Correspondence: Neoclassical Realism and Its Critics
2018; The MIT Press; Volume: 43; Issue: 2 Linguagem: Inglês
10.1162/isec_c_00332
ISSN1531-4804
AutoresDavide Fiammenghi, Sebastian Rosato, Joseph M. Parent, Jeffrey W. Taliaferro, Steven E. Lobell, Norrin M. Ripsman, Kevin Narizny,
Tópico(s)Religion and Society Interactions
ResumoIn his article "On Systemic Paradigms and Domestic Politics," Kevin Narizny criticizes neoclassical realism for incorporating domestic variables that are inconsistent with realist assumptions.1 Below, I argue that Narizny's understanding of realism is flawed and that his recommendation that realists should either say little about state preferences or abandon realism altogether is misplaced (pp. 188–190).To begin, Narizny claims that absence of functional differentiation means that states have the same preferences (p. 162). Absence of functional differentiation means that the international division of labor pales compared to the division of labor within states. Every state strives to perform the same functions: every state has an army, police, schools, and so on. Only in this limited sense are states "like units."2 Unlike states, individuals have specialized functions (e.g., construction worker, lawyer, or soldier).3Next, Narizny argues that Waltz "offers only three hypotheses: balances form; successful innovations are emulated; and bipolarity is more stable than multipolarity" (p. 159). In fact, Waltz posits many more hypotheses, including the following:These additional hypotheses challenge the notion that "in a realist world, states … cannot vary in their preferences" and that "states must define security the same way. Otherwise, they would be functionally differentiated" (p. 162). For Waltz, the pursuit of security is compatible with states having different preferences (i.e., isolationism, pacifism, revisionism, the status quo, or unlimited revisionism). Moreover, the structural constraints to which states are subject vary with their preferences (i.e., for unlimited revisionists, the pursuit of security in most cases leads to ideological moderation [H.2]; for isolationists, to rearmament [H.1.2]). Such restraints also vary with a state's geography (H.2.1), and with its size (H.6.2.1 and 6.2.2). Finally, changes in a state's power position reshape its preferences (former great powers become peaceful [H.1.3]; growing powers become more active abroad [H.1.1]; and sometimes they become revisionists [H.6.2.1].Future scholarship should both refine Waltz's hypotheses to make them falsifiable and investigate (1) how states' preferences vary, as well as how and why states having different preferences are subject to different security constraints; (2) how a state's geography and size affect the security constraints to which that state is subject; and (3) how changes in a state's power position reshape its preferences.—Davide FiammenghiBologna, ItalyIn "On Systemic Paradigms and Domestic Politics," Kevin Narizny argues that the past three decades of scholarship have done little to advance scholars' understanding of foreign policy and, as the bulk of his examples make clear, of great power balancing behavior in particular.1 He targets one problem and one paradigm. To his mind, neoclassical realism is "fundamentally flawed" because it does not assign causal priority to either systemic or domestic variables (p. 156). The solution, he suggests, is to explore bottom-up theories that "tak[e] preferences seriously" (p. 190).We share Narizny's dissatisfaction over the state of the debate. Indeed, we contend that there are two further problems that apply not only to neoclassical realism, but also to realism and liberalism. Yet we dissent from his conclusions. It is too early to tell with regard to foreign policy in general, but as far as balancing is concerned, explanatory success is more likely by taking structure rather than preferences seriously.Although Narizny's discussion centers on neoclassical realism and causal priority, it elliptically reveals two other critical issues that apply to all participants in the debate. One issue is a failure to adhere to what Narizny refers to in passing as "standards of consistency and deductive logic" (p. 189). For example, his description of the literature captures how realists erred early on by focusing almost exclusively on alliances when logic suggests that states have several options if they want to build their capabilities (p. 159). Meanwhile, his account strongly suggests that neoclassical realists have gone even more astray (pp. 183–184), and he is clearly concerned that liberals have not adhered to the requisite standards either (pp. 156, 189). The other issue is a lack of systematic empirical evidence, without which a comparison of the explanatory power of competing theories is impossible. Realists suggest that states "balance effectively against threats," but they give little sense of what the relevant universe of cases looks like (p. 159). At the same time, neoclassical realists and liberals merely assert that states balance less effectively than realists claim (pp. 180, 184). In sum, there is substantial confusion among scholars studying balance of power politics about what independent variables matter most, how they matter, and how much they matter.How far might causal priority, deductive consistency, and systematic evidence take us? Quite far, potentially. Although top-down and bottom-up approaches are equally valid points of departure, we begin with systemic independent variables (p. 160). The central behavioral prediction to be deduced from top-down or realist assumptions is "self-help."2 Without a central authority to protect them and knowing that others have the capability to hurt them, states conclude that they must procure the means to defend themselves. Therefore, they will have a marked propensity to engage in internal balancing—that is, arming in response to the arming of others and imitating competitors' successful military practices. Both activities constitute true self-help. Yet states will rarely engage in external balancing, because alliances are help from others. Thus, external balancing will occur only under the pressure of war and sometimes not even then.3 What do "marked propensity" and "rarely" mean? Our analysis of the great powers from 1816 to 1990 reveals that they reacted routinely and promptly to the military buildups and innovations of others approximately 80 percent of the time. They did so because they worried about their security in an anarchic system, closely monitored each other's capabilities, and worked hard to repair any erosion of their relative military strength. During the same period, there were only six clear-cut examples of external balancing, five of which occurred during wartime. At least where great power balancing is concerned, one could dispense with domestic variables; strictly systemic factors would still explain most of the historical record.4Where does this account leave theories that incorporate domestic politics into international relations theory? On the subject of balancing, it leaves them in trouble. Neoclassical realists can do no more than introduce systemically derived domestic variables to account for the few anomalous cases in which great powers failed to respond promptly to the military advances of their peers (pp. 161–164). Liberals do not need to derive their explanations from structure and are free to offer competing domestic arguments for the strong tendency of states to balance internally. Yet, they will likely have to jettison Narizny's interpretation of the paradigm. In his telling, liberalism envisages a system composed of "multiple states with multiple preferences," and the resulting "complexity may be too high to generate reliable predictions about the likelihood and nature of conflict." Indeed, such "a liberal theory would not suggest that the international system has inherent tendencies toward balancing" (pp. 184–185). A different bottom-up approach, one that identifies preferences other than security that are common to most states, has better prospects.Of course, foreign policy is about much more than balancing. As scholars seek to explain other behaviors, we hope they learn from the missteps of the past and stipulate causal priority, employ consistent deductive reasoning, and adduce systematic empirical evidence. If they do so, they will be able to pit peer theories against each other, evaluate their relative strengths and weaknesses, and improve our collective understanding of state behavior.—Sebastian RosatoNotre Dame, Indiana—Joseph M. ParentNotre Dame, IndianaKevin Narizny's "On Systemic Paradigms and Domestic Politics" raises several challenges to neoclassical realism, an approach that he acknowledges has gained considerable traction in the field of international relations in the last two decades.1 His central arguments, echoing Jeffrey Legro and Andrew Moravcsik, are that neoclassical realism is not "realist" and that it violates what he calls a central assumption of realism—namely, that states all have the same goals and are not functionally differentiated.2 There are too many problems with his argument and his inaccurate caricature of neoclassical realism to take up in this short response, but we address two key problems herein.First, Narizny's principal argument is that neoclassical realism is inconsistent with what he asserts is the hard core of realism—a set of core propositions advanced by Patrick James.3 Why Narizny settled on James's list is unclear, especially given that numerous scholars have offered competing statements of realism's core assumptions.4 Notably, these scholars do not include Kenneth Waltz's central tenet that units must be functionally undifferentiated.Moreover, we disagree with Narizny's interpretation of his R5—namely, that realism requires states not to vary in their preferences. Realism, even Waltz's structural realism, does not stipulate that states have completely identical preferences. It merely asserts that, under anarchy, states prioritize security and survival above all other preferences because, without security from other states, all other preferences would be unobtainable. Consequently, although states and their leaders may have a variety of goals, they are alike in that the international system compels them to be security-seeking actors above all. Nonetheless, it would be foolhardy to suggest either that all states have identical interests or that the international system determines what these interests are.Neoclassical realism stipulates that the foreign policy executive has three sets of goals: (1) to preserve the state's physical survival and political autonomy; (2) to maintain its power position; and (3) to safeguard all other ideological, religious, political, social, and economic goals they may possess. The foreign policy executive will prioritize the first set of goals. Nonetheless, when a state faces a permissive international security environment, the foreign policy executive may have the luxury of focusing on the second and third sets. Furthermore, in extreme circumstances, when they face the immediate threat of losing power, leaders may temporarily trade off security interests to solidify their domestic power positions, if they believe the damage to national security will not be too great.5Neoclassical realism represents an advance on liberal approaches, which do not posit any independent effect of the international system. Liberalism cannot efficiently explain why states with different regime types and dominated by different underlying coalitions make fundamentally similar choices when faced with similar international circumstances. For example, in the immediate aftermath of the terrorist attacks of September 11, 2001, when President George W. Bush declared, "Either you are with us, or you are with the terrorists,"6 states as disparate as Australia, France, Pakistan, and Saudi Arabia joined the U.S. "war on terror." Thus, despite diverging preferences, systemic constraints exerted an independent effect on states' foreign policies.Second, despite Narizny's unsupported claim, liberalism does not incorporate the international system in any meaningful way. Instead of advancing liberalism, Narizny merely restates Moravcsik's core assumptions.7 Neoclassical realism is more appealing than liberalism to scholars around the globe, and especially beyond members of the Organization for Economic Cooperation and Development, because it is better equipped to capture systemic- and regional-level constraints and opportunities states face. The problems with liberalism include (1) it is not very useful in explaining the behavior of non-liberal states, which do not face the same societal and transnational binding constraints as liberal states; (2) it is not very useful for liberal states in non-liberal regions because the former know that their neighbors do not face similar constraints and, therefore, liberalism does not reduce mistrust through greater transparency, the creation of shared norms, or societal, political, or economic constraints on the use of force; and (3) it is not very useful for middle powers and especially weak states, because it ignores regional distributions of power, which are often of great concern to them. Neoclassical realism is more generalizable than liberalism (and structural realism) in explaining interstate behavior across different political regime types, levels of interdependence, and thickness of regional multilateral institutions.Moreover, unlike neoclassical realism, liberalism is both teleological and normative. Liberalism proceeds from the ontological suppositions that the arc of human history is progressive and enlightened; that liberal states are preferable to all other regime types; and that eventually, through mechanisms including economic interdependence, democratization, and international institutions, states will eventually converge upon the liberal model.8 Conversely, neoclassical realism neither privileges any domestic regime type nor expects states to converge upon a single model. It purports to explain foreign policy and international politics as they are, rather than evaluating them from the scholar's normative standard of what they ought to be.—Jeffrey W. TaliaferroMedford, Massachusetts—Steven E. LobellSalt Lake City, Utah—Norrin M. RipsmanBethlehem, PennsylvaniaIn my article "On Systemic Paradigms and Domestic Politics," I criticized neoclassical realism on several counts.1 I argued that it attempts to reconcile the irreconcilable: on the one hand, a commitment to the analytic priority of systemic pressures and, on the other hand, an open-ended engagement with domestic politics. Some neoclassical realists attempt to specify rules for crossing the two levels of analysis, but the result is inevitably arbitrary and incoherent. Lacking a solid deductive foundation, the paradigm induces the commission of methodological errors and hinders the production of knowledge about both systemic pressures and domestic politics.Of the three responses to my article, only that of Jeffrey Taliaferro, Steven Lobell, and Norrin Ripsman directly defends neoclassical realism. It does not resolve any of the problems summarized above; indeed, it does not even address them. Instead, it makes two main claims: first, that not all scholars assign to realism the assumption that states are functionally undifferentiated, "like units" with the same underlying preferences; and second, that liberalism is too flawed to constitute a useful alternative to neoclassical realism.On the first point, Taliaferro, Lobell, and Ripsman wonder why I rely on Patrick James's definition of realism.2 The answer should be obvious: James's approach is guided by philosophy of science, the field of inquiry devoted to evaluating the coherence and progressivity of theoretical paradigms. The purpose of my article was to evaluate the coherence and progressivity of neoclassical realism; thus, James's definition was a natural fit. Furthermore, James is attentive to the distinction between modern realism, which is based on an explicit and consistent set of assumptions, and classical realism, which is not.Most definitions of realism are not concerned with these issues and, therefore, are less systematic or reflect a particular agenda. Consider the four alternatives that Taliaferro, Lobell, and Ripsman suggest in a footnote: one is from a self-declared neoclassical realist (Randall Schweller), and two are from scholars writing about the classical realist tradition (Benjamin Frankel, Robert Gilpin). Only one, from John Mearsheimer, is a suitable arbiter of the definition of modern realism. Mearsheimer, however, is of no help to Taliaferro, Lobell, and Ripsman. He states that in realism, "great powers are like billiard balls that vary only in size"—"like units" that do not vary in their preferences.3Regardless, my critique of neoclassical realism does not depend on how realism is defined. Rather, I argue that neoclassical realism is incoherent by its own standards, on its own terms. I demonstrate this in my article, and it is evident in Taliaferro, Lobell, and Ripsman's letter. In their reply, the authors assert that "states and their leaders … are alike in that the international system compels them to be security-seeking actors above all. Nonetheless, it would be foolhardy to suggest either that all states have identical interests or that the international system determines what these interests are." These two sentences are plainly contradictory. If states seek security "above all," then the international system determines what their interests are. Taliaferro, Lobell, and Ripsman's next paragraph further belies their claim that states are "security-seeking actors above all": "in extreme circumstances, when they face the immediate threat of losing power, leaders may temporarily trade off security interests to solidify their domestic power positions, if they believe the damage to national security will not be too great."Consistency is not the only problem with this last sentence. As an attempt to specify rules for crossing levels of analysis, it irremediably is question-begging. What does "the immediate threat of losing power" mean, operationally? In most states, this is not an extreme circumstance. In a parliamentary regime such as the United Kingdom, any decision can bring down a government and force new elections at any time. The same applies to authoritarian regimes, in which leaders must constantly navigate between the Scylla of coups and the Charybdis of popular uprisings. In the United States, the president is elected every four years, but his ability to enact most of his policy agenda depends on support from Congress, for which there are elections every two years. How far in advance of such elections is "immediate"? Two weeks, two months—or two years? Finally, everything hinges on the leader's subjective beliefs about whether "the damage to national security will not be too great." With such a proviso, Taliaferro, Lobell, and Ripsman's argument crosses the boundaries of social science into the realm of unfalsifiability.There is one point on which I agree with Taliaferro, Lobell, and Ripsman: that "it would be foolhardy to suggest either that all states have identical interests or that the international system determines what these interests are." In my article, I argued that liberalism is the best tool for the analysis of such issues. Moreover, I demonstrated that neoclassical realists' standard criticism of liberalism, that it precludes consideration of systemic factors, is based on a straw man. In fact, liberalism requires consideration of how state leaders will adjust to international pressures that affect their societal coalition's interests (pp. 168, 184–185). As Andrew Moravcsik notes, "Each state seeks to realize its distinctive preferences under varying constraints imposed by the preferences of other states."4 Thus, liberalism poses no barrier to explaining, per Taliaferro, Lobell, and Ripsman, "why states with different regime types and dominated by different underlying coalitions make fundamentally similar choices when faced with similar international circumstances."Consider the example provided by Taliaferro, Lobell, and Ripsman, international support for the U.S. "war on terror" after the al-Qaida attacks of September 11, 2011. Liberalism is more useful than realism in explaining this phenomenon, because it suggests why the United States had to employ different levels and forms of pressure to co-opt different types of countries—from mere suasion vis-à-vis Australia to massive bribes of foreign aid vis-à-vis Pakistan. In other words, different preferences necessitated the application of different kinds of systemic pressure to produce similar results.Taliaferro, Lobell, and Ripsman further assert that liberalism is "not very useful in explaining the behavior of non-liberal states." This is a bizarre claim. The modern liberal paradigm, as developed by Moravcsik (see below), is not limited to the study of "liberal" states. Rather, it is founded on the analysis of the preferences of societal coalitions, whether authoritarian or democratic.5 In recent years, there has been an explosion of research on variance in the foreign policy behavior of different types of authoritarian states, and almost all of it is compatible with liberalism.6Taliaferro, Lobell, and Ripsman's final criticism of liberalism, that it is "teleological and normative," not only misrepresents Moravcsik but also ignores my discussion of this point. As I stated in my article, one must distinguish between Moravcsik's formulation of the modern liberal paradigm, which is self-consciously scientific, and the classical liberal worldview, "a hodgepodge of theories about the causes of peace and prosperity" that "is distinguished by its optimism rather than its internal coherence" (p. 158). I have argued only for the former; Taliaferro, Lobell, and Ripsman's point applies only to the latter. Their criticism, therefore, is of a straw man.Another response to my article, by Davide Fiammenghi, does not explicitly defend neoclassical realism. Nevertheless, it interprets Kenneth Waltz's Theory of International Politics as licensing realists to incorporate variation in state preferences.7 Fiammenghi makes three points. First, like Taliaferro, Lobell, and Ripsman, he argues that Waltz's assertion that states are "functionally undifferentiated" does not imply that all states should have the same preferences. I addressed this claim above.Second, Fiammenghi notes that one can extract from Theory of International Politics many more hypotheses than the three that Waltz emphasizes and that I mentioned in my original article. This is a fair point, but it does not alter my conclusions. The purpose of my discussion of Waltz was to point out that the modern realist paradigm has rejected his evolutionary logic, in which "structure selects" for states that prioritize survival. Instead, it assumes that states are rational actors. This assumption allows realists to theorize not just systemic outcomes, but also foreign policy, and consequently has proved exceptionally useful.8 As for Fiammenghi's hypotheses, some of them could be derived from an evolutionary logic, but none of them depends on it. Nor do they employ domestic politics as an explanatory variable. Instead, they are fully consistent with the modern realist paradigm and the criteria established by James. As such, they are irrelevant to my argument and provide no justification for neoclassical realism.Third, Fiammenghi argues that "the pursuit of security is compatible with states having different preferences (i.e., isolationism, pacifism, revisionism, the status quo, or unlimited revisionism)." This assertion is based on a simple conceptual error: the conflation of preferences, which are actors' underlying goals, with strategies, which are the means with which goals are pursued. All the "preferences" cited by Fiammenghi are actually strategies. This distinction is discussed at length by Jeffry Frieden in an essay that I described as "required reading for any scholar whose research crosses levels of analysis" (p. 189).9 Waltz has no truck with preferences; he attempts only to explain strategies. Again, Theory of International Politics provides no justification for neoclassical realism.Finally, the letter by Sebastian Rosato and Joseph Parent does not defend neoclassical realism and raises some issues that are tangential to my article. Consequently, I will not respond to it in full but rather will touch on two points of concern. One is its opening sentence: "Narizny argues that the past three decades of scholarship have done little to advance scholars' understanding of foreign policy." This is a mischaracterization of my position. I focus on the period since the late 1990s, which is two decades rather than three, and I do not dismiss scholarship as lacking progress. To be clear, my argument is not that research done within the analytic framework of neoclassical realism is valueless; rather, my argument is that the use of neoclassical realism as an analytic framework limits the value of that research.My other concern is with Rosato and Parent's suggestion that their own finding, that great powers "reacted routinely and promptly to the military buildups and innovations of others approximately 80 percent of the time," is dispositive of the debate between realism and liberalism. The purpose of liberal theories is not simply to explain how states respond to threats to their interests. More fundamentally, it is to determine how states define their interests. What motivates states to expand, why do they value different forms of influence and targets of expansion, and what accounts for variation over time and space? Per Moravcsik, "Preferences determine the nature and intensity of the game that states are playing and thus are a primary determinant of which systemic theory is appropriate and how it should be specified."10 Parent and Rosato may be right about "which systemic theory is appropriate" for their cases, but their research design leaves the deeper questions unanswered.—Kevin NariznyBethlehem, Pennsylvania
Referência(s)