Logic

The positivist-empiricist method, Moravcsik originally stated, provides a solution to the ‘ideological’ and ‘utopian’ formulations of liberalism in IR. It promises logic and consistency in place of the ‘disparate views’ one finds in traditional liberal theory based on classical liberal thought (1997: 514). Positivism achieves this goal by ‘proposing a set of core assumptions on which a general restatement of positive liberal IR theory can be grounded’ (1997: 515); linking the specific claims of mid-range theories to such general assumptions. In showing that his mid-range theories are not rigorously derived from his general assumptions (Jahn, Reference Jahn2009: 416–9) I have applied a standard to this theory that, Moravcsik argues, he explicitly rejects. In short, I have misunderstood his method. Yet, what I intended to show, and hope to clarify here, is that once that standard is rejected, the resultant method, when adjusted to the context of IR, does not live up to the promise of superior logic.

The Lakatosian method, according to Moravcsik, entails the claim that a ‘single set of microfoundational assumptions’ may be universally applicable. Yet he also holds that world politics ‘at least at its current state of development’ does not support such an assumption (2003a: 198). Moreover, even the broadest research programs in IR shy away from stating such a ‘claim to universality, even within a circumscribed domain’ (2003a: 198). This fragmentation of the world of international politics and theory leads Moravcsik to argue that the Lakatosian assumption of a zero-sum, knock out competition between different theories is unrealistic for IR. ‘There is no a priori reason to believe that such a universal claim would be valid’ (2003a: 198). Hence, we have to consider the possibility that different theories may be ‘differentially applicable across different specific empirical domains of world politics’ (2003a: 199). Such theories with a limited area of applicability can therefore not be unequivocally derived from a set of universal core claims. Hence, ‘in order to derive individual liberal theories precisely and to circumscribe their empirical scope (…) one needs auxiliary assumptions’ and consequently, these theories ‘need only be consistent with paradigmatic assumptions, not deduced from them’ (2010). Thus, Moravcsik rightly points out that he explicitly rejects the notion that mid-range theories have to be rigorously derived from the microfoundational assumptions (2003a: 176). His is a ‘soft social scientific position’ (2003b: 134) characterized by the loosening of the original strict logical link between paradigmatic assumptions and mid-range theories.

Thus, the question I investigated by assessing the logical connection between Moravcsik’s core assumptions and his mid-range theories was whether, and in how far, this ‘soft positivism’ was able to generate a ‘logically coherent, theoretically distinct, empirically generalizable’ theory (1997: 547). My argument here concerned solely the link between paradigmatic assumptions and mid-range theories – not, as Moravcsik mistakenly assumes, the validity or distinctness of his mid-range theories in comparison with neorealist and institutionalist approaches.Footnote 2 The claim that rational and risk-averse individuals and private groups are the fundamental actors in world politics does not logically imply, I argued, the particular liberal forms of socio-economic organization (market democracy, for instance) or of decidedly liberal foreign policies such as free trade (2009). Indeed, as Moravcsik himself points out, under certain circumstances it can be perfectly rational to support imperialism or protectionism (1997: 528, 529).

In short, once the connection is loosened, as it is when mid-range theories are not any longer derived from core assumptions but merely consistent with them, this method does not deliver any longer the superior logic and consistency of its original formulation. Yet, Moravcsik wants to have his cake and eat it. While sometimes insisting that mid-range theories only have to be ‘consistent’ with paradigmatic assumptions, at other times he clearly and repeatedly states that a theory is defined by ‘a set of positive assumptions from which arguments, explanations, and predictions can be derived’ (1997: 514); he ‘derives from them (the three core assumptions) three variants of liberal theory’ (1997: 513); his theory, he claims, ‘follows from explicit assumptions’ (1997: 547). This ambiguity arises out of Moravcsik’s desire to hold on to the promises of a Lakatosian approach – logic and consistency – while at the same time taking into account that the assumptions underlying this Lakatosian method do not apply in the sphere of IR.

In sum, when I tested the logical connection between Moravcsik’s core assumptions and his mid-range theories, I did hold his theory up to one of his clearly stated criteria: logic. And by doing so I showed that his second criterion – consistency – cannot deliver the superior logic for which he has chosen the positivist method in the first place.

Testability

This loosening of the logical connection between microfoundational claims and mid-range theories also has implications for the second major promise of the positivist method – which lies according to Moravcsik in making theoretical claims empirically testable: ‘it is important to formulate theories in a hypothetically generalizable way, so we can seek to determine the scope of potentially general claims under hypothetical conditions’ (2010). Here again I have supposedly misunderstood the positivist method, which does not hold that these claims have ‘a de facto universal scope’ – and hence cannot be tested directly but only via ‘empirical competition among potentially generalizable claims’, that is, by comparing the explanatory range and power of competing theories empirically (2010). There are good reasons, however, for not following this procedure, two of which I will set out below.

First, if, as I have argued above, the positivist method adapted to the field of IR does not any longer allow a logical derivation of mid-range theories from paradigmatic assumptions, then those general assumptions cannot be tested by comparing the explanatory power of mid-range theories. A successful empirical challenge to the Democratic Peace thesis, for exampleFootnote 3, does not simultaneously undermine the paradigmatic claim that rational individuals are the core actors in international affairs. More generally, and in Moravcsik’s own words, these assumptions ‘do not define a single unambiguous model or set of hypotheses, not least because they do not specify precise sources of state preferences’ (1997: 524). This lack of an unambiguous connection between paradigmatic claims and mid-range theories thus makes it impossible to test the validity and range of the former through the performance of the latter.

Second, the positivist method insists on the hypothetical generalization of claims because, before ‘theory-guided empirical analysis’ we cannot know where the limits of these claims lie (Moravcsik, Reference Moravcsik2010). The general formulation of claims is meant to lay these claims open to the widest range of empirical testing and thus constitutes the ‘agnostic approach of the true social scientist’ (2010) – that is, an approach that does not prejudge the limits of theoretical claims a priori. Yet, this argument overlooks that empirical facts do not unequivocally speak for themselves. Instead, they require a theoretical language in which they can be communicated (Kratochwil, Reference Kratochwil2003: 124). Thus, whether or not the First World War is a case supporting or undermining the democratic peace thesis depends on whether Germany at the time was a democracy or not. This decision, in turn, is entirely dependent on the definition of democracy – a theoretical, not an empirical exercise (Spiro, Reference Spiro1996). Moreover, whether such a case constitutes a ‘refutation’ of the democratic peace thesis or just an ‘anomaly that … does not fundamentally challenge’ it, cannot be decided on empirical grounds (Kratochwil, Reference Kratochwil2003: 125).

Hence, the positivist method cannot provide a clear statement of the nature and scope of empirical evidence that would suffice in practice to refute its claims. This problem is compounded by the general formulation of its theoretical claims, which in principle require equally wide ranging empirical studies for their substantiation or refutation. There is, then, to all intents and purposes, no limit to the empirical tests that could be required. Again, the democratic peace thesis is a case in point. Moravcsik argues that this thesis has been open to constant challenges concerning its veracity and causal logic (2010), demonstrating the ‘agnostic’ nature and the fruitfulness of the positivist approach. Contra Moravcsik, I would argue, however, that the longevity of this thesis, in spite of the range and seriousness of these challengesFootnote 4 and the reams of paper used up in the endless attempts to ‘prove’ or ‘disprove’ it, demonstrates nothing but the impossibility of a purely empirical refutation of such theoretical claims. Moravcsik is right, of course, to point out that this thesis invites empirical testing. But since the conditions under which the thesis could be refuted cannot be specified, this method sends scholars off onto a wild goose chase for empirical evidence that can never reach its goal – whether for or against.

In sum, the lack of a logical connection between core assumptions and mid-range theories, combined with the lack of clear criteria for what counts as relevant empirical data, entails that this positivism ironically ends up removing its general theoretical claims from testing.

It is precisely because this method does not offer the possibility of testing its core claims that I chose to address its underlying assumptions directly. This exercise, I want to show now, is perfectly in line with Moravcsik’s own procedure; indeed, its findings are also substantively supported by Moravcsik’s own work – although in ways which undermine his general claims. Moravcsik argues that my ‘casual efforts to use a single thinly documented historical interpretation – for example, the claim that Early Modern Europeans were not “rational”, based on a secondary interpretation of John LockeFootnote 5 – to support a blanket refutation of the rationality assumption for pre-modern situations’ do not suffice. ‘Locke’s work simply cannot be made to bear such explanatory weight’ (2010). So what exactly did I do? I took the work of an eminent liberal (or protoliberal) thinker, John Locke, and showed that this contains two different claims about rationality: a general claim that all human beings are rational, and a historical or empirical claim that this rationality did not result in general support for what one might for the sake of brevity call ‘liberal’ polities and policies. Locke’s proposed solution to this problem, I showed, was to create the circumstances, the social and political conditions, under which the potential rationality of all people could be expected to result in specifically ‘liberal’ policies. Finally, I provided a range of evidence suggesting that Locke’s solution was indeed translated into political practice over several centuries both domestically and internationally. In short, I distilled a theoretical claim from the work of John Locke and then provided empirical evidence of its centrality to subsequent liberal political practice (2009).

This general procedure does not depart in the slightest from Moravcsik’s own. If we leave aside his formulation of general paradigmatic claims, (from which, as I have shown, his mid-range theories are not deduced), we are left with the way in which he formulates his mid-range theories. To begin with, he takes the work of a number of classical (and more contemporary) liberal authors such as John Stuart Mill, Giuseppe Mazzini, Woodrow Wilson, and Adam Smith from which he distills some core claims that characterize, according to Moravcsik, ideational, commercial, and republican versions of liberal theory. He then provides a number of empirical examples in support of these claims. There is, then, no systematic difference between Moravcsik’s actual procedure for arriving at mid-range theories and my own; they are both ‘theory-guided empirical analyses’ (Moravcsik, Reference Moravcsik2010). The only difference lies in the fact, as I am happy to concede, that he provides a wider range of evidence than I – largely due to the fact that the main purpose of my article was the discussion of ideology and not the development of a full-blown alternative theory of liberal internationalism.

More importantly, however, Moravcsik’s own work provides illustration and substantive support for my thesis. Moravcsik, like Locke, formulates the general claim that human beings are rational. Moravcsik, like Locke, notes that this rationality does not necessarily translate into ‘liberal’ policies. Thus, for Moravcsik as for Locke, it is the social and political context that ultimately determines, which form this rationality takes in political life. Thus, liberal theory holds in principle that free trade is a more efficient, and rational, way to maximize economic benefits; yet market structures, argues Moravcsik, can create ‘incentives for both openness and closure’ (1997: 529). In a non-liberal or partially liberal context – ‘where the main sources of economic profit, such as farmland, slave labor, raw materials, or formal monopoly could be easily controlled in conquered or colonial economies’ – it may be perfectly rational to pursue economic preferences by means of war or conquest (1997: 530). The same holds for the political context: when ‘the military, uncompetitive foreign investors and traders, jingoistic political elites … are particularly well-placed to influence policy’ even liberal states can embark on imperialist policies (1997: 532). Indeed, it is this very point – that what is rational depends on the context – that Moravcsik cites as making his theory superior to the ‘ideological’ formulations of liberalism (1997: 528–9, 532). This insight is perfectly in line with my argument: if the particular form rationality takes is determined by the circumstances, this undermines the general assumption of rationality as a meaningful claim. If the implications of rationality can range from free trade to protectionism, from imperialism to self-determination, or, for that matter, from empiricism to ideology critique, then this claim does not provide some common ground either for political action or its theorization.

In sum, Moravcsik’s positivist method does not fulfill its promises. The promise of superior logic rests on the logical linkage of paradigmatic assumptions and mid-range theories – a linkage that Moravcsik has loosened to one of mere consistency, first, with the result that the watered down version does not deliver the logic of the strict version any longer and, second, with the result that the paradigmatic claims are not any longer open to testing through the mid-range theories. In addition, since this method fails to specify the nature and scope of evidence necessary for its refutation, it removes its general theoretical claims from such challenges, even while it apparently lays them open to empirical testing. It is because this method does not offer what it promises, I would argue, that Moravcsik in practice proceeds in exactly the same way I do. What is more, by arguing that the particular form rationality takes in politics is context dependent, Moravcsik provides substantive support not only for my interpretation of Locke (and its theoretical and political implications) but also, and more importantly, for the concept of ideology and its implications.

Evidence

Moravcsik subscribes to the positivist method because it promises to overcome the ‘subjective interests and values’ of the scholar, to provide a solution to the problem of context-dependent rationality, or ideology. It imposes

In contrast to this ideal of an inter-subjective language, Moravcsik suggests, I am committed to relativism. And instead of accepting the need for empirical testing, I make judgements on the basis of ‘methodological and philosophical fiat’ (2010). Hence, I ‘overlook literally hundreds of pages of directly relevant work’ and am ‘flagrant’ in my ‘unwillingness to provide any logical or empirical support’ for my claims (2010). Yet, in making these accusations, Moravcsik ignores the common ground we occupy – the recognition of political and theoretical fragmentation – and now suddenly treats his method as an uncontested universal language.

The charge that I fail to provide either arguments or empirical evidence to back up my own claims requires clarification, simply because I have provided these, for example, concerning Locke’s conception of rationality and its political consequences.Footnote 6 In fact, even Moravcsik acknowledges that I do empirical work ‘of a bracing and provocative’ nature – though not enough of it. The problem is thus not the lack of evidence but the nature of this evidence. Moravcsik holds that where I do empirical work, it is ‘conjectural’ and this disqualifies it: ‘in order for it to be coherent and credible, it must be tightly disciplined. It must involve concrete testing of competing mid-range theories, using objective data, communicated in a way all can understand’ (2010). In other words, because my arguments and evidence do not follow the rules of the positivist method, they are neither ‘coherent’ nor ‘credible’. Having thus discounted my arguments and evidence, Moravcsik comes to the conclusion that my judgements are based on nothing but ‘methodological and philosophical fiat’; that my ‘criticism floats in a philosophical vacuum’ (2010).

Both Moravcsik and I subscribe to the need to back up our claims with arguments and evidence – and we both do provide these. The difference lies in the manner in which they are arrived at and presented. Yet, it is Moravcsik’s method that falls short of providing a ‘universal’ language that ‘all can understand’. For such a language, as Moravcsik himself states, is ‘an attractive ideal’ – and so far nothing but an ideal. After all, even within the highly circumscribed and largely ‘Western’ field of IR we find a wide range of other languages that do not subscribe to this method, from constructivist, through normative, marxist, gender, poststructuralist to postcolonial theories. In fact, even within the liberal tradition itself we can identify at least three broad approaches – normative (Beitz, Reference Beitz1979, Held, Reference Held1996), historical (Hoffmann, Reference Hoffmann1995; Zacher and Matthew, Reference Zacher and Matthew1995; Richardson, Reference Richardson2001), and Moravcsik’s own empirical or positivist approach. And it was precisely this recognition of the theoretical fragmentation of the field of IR that led Moravcsik to propagate the positivist method as a basis for communication between different approaches.

Yet, instead of then showing how this method is capable of relating to, and communicating with, alternative approaches like mine, Moravcsik suddenly treats the positivist method as the uncontested universal language of scholarly discourse. And on this basis, he explicitly excludes my arguments and evidence simply by virtue of their ‘basic mode of criticism’ (2010). This, surely, is a judgement based on ‘methodological and philosophical fiat’ (2010). Is it not Moravcsik who, though he believes ‘that a scholarly community should be the site for conversations, not soliloquies’ (2010), here refuses conversation on any other but his own ground?

Politics

Finally, Moravcsik holds his own engagement with ‘people, states, power, and other real things’ up against my preference to replace such engagement with discussions of ‘epistemology, methodology and other abstract concepts’ (2010). I ignore Moravcsik’s applied and policy-oriented work; I misrepresent his political position; and I end up proposing that the ‘solution to current problems lies in deep philosophical reflection on social theory’ (2010). Here, too, I think, Moravcsik fails to see the quite considerable common ground between our positions.

To make this clear, it is helpful to recall that my own reflections on ‘epistemology, methodology and other abstract concepts’ in the context of this exchange were triggered by Moravcsik’s 1997 article. The purpose of that article was to replace the ‘ideological’ formulations of liberal theory with a ‘non-ideological and nonutopian form appropriate to empirical social science’, that is, based on a different, positivist, methodology (1997: 513). The 1997 article is teeming with terms like ‘teleology’, ‘microfoundational assumptions’, ‘paradigmatic restatement’, ‘methodological’, ‘theoretical’, ‘empirical’, ‘hypotheses’ and so on (1997: 515, for instance). Moravcsik (Reference Moravcsik1997, Reference Moravcsik2003a, Reference Moravcsikb) is thus no stranger to discussions of ‘epistemology, methodology and other abstract concepts’. Indeed, such reflections constitute an integral part of his work. Moreover, authors like Lakatos and Musgrave (Reference Lakatos and Musgrave1970) and Laudan (Reference Laudan1996), who figure prominently in Moravcsik’s work, are surely no less abstract than Mannheim (Reference Mannheim1960). Indeed, it could be argued that Mannheim’s reflections, based as they are on an empirical history of the development of political thinking, are much less abstracted from the world of politics than reflections on positivist method which originates, after all, in the natural sciences.

So, why is it that Moravcsik, just like me, regularly engages in such abstract reflections? He does so because he is dissatisfied with the limits of particular approaches. Thus, he notes that the traditional formulations of liberal theory in IR cannot accommodate or explain clear instances of ‘non-liberal’ behavior (such as protectionism, imperialism, the use of force) on the part of liberal actors (1997: 528–32), or account for the failure of supposedly liberal institutions like the League of Nations (1997: 546). This weakness of traditional formulations of liberalism, Moravcsik holds, can be rectified by adopting a different methodological approach. In other words, he recognizes that methodological approaches define what does and does not count as evidence, what falls within the purview of a particular analysis and what falls outside it. The turn to abstract theoretical and methodological reflections serves the purpose, then, of finding an approach that promises to overcome the limitations of existing approaches. And it is this goal that leads Moravcsik more generally to argue for a synthesis of rationalist approaches – each of which has certain valuable insights to offer with respect to particular issue areas (2003a: 199). In other words, and perfectly in line with my own argument regarding ideologies, these particular approaches contain some truths, if of a limited nature (2009: 415, 434). A synthesis thus promises to widen the scope of explanation beyond individual approaches without sacrificing their individual strengths.

Hence, both Moravcsik and I engage for exactly the same reason in abstract reflections: in order to overcome the limitations of particular approaches in the analysis of concrete problems in world affairs. In other words, it is the aim of providing the best possible analysis of concrete problems in world affairs that forces scholars on occasion to reflect on the shortcomings of existing approaches and to try to rectify these. Such reflection does not provide solutions for concrete problems, but it is a crucial precondition for it.

The question, then, is whether Moravcsik’s approach actually does encourage the analysis of concrete problems of world affairs. As Moravcsik’s exposition clearly shows, this approach demands empirical analyses not for the purpose of understanding such concrete political problems but for the purpose of testing theories. It encourages studies designed to trace ‘the evolution of the mid-range research’; ‘empirical competition among potentially generalizable claims is the key’ (2010). The objective of empirical research is thus not a concrete problem but rather the empirical performance of a particular theory in competition with other theories. And these theories, in turn, do not necessarily address particularly urgent concrete problems in world affairs. The democratic peace thesis, for example, claims that democracies do not fight each other. This claim quite explicitly entails the recognition that the real problems of world affairs lie elsewhere – and not in the relations between democratic states. The thesis does thus not appear to address, at least not directly, a particularly urgent problem. Yet, given this thesis, the positivist method encourages studies into the question, for example, of whether Germany before WW I was a democracy, or was perceived as a democracy, or not – and thus whether the claim holds for this case (Oren, Reference Oren1996). This, surely, is even less of a burning concrete problem in world affairs. There is nothing in this method, then, that particularly encourages the study of concrete political problems. On the contrary, it may be argued that what it does encourage is the testing of theories, rather than the study of concrete problems.

Nevertheless, Moravcsik (Reference Moravcsik2007a, Reference Moravcsikb), it is true, does engage with concrete political issues. Yet, I would argue he does that, and to his credit, not because, but in spite of, his positivist method.

My reflections on the implications of ideology, in contrast, lead straight back into the analysis of concrete problems in world affairs. An appropriate response to the problem of ideology, I suggested, ‘requires in practice an engagement with (the) conditions of emergence (of political knowledge) and an historical account of its struggle with internal and external competitors’ (2009: 436). Applied to the concrete case of liberalism at hand, it requires an account of the historical conditions of emergence of liberal thought and practice and of its struggle with internal and external competitors. This brings the conflict between liberal and non-liberal forces, as well as the internal competition between different versions of liberalism directly into focus. And these issues are surely relevant for concrete political questions like democracy promotion, intervention, statebuilding – all of which revolve around liberal-non-liberal relations and on all of which we do find politically relevant competing positions within liberalism (Jahn, Reference Jahn2007a, Reference Jahnb). This approach, in short, does encourage the analysis of concrete political problems. And if my own work should not live up to these expectations, as Moravcsik claims, it is, just as in his case, despite and not because of this approach.