Introduction

Over the past decades research on the possible connections between climate change and armed conflict has blossomed. However, this field has failed to establish any clear causal connections between environmental degradation or climate change and conflict in the past. Seemingly assuaging some of the fears concerning climate change and conflict, this research does not allow us to conceptualise, much less operationalise, that actors may break with past patterns and act in new ways in relation to the wholly novel set of phenomena of climate change. Analytical or theoretical deficiencies may be important on their own in ‘purely’ scholarly terms, but the practical implications of these deficiencies create an urgent need for other kinds of analyses. If we cannot adequately understand the connection between the effects of climate change and conflict, we might be caught by surprise by conflicts and miss opportunities to prevent them.

This article advocates hermeneutic understandings of the potential connections between climate change and armed conflict. I outline a perspective focused on actors and their motivations for conflict in light of climate change, which changes the main question of this research from ‘is climate change a security issue’ to ‘how and why can climate change become a security issue?’ The main argument of this article is that climate factors are not the main drivers of conflict under conditions of climate change (and therefore only feature marginally in this article). Instead, the main mechanisms are how actors interpret their historical experiences and roles as guides for future actions and how international structures shape the scope of action in a constitutive fashion. Each mechanism is important on its own and acts in relation with the other. In this article climate change is understood as a social as well as natural fact. As such its political impact is shaped by the actors that interpret it. This understanding entails a focus on local rather than global aspects of climate change.

The article centres on these two mechanisms and illustrates them with an empirical study of the relation between Ethiopia and Eritrea. Taking account of the actors’ perceptions and interests we can narrow down the ways in which armed conflict in connection with climate changes may occur. In some situations, actors’ interests could be served better by cooperation or by refraining from violence. In sum, risks and opportunities can be more realistically gauged by replacing an outside analytical perspective, which treats political actors as passive or disinterested, with an inside perspective that treats actors as intentional and rational subjects.

The perspective on the connection between natural and social events outlined in this article has three elements refining three previously dominant tendencies outlined below: (A) Natural events are not independent from their intersubjective meanings and therefore never impact directly on human actions. Rather, ‘nature’ itself is a social factor. (B) In this light, natural events as social factors become part of a configuration of factors with constitutive effects on social action. In international politics some of the most important social factors are the historically and spatially variable articulations of interests and role constructions. Research on socio-political implications of climate change should focus on change and possible transformation. In the field of international studies, the major question should be whether the structures of the international system could be transformed. The present article contributes to this larger scholarly enterprise by analysing how security actors could think and act in innovative ways. (C) Actors who are rational and intentional and therefore act proactively to serve their interests in relation to a context where political institutional factors, relations with allies and adversaries and anticipated and actual climate change are woven together.

I have chosen the relation between Ethiopia and Eritrea because they are the most important actors in a volatile and strategically important region. Therefore it is crucial to know if they might go to war in connection with climate change or if they will be able to maintain a peaceful co-existence or even cooperate. Not only is it important to know more about the case because of the region itself, but the case illustrates the theoretical arguments of this article and fills gaps in the wider literature on climate change.

First, the article illustrates that conflict in connection with natural events depends on the way those natural events are interpreted and how the international system creates a particular scope of action. The catastrophic drought of the 1980s occurred during an ongoing complex and merciless civil war. Hence, rather than causing conflict, the drought was interpreted by the regime as a weapon against its adversaries. The conditions of bipolarity existing in the international system at the time enabled the regime to receive support despite these policies.

Second, it demonstrates that history is fundamentally ambiguous, as it may be constructed as a story that has to be continued or one that needs to be broken with. Today, the Ethiopian history of drought, famine and insurgencies may produce two radically different experiences and hence two guides to future action. One points towards preventive repression, the other towards the peaceful management of environmental insecurities.

Third, it shows how the international and regional systems have constitutive effects on the scope, possibilities and incentives for conflict. The study shows that Ethiopia and Eritrea have strong incentives and abilities both to maintain the status quo and yet simultaneously to break with the conflict-ridden past. Also, the current configuration of the international system restrains the actors from going to war again and might limit their likelihood to securitise climate change.

Fourth, it highlights the importance of studying how future transformations of the international system could affect risks of regional conflicts related to climate change. If the international configuration would change in a more bellicose direction or towards the securitisation of climate change, then the risk of conflict between Ethiopia and Eritrea might also increase. In sum, cooperation is as plausible as conflict.

I have chosen the form as well as the content of the study in order to illustrate theoretical points and to fill gaps in the existing literature. While earlier research focused on regions or on individual states, I believe studying relations gives a more concrete picture of what motivates actors to fight or cooperate — as demonstrated in another context by David Lake (2009: 4−7, 28−40). Conflict is a social relation and aggression is pursued in relation to goals and risks pertaining to a specific adversary. Analysing relations makes research on climate change and conflict more concrete and less about unspecified risks. Examples of relations that would also be valuable to investigate in relation to the climate change-conflict nexus are Israel–Palestine and India–Pakistan. These relations also take place in volatile and strategically important regions. However, the Ethiopia–Eritrea is suitable as a first illustration of my proposed framework, because it illustrates all the main arguments.

Finally, by focusing on an inter-state relation the study contributes to the literature on environmentally induced conflict. Earlier research, on the Horn of Africa and elsewhere, has focused on sub-state actors. However, states are important to study because of their greater capacity for conflict and their potential to enforce cooperative adaptation policies in the face of climate change.

A reason for the lacuna in previous research is that it is based on a simplified view of the connection between climate and political action. Generally speaking, previous research tends to be organised according to the following conceptions: (1) Climate change is seen as an exogenous independent variable; (2) natural conditions in the areas under study are conceptualised as disconnected from economic, political and social structures; (3) connections between natural phenomena and political action are seen as relatively generalised and have a strong functional and structural cast. In such a causal chain, four central elements of social theory and political action disappear: Socio-political structures that give constitutive shape to action (i.e. permissive causes); intersubjective meanings, interpretations and understandings; constitutive institutions and the intentional actions of political actors (i.e. proximate causes). A result is the obscuring of actors, their motives, interests, identities, experiences, hopes and fears. Hence, it becomes difficult to understand climate change as a change woven into the strategic calculations of different political actors.

The investigation is structured in the following way: the next section provides an in-depth survey of previous literature in the field. I do not seek to refute previous research as much as to complement it and to open perspectives that enable us to ask new questions. Advances made by Baechler (1998), Barnett (2003), Barnett and Adger (2007), Dalby (2002), and Homer-Dixon (1994) — to name some of the more prominent ones — can be taken further through a perspective that stresses the interpretative and intersubjective character of social relations and which emphasises that the rationality and intentionality of political actors are embedded in local contexts. To clarify the need thereof I analyse the foundational assumptions of previous research to argue that these assumptions limit their optics to a specific set of questions, assertions and answers. Given the wealth of material, the survey of literature does not aim at being either exhaustive or conclusive but instead relies on a selection of examples.

The section after that outlines the theoretical argument. The penultimate section illustrates a part thereof by analysing the conflict between Ethiopia and Eritrea. The study deals only with the relationship between the two leading groups of the respective countries, the Ethiopian People's Revolutionary Democratic Front (EPRDF) and the People's Front for Democracy and Justice (PFDJ) in Eritrea. This relation sits in a complex web of relations between other actors, state as well as sub-state in the region. I have not dealt with these actors because a single article is insufficient for the level of concretisation I believe is necessary in order to pinpoint the risks of future conflicts. Hence it must be left to future studies. The last section concludes the article and outlines the implications for future studies.

Previous research: variables and time

The quest for causality

Scholarly research and policy-oriented reports on climate change and conflict are guided by the question ‘will climate change (independent variable) cause violent conflict (dependent variable)?’ There are differences in nuance, often depending on institutional vantage point. Works published within the context of the U.S. policy community phrase the question in terms of threats to the U.S. (Schwartz and Randall 2003; Busby 2007; Campbell et al. 2007; CNA Corporation 2007; US Senate 2007). Scholarly works from the peace research community focus rather on the developing world (e.g. Baechler 1998; Barnett and Adger 2007; WBGU 2007: 157−76), sometimes with an explicit focus on human security (Nordås and Gleditsch 2007: 631). Hitherto, clear connections between climate change and/or environmental degradation and armed conflict have not been demonstrated (Barnett 2000; Matthew et al. 2003). This observation begs several questions: Has the research field actually reached conclusive results and should we therefore close the debate, as proposed by Dyer (2001)? If the answer is negative then we must ask if research so far has been conducted in an adequate manner or whether it has suffered from insufficient methodological sophistication. The answer from within the field is the latter, and remedies are to be sought through perfecting current methodologies. Although research conducted within this perspective has been highly useful in a first stage, we now need to ask different questions. In turn this requires a new perspective. I will now examine the tools with which this quest for causality has been pursued.

Social factors are viewed from the outside

Few scholars now contend that climate change will have direct links to conflict; instead, most envision some kind of combination with social factors (Buhaug et al. 2008). Early on Baechler (1998: 32) argued that socio-political factors are the most important determinants of violence in the context of environmental security, not natural phenomena — for example drought, soil degradation and precipitation — as such. Most often, social factors like poverty and regime type, which studies of civil wars and conflict research have demonstrated to have an impact on conflict, are conceived of as objective variables whose meaning is taken for granted. Economic development, political stability, the quality of governance, ‘bad neighbourhood’ and a history of violence are variables that often feature (Buhaug et al. 2008: 20). For example, Urdal (2005) has discovered a negative link between pressure on potential cropland and conflict, but in another work reports a positive link between population density and conflict (Raleigh and Urdal 2007). The discovery of these connections is important, necessary even. But it is not sufficient. Its merit lies in raising several questions. We are left guessing about the differences in the politics of the lack of available cropland and high population density. In the case of the negative link between pressures on potential croplands and conflict, we can speculate that social identities as farmers and access to cropland do not seem to have been used for political mobilisation. This begs the question if different types of politicians are in power in overpopulated rural and urban areas. This in turn leaves us wondering whether local political constellations in fact determine the risk of conflict. In sum, we do not know what the drivers of conflict were. Even worse, we do not know if and when it might happen again. We would have been better informed if this research had been complemented by intense studies of the conflict areas and their socio-political composition.

The orientation towards variables and causality in the field is an inheritance from the slightly older research field of ‘environmental security’. This perspective has been adopted by high-profile national and international reports (e.g.UN 2009; WBGU 2007). The insight that economical, social and technical characteristics of societies will be important for the capacities to manage and adapt to climate change is commonplace in the debate (e.g. IPCC 2007: 17, 19). A country with less developed physical infrastructure and heavily reliant on subsistence agriculture reacts to a drought differently than an advanced industrial society with a well-developed division of labour, mechanised agriculture and access to an international market. Other variables might be population density, or food and water availability. In this respect climate change is still seen as an independent and exogenous variable. But most importantly the structures with which the analysis deals are not intersubjective, nor do they do not give meaning to political action. We must now turn to such structures.

A perspective that traces conflict to social processes that have been compounded into often quantifiable variables such as ‘ethnicity’, ‘national economic growth’, or ‘democracy’ will easily integrate natural variables like drought and ‘food security’. Although useful in providing surveys and overviews, such a methodology tells us little about the internal motivations of actors. Furthermore, institutions that are intersubjective in their character and essentially interpretative do not feature in the analysis. Linking natural and social factors in causal models is common in sustainability science and systems ecology (e.g. Fraser et al. 2003: 139). However, even sophisticated attempts to map social-ecological systems will not give a full understanding of the possible consequences of climate change (e.g. Janssen et al. 2006) unless the analysis accounts for the different character of social and ecological nodes in the network. Social factors like poverty are tremendously important, but they cannot merely be treated as ‘intervening variables’ that may foster social unrest and conflict (Kasperson and Kasperson 2001). If political factors are compounded and quantified into variables they lose their socially constructed and meaning-dependent character. The social and political meanings of poverty may vary significantly, and this variance may make the difference between social cohesion and violent mobilisation. Studies that inquire about the social construction of poverty might discover that the point where perceived deprivation becomes a potential for conflict may be another than a quantified definition of poverty. As Gurr (1970) has pointed out, perceived relative deprivation may be more important than absolute and quantifiable deprivation. In short, a reliance on quantitative studies that cast poverty as a variable will tell us perilously little about concrete risks of conflict.

Suggestions for further research formulated from within the field centre on refined causal models combining natural and social factors (Buhaug et al. 2008: 37), closer coupling between climate models and formal models of conflict and an increase in the number of variables (Nordås and Gleditsch 2007). Research made so far in the field has laid important groundwork, but I believe that its current framing may have exhausted its capacity to generate answers. This does not mean that the field is not in great need of further exploration and increased understanding.

The absence of actors

One way to increase our understanding is through detailed studies of processes whereby climate change has been or may become implicated in conflictual relations. The existing literature does not describe in detail how political actors may relate to climate change with large-scale violence. Some groups that are frequently pointed out as drivers of conflict, for example refugees or nomadic populations, are silent in the literature (e.g. Molvær 1991; Meier et al. 2007). We seldom see studies of their motives, calculations, assumptions, rationalities and span of options. I am not questioning the focus of research per se but rather its methodology. However, in terms of capabilities and military power, refugees and nomads are of marginal importance compared to states. The relative absence of actors is understandable in two cases. First, with a long-term perspective it may be hard to envision actors and political dynamics (to which communication is central) and therefore tempting to rely on structural explanations. Second, a reluctance to outline details about actors is understandable if the analysis is produced by a government agency, because pointing out actors that might turn out to be militarily threatening will be diplomatically sensitive.

The need to include actors in conflict analysis is evident from the study of civil wars. Structural analyses of internal conflicts sometimes tend towards a ‘no-blame’ view based on assumptions of automatic processes and metaphors like ‘spill-over’, ‘contagion’ or ‘diffusion’ (Brown 1996). A similar reduction of political actors to instruments of structures is found in studies of ‘ethnic’ conflicts. However, as Bronson (1996: 212) notes, ‘[e]thnic and religious diversity does not in and of itself precipitate domestic conflict’. The co-existence of several ethnic groups with a conflictual history or the wrecking of livelihoods by the effects of climate change can be permissive causes of conflict. But to find proximate causes we must look at actors and particularly at those well placed within large organisations capable of wielding coercive power — for example, states.

Consequences of the state of the art

Previously, analyses of climate change and conflict have been functionalist and structuralist in character. They have relied on quantitative studies of conflicts (Buhaug et al. 2008: 23−24, 30) rather than the type of ‘qualitative’ approaches found in anthropology and sociology but perhaps most importantly, IR theory. They also attempt to uncover otherwise obscured causal mechanisms between natural factors and social effects. None of the articles surveyed here make specific reference to scientific ‘laws’, but the methods and approaches suggest affinities with the search for the discovery of law-like correspondences. Once found, such correspondences could be countered by procedures that initiate other, contrary, correspondences that are considered more politically and normatively desirable. If this is the only orientation of research, then the field runs the risk of missing important phenomena, some of which may be harmful and some of which may be more hopeful. The following deleterious effects might result if an orientation towards variable-based nomothetic causation remains the dominant way of analysing climate change and conflict: (1) a privileging of observable causal factors and the omission of constitutive ones — or, expressed in Aristotelian terminology, emphasising final causes but excluding ‘material’ ones (cf. Kurki 2008; Lebow 2009); (2) impaired planning and preventive capacities due to the projection of an implicit rationality onto the actors under study, and (3) an overemphasis on continuities, as variable-based nomothetic analysis tends to rely on a combination of ‘regularity theory of the meaning of a singular causal statement and the covering-law’ of causality (Suganami 2008: 330, 332). Overreliance on generalisations and continuities with past actions can lead to an inability to conceptualise that actors might break with the past and act in new ways; and (4) an inability to conceive of, and outline, future transformations of the constitutive structures of international politics.

The search for ‘law-like’ features steers the investigation of climate change away from understanding potential changes in the structures that are constitutive of strategic decisions about conflict or other means of furthering social and political interests. Climate change, as a compounded political issue as well as through its distinct effects, may create changes where actors change their ways of thinking and acting because of their interests, fears and identities. In this interpretation climate change is not a quasi-acting subject but becomes also an object at the centre of actors’ thinking, planning and acting. Since much of the literature has predictive ambitions, the inability to accommodate actor-driven change is problematic. Moreover, its generally phrased results give little or no direction to military and/or humanitarian planning or guidance between distinct diplomatic options. Because security and to some extent development policies entail relations to specific actors, it is difficult to draw operative lessons from structuralist conclusions.

Retrospective time, not intersubjective history

Campbell et al. (2007) is an example of how an non-historicised view of international politics retains a strong influence in debates on climate change and conflict. It outlines how climate and the environment have had an impact on human societies ‘in history’ (Campbell et al. 2007: 23−33). While laudable, the historical survey deals not at all with how political institutions have varied over the long period that they cover. While correctly pointing to the fact that natural disasters sometimes have led to persecutions of minorities (ibid.: 29−30), no mention is made of the systems of belief or political structures that made such mobilisation possible and attractive to the protagonists. Drought may well have coincided with the rise of slave trade in Africa from the 1550s onward, but we are left guessing as to social and political permissive causes (ibid.: 32). It could be argued that it is too much to ask of a work that is not purely scholarly to take a more nuanced view of historical sociology. I disagree. Doing so would generate a more complex and hence more precise understanding of this complex and potentially dangerous issue.

An implicit materialist logic of action

Explanations of past climate change conflicts (and in particular predictions of potential future ones) that assume material and structural factors are interpreted identically by all actors harbour a more substantial problem for prediction and policy: They contain and transmit implicit assumptions of the rationality that might be driving this kind of conflict. In other words, the conclusions of an outside-retrospective analysis are turned into a prognosticating logic of action. Material deprivation of various kinds, filtered through various social institutions, is a necessary and sufficient motive for the initiation of armed conflict. Understanding the rationality behind a series of actions is obviously important and laudable. However, the problem is that the origins of this rationality lie in retrospective studies of principally materialist factors from an outside perspective. Hence, the assumptions of rationality built into the explanations compound the original problem: Not only do we lack an understanding of the world-views, rationalities and logic of action of relevant actors, but the materialist-structuralist emphasis in previous research implicitly supplies us with an inaccurate image of logics of action. They are inaccurate because they are derived from our own unstated rationality, which we project onto the actors whom we try to understand and whose actions we attempt to predict. A lack of knowledge may entail that we do not know what to look for, but having the wrong knowledge makes us look in the wrong places.

The concrete consequences of working with the wrong assumptions are worth expanding on. Results of scholarly analyses of conflict often find their way into more operative sectors of intelligence-gathering/analysis and policy formulation. Lacking an understanding of why somebody acts in a particular way entails lacking the basis for understanding that actor's future actions, not only with regard to conflict but also to cooperative options. Without a grasp of actual world-views and rationalities, we might come to believe that a situation is ‘safe’ just because the material conditions we believe bring about conflict are not in place. Simply put, if we only judge situations involving other actors on the basis of our own specific rationality (explicitly or implicitly stated), we might miss risks of conflict and opportunities of resolution and cooperation. As underlined by the literature on the risk society (e.g. Beck 1985; Rasmussen 2006), any strategy to avoid risks might produce new ones. This might be unavoidable but uncertainty could be limited by more precise planning.

A hermeneutic alternative: from ‘nature’ to society and from ‘time’ to history

Allegedly, the social sciences have problems conceptualising and explaining connections between ‘nature’ — or rather ‘natural systems’ — and human societies (Deudney 1999). Charting the climate change-conflict nexus is not just about conceptualising relations between ‘nature’ and society’. Natural phenomena never affect human societies directly but are always mediated by different aspects of these societies. Human actions are events that must be expressed and understood in intersubjectively meaningful ways in order to take place at all (Gadamer 2004: 268–91). Hence, social actions cannot be generated by anything other than social factors, although the social interpretation of natural events has in many cases been important in shaping societies. Action is always connected to systems of beliefs; rational human action always proceeds from purposes or motives of some kind, which in turn derive from formal and informal social and political institutions (Scott 2003).

The relation between natural events, interpretation and intentionally acting human subjects can be clarified by means of an analogy with Niklas Luhmann's systems theory, which divides the social world into functional subsystems, for example law, politics and economics. Each system operates according to its own organisational logic and is closed to systems in its environment. Hence an occurrence in one subsystem, for example politics, cannot have a direct effect on another subsystem, for example law. However, the different functional subsystems observe each other through their own terms and logics. An event occurring in one system has to be translated into the terms with which other social systems operate in order to become intelligible (Luhmann 1995). Once observation has taken place systems react to changes in systems in their environment according to their own logic. Let me stress that I will not use a strict Luhmannian framework for analysing climate change in this article. Luhmann does not conceive of either ‘nature’ or natural systems (e.g. the climate system, discrete ecosystems) as functional subsystems. The analogy is used as a heuristic tool to illustrate an epistemological perspective with concrete analytical consequences. Natural systems and events are not conceived of in an essentialist way — to put it plainly, nature never ‘is’. Rather, nature always ‘becomes’ something, subject to human interpretation. This conception allows us to pose different questions than those asked hitherto in the debate.

Climate change is a political and social issue and thus has to be addressed as interwoven with political and social structures. As such, climate change is a social as well as a natural fact (Searle 1995) and part of the intersubjective world-view of an increasing number of political actors. This entails that a variable-based approach becomes difficult to sustain. Focusing on the place of climate change in the world-view of actors, the relevant question becomes not ‘is climate change a security issue’ but rather ‘how and why can climate change become a security issue?’.

The influence of climate change is constitutive rather than causal

A focus on how social and political structures constitutively shape political action takes us away from ‘Newtonian’ causality and thinking in terms of dependent and independent variables. Instead we deal with constitutive or material causality (Dessler 1989: 453; Searle 1995; Wendt 1999: 165) where the composition of elements in a certain structure gives rise to a particular event or process. Since constitutive structures generate and to some extent direct action, omitting them risks misunderstanding the importance of agency and actions (March and Olsen 1989; Searle 1995; Scott 2003; Teschke 2003). What elements are important as constitutive structures in terms of analysing the climate change and conflict nexus? The following are limited suggestions of important structures. The empirical section below demonstrates their capacity to generate analytical mileage.

International systems vary in time and space

International systems produce different kinds of patterns of action, both synchronically and diachronically (Watson 1992; Bobbitt 2002; Teschke 2003). Constructivist works on the shifting historical character of international politics show how variation in international institutions produces different styles of politics, including how interests are formulated (Kratochwil 1982; Onuf 1989; Reus-Smit 1999; Smith 1999). In contrast, neorealism takes for granted that interests are formulated and pursued identically throughout history (Smith 1999; Osiander 2007). In turn, this idea has permeated much of public discourse, including that on climate change and conflict.

If international politics are seen as constant, nuances in climate change are paramount for analysis. However, if we recognise that international politics are shaped by historically contingent structures, then characteristics of and changes in those structures, and not the kind and degree of climate change, will be recognised as the most important factors in determining whether climate change will be politically handled through conflict or cooperation. Clearly, an international system (either regional or global) in which security is understood as divisible, interests are pursued regularly with recourse to military means, and power is understood in a zero-sum fashion, would under conditions of climate change be more likely to produce armed conflict than a more peaceful one. Hence, preserving multilateralism and a broad repertoire of diplomacy and tools of international conflict management is as important as mitigating climate change to avoid conflict.

Spatial variation in international relations and security politics has been dealt with in several ways; one of them is Regional Security Complex Theory (e.g. Buzan 1991; Lake and Morgan 1997; Buzan and Waever 2004). They share the idea that the socio-political constitution of the immediate geographical context determines the shape of security relations but differ with regard to whether Regional Security Complexes (RSCs) are exclusive (Buzan and Waever 2004) or overlapping (Lake and Morgan 1997). I will not go into the details of the rival models but proceed from Buzan and Wæver's definition and divisions while noting that external actors influence world regions (like the U.S. in the Horn of Africa).

The relation between acting in a security region and on the world stage is, however, a two-way street. For a long time social theory worked with an assumption that cultures and societies have well-defined, or at least delineable, boundaries. On closer examination, such assumptions are problematic in historical as well as in contemporary politics (Giddens 1985: 329). Although the institutional milieu in which actors move limits some political actions and favours others, in today's world internationally connected elite groups respond to political impulses from beyond their immediate physical environment. Regional actors engage with the outside world in order to procure material resources in regional conflict as well as to reinforce notions of identity and prestige. Actors do not only act on the world stage to secure advantages in their specific region but also adjust their actions at home in order to secure material advantages and their standing globally. Buzan and Waever (2004: 51−65) define RSCs in a structuralist way explicitly focused on inter-state interaction. Political spheres can also be characterised through a more intersubjective focus, by specific ways of constructing identities and perceptions of the ‘other’. Although location and physical geography matter, the principal landscape of human actors is social and political, and the terrain they navigate is made up of other actors, political issues, their own spaces of experience and horizons of expectations, including hopes, fears, risks, threats and strategic calculations.

Wendt (1999: 258) outlines three different international cultures: the Hobbesian, Lockean and Kantian. Hobbesian cultures tend towards identity constructions where the ‘other’ is seen as an enemy to be vanquished and whose legitimacy is doubtful (Wendt 1999: 270). In Lockean ones the constructions of ‘self’ accord a greater respect for the other and construct him as a ‘rival’, whose right to exist and whose legitimate interests are recognised. In a Kantian culture actors do not construct each other as enemies or rivals, but as ‘friends’. Both the study of RSCs and Wendt's approach imply that we need to study specific relations, not only countries or regions. The empirical study below pursues this line of inquiry by analysing the specific relation between the leaderships of Ethiopia and Eritrea.

Rational and intentional actors will be proactive, not just reactive

Structures do not determine human action, but give constitutive shape to the range of possible actions. They restrain, enable and in some cases force some actions. In addition to not being structural automatons, actors are not only reactive but also proactive and will be so in relation to the effects of climate change. Therefore it is important to gauge how they might take advantage of the openings that climate change provides vis-à-vis their interests, goals and adversaries. These opportunities derive not from material factors but from the configurations of interests, goals and political relations — conflictual as well as cooperative — in which actors are located. Scenarios can be imagined in which politicians securitise (Buzan et al. 1998) climate change and mobilise coercive capacities for internal or external use. Securitisation may originate in genuine threat perceptions and legitimate security interests or from an instrumental strategy whereby an actor seeks to employ the ‘threat’ for ulterior purposes. Security in this sense is relational. If actors prepare for a future characterised by military conflicts and climate changes they might trigger similar reactions among their neighbours. Conversely, if climate change is framed as natural phenomena requiring adaptation by civilian and cooperative means, then security does not automatically deteriorate. However, it is also as important to identify instances where political actors may find that their interests are best served by new cooperative strategies as it is to identify risks. Being proactive sometimes means being innovative and finding new ways of positioning oneself.

Constructed history, not past correlations, provides insights to future changes

We are in danger of missing risks of conflict that could arise due to perceived insecurity or due to fears of coming degradation of resources. At worst, the findings of previous research could give us a false sense of security and lead to failures of the intelligence communities and others engaged in early warning analyses to foresee conflicts. The problems arising from the materialist bias in previous research were outlined above. Even more problematic consequences lie in the way retrospective studies view time and how they neglect the way actors interpret history as a guide for future actions. Urdal (2005) is in some sense right in claiming that the past is our only guide to the connection between climate change and conflict. The actors we study use the past as a guide to the future too. The past is, however, not objective and removed from interpretation. Rather, the past is always interpreted as value-laden history. Therefore we must take care how we construct and study the past. History must be understood as past experiences that contemporary actors interpret. If we only record and report past events abstractly as correlating variables we miss the fact that actors may decide to depart from historical patterns (which might be well documented by previous research) and act in ways that they have never acted before.

Past occurrences are remembered and interpreted to form guides for future action in anticipation of coming events. In this sense we touch upon the need to recognise and investigate the socially constructed need or desire to pre-empt danger rather than to be surprised by it. We must work with modes of understanding that make room for the possibility of that actors learn from and correct their mistakes, or policies in retrospect perceived or constructed as mistakes. The desire of one actor (or many) to break with the past and correct mistakes may take many forms. One way could be to break with a history of defeats and victimisation by bolstering one's collective military capacities — for example in the creation of the state of Israel after World War II. Another could take the form of breaking with a violent past though policies of peaceful cooperation and pacifism — for example in the case of West Germany's foreign and security policy after 1949. Both examples are radical breaks with past patterns and political innovations on the basis of experienced and interpreted history. The creation of both entities with their distinct outlooks on foreign and security policy transformed the Middle East and Europe. Crucially to the present argument, neither development could have been deduced from a quantitative and retrospective study of past events. Indeed, these changes are inexplicable by only looking retrospectively at past events; to understand them we need to understand the interpretation of the past and the construction of history. Serious examples can be given of strategic situations where actors did not act on the basis of the past but rather on the basis of a fear that the future would be radically different from the past and a desire to avoid that future. The most devastating one is the start of World War I, when German and Austro-Hungarian fears of future decline influenced the decision to initiate conflict (Schroeder 2004).

Reinhardt Koselleck's (1985) twin concepts of ‘space of experience’ and ‘horizon of expectation’ capture this tension. Struggles about the meaning of the past become implicated in struggles between rival political interests over the shape of the future. Experience and expectation are not always directly linked; when rapid changes occur that make experiences obsolete, older expectations also become obsolete, forcing actors to innovate and create new expectations for the future (Ifversen 2003; Uhlin 2007). This argument may be highly pertinent in the context of climate change: the prognosticated changes are of a magnitude that may exceed all past experience, such that new expectations of conflict or cooperation will have to be formulated. Hence, continuation along a trajectory of past events may be unlikelier than innovation. The element of uncertainty looms large in Koselleck's writings about action, crisis and expectations. A comprehensive summary of these or other writings about risk are neither possible nor necessary here; suffice it to say that the element of uncertainty in climate change and its potential connections to conflict demand that ‘path-breaking’ (Stråth 2009) must be taken seriously. The section below will chart the possibilities of innovation on the basis of an interpreted past, shaped by regional structures of security, identities and role constructions. The political need to understand how actors experience, interpret and construct history in order to have a basis for future action cannot be stressed enough. Understanding their reasons and inclination to act in new and unexpected ways is of extreme importance for humanitarian NGOs, military intelligence services and diplomatic corps alike. The future cannot be foretold, but we must be ready for changes. Estimating the readiness and ability of an actor to change and innovate is important in order to gauge the risk of future conflict.

The section below illustrates parts of the framework of the article by analysing the relation between the main political actors in Ethiopia and Eritrea, the EPRDF and the PFDJ. The following empirical evidence will be sought in order to assess openings for armed conflict and cooperation: (1) How do relations in the region limit the range of actions? (2) How do actors view neighbours, and what is the likelihood of threat perceptions? (3) What are the incentives for cooperation, conflict and status quo? (4) How liable and capable are the actors to change past patterns of action?

Conflicts and climate change: the Ethiopian–Eritrean duel

This article argues that the central mechanism determining whether climate change might give rise to armed conflict is neither natural factors nor socio-political variables deduced from an objective and reifying reading of past events. Rather, the central mechanism is the construction of the past as a guide for adversarial action and how international structures give constitutive shape to action. The relationship between Ethiopia and Eritrea was chosen to illustrate these mechanisms because the lessons of the past are ambiguous and give reasons for continuity as well as discontinuity with past actions in relation to drought, famine and conflict. International structures in large part provide further incentives to break with the violent past as they offer important gains in power and reputation. Thus the case illustrates the complexity of political possibilities of action in connection with climate change.

As stated in the introduction, environmental or climate factors per se are not the crucial drivers of conflict and hence they will not be analysed in this case study. Some points, however, should be noted. Environmentally, Ethiopia and Eritrea are largely arid, and agriculture is precarious in parts of the country. The country's vulnerability to drought has led to famine on several occasions. Some famines have coincided with armed conflict, but they have not caused conflict; rather, the opposite has been the case. The former emperor Haile Salassie and the dictator who replaced him, Mengistu, used hunger as a weapon against insurgent groups and the populations believed to be supporting them. In the latter case, the government's own counter-insurgency operations made the famine worse by attacking grain-producing areas (Meredith 2005: 332, 334). The repeated failure of rains in 1984 was a calamitous event but the government's initial lack of response and subsequent instrumentalisation turned it into a catastrophe. The then-prevailing situation of international bipolarity enabled the regime to gain support from the Soviet Union despite its destabilising and inhumane policies. Like many other African dictatorships, Ethiopia was in part sustained by the rivalry of the Cold War which led the superpowers to support proxies.

The relation between Ethiopia and Eritrea is highly tense and their recent history is conflict-ridden. Judging by the historical record of war and proxy conflicts, this would seem to be a relation likely to descend into violence once more due to pressures induced by climate change. Their poor relation and the undemocratic characters of both (Eritrea is a dictatorship and Ethiopia has authoritarian tendencies) also seem to increase the risk of inter-state war under the pressures of climate change. The frequent securitisation by both countries’ leaderships of the relation lends further credence to this image. However, given the strategic objectives of both leadership groups, the relation presents an ambiguous picture with respect to the risk of conflict triggered by climate change. The situation harbours more possibilities for co-operation than is revealed by the recent record and present actions of both governments. Below, I will argue that two scenarios are equally likely: (1) the relationship leads to renewed conflict, and (2) the dynamics of the conflict do not entail a high risk of conflict in relation to climate change.

The regional context

In 1998 Ethiopia and Eritrea fought a war involving 500,000 soldiers that resulted in 75,000 casualties and 600,000 displaced civilians (Negash and Tronvoll 2000: 53). The border conflict is still unresolved, and the armies of the two countries are substantial and remain at high alert. The conflict between Ethiopia and Eritrea is part of a wider regional configuration of conflicts and political relations in the Horn of Africa. The civil wars in the Sudan and Ethiopia have previously been intertwined (Buzan and Waever 2004: 245). Both Ethiopia and Eritrea support each other's insurgents. Eritrea supports the Ogaden Liberation Front in Eastern Ethiopia. They are also fighting a proxy war in Somalia through client groups. Eritrea has been providing economical and military backing to Islamist groups like the Union of Islamic Courts (UIC) based on the Hawiye clan as well as the Jihadist al-Shabab (cf. Hansen 2007). With American support, Ethiopia invaded Somalia in December 2006 to combat the UIC (see Quaranto 2008). Even after withdrawal, Ethiopia remains a supporter of the Transitional Federal Government (Menkhaus 2007: 359−360, 363). This situation of latent border conflict and proxy wars would seem to indicate the likelihood of escalation in the face of actual or perceived climate change events.

To ascertain past and future motives for warfare, let us examine the three principal ones behind Eritrea's invasion of Ethiopian territory in 1998: The troubled relation between the ruling parties of Eritrea and Ethiopia, the neo-patrimonial nature of both regimes and the economic problems of Eritrea (Abbink 1998: 552; Lata 2003). Although these motives were present in 1998, it is uncertain whether they remain active drivers of the conflict today and whether they could lead to war in combination with the effects of climate change. It seems uncertain whether the parties to the conflict have any real interest in a resolution of the conflict. Both parties seem content to keep the conflict going at a low level of intensity, since it helps them to mobilise internal support, stifle opposition and remain in power (ICG 2008: 1, 6, 7, 9). Since both Eritrea and Ethiopia stand to gain from keeping the conflict alive and realise that any attempt to force a solution militarily would be costly, it would seem unlikely that either would want to escalate the conflict. In contrast to the North-South conflict in Sudan, where both main actors have unreliable local allies, both Eritrea and Ethiopia are estimated to be in full control over their respective armies (ibid.: 6). This greater degree of control lessens the risk of an uncontrolled ‘flare-up’.

Ambiguous cultures of anarchy

To gain a fuller understanding of the relation we must see it not only in terms of past behaviour and present capabilities but also in terms of constructed relations of ‘self’ and ‘others’. Wendt's scheme of three international cultures (1999) offers a heuristic device to do so. The ‘intensity of mutual hatred and suspicion among the leaders of both governments’ (Negash and Tronvoll 2000: 51), the frequency of mutual denunciationsFootnote 1 and their history of warfare tend towards an interpretation of their relation as ‘Hobbesian’. Indeed this characterisation could be extended to Sudan and the different actors in Somalia as well. The prevalence of proxy wars and mutual support of guerrilla movements are also cases in point. Particularly the PFDJ neither views other actors as friends nor respects them as equals. Rather it sees and treats them as potential tactical allies (Connell 2009: 42) who can be discarded when other more attractive opportunities arise. However, despite the deep enmity and vitriolic tenor characterising relations on the Horn, none of the parties seems to strive for knock-out victories. Rather, risk aversion seems to be a key feature of long-term strategies as well as short-term behaviour. The Eritrean leader Isaias Afewerki and his closest associates perceive the regional system as fraught with dangers to state survival. Although this may look like a state-of-nature interpretation, we must take care to note that it does not preclude alliances or at times the defusing of tensions.

Committing to no one enables a strategy of potentially allying with everyone. Hence, alliances are as evanescent as ideological commitments (Connell 2009: 43). Avoiding alliances can be interpreted as an ability to change course of action and orientation. The lack of principled commitment and willingness of the Eritrean leadership to cooperate in order to secure its own survival may in fact constitute a possible opening for conflict resolution. There may be more chances of cooperation in the face of grave environmental challenges in the region than one might believe if only the current situation is considered. However, Eritrea's equivocation entails long-term risks, as it tends to estrange erstwhile partners.

The combination of Ethiopia's ambitions inside and outside the region and its alliance with the United States act as a restraining force on conflict and hence suggest stalemate rather than escalation. The Ethiopian Prime Minister Meles Zenawi's desire to play a leadership role in the region as well as for the continent means that the country has to act within at least a Lockean culture and pay lip service to the Kantian one which dominates the international community. However, there is a built-in contradiction in Ethiopia's standing since U.S.–Ethiopian relations have so far been based on the latter's role as a partner in the war on terrorism, for which it has received political and military support (Quaranto 2008: 43). The need for Ethiopia's leadership to uphold, but not exaggerate, adversarial relations with the dominant actors in Eritrea and Somalia as well as with some groups within Ethiopia itself could undermine the restraining influence of the extra-regional ambitions of the government.

Judging by the set-up of the region, the risk of renewed war is substantial, particularly since the enmity of the two countries and their involvement in sub-state proxy conflicts keeps the politics of violence alive and creates a number of military risks. This is tempered by the external dimension, by strategic calculations and by the fact that the top priority of both the PFDJ's and the EPRDF is internal control, which points to risk aversion and status quo. As stressed above, a key question is whether this situation is likely to be changed by the main actors. To ascertain that, we must look at how the experience of history influences actors’ expectations of the future and thereby their future action.

History, experience and expectations

Relations between the PFDJ and the EPRDF go back to the 1970s. The two groups cooperated in the civil war against the dictator Mengistu and the junta (the Derg) (Tareke 2009: 45−111). Despite their mutual economic, logistic and military dependence, their relation was fraught with tensions from the start. In the late 1980s, the two groups fell out over political and military differences. Some are now obsolete, like divergent interpretations of Marxist theory and practice. Others retained their relevance when the two groups went on to become dominant in their own states, post-Mengistu Ethiopia and Eritrea, which became independent in 1993. The EPRDF (then under the name the Tigrayan People's Liberation Front) pursued a policy of liberation of all ethnic groups in the multi-ethnic Ethiopia, a policy that has been carried over into Ethiopia's current ethnic federalism. This contrasts with the centralist and non-ethnic vision of the Eritrean People's Liberation Front (EPLF) (which later became the PFDJ), which continued in Eritrea as a state. Because of the many ethnicities of Eritrea, the EPLF/PFDJ perceived and still perceives the ethnic federalism of Ethiopia as threatening (Negash and Tronvoll 2000: 15−17). Of the several things that Ethiopia could do that might provoke Eritrea into escalation, raising the question of autonomy for the many nationalities of Eritrea is one of the most potent. In both Ethiopia and Eritrea the experience of war, first against Menguistu and then against each other, has been seen to produce national unity. For our purposes, it is less relevant whether wars against a foreign ‘other’ actually produce unity. What is relevant is whether the leadership in either or both countries expects that a future war would boost national unity.

The experience of guerrilla struggles against the dictatorship of the Derg and the environmental degradation during the 1980s can be said to produce two different spaces of experience for the Ethiopian leadership. First, hunger, drought and armed regional separatism (de Waal 1994) can lead to state collapse. Hence it is necessary to be proactive against separatist movements as well as external antagonists. In the current context the most significant internal armed opposition is the Oromo Liberation Front (OLF) and the Ogaden National Liberation Front (ONLF). The most significant external antagonist is Eritrea and to a lesser extent the al-Shabab in Somalia. Taking a harder pre-emptive line against the ONLF would escalate not only the internal but also the regional situation. Since Addis Ababa began an offensive against the ONLF in 2007, hardships have increased for the civilian population and destabilised the area (Human Rights Watch 2007; Menkhaus 2007: 387). Conflict in the Ogaden might also enable Eritrea and some Somali actors to escalate the conflict in order to weaken Ethiopia, which would worsen relations between actors in the three states. Hence, this experience could motivate tough measures against oppositionist, minority or actively insurgent groups in connection with or anticipation of adverse effects of climate change. The other space of experience is one of caution rather than proactive repression. The EPRDF emerged as the winner from the volatile situation of famine, drought and civil war of the 1980s. This experience could be interpreted as a need to avoid such a situation from occurring again, by means of political de-escalation and preventive measures against famine, drought and other consequences of climate change. The knowledge that conflict in connection with mismanaged environmental disasters enabled the toppling of the Derg could also induce the leaders of both countries to be cautious in relation to their chief foreign adversary.

It is difficult to say which of these experiences dominates in guiding action, but insights can be gleaned from the current situation. Despite a federal structure based on ethnicity, the country lacks democratic representation and the EPRDF has tightened its grip on power (ICG 2008). The current leadership of Ethiopia has fused the pugnacious world-view of its freedom-fighter past with previous authoritarian traditions of central control. Combined with the vehemence with which it represses the Oromo and Ogadenis, this world-view does not bode well for the prospects of non-violent political solutions to social unrest in the future. However, as we shall see, the attitudes and action shaped by this space of experience may be tempered by the Ethiopian leadership's horizons of expectation.

The PFDJ has based its construction of Eritrea on the experience of the war for national liberation. Since the experience of war has been interpreted as strengthening that state-building project, the PFDJ may be more likely to use an external war to strengthen the country. The isolationist world-view of Isaias and his circle was set during the experience of bitter guerrilla war against the Derg and the space of experience created by fighting and — importantly — by winning alone continues to shape the horizon of expectations of the leadership. Reid (2009) argues that this world-view has been reinforced by even older and deeper roots, dating back to the experiences of Eritrean élites during pre-colonial times of being beset by hostile powers on all fronts. The memory of isolated struggle and perceived abandonment by the international community in the 1970s and 1980s shaped the PFDJ interpretation of how its condemnation by the international community during and after the border war with Ethiopia in 1998–2000 (Healy 2009: 153, 155, 158). The sense of renewed abandonment and betrayal by the outside world would seem to account for the discarding of international diplomacy and the turn towards non-state actors — mostly insurgent groups in Ethiopia, Somalia and Sudan — as partners. It is unclear whether this should be interpreted as a long-term shift or a temporary stratagem in response to the current situation. In order to forestall Eritrean escalation in the face of real or perceived climate change, more diplomatic engagement with the regime would be advisable. That would give the regime a broader range of political options and an increased sense of security. It would also lessen the risk of forcing Eritrea into a crisis where military action is the only option. Of course, it might be difficult for the West to adopt a conciliatory course of action because of the Eritrean regime's weak human rights record.

The horizons of expectation of the two leadership groups are radically different. While Ethiopia receives recognition and advantages from integration in the international community, Eritrea suffers from international isolation and a faltering economy. One might therefore assume that Eritrea is likelier to escalate vis-à-vis Ethiopia than the other way round. Although a high percentage of Eritrea's population is under arms (ICG 2005: 7; ICG 2008: 10), Ethiopia is considered to be the militarily stronger party and hence less threatened (ICG 2008: 7).

Most importantly, as we saw above, Ethiopia is a key ally of the United States in the region and hosts the headquarters of the African Union (AU). The U.S.-led unipolar world creates a more narrow scope of action for African leaders than did the bipolarity of the Cold War. In a unilateral world there is only one principal centre of patronage, power and protection, and certain rules have to be obeyed in order to gain access to it. It is, however, conceivable that the currently increasing influence of China might fundamentally affect the restraining influence of international structures and give actors like Ethiopia's leaders a greater freedom of action and lower the structurally induced costs (in the form losing protection, reputation, trade and aid) of going to war.

The U.S. alliance gives Ethiopia a stronger diplomatic position than Eritrea (for a background see Woodward 2006: 77−91). The appointment of the Ethiopian Prime Minister as the representative of the AU at the climate change conference in Copenhagen in December 2009 is a sign of Ethiopia's current standing.Footnote 2 This means that it does not need securitisation or military escalation in order to gain external support. Given its international reputation and the benefits it can reap from being a good citizen of the international community, it cannot afford adventurism and brinkmanship. In contrast, Eritrea is isolated diplomatically and has attempted ever more unlikely courtships in the search for alliances — witness its application to the Arab League. Finally, Ethiopia is economically stronger than Eritrea, which makes it more secure (CIA 2009).

Ascertaining the horizons of expectations, albeit briefly, has brought nuance to the analysis. The EPRDF in Ethiopia stands to gain from maintaining the status quo since it expects increasing benefits. However, in relation to internal rivals, its spaces of experience entail that it balances between proactive repression and pre-emptive management of social grievances, between armed conflict and political cooperation. For the PFDJ in Eritrea, the future looks bleaker, which could point towards revisionism. However, we must beware of inferring motives and concomitant courses of action on the basis of our own rationalities which we project onto the PFDJ. The PFDJ and Isias Afewerki focus on survival and have demonstrated substantial resourcefulness and a willingness to change courses of action, abandon allies and ideologies for the greater goal of state survival. This creates uncertainty, but it also makes it less likely that actors will continue acting as they have done in the past.

Climate change and the risks of renewed war

The three major questions remain: (1) How could climate change become a part of this configuration? (2) How could the two principal actors use climate change as a political issue? (3) Will armed conflict be a part of these changes? Three ideal-type situations can be imagined in connection with political reactions to and use of climate change: (1) Real effects (e.g. drought) or perceived future risks could be used as a pretext to escalate in the hope that this would legitimate or obfuscate military action. As noted in the analysis above, it is highly doubtful whether any of the parties could draw advantages from military action. (2) Climate change effects might weaken one or both parties, which could lead the other to military action in order to gain an advantage. Although possible, this course of events may be less likely if both parties are more interested in remaining in power than resolving the conflict. (3) Destructive effects of climate change could be seen as a risk in relation to the other party or as a risk in relation to maintaining internal control. In the first case the combination of one's own weakness and the perceived threat posed by the other could prompt a pre-emptive attack. In the second case, the effects of climate change (e.g. storms, droughts, sea-level rise) could lead to social hardships and economic difficulties, which may in turn cause the leadership to fear losing power. The possibility of shoring up legitimacy through a war with the chief foreign adversary might seem like a risk worth taking.

These ideal-type situations can be envisioned, but how probable are they? Would Ethiopia or Eritrea securitise the issue of climate change, either in its entirety or its individual effects? Risk aversion (c.f. Buzan and Herring 1998: 97−100, esp. 99) seems to play a large role in the strategic calculations of both parties and therefore offers a clue to future securitisations of climate change. Judging from their actions between 2000 and 2009, when their border conflict remained frozen but both regimes were active in securing their domestic power base, both Eritrea and Ethiopia perceived the greatest threat to their security as coming from internal opposition. Ethiopia's invasion of Somalia in 2006 was motivated by the risk that an Islamist and revanchist regime headed by the UIC would mobilise ethnic Somalis in Ethiopia's Ogaden region. Ethiopia's largest military action since 2000 was thus undertaken with the intention to forestall internal challenges to the country's unity. However, presently Ethiopia has everything to win by its international behaviour conforming to the values of its American partner.

The pattern of risk-averse policies being directed internally may be more pertinent in Ethiopia's case than in Eritrea's, considering the latter's frequent perceptions of neighbouring countries as security risks. Initiating conflict always involves an element of chance (cf. Clausewitz 1976: 85−86). Consequently a key question is whether perceptions of climate change will spur decision-makers to favour taking the risk of war in order to escape the risks of a changed climate (c.f. Rasmussen 2006: 88). Both EPRDF and PFDJ are ready and capable of going to war if they believe themselves to be threatened and their mutual willingness feeds a spiral of insecurity. Combined with the risks climate change entails, this creates a climate of fundamental insecurity. It is far from clear that these two survival-oriented groups capable of long-term planning are ready to increase the insecurity about the future by recourse to war. The situation stands in contrast to the earlier example of World War I as triggered by expectations of future decline. In that case the future was seen as insecure but not as open and uncertain, since all actors were familiar with the kind of game being played and its possible outcomes. In this case the future is not only characterised by ‘known unknowns’, but by ‘unknown unknowns’ (quoted from Daase and Kessler 2007: 1).

Since the post-conflict relation has been going on for 9 years, we can assume that it has developed other dynamics since 1998. Both parties have been diplomatically active in the border conflict as well as in other issues, seeking regional allies and support from the international community. The habit of engaging internationally in order to gain an advantage over each other makes it possible that either one could try to make political use of climate change for internal mobilisation to gain international support. Even if the strategic calculus — that neither party wants the conflict to turn into a shooting war — remains constant, securitisation of climate change may escalate the situation beyond control.

To summarise, an investigation of the rationalities of both sides does not point towards renewed war nor necessarily towards improving relations and cooperation. Rather, it indicates stalemate. The analysis above has demonstrated that Ethiopia has a strong interest in remaining a U.S. ally and retaining its international position. Although this position rests on the Ethiopian Prime Minister Meles Zenawi taking a bellicose stance against Islamists, it would be jeopardised in the case of an escalation against Eritrea. As for Isaias Afewerki and his circle, they have a capacity to change their tactics, sometimes drastically in order to further the strategic goal of state survival. The attempt to further interests through a war with Ethiopia has yielded few benefits so far. Hypothetically, the calculus of risk and opportunity of the two actors could be drastically changed if the international environment changed in such a way that removed external restraints and opportunities — perhaps a serious collapse in the world economy. If the players of the Horn were left entirely to their own devices, then the dynamics between them would certainly change, although it would be hard to tell in what direction.

Conclusions

This article has outlined a framework for analysis of future conflicts illustrated with a study of the relation between the leadership groups of Ethiopia and Eritrea, the EPRDF and the PFDJ. This relation was chosen partly because of the region's vulnerability and strategic importance and partly because it illustrates the core theoretical arguments of the article and their interrelation: (1) Natural effects are interpreted through socio-political configurations. (2) The past can be construed as a history requiring continuity or discontinuity with past practices of violent repression and inter-state war. (3) The international system constitutively shapes incentive structures and the ways regional actors pursue interests. The future is therefore open and indeterminate since actors may to some extent choose what lessons the past holds for future actions. Also, international structures that determine the scope of action are also subject to change, albeit in far less controllable and predictable ways. (4) Under current conditions, as good a case can be made for the actors refraining from repeating the conflicts of the past as for them going to war in a future characterised by climate change. If these conditions were to change substantially in the future, then actors might choose other and more violent strategies to pursue their interests. To ascertain the risk of future conflicts in connection with climate change, the really important issue is long-term transformation of the international system. Tackling this issue in detail is undoubtedly more than can be accomplished by a single article.

However, discussions and calculations of the effects of climate change on international politics and armed conflict must take into account that the large-scale effects of climate change may lie many decades, even centuries, in the future. They are thus likely to take place in a world whose essential socio-politics structures — which constitute the spaces for political action — may differ substantially from current ones. Without commenting on probabilities, some aspects can be briefly outlined. First, an important issue to study are the ways in which climate change might become ‘securitised’ by major actors like the U.S.A., China, Russia and the European countries, which in turn could lead to worsened security relations among them. The issue raises a keen interest among security agencies, but we have yet to see any of its consequences on inter-state relations. A second issue is whether climate change mitigation and adaptation can reduce the possibilities for cooperation and in fact increase pressures leading to conflicts. If the failure to reach an agreement on a globally inclusive regime for mitigation and adaptation at the Copenhagen summit in 2009 is repeated, it is not unrealistic to think that adaptation to climate change could mutate from a question of common survival and indivisible security to one of competition between individual countries or blocs and be securitised in a divisible way. Repeated failures to perceive climate change as a common danger and to tackle it collectively may lead to increased distrust to the detriment of mitigation and adaptation. Third, in comparison with other systems in world history, the current international system is not a particularly violent one. Many factors mitigate large-scale inter-state conflict, and one of them is that the states rarely use military force as a resource in economic competition. Currently, we do not see this standard feature of international politics between (at least) the seventeenth and early twentieth century in play. The arguments regarding the shape of the world economy and the shape of inter-state conflict cannot be recounted and scrutinised in full in this article. However, arguments linking modes of production, trade and consumption to distinct patterns of collective violence deserve to be taken seriously in connection with predictions that climate change may seriously disrupt the world economy. All three areas outlined in an all-too-brief form above have to be subjects for future articles. However, all three play important roles in a larger scholarly programme that seeks to understand climate change as an issue of socio-political change, or even transformation.