Neoclassical realism(s) vis-à-vis other theories of foreign policy: taking the indistinguishability problem seriously

Introduction

Since Gideon Rose’s foundational review essay ‘Neoclassical Realism and Theories of Foreign Policy’ was published in 1998, ¹ there has been a heated debate regarding neoclassical realism’s (NCR) feasibility. The opponents of NCR have considered it an incoherent and indistinctive approach because it attempts to address the transmission belt between the strategic environment and states’ foreign policy through leaders’ perceptions and domestic institutions. It means to treat the state as an actor with a broad range of preferences beyond those of security seeking or power maximizing and with different functional capabilities, primarily regarding their distinct institutional schemes of resource mobilization at the domestic level. ² In this regard, critics argue that NCR undermines core realist assumptions by introducing auxiliary assumptions about the importance of ideas and domestic politics, ³ indicating its inability to exclude causal mechanisms from alternative existing explanations, such as liberalism and constructivism. ⁴ According to this critique, NCR makes the term realist a misleading label. ⁵ In this regard, criticism addresses NCR’s fuzzily established internal and external conceptual boundaries, which I consider as two central dimensions of this approach’s ‘indistinguishability’ problem.

The proponents of NCR have responded to such criticism by considering it as a realist theoretical breakthrough by precisely including ideational and institutional intervening categories. ⁶ If power is the critical concept for realism within an anarchical positional competition conducted by the states, then how power is mobilized according to different leaders’ perceptions and state structures ends up being realism’s essential theoretical breakthrough. ⁷ They also insist on stating that NCR is an insightful realist approach since it helps understand that there are no ‘exclusive rights to the building blocks of international relations’. ⁸ NCR’s theoretical statement entails that ‘each paradigm may use ideational and institutional factors in a way that serves and reflects the logic of its approach’. ⁹ Ideas and domestic politics are never understood by NCR as independent or constitutive variables, in terms advanced by liberalism and constructivism, but as intervening variables that help explain the problems of rationality and the need to mobilize domestic resources to achieve state goals. ¹⁰ NCR’s insightful proposal should not be ‘known by what it incorporates but rather where it begins’. ¹¹ It begins with identifying as a point of departure the systemic imperatives and, then, it proceeds to operationalize the particular way states respond to them.

Although this debate has already shown the extent to which critics could have provided a straw man depiction of NCR, it has in recent decades had an enticing outcome. ¹² Regarding the allegations of lack of internal boundaries, neoclassical realist scholarship has clarified specific ways in which this approach can indeed privilege systemic variables rather than non-systemic ones as the independent variable of state behaviour, while refining intervening variables’ place within the theoretical framework. Notably, this concern brought about the distinction between NCR as a theory of mistakes (Type I NCR); as a theory of foreign policy (Type II NCR); and as a theory of international politics (Type III NCR). ¹³ Type II has also debated the extent to which NCR could define its internal boundaries by presenting itself as the logical extension of either neorealism or classical realism, giving rise to Type IIA and Type IIB NCR. ¹⁴ As far as the external conceptual boundaries are concerned, neoclassical realist scholarship has worked largely on distinguishing itself from innenpolitik and rationalist approaches to foreign policy; several variants of realism, especially classical realism, neorealism, defensive realism and offensive realism; liberal theories, such as Putnam’s two-level games (TLG), democratic peace theory and Moravcsik’s liberal theory of foreign policy, and constructivist approaches, especially Wendt’s classical constructivism. ¹⁵

Even though NCR has responded in a sophisticated manner to its critics, I argue that such a response still falls short of advancing some subtle strategies to forge NCR identity, with conceptual and methodological implications for NCR research design. First, it has not worked sufficiently on clarifying the extent to which NCR can become a realist theory of foreign policy in its own terms rather than a logical extension of other variants of realism. Secondly, neoclassical realist scholarship also tends to mischaracterize its rivals – which is a claim I cannot take issue with its opponents. ¹⁶ Indeed, NCR and its critics could have been entangled in a mutually advanced straw man depiction within which both sides must take on positions to ratify if not glorify themselves vis-à-vis their counterparts. ¹⁷

To fill these gaps, I provide an analytical framework to compare neoclassical realist theories with each other and other theories of foreign policy. Based on this framework, I intend to shed light on how these theories select variables, levels of analysis and types of rationality in their own terms. I begin with addressing the indistinguishability problem’s internal dimension by clarifying neoclassical realist theories’ common ground and their differences in terms of theoretical claims and research designs. Then, I discuss the indistinguishability problem’s external dimension by comparing these theories with organizational, bureaucratic, realist, liberal and constructivist approaches to foreign policy. Theory selection is based on the same choices made by neoclassical realist scholarship, as mentioned above. Yet some additional approaches, such as Allison’s models and critical constructivism are included to complete the broad spectrum in which neoclassical realist theories must find their place.

Based on this comparison, I argue that neoclassical realist theories provide a top-down, as compared to a bottom-up, approach to the state to explain foreign policy as a policy rather than a process. It also provides a thorough explanation, rather than a parsimonious account, aimed at distinguishing independent and different types of intervening variables. Every type of NCR substantiates their research design by clarifying intervening variables differently, as ‘moderating factors’, ‘complementary factors’, ‘direct causes’ or ‘sufficient causal processes’ – as explained by Götz and Kitchen. ¹⁸ With different levels of intensity, neoclassical realist theories also embrace specific, refined logics of consequentialism rather than appropriateness. The former is the case even when analysts aim to combine such theories with reflexive variants of foreign policy analysis. In this regard, it must uphold the distinction between NCR theories’ consequentialist assumptions and how they are used and combined with others within eclectic frameworks. Finally, neoclassical realist theories are interested in explaining different state preferences rather than assuming power maximization or security seeking as pre-fixed state goals.

Comparing theories of foreign policy: a framework for analysis

I adopt three criteria to compare neoclassical realist theories with each other and other approaches to foreign policy: the type of variables and the levels of analysis selected by every approach and the role that rationality plays in each of them. The next sections explain these approaches’ theoretical claims by addressing their primary variables or categories, whether they be seen as (a) independent, (b) dependent, (c) intervening or (d) constitutive. Identifying independent and dependent variables is crucial for clarifying the primary driving forces of foreign policy formulation and the dimensions of foreign policy implementation. Clarifying the role of intervening and constitutive variables entails explaining or understanding transmission belts linking systemic pressures and foreign policy and the mutual constitution between the strategic environment and agents, respectively. Most approaches attempt to clarify independent and dependent variables, yet neoclassical realist theories, defensive realism and constructivist approaches are the only ones considering intervening and constitutive variables. ¹⁹

I also pinpoint the level(s) of analysis privileged by each approach, be it (a) systemic level, (b) unitary, national level, (c) nonunitary, national level or (d) individual level. The selection of level(s) of analysis enables the researcher to throw light on the level of parsimony or thorough explanation of the approaches at hand. Those approaches adopting a parsimonious account of foreign policy tend to focus on one single level of analysis, normally the systemic level or unitary, national level. In corresponding to the totality of interactions within the international system (in the case of the systemic level) and in considering the state and national governments as single entities (in the case of unitary, national level), those levels lead the analyst to postulate a high degree of uniformity in foreign policy making. Both of them, accordingly, attempt to provide a reasonable basis for prediction. In contrast, those approaches embracing a thorough explanation of foreign policy tend to combine several levels of analysis and to consider functional differences between states, as well as the interaction with their surrounding society. In adopting those levels, the researcher manages to provide a reasonable basis for explanation and understanding. ²⁰ Rather than making a general case in favour of explanation versus parsimony, as if the former entailed taken-for-granted advantages over the latter, I seek to clarify the extent to which neoclassical realist theories become thorough accounts of foreign policy with specific characteristics.

Then, I address the role that rationality plays in each approach under comparison. Figure 1 provides a spectrum with two opposing extremes of rationality as being informed respectively by March and Olsen’s distinction between the logic of consequentialism and the logic of appropriateness. Even if such a typology has been conjured up to underpin the new institutionalism, embracing an ontology of appropriateness that is not necessarily shared by all approaches considered here, including neoclassical realist theories, it helps classify different conceptions of rationality. However, it entails considering consequentialism and appropriateness as equally justifiable, alternative explanations of state behaviour. ²¹ Above all, March and Olsen’s work helps imagine a spectrum classifying rationalities in relation to one another. On one extreme, such a spectrum highlights foreign policy approaches explaining human actors, willing and able to choose among different alternatives by evaluating their consequences in terms of the utility for actors’ goals (consequentialism). On the opposite end of the spectrum, the framework identifies foreign policy approaches attempting to understand human actors following rules that link particular identities with particular situations (appropriateness). ²²

OPEN IN VIEWER

Figure 1.

Rationality spectrum and approaches to foreign policy.

As a conceptual refinement, two types of logics of expected consequences and two types of logics of appropriateness may also be imagined; namely: the classical and refined logics of consequentialism, and the classical and refined logics of appropriateness. The first distinction is derived from Simon and Jervis’ conceptualization of bounded rationality and misperception within the organization theory, political psychology and IR fields. Simon argues that comprehensive rationality, seen here as the classical logic of consequentialism, is attributed to completely informed, self-interested, autonomous, maximizing actors. Such actors assess every course of action in terms of all consequences and utilities to choose the best option. This type of rationality does not simply exist in any human domain. ²³ Instead, rationality is often limited to thinking capacity, uncertainty and inability to access complete, noncontradictory, unified information, which corresponds to a refined logic of consequentialism. ²⁴ Meanwhile, Jervis considers that misperception, being addressed equally by a refined logic of consequentialism, is derived from decision-makers’ tendency to assess the strategic environment in terms of existing images and the inability to adjust their perceptions to new, conflicting information and time pressures. ²⁵ Both Simon and Jervis would replace maximization, seen as the ability to choose the best option, by satisficing, referred to the capability to select the option which is good enough within an uncertain strategic environment. Ultimately, the refined logic of consequentialism, as informed by these thinkers, would consider decision-makers as incompletely informed, self-interested, strategically embedded, satisfying actors.

The second distinction is derived from Wendt’s account of weak and strong understandings of the role of norms, rules and institutions. The former reifies them in treating them as social constructs occurring independently from human agency, ²⁶ whereas the latter considers their socially-generated constitution. ²⁷ In its poststructuralist version, the refined logic of appropriateness leads the researcher to understand critically ‘heterogeneous ways in which people experience [norms, rules and institutions]’ without presupposing ‘master narratives’. ²⁸ Within the rationality spectrum, as developed above to place all theories of foreign policy under consideration, I state that Type IIB NCR and Type III NCR, followed by Type IIA NCR and Type I NCR, manage to provide refined logics of consequentialism. The next section discusses neoclassical realist theories’ common ground and differences with each other, corresponding to the indistinguishability problem’s internal dimension.

Neoclassical realist theories’ common ground and differences

NCR has become seen as a ‘family of diverse theories’ ²⁹ or even a ‘(theoretical) beast’, which can take on several forms, including ‘midrange theory’, ‘grand theory’ or ‘progressive research program’. ³⁰ Still, neoclassical realist theories share the assumption that the analyst ‘gains more insight by combining diverse variables rather considering them separately and exclusively’. ³¹ In terms of research design, neoclassical realist theories share the willingness of looking into complex causal mechanism clarifying, rather than assuming, the specific translation of system imperatives into state behaviour, meaning that the researcher must uphold the distinction of independent, dependent and intervening variables of foreign policy. At the independent variable level, neoclassical realist theories argue that state behaviour corresponds to the adaptation to the international system, as defined by anarchy and the relative distribution of power, with changing threats to the states. ³² As I will show, differences between neoclassical realist theories come out when dimensions other than this basic, common denominator, are added to operationalize the independent variable. At the level of dependent variables, neoclassical realist theories can address different types of state decisions: (i) responses to international crises, expressed in the short time span; (ii) foreign policies and grand strategic adjustments, advanced in the medium term and (iii) systemic adjustment to structural changes, frequently derived from the dramatic transformation of international distribution of power observed in the long term. ³³ At the level of the intervening variables, neoclassical realist theories explore state rationality in a complex way by adopting a refined logic of consequentialism shaped by ideational and institutional factors. Like the independent variable, differences between neoclassical realist theories appear when analysts operationalize this variable through several dimensions.

By introducing ideational factors, these theories aim to study the foreign policy executive (FPE)’s complex perceptions of their strategic environment under the assumption that such perceptions filter systemic pressures. ³⁴ Empirically, the study of such perceptions rests on the need to adapt abstract conceptions of rationality, based exclusively on systemic imperatives, to analyse real decisions, which can be seen as maladaptive or merely choices. To be sure, the FPE’s strategic decisions may be affected not only by systemic pressures but also by cognitive filters. As human beings, they also cannot process information rapidly, not to mention that the information to make decisions tends to be often incomplete and contradictory. ³⁵

Also, neoclassical realist theories affirm that the extent to which a country responds to systemic pressures depends on the FPE’s ability to extract resources from society within a given domestic institutional setting. ³⁶ By incorporating domestic institutional intervening variables, neoclassical realist theories aim to explain the ability to extract and mobilize domestic resources as a central component of state capabilities. Neoclassical realist theories refuse to assume that the FPE have automatic access to human, material and financial resources required to conduct their foreign policy. The more autonomy the FPE enjoys, for instance, vis-à-vis the legislative branch or domestic actors within institutional arrangements, the more room for manoeuvre to conduct preferred foreign policy and to devote resources to those purposes, and vice versa. ³⁷ Similarly, the FPE must overcome hurdles derived from ‘the level of public satisfaction and the degree of elite consensus’ regarding the FPE’s foreign policy’. ³⁸ All in all, neoclassical realist theories argue that there is no direct transmission belt from the international system to foreign policy formulation and implementation. In the real world, such transmission is affected by ideational and institutional intervening factors. ³⁹

To clarify their theoretical claims and research design, NCR has evolved to include different ways to link system imperatives and ideational and institutional intervening factors, which have recently come in the form of Types I, II, IIA, IIB and III. Type I NCR’s works aim to explain suboptimal foreign policy responses to systemic imperatives. This approach attempts to explain why the best option of responding to systemic pressures directly is frequently not achievable. Instead, states often end up following different courses of action based on a leader’s perceptions and domestic politics. Because of these intervening factors, states can underact or overreact to the strategic environment, even if such an environment offers clear information about how national interests can be threatened or what optimal paths must be followed. ⁴⁰ Within the literature, this type is seen as an ‘orthodox’ approach to NCR because it sets out to define what constitutes an optimal foreign policy by using precisely a variant of structural realism as a baseline, whether it be a modified version of Waltz’s balance-of-power theory, defensive realism or offensive realism. These variants helped, for instance, Schweller, Juneau and Layne criticize overbalancing and underbalancing, squandered opportunities and immoderate search for hegemony, as compared to timely and orderly balancing, ⁴¹ opportune power maximization ⁴² and offshore balancing. ⁴³ In this regard, Type I NCR treats intervening variables as moderating factors, which restrict or magnify state response to systemic imperatives. ⁴⁴

Type II NCR seeks to explain state choices among a range of foreign policy options by considering state adaptation to changes in relative power. ⁴⁵ Rather than seeking to assess maladaptive behaviour, it considers that states rarely can discern optimal foreign policy due to the complex information provided by the strategic environment. ⁴⁶ To operationalize the analysis, this type aims to discern the broad range of foreign policy options available for the states according to their strategic environment. ⁴⁷ Then, it explains why and how the FPE select particular foreign policies due to domestic factors. Dueck, for instance, explained why and how American leaders chose, among different options, isolationism, containment, liberal internationalism and American primacy during the last century’s critical moments of strategic adjustment, based on two contrasting cultural drivers of American grand strategy; namely, America’s liberalism and limited liability. Together such traditions have turned Americans into reluctant crusaders, adopting crusading strategies, yet unable to assume the actual cost of such a promotion. ⁴⁸

A caveat is in order here as to Type II NCR’s works, addressing ‘culturally influenced results’, such as that advanced by Dueck, ⁴⁹ and reflexive assumptions on ‘meaning-making’ about power, such as that proposed by Gelot and Weltz. ⁵⁰ On the surface, one could argue that such works challenge the argument that neoclassical realists distinguish themselves by embracing a logic of consequentialism. By bringing about constructivist and poststructuralist claims, they seem to lean more towards a logic of appropriateness. However, a closer examination shows that a fundamental part of these works, as related to NCR, is keen to link several levels of analysis within a refined logic of consequentialism. They grant causal primacy to the international system, which defines the range of state strategic choices in terms of their utility, all the while they acknowledge that strategic cultures and the struggle for meaning-making are crucial to clarify how feasible carrying up specific foreign policy or strategic adjustment will be in given situations. ⁵¹ Although these works introduce the logic of appropriateness by using reflexive theories, they understand it in a subsidiary or supplementary manner vis-á-vis the logic of consequentialism. Rather than distinguishing between soft positivist and reflexive variants, neoclassical realist scholarship would benefit itself from distinguishing between researchers adopting exclusively NCR claims and those combining them with other theories within eclectic frameworks.

Type II NCR has recently evolved to give rise to Types IIA and IIB, depending on the extent to which its proponents use structural realism or classical realism. According to Onea, Type IIA constitutes the ‘semi-orthodox’ approach since, although it acknowledges the causal primacy of system structure in the terms advanced by neorealism, it does not turn, unlike Type I, to intervening factors to explain maladaptive behaviour as one not corresponding to systemic expectations. For its proponents, such as Taliaferro and Wishart, the simultaneous convergence between systemic pressures and intervening variables is more normal than abnormal. ⁵² To be sure, intervening variables are seen as complementary factors, which supplement systemic drivers of foreign policy. ⁵³ Yet this type adopts Waltz’s interpretation of non-structural factors, seen merely as domestic factors, meaning that it does not include states’ interaction with each other as other non-structural factors affecting foreign policy making. ⁵⁴

Type IIB is seen as the ‘revivalist’ approach because of its willingness to understand realist theorizing more broadly, which means to revive classical realism. ⁵⁵ After years of having been silenced by the dominance of structural realism, especially neorealism, ⁵⁶ some scholars, such as Foulon, argue that this version could indeed allow NCR to build bridges with the Global IR literature. For them, classical realism provides more entry points than structural realism for different experiences around the world and even joint collaboration between this theory and others. ⁵⁷

With classical realism insights in mind, the proponents of Type IIB concede an analytical significance to the systemic pressures yet introduce a robust understanding of the strategic environment and the drivers of foreign policy. Such an environment is defined by relative material capabilities but also by several non-structural factors, such as geopolitics, states’ historical interactions, leader’s perceptions about their counterparts’ future decisions, etc. ⁵⁸ Also, some scholars advancing this version, such as Reichwein, suggest turning NCR into a foreign policy guide, by providing ‘normative and political judgments’ during foreign policy making. ⁵⁹ Ultimately, intervening variables are seen by these scholars as direct, yet secondary, causes of state behaviour, as compared to the indirect, yet primary, impact from the international system. ⁶⁰

According to Ripsman, Taliaferro and Lobell, Onea and Desmaele, Type III NCR aims to provide insights into international politics by addressing how states’ decisional contexts and choices, international outcomes or system structure’s changes are inextricably intertwined. Still, unlike other system-based realist theories, especially neorealism, it allows the analyst to understand not only international political dynamics but also grand strategy and foreign policy making. ⁶¹ This type attempts to provide a conceptual refinement vis-à-vis Types I and II by assessing state behaviour with a broader conception of the international and domestic strategic environment in mind. This type breaks down the international strategic environment into several dimensions; namely: (a) the distribution of material capabilities and polarity as the central structural aspects of the international system; (b) material factors, modifying the effects of the central aspects, as mentioned above, including ‘geography, the rates of technological diffusion, and the offense-defence balance in military technologies’, ⁶² which Taliaferro coined as ‘structural modifiers’, ⁶³ (c) the level of clarity, scope and imminence of the threats, the availability of states’ strategies to face such threats, and opportunities offered by the environment for the states, which is also inextricably intertwined with the levels of permissiveness/restrictiveness of the strategic environment. ⁶⁴ This type also assesses the domestic strategic environment in terms of ideational intervening variables – leader images and strategic culture – and institutional intervening variables – state-society relations and domestic institutions. ⁶⁵ With a more refined interpretation of the transmission belt between systemic pressures and foreign policy making in mind, Type III NCR could, according to Kitchen, address intervening variables not only as explanatory factors, such as proposed by Types I and II, but also as ‘sufficient causal processes’, explaining foreign policy. ⁶⁶

The argument, as presented above, shows that neoclassical realist theories provide slightly distinct approaches to research design. Above all, it indicates that such theories operationalize independent and intervening variables with varying levels of depth, allowing the analyst to provide different types of thorough accounts of foreign policy, intervening factors and rationality. Although all types adopt the refined logic of consequentialism, Types III and IIB would adopt the most refined version, followed by Types IIA and I. The placement of these theories within the rationality spectrum does not mean to favour one type over another but clarifies different options available for analysts, depending on their empirical goals.

The following sections aim to address the indistinguishability problem’s external dimension by comparing neoclassical realist theories with other theories of foreign policy. Such comparison helps put limits on the use of critical analytical categories in studies inspired by these theories, such as ideas and domestic politics. Neoclassical realist theories turn to leaders’ perception not under the assumption that reality is socially and intersubjectively constructed – as constructivism would state – but because of the difficulty of perceiving systemic stimuli and access to complete, noncontradictory, unified information. ⁶⁷ Also, it turns to institutional constraints not to understand foreign policy decision-making process broadly seen in terms of a broad set of institutional constraints and the involvement of several domestic and transnational actors or of social meanings – as organizational, bureaucratic, liberal and constructivist approaches would suggest. Instead, it turns to institutional constraints to define foreign policy as a policy led by the FPE as the last-say decision-makers, selecting courses of action to achieve given goals in an uncertain strategic environment. ⁶⁸

Organizational and bureaucratic models

As originally developed by Graham Allison, organizational and bureaucratic-politics approaches to foreign policy (hereinafter referred to as ‘Allison’s Model II’ and ‘Allison’s Model III’, respectively) share with neoclassical realist theories the purpose of looking inside the state to supplement or supplant the Rational Actor Model (Allison’s Model I). Like neoclassical realist theories, Allison’s Model II and III, aimed at focusing primarily on organizational and bargaining processes underlying foreign policy decision-making, disagree with Model I’s view of the state as a black box or monolithic entity and, accordingly, the goal of analysing foreign policy according to a basic unit of analysis or independent variable: the ‘[g]overnmental action as choice’. ⁶⁹ Above all, these theories and models attempt to go further state-centric approaches, considering decision-makers to be unitary and rational actors, with the alleged ability to choose the course of action that meets the state’s goals and objectives in the highest rank. That is to say, they overtly criticize comprehensive rationality or the classical logic of consequentialism. ⁷⁰

Allison’s Model II is based explicitly on bounded rationality, meaning that organizations address strategic problems by pinpointing criteria to meet their goals, and clarifying procedures to deal with uncertainty at the international level. Firstly, strategic problems are factorized, that is, they are split up to assess a limited number of aspects, and they are assigned as sub-goals to organizational subunits. Secondly, maximizing, as the ability to consider all courses of action and to choose the one that meets organizational goals optimally, is replaced by satisficing, as the ability to pick up the action, which is good enough to achieve these goals. Thirdly, organizations define repertories of action which are acceptable in given recurring situations. These characteristics of bounded rationality informing organizational behaviour result in standard operating procedures affecting foreign policy strategic options. That is to say, organizations’ inability to adopt comprehensive rationality forces them to adopt such procedures as an inescapable, plausible, reasonable alternative. ⁷¹

Allison’s Model III deepens the impossibility of comprehensive rationality in assessing bargaining games. Government behaviour is much further from being rational not only in the sense of the first model but also of common parlance. After all, foreign policy options are ‘not chosen as a solution to a problem but rather results from compromise, conflict, and confusion of officials with diverse interests and unequal influence’. ⁷² The model is far from providing any explanation of rational behaviour either as exhaustive calculation or adoption of operating procedures. It only attempts to rationalize the rules of the game irrespective of the rationality of its political resultant. Such rules define, among other issues, who are the primary players (according to their position), what determines their stand (parochial priorities, perceptions, goals, interests, deadlines) and what determines their ability to impact results (position, formal authority, control over resources, expertise, access to information, influence over other actors). ⁷³ In not having to mention bounded rationality to underpin its insights, this model would be one of ambiguous rationality. At times, some well-positioned actors can make complex strategic calculations similar to those expected by Allison’s Model I. Other times actors are seen as boundedly rational. Sometimes presidents or prime ministers are seen as superpowers among other lesser powers and, at the aggregate level, nobody seems to be capable of controlling final outcomes. ⁷⁴

Despite their shared criticism against state-centric approaches, Allison’s Model II and III and neoclassical realist theories embrace different views of the significance of the nonunitary, national level of analysis, with the first two considering it the only explanation of foreign policy and the former including it in a more complex causality. Above all, their research designs entail distinct outcomes. As we can infer from the previous reasoning, Allison’s Model II suggests focusing solely on large organizations’ outputs and bureaucracies functioning according to standard operating procedures. ⁷⁵ It means that its primary unit of analysis, seen as the independent variable, is the ‘[g]overnmental action as organizational output [which] is determined by previously established procedures’. ⁷⁶ Allison’s Model III draws attention to the bargaining process between organizational players whose power is derived from their place in the organizations. In this regard, its primary unit of analysis, seen as the independent variable, is the ‘[g]overnmental action as political resultant’. ⁷⁷

Allison’s Models II and III are based on structural and agency-oriented views of the nonunitary, national level of analysis. ⁷⁸ In particular, they see this level in terms of the whole foreign policy bureaucratic team including, but not limited to, the foreign, defence and intelligence apparatus. In contrast, neoclassical realist theories provide a more centred analysis of the foreign policy decision-making process by focusing on the last-say decision-makers or the FPE. By brushing aside organizational and bureaucratic variables, these theories can afford to include other variables at the systemic and domestic levels without compromising on analytical quality. Ultimately, organizational and bureaucratic models and neoclassical realist theories end up understanding foreign policy as process and policy, respectively.

Realist approaches

Like neoclassical realist theories, classical realism and structural realism – including neorealism, defensive realism and offensive realism – explain foreign policy making in terms of power capabilities and the competition between states ‘under general conditions of scarcity’. ⁷⁹ However, realists disagree about the primary sources of foreign policy, including human motivations, national imperatives and the international distribution of state capabilities. ⁸⁰ Neoclassical realists attempt to address them simultaneously, whereas the other approaches tend to focus solely on one or two variables.

Morgenthau’s classical realism highlights human motivations and national imperatives, as studied on the individual and unitary, national levels. Drawing on the assumption that ‘politics [. . .] is governed by objective laws that have their roots in human nature’, driving political actors to define their interests ‘in terms of power’, ⁸¹ this theory argues that states turn into power-maximizers. Their interests collide with their rivals and enemies, making them entangled in international conflicts and war. ⁸² By introducing ideational and institutional variables, neoclassical realist theories explain state preferences other than merely power maximization.

Another disagreement between classical realism and neoclassical realist theories would be their different understandings of consequentialism. Before clarifying this disagreement, a caveat is in order: On the surface, classical realism would adopt a classical logic of consequentialism. Not only does such an account consider that states have fixed preferences as self-interested, maximizing actors, but they also must assess every course of action in terms of its consequences in a world governed by selfishness and power. ⁸³ Still, a closer analysis would reveal that it would have adopted a refined logic of consequentialism because foreign policy is shaped by specific political circumstances informing cost-benefit calculations. Foreign policy is also judged by its ability to acknowledge that ‘the moral aspirations of a particular nation’ are not superior, corresponding to moderation or political morality. ⁸⁴

Even if classical realism’s refined logic of consequentialism is apparent, this theory and neoclassical realist theories differ in two aspects. First, classical realism solely addresses consequentialism at the international level, which is to say that it brushes aside domestic factors affecting the FPE’s calculation of expected consequences. Classical realism could consider domestic politics, yet this theory did not operationalize it to understand its effects on the FPE’ calculations. Second, contrary to classical realism, Type I, IIA and III NCR did not unpack the moral assessment of foreign policy. For these variants, constraints on the FPE’s decisions come more from international and domestic systems than political morality. However, Type IIB has paved the way for advancing dialogue with classical realism by introducing normative considerations about foreign policy.

Structural realism focuses mainly on the effects of systemic imperatives on state behaviour, meaning that it privileges the system level of analysis. Waltz’s neorealism aims to explain the international domain as a whole, which is defined by its structure, seen as the arrangement of the states. ⁸⁵ Above all, Waltz defines the international system in terms of anarchy, states’ functional sameness and the distribution of state capabilities. ⁸⁶ Reasoning by analogy, Waltz argues that a theory of international politics is similar to the elegant and parsimonious economic theory of the market. It clarifies how the organization of realm (the international system functioning as a market) constrains the interacting units (states acting as firms), and accordingly, addresses how the states are pressed by structural forces (such as market forces) to behave in given ways. ⁸⁷

By addressing such a level of generality about the functioning of the international system, this version of structural realism would have nothing to say about foreign policy’s particular features. ⁸⁸ However, Elman showcases how structural realism – depicted also by Mearsheimer, Gilpin, Grieco and Walt – already provide foreign-policy explanations, which makes a comparison between structural realism and NCR a necessary endeavour. Such explanations help analysts (a) operationalize key concepts of structural realism – anarchy, polarity, alliance, balancing, bandwagoning, offense-defence balance – to understand foreign policy making; (b) merge systemic-level causes with secondary assumptions regarding, for instance, the likelihood of cooperation; and (c) clarify states’ specific behaviour derived from systemic pressures. ⁸⁹ The dialogue between structural realism and neoclassical realist theories would also be even richer if one recognizes that the former embraces conceptions of comprehensive and bounded rationality implicitly. ⁹⁰ Although at first glance structural realism merely sets out to explain states’ capability to assess the strategic environment objectively, it turns out to address states’ ability to compete and adjust to such environment, which constrains, yet not determine, state behaviour. As a result, states manage different levels of success, at times far from optimal behaviour. ⁹¹ To advance this explanation, structural realism would need a foreign-policy explanation aimed at introducing ideational and institutional variables, meaning to turn to neoclassical realist theories.

Like neoclassical realist theories, defensive realism and offensive realism have sought to elaborate more thoroughly on the assumption that the strategic environment shapes foreign policies, yet providing different conceptions of the international system and consequentialism. Defensive realism maintains that anarchy and the gross distribution of state capabilities, along with ideational and material structural modifiers, shape the strategic environment. Structural modifiers are seen as factors mitigating or intensifying the likelihood of conflict by modifying state rationality in ways that detach state behaviour from systemic pressures. These factors, as mentioned, can include leaders’ perceptions, military technology, geopolitical positions or international economy pressures. ⁹² With this conception of the strategic environment, defensive realists see the state as a self-interested, nonunitary, satisfier but not an autonomous, completely informed actor, whose primary goal is security seeking. Offensive realism considers that the strategic environment is, of course, shaped by anarchy and the distribution of state capabilities. But it is also conditioned by the possibility that great powers hurt each other due to their offensive military capability, the uncertainty about other states’ intentions and states’ need to survive. Above all, the state rationality assumption goes hand in hand with those of pervasive uncertainty and intensive fear in a world governed by the struggle for power among potentially aggressive nations. ⁹³ This variant of realism holds that becoming the most powerful state is the primary goal. However, this objective is impossible to achieve, making regional hegemony the best and most likely outcome for great powers, and the containment of countries with hegemonic aspirations in other regions the best strategy. ⁹⁴ Ultimately, offensive realism sees the state as a self-interested, unitary, maximizer, but not an autonomous, completely informed actor, whose primary goal is power maximizing.

Neoclassical realist theories differ from both offensive realism and defensive realism’s refined logic of consequentialism. For neoclassical realists, neither the state is doomed to be a power-maximizing entity nor it must always be a security-maximizing entity. Instead, the nature and scope of a state’s ambitions depend on its place in the international system: the more relative power the state has internationally and domestically, the broader its interests and need to influence. ⁹⁵

The comparison between neoclassical realist theories and realist approaches, indicating their commonalities and differences, helps treat neoclassical realist theories as realist theories, informing specific, reliable research designs on their own rather than as merely logical extensions of other variants of realism. Recognizing such independence from its realist counterparts is critical to taking the indistinguishability problem’s internal dimension seriously. Some critical aspects must be in place to make it.

By discussing with neorealism, Type I must clarify that its purpose has merely been to turn to systemic imperatives to identify selectors rewarding or punishing states’ behaviour, depending on their ability to observe and adapt themselves to such imperatives. ⁹⁶ Although Type I NCR can discuss with neorealism what system imperatives means and entails, this type must be careful in using it as a baseline. Indeed, neoclassical realist scholarship must prevent itself from undoing what Waltz attempted to carry out: providing an idealized, yet helpful, depiction of the international realm as a whole rather than a theoretical construct to empirically falsify hypothetical propositions about foreign policy. ⁹⁷ By explaining how intervening variables can lead states to detract from system-oriented realist expectation, Type I can, as shown by Massie, defines such expectation more broadly and pragmatically. ⁹⁸ Type IIA equally needs to consider that using Waltz’s approach to the international system can drive the analyst to brush aside states’ interaction with each other as critical non-structural factors. After all, Waltz viewed international politics as a realm which must be studied by addressing the placement of the states in the international system rather than their interactions with each other. ⁹⁹

Although Type IIB seeks to turn to classical realism to solve the problems of Types I and IIA’s neorealism-based assumptions, it needs to carry out this dialogue more carefully to prevent neoclassical realism’s complex causality from being indeterminate. Specifically, it ought to clarify under what systemic circumstances domestic factors can become direct causes of foreign policy. Ripsman advanced this discussion by arguing that such factors can indeed become direct causes when states face a ‘low-threat international environment’ or ‘ambiguity or confusion reigns in the policy environment’. ¹⁰⁰

Some scholars, revising Type III NCR, have echoed the argument presented here by stating that it ‘should not proceed from the starting point of modifying any particular realist variant [. . .] This would be more in keeping with Waltz’s insistence on the proper pursuit of the theoretical enterprise than NCR’s current strategy of extending structural realism’. ¹⁰¹ Above all, Type III’s theoretical refinement will depend on clarifying the extent to which intervening variables, seen as ‘sufficient causal process rather than explanatory factors’, can elucidate international outcomes and grand strategies. ¹⁰² Ultimately, it means that scholars operationalizing Type III must be more interested in addressing NCR’s historical turn than turning to structural realism. ¹⁰³

Liberal approaches

Liberal approaches to foreign policy, such as democratic peace theory, Putnam’s two-level game (TLG) and Moravcsik’s liberal theory of foreign policy constitute other theories that can be used to assess neoclassical realist theories’ distinctiveness. Unlike these theories, seen as ‘top-down, international system-based theor[ies]’, those liberal variants constitute ‘bottom-up, domestic society-based theor[ies]’ of foreign policy. ¹⁰⁴ Specifically, they explain foreign policy according to the multiple domestic institutional and ideational sources of state preferences and the intertwine between domestic and international factors. Moravcsik and Putnam also address the entanglement of state preferences at the international level. ¹⁰⁵

By contrasting democratic and non-democratic regimes, democratic peace theory makes the regime type its independent variable to explain foreign policy. The primary contention is that liberal democracies are able and willing to build peaceful relations when interacting with each other and fight solely against authoritarian regimes. ¹⁰⁶ Thus, democratic peace theory’s original argument entails embracing a single, undifferentiated conception of democracy. Nevertheless, some thinkers have sought to propose a refined version of democratic peace theory, highlighting differences between democracies. ¹⁰⁷

Peace democratic theory’s original argument embraces a classical logic of consequentialism, whereas the revised version adopts a refined logic of expected consequences, meaning that the latter shares neoclassical realist theories’ rationality to a certain extent. Yet a critical assessment of their assumptions could also reveal that both liberal variants embrace a logic of appropriateness as democratic collective identity ends up being crucial to conducting foreign policy. Appropriateness would define the overall range of actions, whereas consequentialism would define particular strategic options. The classical argument attempting to showcase that ‘the classical liberals were right’ ¹⁰⁸ and, accordingly, making a case for the virtues of democracy and internationalism, could have led their proponents to adopt a classical logic of appropriateness aimed at reifying such virtues. For its part, the refined argument would have the potential to involve the refined logic of appropriateness by explaining variations in ideational orientations and preferences within domestic and international institutional settings. ¹⁰⁹

In any case, this assessment goes beyond the purpose of this review. Suffice it to say that when the assumptions of both of these approaches to democratic peace are considered according to their own proponents, consequentialism is dominant. On the one hand, classical proponents argued that war is costly and risky by precisely adopting consequential rationality and complete information assumption. Decision-makers know that a war can turn them out of office by audiences questioning their motives and opposing parties highlighting costs in blood and treasure derived from fighting wars. Also, democracies, as opposed to non-democracies, can bargain more easily due to their informational advantages. Specifically, democratic political systems reduce uncertainty in an imperfectly informed strategic scenario by frequently revealing their motives. ¹¹⁰ On the other hand, the proponents of the modified version embrace imperfect rationality in addressing the different ideational orientations of democratic regimes, ¹¹¹ as well as their different institutional arrangements – in ‘presidential, coalitional parliamentary, Westminster parliamentary, and semi-presidential democratic systems’ – as the independent variable, affecting the way democracies undertake their foreign policies, in general, and their propensity for waving the war, in particular. ¹¹² Although neoclassical realist theories consider ideational and institutional variables above critical to developing its analysis, such variables are treated as intervening variables rather than independent variables, within a framework embracing several levels of analysis.

Drawing on Putnam and Moravcsik’s works, liberals could reply to neoclassical realist theories that a liberal approach to foreign policy is well-equipped to carry out a thorough explanation by addressing a myriad of variables at different levels of analysis. To begin with, Putnam captures the complex interplay between domestic and international politics by developing ‘the logic of two-level games’. For Putnam, foreign policy making, mostly in the realm of international negotiation, depends on chief negotiators bargaining with their counterparts at the international level to achieve a tentative agreement (Level I). Such negotiators must also discuss with domestic actors to ratify the agreement, as mentioned above (Level II). The logic of consequentialism is apparent in this theory because the possibility of arriving at an international agreement depends on the overlapping of each negotiating party’s ‘win-sets’. ¹¹³ In turn, the size of these win-sets relies on (a) ‘the distribution of power, preferences, and possible coalitions among Level II constituents’, (b) ‘the Level II political institutions’ and (c) ‘the strategies of the Level I negotiators’. ¹¹⁴ Although TLG approach acknowledges international and domestic variables, some of these being implicitly introduced as intervening factors, within a refined logic of consequentialism aimed at addressing negotiators’ bounded rationality, it is not interested, unlike neoclassical realist theories, in embracing the primacy of systemic stimuli.

Moravcsik’s liberal theory of foreign policy would explain foreign policy as the result of different independent variables, such as societal ideas, transnational economic exchanges and domestic representation schemes. ¹¹⁵ Specifically, Moravcsik conceives foreign policy as ‘a two-stage process of constrained social choice’. ¹¹⁶ The first stage would be to analyse the configuration of state preferences as ‘the fundamental social purposes underlying the strategic calculations’. ¹¹⁷ At this level, it would not consider the FPE’s perceptions and structural autonomy – NCR’s core domestic variables – as the most important domestic factors. Instead, it would include a myriad of variables to understand states’ preferences as the result of state-society complex interactions. What is more, sources of state preferences – power, ideational and institutional factors – are addressed as having the same analytical relevance. ¹¹⁸ The second stage would account for strategic interaction with other nations operating with similar constraints. ¹¹⁹ Unlike NCR, this liberal approach would adopt a state-state approach rather than an actual systemic analysis. At this level, it is clear that, like Putnan, Moravcsik does not aim to establish a causal mechanism in which systemic pressures were the starting point.

Constructivism’s classical and critical variants

Classical and critical variants of constructivism disagree with several aspects of neoclassical realist theories. Whilst the latter combines ideational and material factors within a realist ontology, the former adopts an ideational approach to understand the constitutive relationship between agents – mainly states – and structures – the social configurations of the international system. By adopting a social interpretation of international politics, constructivism aims to explain the relationship between strategic culture and identity, on the one hand, and foreign policy, on the other hand. ¹²⁰

Specifically, Wendt’s constructivism aims to explain foreign policy through a constitutive-making rationale drawn through several steps. Firstly, states are seen as actors, whose strategic identity, ends and means are socially acquired through socialization processes. Secondly, as a way of learning, socialization allows states to acquire common and collective knowledge amid micro and macro structures of interaction. Thirdly, taking the causal and constitutive effects of common and collective knowledge seriously allows the analyst to define structures in cultural terms, which reproduce themselves through shared knowledge. ¹²¹ Finally, the emergence of international cultures reaffirms state agency, repeating the loop of the social constitution. ¹²² However, international cultures defining foreign policies can change, considering the different roles and meanings which dominate the international system. ¹²³ In contrast, neoclassical realist theories, adopting a top-down conception of the state, view states’ agency and strategic interactions only in terms of the FPE’s decisions derived from their perceptions of the strategic environment and room for manoeuvre, as defined mainly by the international distribution of state capabilities and domestic structural autonomy.

In terms of its conception of rationality, constructivist approaches to foreign policy seek, accordingly, to embrace the distinction between the logics of consequentialism and appropriateness to make a case for the latter, which has distanced constructivism even more from NCR. The constructivist point of departure is that approaches embracing consequentialism, as represented by realism and liberalism, ‘treat the interests and preferences of actors as mostly fixed during the process of interaction’. ¹²⁴ According to constructivism, such consequentialist approaches adopt instrumental rationality focused solely on maximization, leaving out the role of identities, norms and institutions, shaping human actions. To fill this gap, social constructivism adopts normative rationality to indicate how human action is a ‘rule-guided behavior [which] differs from instrumental rational behavior in that actors try to “do the right thing” rather than maximizing or optimizing their given preferences’. ¹²⁵ It follows that rules, norms and institutions not only regulate human behaviour but they also bring about human agency in the sense the latter does not have meaning without referring the former. At the same time, human agency in the form of social interactions is critical to reinforce, transform or qualify the social understanding that gives meaning to the social reality. ¹²⁶ Still, by discussing in the previous sections how consequentialism has been indistinctly adopted by realist and liberal approaches, I argue that constructivists have oversimplified the way these approaches consider rationality. Above all, they have not captured NCR’s refined logic of consequentialism appropriately.

By drawing attention to the disciplining effects of discourses and narratives on foreign policy, critical approaches to constructivism have distanced themselves even more from NCR. ¹²⁷ Critical constructivism argues that discourses and narratives are essential to thinking of foreign policy since they tell agents about how the social world works, their possibilities for taking action according to what constitutes legitimate behaviour, and how they could assume commitments. Foreign policy is seen as a realm of intersubjective contestation and negotiation between several discourses and narratives about political communities’ core values, what threatens such values, and the means to preserve them. ¹²⁸ Critical constructivism attempts to provide a broader conception of normative rationality and social agency by considering human actions as involving a complex mental process aimed at interpreting a rule. According to this perspective, human consciousness and ‘rationality’ would be derived not only from the structures of the language but also eventually from contradictory reasons which at times are ignored by actors, such as the premises of psychoanalysis would suggest. ¹²⁹ Even if neoclassical realist theories could address contradictions informing foreign policy, it is uninterested in addressing the FPE’s perceptions and decisions at the psychological and sociological levels, as understood by critical constructivism.

Drawing on constructivism’s classical and critical variants, the analysis of foreign policies would be oriented to addressing essential ideational factors, namely, strategic culture and identity. It means that constructivism diverts attention from material constraints at the international and domestic levels, such as the relative distribution of power between states and domestic institutional hurdles to mobilize resources. In a nutshell, constructivism and neoclassical realist theories provide different understandings of the strategic context in which foreign policy takes place. Still, NCR critics seem to ignore such differences, clearly established by both sides.

Conclusion

In this article, I aimed to take NCR’s indistinguishability problem seriously, especially in terms of its internal and external conceptual boundaries. To this end, I provided an analytical framework aimed at comparing neoclassical realist variants (I, II, IIA, IIB and III) with each other and with other theories of foreign policy. Although neoclassical realist scholarship has to a certain extant discussed such boundaries to respond to its critics explicitly and implicitly, I argued that it falls short of illustrating how neoclassical realist scholars could define their research design on their own and find their place in foreign policy analysis scholarship without glorifying themselves or implicitly neglect the worthiness of their counterparts.

Even if neoclassical realist theories embrace a common causal mechanism, I showed that they entail distinct types of research design, defining the independent, intervening and dependent variables differently. Such clarification is crucial to understanding the options available to analysts interested in substantiating their work on NCR. Type I operationalizes the independent variable in terms of structural realism’s expectations so that intervening variables are seen as factors detracting states from optimal responses to system pressures. Although Type IIA defines the independent variable like Type II, it acknowledges that intervening variables are seen as factors supplementing systemic pressures more likely than expected by Type I. Both Type I and Type IIA study non-structural factors only at the domestic level, meaning that international non-structural factors, such as states’ interaction with each other are not included in their research designs. Type IIB operationalizes the independent variable in terms both of relative material capabilities and several non-structural factors. This variant can also include normative considerations about foreign policy, as inspired by ethics of responsibility. With a more robust understanding of international and domestic strategic context, this type treats intervening variables as direct, yet secondary, causes of state behaviour. Type III goes further by broadening the independent and intervening variables’ scope. The former is seen in terms of the distribution of material capabilities and polarity, structural modifiers and the level of clarity and permissiveness of the strategic environment. The latter is analysed according to leader images, strategic culture, state-society relations and domestic institutions, which the most refined version of Type III would conceive as sufficient causal processes informing grand strategy or foreign policy making.

Still, I argued that, to deal with the indistinguishability problem, neoclassical realist theories need to overcome their position of being merely a logical extension of other variants of realism. On the one hand, NCR needs to revise its link with neorealism to prevent itself from undoing what Waltz attempted to carry out in formulating neorealism. Such a link, which has been uncritically established, has undermined the likelihood for some NCR works to examine non-structural factors, such as states’ interaction with each other. On the other hand, neoclassical realist scholarship needs to use classical realism carefully to clarify under what systemic circumstances intervening domestic factors can become direct causes of foreign policy. Finally, the development of a neoclassical realist theory of international politics, which depends on understanding intervening variables as sufficient causal processes, means embracing NCR’s recent historical turn rather than attempting to supplement structural realism.

By placing neoclassical realist theories within the analyses of foreign policy, I also made a case for their distinctiveness and external coherence in a broad, comparative way, by assessing these theories and their counterparts’ core assumptions in terms of their privileged variables, levels of analysis and types of rationality. I showed that most of the approaches under consideration pay attention to independent and dependent variables to examine foreign policy. Yet neoclassical realist theories, defensive realism and constructivist approaches are the only ones that include intervening and constitutive variables. In comparing theories of foreign policy in terms of their level of analysis, I stated that Allison’s models, refined democratic peace theory, and offensive realism draw attention to a single level of analysis. In contrast, classical democratic peace theory, liberal theory of foreign policy, classical realism and defensive realism highlight two levels of analysis. It means that neoclassical realist theories are the only ones addressing several levels of analysis to consider capabilities, perceptions and domestic politics.

I also divided the theories under scrutiny into two categories by considering their types of rationality. The first category corresponds to theories embracing a logic of consequentialism, whereas the second category comprises theories adopting a logic of appropriateness. Such classification is much more accurate by considering classical and refined logics of consequentialism and appropriateness. Allison’s Model I and classical democratic peace theory are the only ones embracing a classical logic of consequentialism. In contrast, Allison’s Models II and III, the liberal theory of foreign policy, and realist approaches adopt a refined logic of consequentialism. Constructivist variants are, by definition, approaches including the refined logic of appropriateness. Within the approaches adopting the logic of consequentialism, neoclassical realist theories adopt imperfect rationality due to systemic pressures, the FPE’s perceptions and domestic institutional constraints. It means that it does not adopt comprehensive rationality, bounded rationality within organizational routines or undetermined rationality (like Allison’s Models), aggregated, endogenously defined rationality (like the liberal theory of foreign policy and TLG approach) or simplistic constrained rationality (like neorealism and offensive realism), and undetermined rationality due to domestic factors (like defensive realism).

All in all, I made a case for neoclassical realist theories’ distinctiveness and coherence by explaining what it is and is not. Achieving this goal did not entail making a case for a specific conception of foreign policy as a policy versus process, thorough explanation versus a parsimonious account, realism versus liberalism and constructivism or consequentialism versus appropriateness. Rather than showing these theories’ conceptual and methodological superiority, I aimed to clarify what is at stake in using them vis-à-vis other theories of foreign policy and shed light on their distinctive, coherent aim, among other relevant, justifiable aims of foreign policy research designs.