very similar). REO = regional economic organization.
Beginning with internal political dynamics, REO Rivalry has a strong negative effect on the prospects of security cooperation within REOs. The coefficient of this variable is highly statistically significant in all models irrespective of the specification. The substantive effect of this variable is sizable as well. As Table 4 shows, moving from zero (nonrivalry) to one (rivalry) decreases the probability of any security cooperation by 24 percent and puts the likelihood of deep security cooperation at virtually zero. Similarly, the coefficient of Ideal Point STD is negative and statistically significant in most models. This suggests that similar voting patterns in the UNGA are associated with more robust security cooperation among REO members. Substantively, moving from one standard deviation above the mean to one standard deviation below the mean increases the probability of deep security cooperation by 18 percent.
Consistent with these findings, the most notable cases of deep security cooperation, e.g., the EU, ECOWAS, and Southern African Development Community (SADC), do not suffer from major political rivalries. On the other hand, such rivalries and conflicting interests were significant impediments to security cooperation in REOs such as SAARC and CAN. Thus, the existence of intraregional political rivalry appears to be an important hurdle to a meaningful expansion of REOs into the security realm. The results provide strong support for the notion that diverging political interests and strategic rivalries affect the prospects of security cooperation and its depth.
Turning to organizational overlap, it is apparent that the presence of an RSO does not prevent REOs from expanding into the security domain. Surprisingly, perhaps, the estimate of RSO Overlap is positive, indicating that, if anything, REOs that overlap with existing RSOs are more likely to engage in security cooperation. The estimate does not reach conventional levels of statistical significance, however. The results on High-Performance RSO Overlap are more consistent with conventional expectations. The negative coefficient suggests that high-performing RSOs discourage security cooperation within overlapping REOs. Again, though, the estimate is not statistically significant. In and of themselves, then, existing RSOs in a given region do not seem to have a systematic effect on the spillover of economic IOs into matters of the regional security.
As we discussed earlier, one should specify the conditions under which RSOs affect security cooperation within overlapping REOs. We argued that rivalry within an existing RSO will propel a smaller group of like-minded states to embark on security cooperation within an existing REO. Models 3 and 4 in Table 2 provide strong empirical support for this conjecture. As expected, Rivalry Gap is positive and statistically significant, irrespective the coding of NATO as a rivalrous RSO (although the results are stronger when it is treated as such). Substantively, an increase from zero (indicating that either both ROs are rivalrous or both are nonrivalrous) to one (indicating that the overlapping RSO is rivalrous but the REO is not) increases the probability of any security cooperation by more than 35 percent and of a high level of security cooperation by about 13 percent. As Table 3 shows, replacing Rivalry Gap with RSO Rivalry, alongside REO Rivalry, points to the same conclusion. The coefficient of RSO Rivalry is positive and statistically significant, indicating that rivalrous RSOs result in deeper security cooperation of the overlapping REO. Switching to a binary-dependent variable that not affect this result.
Thus, RSOs with strategic rivalries give rise to REOs that take on security tasks, often with fewer members. This scope expansion is especially likely to occur when this smaller group is itself like-minded and free of significant political rivalries. This finding explains, perhaps, why GCC members have opted to address security concerns within this REO (until the outbreak of rivalry between Saudi Arabia and Qatar in the middle 2010s) rather than the highly rivalrous LAS. It may also account for the lack of security cooperation within WAEMU and SACU, which are nested within ECOWAS and SADC, respectively. The latter two tackle security matters and contain a relatively small number of rivalries (themselves nested within the AU that includes numerous rivalries). This result underscores the need to consider the effect of political dynamics in one IO on the broader institutional environment, particularly on IOs with overlapping membership.
The results on the control variables are consistent with previous research. The estimate of Conflict is mostly positive and statistically significant in several models, indicating that violent disputes create demand for security cooperation. Hegemony is also positive but never approaches statistical significance. Here, contradictory effects of a dominant regional power might wash each other out. Delegation is always positive and highly statistically significant. Presumably, member-states take advantage of existing institutional infrastructure, knowledge, and experience when they confront sensitive political and military problems (Haftel and Hofmann 2017; Wallander 2000). This might explain why many like-minded states expand the mandate of existing REOs rather than creating new RSOs.
The effect of Members is positive and statistically significant in some of the models. This finding (weakly) suggests that larger REOs are more likely to engage in security cooperation, perhaps due to advantages to scale. This finding may appear to be inconsistent with our general emphasis on like-mindedness and cohesion, which are more likely in smaller groups. We do not think so. Many of the REOs in our sample are subregional and as such have limited membership (the average on this variable is about seven with a standard deviation of four). Possibly, the advantages that come with larger membership decline with mega-regional and continental IOs, a label that fits several RSOs, such as the AU, OAS, and NATO. This suggests that a medium size organization (perhaps ten to fifteen member-states) is the most appropriate for security cooperation.
As expected, we find empirical support for the notion that REOs in the developing world are more likely to address security issues compared with their counterparts in the developed world. The substantive analysis shows that security substructures, especially shallow ones, are much more likely to be erected by South–South REOs. Finally, the coefficients of the time fixed effects (not presented in the tables) corroborate the observation that security substructures within REOs are becoming increasingly common over time.
Conclusion
In the shadow of performing RSOs, some REOs have “trespassed” into the security policy domain. Especially since the end of the Cold War, more and more REOs have taken on global and regional security tasks. This scope expansion can create instances of organizational overlap that are not functionally necessary at first sight. As we have shown, the current landscape of ROs indicates that, indeed, many REOs that deal with security matters operate in regions where RSOs are not only present but also perform important security functions. As a result, several ROs claim authority over the security policy domain and, intentionally or not, contest the organizational scope and membership in many world regions. Despite the potential significance of such overlap for regional stability and peace, as well as organizational performance, the sources of this phenomenon are poorly understood.
This study is perhaps the first exploration of this question in a systematic manner. It develops an argument with respect to the reasons for scope expansion and overlap, with an emphasis on the presence of strategic rivalries, or lack thereof, inside ROs. It then presents a typology of security institutions nested within REOs and utilizes it to code a large number of ROs around the world on the depth of security cooperation. Utilizing this new data set in a statistical analysis, we find that overlapping RSOs do not prevent REOs from expanding into the security realm. Our findings point to the importance of strategic rivalries within IOs. In particular, they suggest that rivalry in RSOs lead to the expansion of more cohesive REOs into the security realm. Such political dynamics have yet to find inroads to the study of IOs at large.
Given that many overlapping ROs are here to stay, rivalries within and across these organizations have significant policy implications. We should expect a variety of political dynamics to either push for coordination among the overlapping ROs or create competitive dynamics (Hofmann 2009). The latter can lead to wasted resources, for example. And if the REO’s scope expansion is not only successful, but also performs significant security functions (e.g., ECOWAS, SADC), it can compete with the RSO for resources and expertise. Over time, we can envisage three scenarios: political dynamics could either establish (a) a functional or geographic division of labor, (b) start the demise of one of the two ROs, or (c) a long-term entrenchment. Future research could illuminate the conditions under which these scenarios are likely to play out and how overlap impacts the political landscape of member-states that have to cope with these rivalries and politically cohesive subgroups.
Finally, the logic of our argument may “travel” beyond the security policy domain. As recent research demonstrates, REOs and other economic agreements have expanded their scope beyond pure economic matters and include a variety of nontrade issues (NTIs). Thus, many trade agreements tackle NTIs such as human rights, labor, and the environment (Milewicz et al. 2018). This expansion may be driven, at least in part, by rivalries in multilateral bodies that were supposed to address such issues. It seems plausible that the deadlock in the World Trade Organization and the UN, not least because of persistent rivalries among some of their leading members, drove smaller groups of like-minded states to adopt NTIs in their regional IOs or bilateral agreements. That being said, such extensions require some caution. Cooperation on international security is different from most other NTIs, which deal mostly with domestic regulation. As such, they present member-states with different dilemmas and costs, compared to crisis management, peace operations, and the like. In addition, unlike the security domain, there are no obvious regional IOs that address other NTIs. Instead, the choice is between the regional REO and a trade agreement, on the hand, or a multilateral framework, on the other. This reality probably requires an adjustment of the theoretical framework. Be that as it may, accounting for the effect of political dynamics in one IO on the broader institutional environment is a promising avenue of further research.
Supplemental Material
Supplemental Material, Appendix - Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations
Supplemental Material, Appendix for Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations by Yoram Z. Haftel and Stephanie C. Hofmann in Journal of Conflict Resolution
Supplemental Material
Supplemental Material, Information_Sheet_on_variables - Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations
Supplemental Material, Information_Sheet_on_variables for Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations by Yoram Z. Haftel and Stephanie C. Hofmann in Journal of Conflict Resolution
Supplemental Material
Supplemental Material, Information_Sheet_on_variables - Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations
Supplemental Material, Information_Sheet_on_variables for Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations by Yoram Z. Haftel and Stephanie C. Hofmann in Journal of Conflict Resolution
Supplemental Material
Supplemental Material, Master_JCR_Jan_2019 - Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations
Supplemental Material, Master_JCR_Jan_2019 for Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations by Yoram Z. Haftel and Stephanie C. Hofmann in Journal of Conflict Resolution
Supplemental Material
Supplemental Material, Results_CJR_Final_Jan_2019 - Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations
Supplemental Material, Results_CJR_Final_Jan_2019 for Rivalry and Overlap: Why Regional Economic Organizations Encroach on Security Organizations by Yoram Z. Haftel and Stephanie C. Hofmann in Journal of Conflict Resolution
Footnotes
Authors’ Note
Authors’ names appear in alphabetical order and indicate equal co-authorship.
Acknowledgments
We thank Vincent Arel-Bundock, Leonardo Baccini, the two anonymous reviewers, and the editor for their helpful comments and suggestions.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Israel Science Foundation (Grant No. 1109/14).
Supplemental Material
Supplemental material for this article is available online.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
