Moving beyond democratic exceptionalism: the peace among similar political systems
Evolution of a concept
We have seen that monarchies are more peaceful toward each other but not necessarily toward other types of systems. To explain both sides of monarchic foreign policy, we need an explanation that is dyadic, one that takes both conflict actors into consideration at the same time. Dyadic explanations do not necessarily depend on the specificities of the two regimes that interact but rather focus on characteristics of the interaction itself. If there are different sets of behaviors of monarchies toward other monarchies than toward republics, the conflicting sides must at first recognize the other as different or similar to themselves. This means that they might adapt their behavior depending on what they know of their opponent. Given that knowledge is not an objective and unchangeable monolith but highly constructed, social constructivist factors need to be taken into account. This book therefore uses approaches based on social identity theory (SIT). SIT's advantage is that it is not restricted to the specificities of monarchism but instead enables the explanation of a broader phenomenon - the similar political systems peace (SPSP) - of which the monarchic peace (MP) and the related democratic peace (DP) are subtypes.
SPSP approaches, mostly derived from the DPT literature, have been proposed numerous times by various scholars but could not gain traction because of a misleading specification and operationalization relating to the understanding of "autocracy” as a type of political system and "similarity”. This book is based on an adapted alternative that uses a more nuanced concept of autocracy as well as clearly defined criteria for (political system) similarity.
The democratic peace is the closest thing “we have to an empirical law in international relations” (Levy 1988, 662), or “one of the strongest nontrivial and nontautological generalizations that can be made about international relations” (Russett 1990, 123). It is possibly the most strongly established empirical phenomenon in international relations (IR). All the more surprising, then, that there is still no consensus or even cautious agreement on why it exists (Choi 2016; Ember. Ember, and Russett 1992). This suggests that while the empirical phenomenon persists, the framing and specification might be misconceived and should be re-examined. These previous misspecifications are in large part to blame for theМоу>ing beyond democratic exceptionalism 19 inability of the alternative conceptualization of the SPSP theorem to reach a wider consensus. The following is an attempt to construct an SPSP concept that does more justice to the empirical phenomenon and thereby revitalizes the concept that can in turn be used to explain the monarchic peace.
After the phenomenon that no two democracies go to war against one another was established as a virtually undisputed empirical fact, the most frequent criticism of DPT revolved around the question of misspecification. The criticism was clustered mostly along three varieties: confounding; (horizontal) special case; and (vertical) sub-case.
The confounding-factor critics have claimed that the democratic peace is not that after all but in fact a capitalist peace (Gartzke 2007), a trade and development peace (Mousseau, Hegre, and O'neal 2003), a cold war or common-interest peace (Farber and Gowa 1995, 1997), or a stable border peace (Gibler 2014): better explained by third, confounding factors that are usually found among democratic states without necessarily being directly related to democracy.
The second type of misspecification criticism is that the DPT may be one special case of many geographical and historical "zones of peace” (Kacowicz 1998, Ch. 1), e.g. in Northern Europe (Archer 1996), East Asia (Kelly 2012), Africa (Herbst 1990), or South America (Martin 2006), as democracies are also highly clustered geographically (cf. Gartzke and Weisiger 2013, 173).
This book ascribes to the third type of criticism, which asserts that the DPT may be part of a broader phenomenon of which it is a sub-case, a peace among politically similar systems (Peceny, Beer, and Sanchez-Terry 2002; Souva 2004; Werner 2000). This concept is referred to here as similar political systems peace (SPSP).1 The SPSP is a more encompassing theorem than is the DPT, and it can explain a larger universe of cases and events and leads to discoveries of regularities among other than democratic dyads. Therefore, Imre Lakatos’s criteria for a "progressive” research program to supersede an older one are met (Lakatos 1970) and the explanatory power of the DPT should be reevaluated.
This by no means implies that we need to throw out all the research and insights that the DPT has helped to uncover with the bathwater of its narrow scope on a subset of a larger class of instances. As Bennett correctly points out, the DPT might still have idiosyncratic features that could be understood and studied independently of the general SPSP (2006, 317). Even if the DPT were a special subcase of the SPSP, it would not become obsolete, just as Newtonian mechanics has rightly remained in school curricula after the discovery of Einstein’s relativity theory. However, the SPSP provides a broader framework helping explain a larger universe of cases of peacefill groups of dyads.
The SPSP is not a novel theory. First hints are already found in early reviews of the DPT literature where a joint institutional similarity effect is found even for dyads that are not democracies. Studies since the 1990s have found evidence of an autocratic peace (e.g. Gleditsch and Hegre 1997; Raknemd and Hegre 1997).
In a 2000 article, Werner made that link explicit and analyzed “political similarity”, finding a strong effect even when controlling for joint democracy (Werner 2000). Peceny, Beer, and Sanchez-Terry looked at subtypes of autocracies andfound evidence of a dictatorial peace comparable in effect to democratic dyads at least among some subtypes of dictatorships (Peceny, Beer, and Sanchez-Terry 2002). Souva later found that economic similarity is also important for peacefulness, even among politically dissimilar states (Souva 2004). Henderson asserts in probably the strongest terms that the DPT is "simply a subset of peace among politically similar states'’ (Henderson 2002, 42), after modifying a classic DPT study to include a measure for political dissimilarity. Gartzke and Weisiger provide a more nuanced argument of the dynamic nature of democratic and autocratic peace relative to the overall frequency of democracies or autocracies in the system (2013).
However, the SPSP phenomenon has encountered controversy, with criticism miming the gamut from highlighting the greater impact of democratic similarity to a dismissal of the phenomenon altogether. It can be divided into three main groupings: First, an SPSP effect is discernable, but considerably smaller than that of joint democracy (Bennett 2006; Conrad and Souva 2011; Maoz and Abdolali 1989). A second, often overlapping criticism identifies strong regime similarity effects for “coherent” systems, i.e. among highly autocratic or highly democratic regimes (N. Beck, King, and Zeng 2004; Bennett 2006; Kinsella 2005) with a U-distribution. The third group of critics deny any SPSP effect altogether (Bausch 2015; Ray 2005)?
This mixed picture points us to the central problem of the specification and operationalization of what scholars referred to as political system similarity (Werner) or institutional similarity (Souva). Werner and Souva, along with most authors who did not find a separate effect of an autocratic or SPSP, do not actually operationalize similar political systems or institutions, but only different degrees of autocracy/democracy, on the basis of the polity scale data set (Marshall, Gurr, and Jaggers 2014 and previous versions). Even-more-recent articles still employ this measure even if they recognize this shortcoming themselves (see e.g. Choi 2016; Dafoe, Oneal, and Russett 2013). Despite various modifications of this measure,3 none of it is a valid operationalization of political similarity, because it merely differentiates between the degree of authoritarianism.
Authoritarian regimes, however, are highly diverse, much more so than democracies. Since the DPT is still paramount in IR, along with the realist dictum that is blind toward regime differences in general, autocracies are rarely studied with the same degree of nuance and sophistication as democracies are, in striking difference to the sub-field of CP.4
Most democracies (especially "coherent” ones) undoubtedly share various similarities, being politically pluralist and participatory, with similar institutions, mostly Western and mostly economically powerful states, clustering in certain geographic regions. This gives rise to the frequent criticism invoked at the beginning that it is a (confounding-factor) similarity different from "democraticness” that induces dyadic peacefulness.
Looking beyond the democratic backyard, there is not nearly the same level of similarity among systems as diverse as North Korea, Morocco, or the misleadingly named Democratic Republic of Congo. What has long been recognized and
Moving beyond democratic exceptionalism 21 ameliorated in autocracy research, namely that the great scope and heterogeneity of the different systems make their grouping together as autocracies or authoritarian states inadequate to explain their behavior and vulnerabilities (Geddes 1999; cf. Levitsky and Way 2002; Schedler 2006), has so far not taken hold in IR and FPA literature. Geddes’s famous dictum that "different kinds of authoritarianism break down in characteristically different ways” (Geddes 1999, 117) is applicable to foreign policy behavior as well: "different kinds of authoritarianism make war and peace in characteristically different ways” too.5 The criticism that autocracy is more diverse and heterogeneous than democracy and cannot be adequately covered by a dichotomous or continuous operationalization was already voiced as early as 1995 by Hermann and Kegley (1995, 519-520), before even the first consolidated SPSP study, but could so far not shake the deficient operationalization along the democracy-autocracy continuum or dichotomy. Bridging the fields of international relations and comparative politics is therefore an important task.
This misspecification explains the mixed review that the SPSP received in IR literature and the common finding that the effect is strongest on the edges - highly jointly democratic and highly jointly autocratic states - itself another hint of the lack of differentiation among autocratic regimes. While extremely autocratic regimes might indeed resemble each other more since a high level of autocratic control and lack of audience costs might produce structurally similar, e.g. personalists systems, this would not apply to less-autocratic regimes per se, which would still be more heterogeneous.
In the words of Steve Chan, "how similar or different the regimes in a dyad are from each other offers itself as a relevant factor for future analysis” (Chan 1997, 83). But what exactly constitutes difference or similarity? Since autocracies are more heterogeneous than democracies are, we must find more adequate measures for similarity by looking at different regime subtype dyads as possible other subcases of the SPSP, along with the DPT. Two of the studies most favorably disposed toward a possible SPSP are, as would be expected from the argument made earlier, those with a more sophisticated understanding of similarity.
Peceny et al., following Geddes (1999), break up the autocracies into three types: personalist, military, and single party (2002). Because the Geddes data set used by the authors does not yet include monarchies, they are excluded or lumped into the “mixed” category. The results show evidence for a sort of dictatorial peace, especially among military and personalist systems. The caveats that the authors apply are related mainly to the rarity of war and some of the regime subtypes and therefore the tenuousness of the statistical effect, not to any lack of effect per se.
Souva then reverts to the polity scale to measure political similarity but employs an additional measure of economic similarity and finds that systems that are both politically and economically similar are especially unlikely to confront each other militarily (2004). Having two measures of similarity enhances the chances of having a valid depiction of similar regimes as it adds more markers of similarity. Using such a broadened measure of similarity would therefore clearly distinguish between regimes like Saudi Arabia and North Korea, which have similarpolity scores but dissimilar economic (and political) systems. While presenting a more valid and robust operationalization for large-n studies that necessitate broad encompassing categories, it is still far from a fleshed-out concept of similarity.
There is a limited number of studies critical of the SPSP that recognize and tackle this major problem. The most nuanced critique is provided by Conrad and Souva (2011), who only find consistent results for joint democracy but inconsistent results for other kinds of joint similarity. Recognizing the problems in lumping all autocracies together, the authors test not only joint coherence, on the basis of polity scores, but also two other operationalizations - that is, institutional regime similarity, on the basis of a typology consisting of democracy, monarchy, singleparty system, military dictatorship, and personalist system, and audience costs, on the basis of a slightly different typology with mixed non-democracy replacing personalist and dynastic monarchy replacing monarchy. This leads to a more nuanced although still-not-determining picture: the evidence points to peaceful effects among socialist dyads, single-party dyads, and monarchic dyads, although they are not seen across all models and specifications employed, due mostly to the small number of cases in some types of non-democratic dyads.6 The weaker evidence for other kinds of SPSP is therefore a problem of data availability, not of lack of effect per se. Joint democracy is always significant, in part because there is a far larger set of onset cases to draw inferences from. A lack of significance might thus rather be a consequence of the data in use than of a random distribution.7
AN SPSP concept based on social identity’ theory: ingroup/outgroup identification influences foreign policy behavior
In light of the discussed methodological problems, the evidence against SPSP does not seem that strong after all. To obtain a valid test, we need to adapt the specification and operationalization of political similarity among autocracies. Reconceptualizing the empirical patterns as SPSP and respecifying similarity might help retest the SPSP in quantitative studies similar to those already conducted with the old conceptualization. It is, however, not enough to revitalize the struggling theorem; a convincing causal pathway is called for. Distinct approaches attempting to explain an SPSP are circulating, many of them adopted and adapted from the DPT.
If there is a common mechanism that explains DP and MP. it cannot be exclusively applicable to democracy (or monarchy).8 Conrad and Souva identify three strands of theoretical approaches applicable to other than just joint democracies:
Coherence approaches have already been shown to inadequately portray the heterogeneity in autocratic systems. Audience cost approaches are additive and
Moving beyond democratic exceptionalism 23 not inherently dyadic. In general, normative and institutionalist approaches are mostly democracy-specific and exhibit an additive effect - not an inherently dyadic one.9
Ideational or social constructivist approaches, embedded in the political similarity literature, have the greatest potential to explain dyadic behavior because they explicitly entail the dimension of interaction and mutual perception. Theorists of this brand of IR assume that “soft” factors like identity, role, and norms are at least as important for the behavior of states as are institutions and power relations (Reus-Smit 2009; Wendt 1992).
How states act toward each other and in general is significantly shaped by their self-image/identity and their image of their interaction partner, the “other”. The most likely candidate to explain behavior driven by identity is social identity theory (SIT) (Conrad and Souva 2011; Gartzke and Weisiger 2013; Hermann and Kegley 1995; Tajfel 1974, 1982; Weart 1998). Social identities are "sets of meanings that an actor attributes to itself while taking the perspective of others, that is, as a social object” (MCCall and Simmons 1978, 61-100, cited in: Wendt 1994, 385) or, following a liberalist view a "set of preferences shared by individuals” that shapes a set of interests representing these rational actors’ preferences (Moravcsik 1997, 517, 525).
Originally a micro-level psychological concept, the point of departure of SIT is that identity is constructed against a “significant other” with collective identity distinguishing between the ingroup and the outgroup (Dovidio, Gaertner, and Validzic 1998; Tajfel 1974). And “apparently, the mere fact of division into groups is enough to trigger discriminating behavior”, even when that division is based on arbitrary factors such as a preference for the modern painters Paul Klee or Wassily Kandinsky (Tajfel 1970, 96) or in which of two buses one happened to arrive (Sherif 2010). This concept can be applied to large-scale social groups, including nations and states - as Tajfel intended from the outset (Robinson and Tajfel 1996, 66). This lends itself handily to an analysis of groups of similar political systems who form the ingroup, while non-similar states are conceptualized as the outgroup.
These processes are set in motion even without direct contact or even personal knowledge of the other, designation as other being enough. As self-worth is linked more closely to the group identity, the incentive to protect the group becomes stronger (cf. Hermann and Kegley 1995, 517-518). Group identity has real behavioral consequences, with ingroup favoritism and outgroup discrimination and hostility often occurring simultaneously - ingroup "love” and outgroup "hate”, in Marilynn Brewer’s terms (1999, 442). Although outgroup hostility is not an unavoidable outcome of ingroup identity building (Brewer 1999; Oakes 2002), it can accelerate in the event of a common threat or a salient external pressure (cf. Gibler, Hutchison, and Miller 2012,1657), which has to be perceived "by the population at large as a serious societal danger” (Gibler, Hutchison, and Miller 2012, 1659). External pressure does not mean here that the threat cannot have an internal source, even if it has been triggered from outside. This is especially germane for the Middle East, where regional foreign policy was “chiefly used tocounter domestic threats: either regimes used anti-imperialistic rhetoric to shore up their fragile legitimacy or they sought protection from the Western powers against domestic opposition” (Hinnebusch and Ehteshami 2002, 34).
Inside an ingroup, violence has ceased to be a viable option because "interaction has literally eliminated defection (war) as a possibility-where there exists the ‘impossibility of imagining violence”' (Abdelal et al. 2006, 697), with the potential to explain peace inside the monarchic ingroup. Ingroup bias is greater the more important the relevant attribute to the social identity of its members is and the more comparable the outgroup is to the ingroup (Tajfel 1978, 250), a finding that can be applied to political systems and regime types that are easily compared to each other.
However, an ingroup identity does not depend on complete sameness or assimilation; rather, intragroup differences exist and are vital for the individual members. Belonging to an ingroup also does not mean that every member is equally important or that the group is acephalous (see e.g. Hogg 1996, 80-83). In short, the sameness is perceived in terms of an "equality in kind”, not an “equality in power” or importance.
Transferring a micro-level concept onto the state level
But how can a psychological or sociological theory for groups of individuals be translated from the first to the third image (Waltz 2001) - as states are not just "people, too”? Under certain conditions, applying anthropomorphous traits to states can be appropriate. First, Tajfel himself intended his theory of intergroup relations to apply to large-scale social groups, including nations, from the outset (Robinson and Tajfel 1996, 66). Second, social constructivist theoreticians in IR have already applied social identity theory to states - Alexander Wendt explicitly notes the possible impact of collective identification, i.e. “identification with the fate of the other” (Wendt 1994. 385, emphasis in original) through "shared norms and political culture” (1994, 386). This can form a positive collective identity - the basis for "friendship” (Wendt 1999, ch. 7). The literature on security communities takes this at its point of departure (Deutsch 1957) and recognizes the importance of identity building against an "other” (Adler and Barnett 1998, 56). Michael Barnett carved out the importance of group identities for the Arab alliance system (Barnett 1996), and transnational and subnational identities have been shown to affect foreign policy in the Middle East (Telhami and Barnett 2002). Role-theoretical analyses also apply these group-centered identification processes to state behavior (cf. Harnisch. Frank, and Maull 2011).
Ingroup-outgroup explanations have been used explicitly by proponents of DPT and SPSP (Gartzke and Weisiger 2013; Hermann and Kegley 1995; Souva 2004; Weart 1998). As Owen states, “democracies recognize one another and refuse to fight on that basis” (J. M. Owen 1994, 96). One of the most extensive and comprehensive accounts of the DPT takes ingroup identification as the primary explanatory prism and claims that certain states are peaceful toward each other because they recognize each other as equal (in kind) (Weart 1994). This examination also shows that ingroup-outgroup identification is not exclusive to
Moving beyond democratic exceptionalism 25 democracies. Spencer Weart claims in his work, elaborated in a course of extensive historical examination and covering sources from antiquity to the present, that established territorial republics of the same type, democracy or oligarchy, (almost) never wage war against one another (1998, 14).10
In addition, empirical findings from other fields substantiate the effect of large-group similarity on foreign policy behavior. Experiments confirm that popular acceptability to use force against a country becomes stronger when it is not perceived as "one of their kind" (Mintz and Geva 1993), leading back to the original micro-foundation of SIT. This conforms to the findings on the tendency to ally and refrain from force among politically similar (via common culture or common regime type) systems (Lai and Reiter 2000; Siverson and Emmons 1991), which is corroborated for democracies and for autocracies (Koschut 2012).
Finally, micro-level approaches are directly applicable to leader-centric analysis and regime-centric analysis, as is the case here. The importance of individual leaders is often underestimated in IR (Hermann and Kegley 1995; Lebow 2010). This is particularly true for autocracies, because ultimately, the leaders' perception is decisive.
This is one of the reasons why Middle Eastern monarchies lend themselves especially well to SIT-style approaches. Since they exhibit an especially high level of personalization, comparable only to the “presidents for life” in some Arab republics (R. Owen 2012), they are especially suited to a leader-centric analysis. Monarchs often stayed in power for decades, shaping the modem states decisively and rarely delegating too many powers. Personal relationships often form the basis of agreements and formal relations, and thus, the human traits of the monarch and their inner foreign policy circle are more easily transferrable to the state level. In addition, the monarch is also usually the prime decision maker, especially in foreign policy. Even in more open and liberalized states, foreign policy is usually conducted by a small circle of people; this is even more true for the highly untransparent authoritarian monarchies in the MENA. As Ayubi states, "Foreign relations are obviously a domain reserved exclusively for the family”, where even the office of the foreign and defense minister is in general either occupied by the monarch or by a close relative (Ayubi 1995, 230). In this analysis, the monarchs and their immediate regime members are the identification actors because their perception of their counterparts counts in foreign policy decision-making, especially in tunes of crisis.
This work therefore follows the call by Valerie Hudson: since the particularities of decision makers are vital to understanding foreign policy choice they "should not remain as undigested idiosyncracies (as in traditional single-country studies) but should rather be incorporated as instances of larger categories of variation in the process of actor-specific theory-building” (Hudson 2005, 7).
A definition of political similarity
But what exactly could constitute an ingroup among states? How is an "other” constructed? In other words, what constitutes political systems similarity thatengenders peace? The first criterion is the most obvious one and one that recurs in large parts of the earlier-cited research: politically similar regimes must be institutionally similar. This does not mean, however, that their degree of democraticness or autocraticness is the primary distinction or of any importance at all. Rather, a more complex view of institutional similarity revolves around the specificities of the political system. Second, regimes must be ideationally similar, which includes immaterial factors such as a common culture, history, religion, etc. Third, and most importantly, they must recognize each other as similar - perception of similarity becoming the third criterion.
The type of political system or regime (sub)type is a particularly salient similarity characteristic and therefore a relevant criterion of difference and similarity for various reasons. First, it is clearly visible across borders and combines "malleability and staying power” (Gartzke and Weisiger 2013, 175).11 Second, a rival or enemy with a different type of political system than one's own can appear particularly threatening. As illustrated by Stalin’s infamous dictum '’Whoever occupies a territory also imposes his own social system on where his army arrives. It could not be any different”,12 such an opponent is a double threat: to both the territorial integrity of the state and the stability of the regime. They can do this by military conquest, by increasing "the costs of enforcing the state’s particular set of institutions” (Werner and Lemke 1997, 532), or by promoting the interests of internal rivals of the leader (Werner and Lemke 1997, 533). Third, given that similar institutions are either the consequence of or the origin of a similar worldview or ideology, institutionally similar states have a similar outlook on the world and thus simply fewer issues of disagreement (Souva 2004; Werner 2000). Fourth, regime stability as a prime driver of foreign policy is tied to regime legitimation: the imperative to be seen as a “good” system is important for regime survival. This leads not only to war avoidance but also to alliance: if regimes are "good”, then helping out a similar system must be good too. Also, being part of a broader movement or club bolsters legitimacy because it justifies its own political principles as accepted and prevalent (Walt 1985, 20). A last but crucial point why regime type similarity rather than any other important similarity between two states should be examined within this particular framework is that regimes are responsible for conducting foreign policy and are thus especially likely to be a reference for decision makers who are responsible for foreign policy and ultimately for war and peace.
But bare institutional similarity is not enough: Conrad and Souva emphasize that to better account for similarity, instead of preference similarity from similar political institutions “a broader institutional similarity, common culture, shared threat, economic wealth, or some combination of these factors may better influence preference similarity and the likelihood of conflict” (Conrad and Souva 2011,26).
Some of these immaterial traits have been studied as a basis for a feeling of commonality and explanatory lenses for bilateral relations, such as the classic
Anglo-American special relationship. Cultural and linguistic commonality was invoked: "though the American people are very largely foreign, both in origin and in modes of thought, their rulers are almost exclusively Anglo-Saxons, and share our political ideas” (Lord Cecil 1917, cited in: Reynolds 1985, 2). Common history was also invoked: “History, tradition, affinity have been cmcial to the alliance, rather than peripheral” (Dawson and Rosecrance 1966, 41).
A study focusing on the Confucian Long Peace, from 1644 to 1839, finds a separate peace between the similar political systems of China and its Confucian neighbors (Kelly 2012). Kelly claims that their common culture generates a “we-ness”, a self-perception of an "imagined community” that, combined with the Confucian anti-war ethic, leads to peacefol behavior within this group of states not present in encounters outside the group. The author even mentions the Arab state system as a possible candidate for another "zone of peace” via cultural similarity and finds corroborating evidence. However, the evidence is weak, because all the different types of Arab states are lumped together. While pointing to the importance of a cultural link, this strengthens the argument that common culture, language, and history are apparently not sufficient for a separate peace, same as institutional "coherence”.
PERCEPTION OF SIMILARITY AND SALIENCE
Institutional and ideational or cultural similarity need to be combined to find similar political systems. In the end, however, the only definitional hallmark of political similarity is whether the leaders of states (or their publics) consider then-counterpart as similar to themselves or not (cf. Hermann and Kegley 1995) -perception of similarity - under the condition that they "believe” that ideology shapes international alignment (Walt 1985, 25). The other criteria merely help delimitate where to look for this perception but only form can necessary, not sufficient. definitional conditions.
Perception of similarity would be sufficient to classify the members of a dyad as politically similar, but it is notoriously hard to study and prove. Nevertheless, it has also been highlighted in DPT literature when controversial cases of possible war among democracies were resolved by establishing that the rivals did indeed not recognize each other as democracies, a perception contributing to war (J. M. Owen 1994; Weart 1998).
States can be similar in countless ways, e.g. culturally, linguistically, by religion, by geography and climate, economically, by political organization, by institutions, and by culture. But the perception of similarity is crucial to understanding which commonality is salient enough to affect foreign policy behavior in the particular context at the particular time under consideration. It is one characteristic that can help us distinguish and understand when to use a typology consisting of military, single-party, personalist, and monarchic regimes, when to look at Muslim, Catholic, or Hindu identities, and when at Russophone and Ukrainophone polities.
Even when looking exclusively at different regime types for the reasons just cited, salience, i.e. the simultaneous existence of a comparable or contrasting category (Turner and Oakes 1986), defines when a democracy/non-democracy difference has become politically relevant or when regime subtype distinctions matter more. If an identity has a higher salience, it is more likely to be acted on. A higher salience of a shared quality enhances the perception of belonging to that category with others. Fluidity and amenability to change with different identities alternating and overlapping is an essential characteristic of ingroup and outgroup identity, and capturing this change is a prime challenge to any theoretical framework.
In the following case studies, political system similarity pertaining to the structural dimension of similarity as opposed to perception will be operationalized by three indicators, to capture a multidimensional similarity between two states (or, more precisely, their ruling and decision-making elites). In case of dissimilarity, numerous dimensions are absent or underdeveloped. This step is important because identity, while constructed, is not infinitely flexible but rather is based on and restricted by physical and material attributes that are resistant to change, including geography, size, language, ethnicity, etc. (Chafetz, Spirtas, and Frankel 1998, xi). The three indicators are as follows:
In the cases analyzed here, a similar political system is defined by regime subtype. For monarchies, this implies that their heads of state have a hereditary title and rule and reign (authoritarian monarchies). Other state institutions, such as parliaments or consultative assemblies, might be present but must be subordinated to the monarch and/or the ruling family.
1.3 Similar economic system
A comparable economic system often goes hand in hand with a similar political system. Nevertheless, having a similar economic system helps in establishing clear expectations by ingroup members and fostering similar interest preferences.
Constraints and catalysts to monarchic salience: the waxing and waning of the “royal club ”
Although neither institutional similarity nor a common culture necessarily brings about a perception of similarity, three main conditions can be identified that either catalyze or obstruct the perception of similarity of institutionally similar systems and thus condition the salience of the type of political system shared by the ingroup members:
1 Common threat. This binds an ingroup together against a common enemy and induces a “rally-around-the-flag” (or, in this case, throne) effect. A common threat against monarchies either threatens the survival of their states
Moving beyond democratic exceptionalism 29 or the survival of their regimes or ruling dynasties and thus can have external (e.g. radical pan-Arabism incited by republican presidents) or internal sources (e.g. opposition groups that aim to overthrow the monarchy). In case of mixed dyads in the “quasi-experiment” cases, this category indicates whether there is a common threat against the group of states of the alternative potentially shared identity post-regime change (e.g. against Arab or Muslim states). Although this factor is often emphasized in the literature, it is not a necessary condition for a shared identity to emerge (Chafetz, Spirtas, and Frankel 1998, xiii).
This principle of dynamic shifting allegiances via varying identification is further backed by ethnographic evidence and known as "fission and fusion” (Neel and Salzano 1967) in cultural anthropology. It is best embodied in the Arab proverb "me against my brother, me and my brother against my cousin, me and my cousin against the stranger”14 and refers to the dynamics of large units and subunits in interaction with internal or external units where allegiances change as the reference category changes, usually in tribal or clan societies consisting of multiple lineages.15 Who bonds together is thus determined by who the “other” is in the given context.
Summing up the first step of the causal mechanism, similar political systems are characterized by similar regime types; a common culture, language, and history; a similar economic system; and the perception of belonging into the same category. Furthermore, a shared threat, a low prevalence of the respective type of political system, and lack of divisive ideology strengthen that perception, although not all factors have to be present at all times. This definition is a highly flexible one, characterized by a Wittgensteinian family resemblancerather than by a monothetic classification. This makes perceived political similarity a multidimensional concept combining institutional and ideational elements that can accommodate and explain changes over time better than static approaches can.
When we apply this definition to the monarchic peace, authoritarian monarchies in general undoubtedly become similar political systems. They share a regime type, a political culture, a common alliance with the West, and for the most part similar economic systems. In the Middle East, the area with the highest remaining concentration of monarchies, they also accrue a common history, religion, culture, and (excluding monarchic Iran) language.
But that does not automatically mean that monarchy becomes a salient joint characteristic. In case of a low salience of monarchic identity, other commu-nalities might be more relevant for foreign policy behavior (Arab ethnicity, Islam, local or tribal identities, etc.), and the monarchic identity does not necessarily translate into foreign policy behavior of any kind. If its salience is high, it is expected to influence foreign policy behavior, most notably vital decisions like war initiation. The higher the salience, the more politically relevant the identity becomes. This raises the likelihood that foreign policy decisions of a lower level are also affected, resulting in secondary implications of a monarchic peace, such as monarchic solidarity. As there is no clearly defined threshold for salience "sufficient” for an ingroup identification to arise, and the effect of the enumerated factors is additive, with none expected to be automatically stronger than the others; any combination of them might be sufficient for a salient monarchic (or otherwise similar in case of studies of mixed dyads) identity to emerge.
How that translates into perception of similarity and how this perception is developed and affirmed will be examined in the following section and illustrated empirically in Chapter 3.
How ingroup identification leads to de-escalation
While the statistics in the introduction showed a clear negative peace as an absence of war, the SPSP framework also implies a positive definition of a stable peace as “a situation in which the probability of war is so small that is does not really enter into the calculations of any of the people involved” (Boulding 1978, 13). If similar political systems do not wage war against each other, they must either have less conflictual relations (e.g. due to fewer issues of disagreement, as prior SPSP researchers suggested [Souva 2004; Werner 2000]) or, as is the interpretation followed by this book, have additional ways to de-escalate crises in existing conflicts that dissimilar systems do not necessarily possess.
This second interpretation is easier to trace because fewer cases are needed and because the comparison of levels of non-military conflict would pose significant methodological and ontological challenges, but more importantly, it emphasizes that absence of war does not mean absence of conflict - even with fewer conflict issues, there always remains the possibility of at least a few serious conflicts to arise, which seems to conform to reality. The Middle East, and especially the Arabian Peninsula, is a highly conflict-ridden region - yet conflicts still fail to escalate into wars. Third. SIT implies a different code of conduct in case of conflict rather than the elimination of conflict issues - as “interaction has literally eliminated defection (war) as a possibility-where there exists the ‘impossibility of imagining violence’” (Abdelal et al. 2006, 697) inside an ingroup. It is not the absence of conflict that calls for an explanation - the actual puzzle is why all of these conflicts have never escalated to the extent of overstepping the threshold of war and rarely led to any sort of military confrontation, whereas other conflicts in the region frequently did.
Given that war is a rare event - even among the most dissimilar of systems -there are, of course, multitudes of pathways toward peace. Ingroup identification is only one additional path to de-escalation. It is not non-war per se that needs explaining but rather the “concentration” or “overrepresentation” of non-war among certain types of states. Escalation to war is not a given for dissimilar dyads in the same way that alliance is not a given for similar ones. In fact, the conflicts on the Arabian Peninsula often hamper intra-monarchic alliances: “Cooperation among the Gulf monarchies themselves - a natural tiling given their common domestic backgrounds, regional vulnerabilities, and international orientation - is complicated by historical disputes among the dynasties” (Gause 1994, 120-121). However, the same mechanism that supports de-escalation should also make cooperation and long-term alignment among ingroup members more likely, albeit not assured. To trace the effects of a different code of conduct inside the ingroup, we need to identify the social processes of ingroup identification. Once there are enough structural similarities and monarchic identity is salient, an ingroup (via perception as equals) is likely to develop. Once it develops, it needs to be sustained to remain relevant. To ascertain whether this development has indeed taken place, indicators capturing both ingroup development and affirmation need to be found. Given that perception is difficult to measure, the indicators should be highly varied to allow for an intersubjective analysis. The numbered elements here follow a general causal and roughly chronological sequence, as expected from the theoretical framework.
1 Mutual recognition
Mutual recognition is often the first step toward an ingroup development because it indicates an acceptance and recognition of equality of status between two states. Yet it is not a sufficient condition for ingroup development. because it is the default in international relations, regardless of the level of conflict between states. States that mutually recognize each other diplomatically might still see each other as illegitimate or fair game. Recognition is preceded by the consolidation of borders and state institutions, which is a scope condition (see section 2.3). Before the consolidation of the nation-state, imperial expansionary logic trumps nation-state territorial logic and makes conflict much more likely even in the case of an existing ingroup identification.
In the case of regime change, this indicator also clarifies whether there has been a cut or downgrading of diplomatic relations. Similar to the limited expressiveness of the absence of mutual recognition, a rupture might merely be a short-lived escalation that does not indicate longer-term deterioration of relations in a dyad, but merely a restricted crisis. A withdrawal of recognition is still a strong indicator of escalation, but an existing recognition is a weak indicator of a joint identification.
2 Personalization of bonds between ruling elites (especially riders) Personalized bonds between rulers and other decision makers are the basis that fosters ingroup development. Intensive personal bonds regularize and normalize interaction and are instrumental in clearing misunderstandings directly before they can become detrimental to the bilateral relationship. There are numerous avenues by which personalization happens, but the analysis will focus on the following three:
Frequent high-level state visits
Frequent high-level indicate regular and normalized contact between elites. The more personal interactions on the political level, the easier it is for other types of bonds, like intermarriage and friendship, to ensue. On the other hand, other types of bonds might induce more frequent visits. If bilateral relations sour, particularly among dissimilar regimes that have little incentive to prioritize cooperation over conflict, less-direct exchange between rulers or their representatives is expected. However, this is a weak indicator only for a rupture because visit frequency in general depends on the quality of bilateral relations at a given time. Explicitly boycotting visits indicate tense relations.
Kinship, intermarriage, and friendship bonds
Personalized bonds beyond the political sphere entangles political elites, who become more interdependent, and if groups of people overlap, the similarities between different polities and regimes are heightened as well, and thus, alliance becomes more likely. Intimate bonds also humanize the "other”, and trust can develop more easily, especially in personalized political systems like monarchies in the Middle East, where personal relations are valued more highly than institutional formalities and official positions and ranks. Each of these makes violence more likely to be seen as an illegitimate instrument as the other becomes part of an actual or metaphorical "family”. It is no accident that marriages were often pursued for political reasons, both in the high times of European monarchies and among royal elites in the Middle East today.
The personalization level between ruling elites also decreases if no or little new close unofficial relations develop among them or if old existing relations rupture. This is especially likely in the case of monarchic breakdown, because monarchies often exhibit a degree of intermarriage that is not continued by republican elites.
Personal relationships and a shared education system or process shape the worldview of the decision-making elite, leading to a more similar outlook and ruling ideology among individuals who went through the same socialization. It shapes perceptions of the other and fosters common ideas, preferences, and conflict-resolution approaches (Peceny, Beer, and Sanchez-Terry 2002, 19). This in turn makes ingroup identification more likely. Socialization can take place in palace schools, the military, or traditional education, forging elite solidarity at the top (cf. Ayubi 1995, 245).
There are two main ways that the socialization of (ruling) elites are hampered after a regime change in one member of a dyad. First, the new ideology used to legitimize the new system is incompatible with prior ways of preparing, e.g. future monarchs for ruling (e.g. the changes in power access, recruitment, social composition, and desirable traits and values brought about by the installation of Arab nationalist ideologies after coups). Second, there is a different predisposition toward the prestige of education institutions (e.g. the sending of political elites to British educational facilities preferred by the monarchs was transformed into a preference for education facilities in Arab nationalist states, e.g. Egypt, in the first decades following anti-monarchic and anticolonial revolutions).
3 Affirmation (or denial) of commonality
Because identification is a fluid and social process, it is not a given once it has developed, but rather, it has to be strengthened and reaffirmed to remain normatively relevant. Affirmation serves the double purpose of signaling to oneself and others (both other ingroup members and sometimes outgroups) who belongs together and on what grounds and of reaffirming and reifying an existing community or ingroup.16 It defines and shapes common identity over time. There are many ways that a shared identity can be reified, and the following three will form the main indicators used here.
The breakdown of the dyadic relationship not only means a passive lack of reification of communality but in general also an active othering process, i.e. the negation of similarity and the emphasis and consolidation of difference.
Kinship and family references
“Family” is one of the most exclusive and solidarity-demanding "clubs” there is. Such references between elites are therefore vital to identifying where such a solidarity club has formed. Of special importance are repeated appeals to family membership toward rivals or opponents. To enable war against an “other”, a process of “othering” must take place, usually in a dehumanizing or demonizing maimer. If. however, opponents consistently refer to each other as family, especially in fraternal terms (as these imply equality in contrast to, e.g., parents-offspring asymmetry), de-escalation is more likely. Family members might be rivals, but not enemies, and therefore, military attack is unacceptable to them, but it may be possible against outsiders. It is especially often found among the Gulf monarchies, who "resort to ‘organic’ concepts such as family, kind, neighborhood and community, in the attachment to custom and tradition, and in the nostalgia for the past” (Ayubi 1995, 244), but it can be seen across other states as well.17
Family rhetoric toward populaces, a frequent occurrence in republics rather than between regimes and its elites, serves a different function and is therefore not counted as an instance of kinship rhetoric here. Popular interests are different from regime interests and can be played against each other, which does not necessarily preclude interstate conflict.
Emphasis on similarity over difference
Similar arguments can be used for this indicator as a somewhat weaker measure of "friendship” instead of "family”. An othering discourse and following escalation are far less likely if acknowledged similarities outweigh perceived and acknowledged differences. If the level of conflict rises, usually differences are accentuated more than similarities. If this does not occur, the ingroup must be sufficiently well developed to recognize that conflict is present but can (and should) be contained. The valuing of similarity over difference can be seen in, e.g., refraining from othering by, e.g., emphasizing different ethnic or confessional differences.
The opposite, and an indicator for the breakdown of ingroup identity, are instances of othering, e.g. excommunication of the community and, at the most extreme, dehumanizing and demonizing rhetoric. In these cases, differences are more likely to be emphasized or exaggerated, especially along salient identity lines like ethnicity or sect/religion.
References to shared historical narratives
The importance of this indicator lies in that it is not merely the mere historiographic past and similar events that foster a joint identity but how the past is pictured and perceived, i.e. if historical narratives, such as about (anti)colonialism and shared alignment patterns, are similar. Monarchies and republics in the Middle East have highly intertwined histories but interpret and value different events and developments differently. This is a result of a different worldview that might hinder ingroup identification. A similar worldview will more likely result in more similar alliance patterns and political priorities fostering mutual peacefulness. A shared view of history leads to a similar outlook on the future and possibly even a “community of fate”. Given that the worldview shapes preferences and identity, it is an important factor in determining whether regimes are similar enough to develop a shared identity.
In mixed dyads with no ingroup identification, another indicator for disassociation is the use of competing constructions of identity, especially in historical narratives.
Common ceremonies and shared institutions based on identity
To make up for the problems associated with measuring similarity perception via public speeches, another type of indicator is also used: the affirmation of common bonds via shared ceremonies that reify shared identity. As Lisa Wedeen notes, spectacles, a subtype of public ceremonies, are "systems of signification and community” and "functional strategies to enforce dominance and construct community” (Wedeen 1999, 13). Ceremonies and rituals can vary in their level of regulation and strucnire. They can be singular events, like the Persepolis Celebrations of the Shah of Iran, celebrating, defining, and reifying monarchic tradition or regular events with established protocols, like succession festivities, royal weddings, or funerals. On the most institutionalized end of the spectrum, rituals, ceremonies, and protocols can find regularized expression in common institutions and organizations whose memberships are at least partly defined by identity, e.g. the Gulf Cooperation Council.
Connecting the first step in the SPSP pathway to its predicted outcome, we can claim that similar political systems that develop the perception of belonging to the same group do not go to war against one another, because this option is eliminated as unacceptable inside an ingroup.
This does not preclude members from going to war against "outsiders”, where violence is not ruled out. If the ingroup identity is particularly strong, the same mechanisms could also lead to ingroup alliance. This perspective not only explains peace among democracies but, in addition, peace among dyads of other political systems, such as monarchies.
Once ingroup identity has been established and affirmatively sustained, it is expected that the ingroup members share core norms of behavior, most notably nonviolence toward ingroup members. If the identity is especially central or if there is a strong common threat, positive norms might become guiding principles as well and lead to sustained cooperation and mutual support. To test nonviolence norms, it is insufficient to look at general relations. Decisive for the strength of a norm is whether it still holds in circumstances when strong circumstances or interests (e.g. survival or hegemony) play against it. For the norm on nonviolence, this would be a serious crisis, which might emerge even among "family” members. In that case, ingroup identity enables and prioritizes de-escalation and cooperation, whereas a lack thereof makes escalation to war more likely. The following three main indicators serve to measure these different possible outcomes:
1 Military restraint
This first and foremost means refraining from military action - war and MID - and disavowal of violence in general within the ingroup. Generally, the definitions of the CoW and MID data sets will be employed to define the military clashes to include in the analysis. The CoW data set uses the threshold of 1000 battle-related deaths in the span of a year to record a war, whereas the definition of “MID” is “a set of interactions between or among states involving the threat, display, or use of military force in short temporal intervals” that is overt, non-accidental, government sanctioned, and government directed (Maoz and Abdolali 1989, 13). Only MIDs with a hostility level of at least 4 (which means “use of force”, whereas 5 means "war”) will be considered instances of military actions, although others are still noted in the analysis.18
2 Non-military restraint
War and even militarized disputes are rare in interstate interaction, whether of similar or dissimilar political systems, and we thus need additional indicators to measure restraint from escalation of monarchic elites. They are only the most dramatic culminations of escalation spirals and crisis slides. This set of indicators serves to identify the actions that are most likely to lead to an escalation to war by analyzing dynamics just below the threshold for interstate violence. It is rare that war breaks out without any previous signs or buildup and consistently poor prior relations. Rather, severe militarized conflict is usually preceded by some of the following steps: delegitimization, subversion, and rhetorical escalation. Secondary implications of monarchic de-escalation thus would be refraining from such measures and instead framing the conflict in terms of disagreements instead of existential incompatibility.
Refrainingfrom delegitimizing the opponent’s regime
To legitimize an act of war against an opponent usually also means justifying one’s own cause and delegitimizing the other to ensure backing by the populace and allies. The delegitimization of the opponent's regime at the least (or, in some cases, the whole state) indicates a break of ingroup norms in that it legitimizes its removal or transformation altogether, acting directly against the prime drive of all leaders: regime survival. It is distinct from mere criticism of individual policies or personal criticism of the ruler or some of their individual associates. The personal delegiti-mization of the ruler is also strongly confrontational, but the removal of the ruler without the removal of the regime might (and in Middle East monarchies often does) mean mere retirement and often comfortable asylum in the near abroad. Removing the regime, however, often means incarceration or death and the discontinuation of the rule of all the ruler’s allies (and, in case of a monarchy, relatives altogether). SIT posits that a mutual recognition of equality (in kind, not in power) is a hallmark of ingroup identification. The theoretical expectation is thus that once ingroup identification is present, delegitimizing the regime or revoking previous recognition of legitimacy of the regime should be unacceptable inside an ingroup because it would also mean a pars pro toto delegitimization of the club itself. Delegitimization can take place on the basis of the disavowal or discrediting of the ruling ideology, on the basis of sovereignty of a regime ("puppet regime” or foreign agent) or on the state level by claims that the state itself is not a viable or “real” state.
It is impossible to clearly separate this indicator from the indicators for social identity processes, because there is a feedback effect between the successful settling of a conflict or de-escalation and an affirmation of monarchic identity. Similarly, there is also a feedback loop between military escalation and the othering of an opponent. For the sake of the concision of the analysis, these feedback effects are ignored. In general, social-process indicators (especially the emphasis of similarity over difference) are taken to mean "typical” behavior in "normal” times, whereas the delegitimization of regime or the refraining from doing so is measured during ongoing crises.
Refraining from subverting the opponent’s regime
For the same reason, subversion, i.e. the attempt to undermine and change the regime/political system of the other, via e.g. coups or the support of subversive opposition groups in the target country, should also be off the table in the case of a monarchic ingroup. If threats to state and regime survival are the prime threats that foreign policy decision makers have to cope with, subversion as being directed against regime security is one step removed from militarized action (directed against state integrity).
Casting the conflict in non-existential terms helps prevent escalation spirals with rhetorical reprisals that might spill over into action. This restraint might be, e.g., in territorial conflicts seen when staking claims to small parts of the opponent's territory, but not the whole, emphasizing that the disagreement is over issues, not over identity or mutual incompatibilities or a question of essential security and refraining from threats of violence. Escalating rhetorically and securitizing a conflict raises the stakes for all parties and can lead to conflict escalation. In contrast, rhetoric restraint signals that it can be solved peacefully and is a de-escalatory tactic that favors cooperation and compromise.
In contrast to ingroup members who tend to frame conflict issues in de-securitizing ways, ruling elites who perceive each other as belonging to different groups are more likely to inflate, securitize, and dramatize conflict issues.
3 Alliance and solidarity
Although the focus lies on the de-escalation of conflict and therefore an ex negative concept of non-war, an important consequence of ingroup identification is positive - namely monarchic solidarity, expressed e.g. in persistent alliances (in contrast to ad hoc coalitions that merely indicate convergence of interest) (for the difference between alliances and coalitions, cf. Kober 2002, 1). The key difference between overlapping interests and joint ingroup identification resulting in a foreign policy convergence and coordination and nonviolent dispute resolution is that the former is episodic and ephemeral, i.e. it can disappear as soon as the interest convergence vanishes. The latter, however, can bridge difficult periods when members of the same “club” have
Figure 2.1 Causal chain links and steps 1-4
clashes over interests and still refrain from violent escalation. One example of ephemeral interest convergence is the foreign policy of Iraq under Saddam Hussein, which was highly cooperative toward the Gulf monarchies during periods of shared interests, e.g. during the Iran-Iraq War, but it attacked one of its major supporters and hinders, Kuwait, after the war and the convergence of interests passed.
According to the underlying theoretical expectations, a longer-term alliance (in contrast to an ad hoc coalition due to interest convergence) is less likely to occur and persist among mixed dyads (but evidently not impossible). Especially strong indications of a "club” are an institutionalized form of security and defense-related formal alliances, e.g. intragovernmental organization with security-related provisions.
The causal mechanism as adapted for the MP and broken up into four causal chain links as a subtype of the SPSP is illustrated in Figure 2.1.