リアリズムが国際政治学をダメにした: ケネス・ウォルツの勢力均衡論

Pocket

John A. VasquezThe Power of Power Politics: From Classical Realism to Neotraditionalism (Cambridge University Press, 1999)の第11章でケネス・N・ウォルツの勢力均衡論が強く批判しています。第11章は本著で一番面白い章です。

第11章: 退行的リサーチ・プログラムとしてのリアリスト・パラダイム ─ ネオ伝統主義とウォルツの均衡に関する命題

“If there is any distinctively political theory of international politics, balance-of-power theory is it.” (Waltz, Theory of International Politics: 117)

ネオリアリストのバイブルであるケネス・ウォルツの『国際政治の理論』の中心的議論は、中央政府の存在しない国際政治では国家は必然的に勢力均衡政策をとらざるを得なくなるというものです。本章でVasquezはイムレ・ラカトシュのリサーチプログラム論に基づき、国際政治学におけるウォルツの勢力均衡論に伴う議論が前進的か退行的か分析しています。ラカトシュの議論は下記文献で確認できます。翻訳物もあります。
Lakatos, Imre (1970) “Falsification and the Methodology of Scientific Research Programmes” in I. Lakatos and A. Musgrave (eds.) Criticism and the Growth of Knowledge, Cambridge: Cambridge University Press, pp. 91-196.

ラカトシュの議論を長々引用・説明するのは面倒なので彼のリサーチプラグラム論を説明している日本語のサイトを確認しましたが、すぐには良いサイトを見つけることができませでした。ということで長々と引用します。

「退行的な」リサーチ・プログラムとはどのようなものか?
How can we tell whether a series of theories that comes out of a research program is degenerating? First, the movement from T to T’ may indicate a degenerating tendency if the revision of T involves primarily the introduction of new concepts or some other reformulation that attempts to explain away discrepant evidence. Second, this would be seen as degenerating if this reformulating never pointed to any novel unexpected facts, by which Lakatos means that T’ should tell us something about the world and its regularities other than what was uncovered by the discrepant evidence. Taken together, these two indicators are one type of ad hoc explanation – what Lakatos (1970: 175 note 2) calls ad hoC(see Elman and Elman 1997: 923). Third, if T’ does not have any of its new propositions successfully tested or lacks new propositions (other than those offered to explain away discrepant evidence), then it does not have corroborated excess empirical content over T, which can be an indicator of a degenerating tendency in the research program – what Lakatos (1970: 175 note 2) calls an ad hoc2 explanation (see Elman and Elman 1997: 923). Fourth, if a research program goes through a number of theory-shifts, all of which have one or more of the above characteristics and the end result of these theory-shifts is that collectively the family of theories has fielded a set of contradictory hypotheses that greatly increase the probability that at least one of them will pass empirical testing, then a research program can be appraised as degenerating. (pp.244-5)

要するに、退行的なリサーチ・プラグラムとは、現実が理論から導き出される予測と異なるときにその場しのぎの説明(ad hoc explanation)を加えることで、どんな現象でも事後的に説明できるようにし、理論の反証可能性が低下し続けるリサーチ・プログラ ムのことを指します。例えば、リアリストは冷戦の突然の終焉を予測できませんでしたが、適当な理屈を後からつけ加えることで、冷戦終焉はリアリズムの議論と矛盾しないと言い始めるのが「退行的」。逆に、リアリズムが冷戦終焉をうまく説明できないことを認めたうえで、リアリズム理論が適用可能な国際政治現象の領域を理由付きで限定することは、リアリズムが冷戦の終わりを予測できなかった理由も明らかになり、「前進的」なリサーチ・プログラムとなります。「退行的」なものとしてVasquezが例に挙げるのが、パラダイム内に(1)戦争はパワーが不均衡な時に生じやすい、(2)戦争はパワーが均衡しているときに生じやすい、という矛盾する命題双方を含む場合です。ひとつのパラダイム内に矛盾する命題を抱えれば反証不能になります。
An effect of the repeated semantic changes that are not progressive is that they focus almost entirely on trying to deal with experimental outcomes or the identification of empirical patterns that are contrary to the initial predictions of the theory. One consequence of this kind of scholarly practice is that collectively the paradigm begins to embody contradictory propositions such as (1) war is likely when power is not balanced and one side is preponderant and (2) war is likely when power is relatively equal. The development of two or more propositions that are contradictory increases the probability that at least one of the paradigm’s propositions will pass an empirical test. When a series of theories, all of which are derived from the same paradigm (and claim a family resemblance, such as by using the same name – e.g., Freudian, Marxist, or realist), predict one of several contradictory outcomes as providing support for the paradigm, then this would be an example of the fourth indicator. Carried to an extreme, the paradigm could prevent any kind of falsification, because collectively its propositions would in effect be posing the bet, “Heads, I win; tails, you lose.” A research program could be considered blatantly degenerative if one or more of the behaviors predicted were only predicted after the fact. (p.246)

ネオリアリズムはまさにそのような退行的なリサーチ・プログラムであるとVasquezは主張します。
It will be argued that what some see as an apparent theoretical fertility and development of the realist paradigm is really a proliferation of emendations that prevent it from being falsified. It will be shown that the realist paradigm has exhibited (1) a protean character of theoretical development, which plays into (2) an unwillingness to specify what form(s) of the theory constitutes the true theory that if falsified would lead to a rejection of the paradigm, (3) a continual and persistent adoption of auxiliary propositions to explain away empirical and theoretical flaws that greatly exceeds the ability of researchers to test the propositions, and 4) a general dearth of strong empirical findings. Each of these four characteristics can be seen as “the facts” that need to be established or denied to make a decision about whether a given research program is degenerating. (p.248)

ウォルツの議論の要点は、中央政府の存在しない自助システムである国際政治では、国家アクターは均衡行動(balancing behavior)をし、バランス・オブ・パワーが自ずから生まれるという主張です。
He maintains that a self-help system “stimulates states to behave in ways that tend toward the creation of balances of power” (Waltz 1979: 118) and that these balances “tend to form whether some or all states consciously aim to establish “them” …”(Waltz 1979: 119). This law or regularity is what the first six of the nine chapters in Theory of International Politics are trying to explain (see, in particular, Waltz 1979:116-128). (p.251)

本当に国家は均衡行動に出るのか。ウォルツ以後、数多くのリアリストが実証分析をします。それらの研究は「前進的」or「退行的」のいずれか?、というのが本章の主題です。

ウォルトの脅威の均衡論

強い国に対抗するのがbalancing、強い国につくのがbandwagoningです。ウォルツは中央政府の存在しない自助システムでは国家はbalancingの行動に出ると主張します。何にbalancingするかというと、パワーに対してです。国家はパワーを持つ国に対して対抗しようとするのがウォルツの勢力均衡論です。

Stephen WaltThe Origins of Alliances (Ithaca: Cornell University Press, 1987)で、ウォルツが主張するように国家はbandwagonではなく、balancingに出ることを明らかにしますが、何に対してbalancingするかというと、ウォルツとは異なり、powerではなくthreatであると主張します。ここで注目すべきことは、ウォルトの主張は明らかにウォルツの勢力均衡論に反しているにもかかわらず、彼はネオリアリズムの議論を洗練させただけであると主張していることです。
Here, then, is a clear falsification of Waltz – in the naive falsification sense of Popper (Lakatos 1970:116) – but how does Walt deal with this counterevidence, or counterinstances as Lakatos would term it? He takes a very incrementalist position. He explicitly maintains that balance of threat “should be viewed as a refinement of traditional balance of power theory” (Walt 1987: 263). Yet in what way is this a “refinement” and not an unexpected anomalous finding given Waltz’s prediction? (p.255)
(Kenneth N. Waltzのwikiを見ると、These recurring principles of behavior include balancing of power (the theory was refined by Stephen Walt, modifying the “balance of power” concept to “balance of threat”) …と書かれています)

し かし、国家がバランスしようとするものがパワーではなく脅威という主張はそれまでのリアリストの議論に反します。リアリストが脅威ではなくパワーを重視するのは、相手の意図がいつ変わるかわからないためです。だから国家は隣国と大きな対立が現時点ではなくても、隣国に対抗して軍事力をもとうとするとリアリストは主張します。他国が自国を破壊する能力があったとしてもその国と友好関係にあれば、つまり脅威でなければ対抗する必要がないというのはリベラリストの考えです。 「パワー」は「能力」、「脅威」は「能力」と「意図」の変数の関数であり、「パワー」のみを重視するのがリアリスト、「能力」と「意図」を重視するのがリ ベラリスト、「意図」のみを重視するのがユートピアンなのですから。
For Morgenthau and Waltz, the greatest source of threat to a state comes from possible power advantages that another state may have over it. In a world that is assumed to be a struggle for power and a self-help system, a state capable of making a threat must be guarded against because no one can be assured when it may actualize that potential. Hence, states must balance against power regardless of immediate threat. If, however, power and threat are independent, as they are perceived to be by the states in Walt’s sample, then something may be awry in the realist world. The only thing that reduces the anomalous nature of the finding is that it has not been shown to hold for the central system of major states, i.e., modern Europe. If it were there that states balanced threat and not power, then that would be serious, if not devastating, for neorealism and the paradigm. (pp.255-6)

ウォルトが「脅威の均衡」論をもってリアリストの議論を補強しようとする試みはリアリズムが「退行的な」リサーチ・プログラムであることの表れであるとVasquezは主張します。リアリストの理論がより反証しにくくなるためです。
To show that the concept of the balance of power lacks empirical significance must be devastating for the paradigm because it is a central concept. Each emendation tries to salvage something, but does so by moving further and further away from the original conception of balance of power. (p.263)

Ultimately under the fourth indicator, such theory-shifts are also degenerating because they increase the probability that the realist paradigm will pass some test, since now three kinds of behavior can be seen as confirmatory. While any one version of realism (balance of power, balancing power, balance of threats, balance of interests) may be falsified, the paradigm itself will live on, and indeed be seen as theoretically robust. In fact, however, the protean character of realism is preventing the paradigm from being falsified because as soon as one theoretical variant is falsified, another variant pops up to replace it as the “true realism” or the “new realism.” (p.264)

クリステンセンとスナイダーのbuck-passingとchain-ganging

Christensen, Thomas J., and Jack Snyder (1990) “Chain Gangs and Passed Bucks: Predicting Alliance Patterns in Multipolarity,” International Organization 44 (Spring): 137-168.

buck-passingは「責任転嫁」、chain-gangingはa group of prisoners chained together and forced to work、つまり1つの鎖につながれて屋外労働をさせられる囚人のことを指します。無駄にわかりにくい用語を使っていますね。この論文でクリステンセンとスナイダーは国家はbalancingの行動に出ないこともあると主張します。それに影響を与えるのが「攻撃─防御バランス」です。2人によれば、第一次世界大戦前の同盟行動はchain-gangin、つまり、同盟の縛りに拘束され、第二次世界大戦前は、各国が同盟を結んで脅威であったドイツに対抗できず、buck-passingの行動に出ます。それは第一次世界大戦前は攻撃有利と見られ、第二次世界大戦は防御有利であると見られたためだそうです。ちなみにこれらの行動は多極でしか生じないとのことです。
The bandwagoning research program is not the only way in which the protean character of realism has been revealed. Another and perhaps even more powerful example is the way in which Christensen and Snyder (1990) have dealt with the failure of states to balance. They begin by criticizing Waltz for being too parsimonious and making indeterminate predictions about balancing under multipolarity. They then seek to correct this defect within realism by specifying that states will engage in chain-ganging or buck-passing depending on the perceived balance between offense and defense. Chain-ganging is when states, especially strong states, commit “themselves unconditionally to reckless allies whose survival is seen to be indispensable to the maintenance of the balance”; whereas buck-passing is a failure to balance and “counting on third parties to bear the costs of stopping a rising hegemon” (Christensen and Snyder 1990: 138). The alliance pattern that led to World War I is given as an example of chain-ganging, and Europe in the 1930s is given as an example of buck-passing. The chain-ganging/buck-passing proposition is applied only to multipolarity because in bipolarity balancing is seen as unproblematic. (p.265)

国家がパワーに対してバランス行動に出ないこともあるという主張は明らかにウォルツの議論を根本的に否定するものですが、ネオリアリズムはびくともしません。ネオリアリズムによれば、もともと多極は不安定なシステムだからそうです。
On closer inspection, there are two major problems that reveal the degenerating character of this emendation. First, and most importantly, the argument that states will engage in either buck-passing or chain-ganging under multipolarity is an admission that in important instances, like in the 1930s, states fail to balance the way Waltz (1979) says they must because of the system’s structure. Recall Waltz’s (1979: 121) clear prediction that “Balance-of-power politics prevail wherever two, and only two, requirements are met: anarchy and units wishing to survive.” Surely, these requirements were met in the pre-World War II period, and therefore failure to balance should be taken as falsifying evidence. (p.266)

My objection to Christensen and to Snyder is not with their criticisms of neorealism, but their apparent unwillingness to admit the extent to which the evidence they present poses such a damning portrayal of Waltz’s balancing proposition. I can agree that balancing often does not occur because of misperception (generally), because of uncertainty (i.e., low information), because of the role of domestic politics, but for those outside the realist paradigm all this is another way of saying that balancing never occurs in the systematic fashion that Waltz believes, and “power” is not the core of what shapes world politics. (pp.270-1)

歴史的事例研究

Rosecrance, Richard, and Arthur A. Stein (eds.) (1993) The Domestic Bases of Grand Strategy, Ithaca: Cornell University Press.

リチャード・ローゼクランスとアーサー・A・ステイン編著の『グランドストラテジーの国内的基盤』も国内的要因をブラックボックス化しているウォルツの議論を真っ向から否定し、国家戦略の設定には国内的要因の影響が強いことを明らかにしますが、ネオリアリズムのリサーチ・プログラムに打撃を与えることはできませんでした。
Unlike the above explicitly sympathetic work, there have been several historical case studies that have focused on the balancing hypothesis and which have given rise to more severe criticism of realist theory. Rosecrance and Stein (1993: 7) see the balancing proposition as the key prediction of structural realism. In a series of case studies, they challenge the idea that balancing power actually occurs or explains very much of the grand strategy of twentieth-century major states they examine; to explain grand strategy for them requires examining domestic politics (Rosecrance and Stein 1993: 10, 17-21). In contradiction to structural realism, they find that balance-of-power concerns do not take “precedence over domestic factors or restraints” (Rosecrance and Stein 1993: 17). Britain in 1938, the United States in 1940, and even the Soviet Union facing Reagan in 1985 fail to meet powerful external challenges, in part because of domestic political factors (Rosecrance and Stein 1993: 18; see also the related case studies in chs. 5-7). States sometimes under- or over-balance. As Rosecrance (1995: 145) maintains, states rarely get it right – they either commit too much or too little, or they become so concerned with the periphery they overlook what is happening to the core (see Kupchan 1994; Thompson and Zuk 1986). And of course they do this because they are not the unitary rational actors the realist paradigm holds them to be. Contrary to Waltz, and even Morgenthau, states engage in much more variegated behavior than the realist paradigm suggests. (pp.272-3)

Elman and Elmanのネオリズム擁護

ネオリアリズムの主張が実証的にまったく支持されないことを明らかにしたのが歴史学者のポール・シュローダーです。
Schroeder, Paul, W. (1994a) “Historical Reality vs. Neo-realist Theory,” International Security 19 (Summer): 108-148.

しかし、2人のElman(夫妻?)はウォルツ擁護に走ります。
Elman, Colin, and Miriam Fendius Elman (1995) “History vs. Neo-realism: A Second Look,” International Security 20 (Summer): 182-193.
(1997) “Lakatos and Neorealism: A Reply to Vasquez,” American Political Science Review 91 (December): 923-926

2人は国家がbalancing(強者に対抗すること)ではなくbandwagon(勝ち馬に乗ること)することを示したシュローダーに対し、balance-of-powerであろうと、bandwagonであろうと、国家が利己的に行動する限りリアリズムに反しないと暴論を吐きます。つまり、リアリズムにとっての反証事例は国家が利他的行動に出る時のみだということです。
Instead, they conclude of Schroeder’s (1994a) historical evidence that “No evidence could be more compatible with a neo-realist reading of international relations” (Elman and Elman 1995: 184; constrast this with Mastanduno 1997: 86 note 115). They conclude this because each of these strategies (bandwagoning, etc.) does not challenge the realist conception of a rational actor behaving in a situation of competition and opportunity. For them, so long as states choose strategies that are “consistent with their position in the global power structure and pursue policies that are likely to provide them with greater benefits than costs” (Elman and Elman 1995: 184), then this is seen as evidence supporting the broad realist approach. Only Wendt’s (1992) claim that states could be “other-regarding” as opposed to “self-regarding” is seen as discrepant evidence (see also Elman 1997: 50-51). (p.277)

What kinds of political actors would, for example, consciously pursue policies that are “likely to provide” them with greater costs than benefits? To see only “other-regarding” behavior as falsifying leaves a rather vast and variegated stream of behaviors as supportive of the paradigm. Schroeder (1995: 194) has a legitimate complaint when he says in reply, “The Elman argument … appropriates every possible tenable position in IR theory and history for the neo-realist camp.” He concludes, “… their whole case that history fits the neorealist paradigm falls to the ground because they fail to see that it is their neo-realist assumptions, as they understand and use them, which simply put all state action in the state system into a neo-realist mold and neo-realist boxes, by definition” (emphasis in the original, Schroeder 1995: 194).

Elmanによれば、二極よりも多極の方が勢力均衡政策を維持するのが難しいのだから、多極であった第一次世界大戦と第二次世界大戦時にそれをしなかった国があったり、うまくできなかった国があっても、それはネオリアリズムに対する反証事例にはならないのだそうです。
The cause of this problem is the lack of rigor in the field in appraising theories. The nature of this problem can be seen in Elman and Elman’s (1995) second point against Schroeder. Drawing upon Christensen and Snyder (1990), they note that balancing under multipolarity, for Waltz, is more difficult than balancing under bipolarity: “Thus Schroeder’s finding that states failed to balance prior to World War I (pp. 122-123) and World War II (pp. 123-124) does not disconfirm Waltz argument … In short, a failure to balance is not a failure of balance of power theory if systemic conditions are likely to generate this sort of outcome in the first place” (Elman and Elman 1995: 190-191). This sets up a situation where any failure to balance under multipolarity can be taken as confirmatory evidence because, according to Elman and Elman (1995: 90), “… Waltz’s theory also predicts balancing failures” (emphasis in the original). This again poses an “I win, you lose” bet. If the periods before World War I and World War II are not legitimate tests of Waltz’s prediction of balancing, then what would be? The implication is that balancing can occur only under bipolarity, when external balancing is structurally impossible by definition. If this is the case, how is balancing a “law,” or the main outcome of anarchy? This is especially problematic because there is a tendency in Waltz to see only the post-1945 period as a true bipolarity (see Nye 1988: 244), which means the rest of history is multipolar and subject to balancing failures. In the end, Elman and Elman (1995: 192) concede that Waltz does believe that “on aggregate” states should balance, so that: “Schroeder’s evidence that states rarely balance does indeed pose a problem for Waltz’s theory.” (p.278)

このようにウォルツのネオリアリズム論はいくら反証事例が示されてもびくともしません。ポパーの反証主義がまったく無視されているわけです。
This is a much better reaction to Schroeder than what Waltz (1997a), Walt (1997) and even Christensen and Snyder (1997) say. Waltz (1997a: 914) says, “What Vasquez takes to be Schroeder’s ‘devastating evidence’ turns out to be a melange of irrelevant diplomatic lore.” From the scholar who became renowned for dismissing all quantitative evidence as “correlational labors” (Waltz 1979: 12), we now find that serious discussion of the historical record is simply “diplomatic lore.” Will any evidence ever be sufficient to test his proposition? (pp.278-9)
Waltz, Kenneth N. (1997a) “Evaluating Theories,” American Political Science Review 91 (December): 913-917.

Vasquezのウォルツ批判は続きます。
Let us suppose that research shows that 50 percent of the cases support the proposition and 50 percent do not, clearly a random relationship. A reformulation that was degenerative would simply group the cases that supported the proposition, give them a label (e.g., balancing states or prudent states) and say that this is the domain of proposition; whereas the other group could be given another name (e.g., impaired states) and it could be “hypothesized” that in this domain the proposition did not hold. If the labels lack any ability to predict (before examining the evidence) which states will be prudent and which impaired, then the reformulation is just semantic relabelling. Likewise, if the labels are merely hunches that borrow the paradigm’s phraseology, but lack a clear theoretical explication distinguishing what the cases have in common that makes them behave in two different ways, then the reformulation is degenerative. Lastly, if the theoretical explication of why one set of cases behaves one way and the other set the opposite way can only explain the outcome of this experiment, then its explanatory power is ex post facto (and it fails to provide any novel facts). Likewise, if it can explain other patterns or predict novel facts, but these new explanations and predictions do not survive testing, then the reformulation lacks excess corroborated empirical content and is degenerative. (p.281)

ネオリズムが「退行的な」リサーチ・プログラムであることが示されています。
This has not been the case with neotraditional research on Waltz’s balancing proposition. No basic pattern has been established. Nor has this research been very successful in establishing a domain where the proposition holds. Walt (1987) does not find that under certain circumstances states balance power and in others they balance threat. He maintains that they generally balance threat and do not balance power unless threat is present. Schweller (1997: 928-929) clearly specifies a domain for his hypothesis about bandwagoning (namely revisionist major states), but remains neutral as to whether states balance outside this domain. Christensen and Snyder (1990, 1997) come closest in specifying a domain – namely, multipolar systems when the offensive has the advantage. In this domain, they predict that chain-ganging tends to occur; however, this term has a negative connotation, suggesting something aberrant though not quite pathological. In part, this is because they are concerned with instabilities within multipolarity; yet nowhere do they specify where “normal” (what Morgenthau would call prudent) balancing occurs. Christensen (1997: 69) further delimits this domain to frontline states whose perceived power is not superior to its opponent and the offensive has the advantage. Such an emendation is required because there are many states (like Britain in 1914) that do not balance even under conditions where the offensive is perceived to have the advantage in a multipolar system. Yet, it is far from clear why this further specification of domain is not just a reaction to the discovery of additional discrepant cases; i.e., why is it not a degenerative shift? (pp.282-3)
Schweller, Randall L. (1997) “New Realist Research on Alliances: Refining, Not Refuting, Waltz’s Balancing Proposition,” American Political Science Review 91 (December): 927-930
Christensen, Thomas J. (1997) “Perceptions and Alliances in Europe, 1865-1940,” International Organization 51 (Winter): 65-97.

国際構造はアナーキーであるから、生存が第一目標になり、国家は自国を守ることために勢力均衡策をとる、というネオリアリズムの議論は根本的に間違っているという事です。でも、いくら実証的に否定されてもこの議論が国際政治学というディシプリンから消えることはありません。それは、国際政治学の主流であるリアリズムが「退行的」なリサーチ・プログラムのためだというのがVasquezの主張です。
The narrow conclusion one could draw from this appraisal of the neotraditional research program on balancing is that Waltz’s attempt to explain what he regarded as the major behavioral regularity of international politics was premature because states simply do not engage in balancing with anywhere near the regularity he assumed. The defect of his theory may not be so much theoretical as it is empirical. It is the failure of neotraditional researchers and historians to clearly establish the empirical accuracy of Waltz’s balancing proposition that so hurts his theory. If the logical connection between anarchy (as a systemic structure) and balancing is what he claims it to be, then this empirical anomaly must indicate some theoretical deficiency, within either the internal logic or the broader paradigmatic view of the world the theory is taking for granted. (p.284)

多分、本章でのウォルツの勢力均衡論に対する批判は正しいのかと思います。でもいくらウォルツの勢力均衡論が実証的におかしいことが示されても、国際政治学の徒は皆、ウォルツの議論を学ばされるでしょう。それをpathologicalと思うか思わないかはあなた次第です。

12月 29, 2014 · Pukuro · No Comments
Posted in: ☆社会科学

Leave a Reply