Critique of Rational-Choice - Stanford University



Todd Sechser

Political Science 311

Response Paper #4

6 December 2001

Rational Choice Pathologies in Fearon’s Explanation for Ethnic Conflict

As international relations scholars have paid increasing attention to ethnic conflict, they have imported well-developed sets of tools from other branches of political science in order to explain this phenomenon. One such tool set is rational choice “theory,” which argues that individuals’ attempts to maximize their personal expected utility can explain broad patterns of political behavior. This essay concentrates on one particular rational choice explanation for ethnic conflict – James Fearon’s “commitment problem” – and argues that his theory suffers from some of the more serious methodological pathologies of the rational choice camp. The essay proceeds in two parts: first, I review Green and Shapiro’s Pathologies of Rational Choice Theory in order to summarize many of the criticisms brought against rational choice theory; second, I examine the applicability of these arguments to Fearon’s model of credible commitments, and argue that his theory requires a great deal of reformulation and empirical verification in order to be considered valid.

Green and Shapiro’s Critique

Green and Shapiro advance three fundamental points in their critique of rational choice theory. First, rational choice theory suffers from post-hoc theoretical development, in which scholars toy with the assumptions of stylized models in order to create the conditions that explain why a particular event occurred in the past. Second, rational choice theory fails to formulate effective empirical tests for its “slippery” predictions. Third, scholars of the rational choice tradition are selective and biased in their use of evidence.

Post-hoc theory development

The first line of argument in Pathologies of Rational Choice Theory is that rational choice scholars seek to explain historical events rather than predict future outcomes. This is another way of saying that rational choice theorists select on the dependent variable – all of their cases contain outcomes predicted by their models. Instead of testing outcomes against predictions, Green and Shapiro argue, rational choice scholars first observe an empirical regularity and then design the assumptions of their model so that the model “predicts” the outcome that is observed. These scholars make no predictions about the future against which their model can be tested.[1]

A consequence of constructing models to fit empirical outcomes is that rational choice hypotheses become unfalsifiable.[2] In many cases, models are tested against absurd null hypotheses that have no chance of being true; some studies use no null hypothesis at all. Without a valid null hypothesis against which to test the model’s predictions, the model’s predictions become tautological – it is unclear what, if any, hypothetical piece of evidence could possibly disconfirm the theory. As a result, such a theory is not useful.

Furthermore, Green and Shapiro argue, rational choice scholars fail to consider alternative explanations for the empirical regularity they seek to explain.[3] Without addressing other processes that might have produced the observed result, the theory’s strength is limited.

Test formulation

A second contention of the book is that rational choice theory fails to construct adequate tests for its claims. This problem manifests itself in two ways.

First, the predictions of rational choice theory are too “slippery.” When empirical outcomes fail to match the predictions of a particular rational choice explanation, the theory’s proponents generally resort to the argument that some aberration prevented the model’s assumptions from being met.[4] Used too many times, this argument undermines the applicability of the theory – if its conditions never hold, then the usefulness of the theory is questionable.

Poorly specified predictions also produce testability problems. Without distinct quantitative predictions, any empirical outcome can often be construed as support for the theory. Collective-action theorists, for example, claim that the absence of selective incentives will preclude cooperation to solve social dilemmas. Yet in many such dilemmas, some level of collective action exists – individuals contribute to environmental organizations, for example, even though rational choice theory predicts that they should free-ride. Rational choice scholars do not specify beforehand whether this should be considered evidence in favor of the theory (due to the small ratio of contributors to potential beneficiaries of environmental protection) or against it (due to the fact that collective action occurred, if only at a low level). This vagueness renders falsifiability nearly an impossible task.

Second, rational choice theory suffers from predictions that are “vaguely operationalized” and difficult to verify.[5] Green and Shapiro argue that this vagueness results from elusive switching between predictions about specific equilibrium outcomes (point predictions) and predictions about an equilibrium’s direction of movement (marginal predictions) after a change in an exogenous variable. When a theory’s point predictions fail, its proponents cite successful marginal predictions, and vice versa. Confirmation of the theory becomes too easy; falsifiability becomes impossible.

Selecting and interpreting evidence

The third broad pathology identified by Green and Shapiro relates to the search for and use of evidence to test rational choice theory. The authors identify three objectionable practices within this class of problems.

First, rational choice scholars search actively for evidence confirming their theory by collecting illustrations that support their conclusions while neglecting disconfirming evidence. Green and Shapiro liken this procedure to collecting “advertisements that show one’s brand achievements while mentioning neither its failings nor the achievements of its competitors.”[6]

Similarly, scholars sometimes conduct lab experiments with no control group, seeking only an example to help confirm their theory.[7] But without a control group to assess the consequences of an absence of one’s independent variable, it is impossible to evaluating with any certainty the consequences of its presence.

Second, rational choice theorists “project” evidence from their theories: either they imagine data that would be consistent with their theory and assume it to be empirically true, or they assert that a particular piece of evidence confirms the theory without fully explaining the evidence.[8] This does not constitute adequate theory-testing.

Third, rational choice theories often suffer from arbitrarily-restricted domains of application.[9] Specifically, a theory’s proponents sometimes deal with disconfirming evidence by excluding the situations found in that evidence from the set of situations that the theory claims to explain. The domain of applications is thus limited to instances where the theory appears to hold, although no explanation is given for this restriction. Green and Shapiro argue that this practice leads rational choice scholars to neglect disconfirming evidence and disproportionately emphasize supporting illustrations.

Application: Fearon’s Explanation for Ethnic Conflict

Below I assess the validity of Green and Shapiro’s criticisms with regard to a well-known rational choice explanation for ethnic conflict: James Fearon’s piece on commitment problems and ethnic warfare.[10] A brief glance at this work demonstrates that it suffers from many of the deficiencies outlined in Pathologies of Rational Choice Theory.

Fearon’s “Commitment Problems and the Spread of Ethnic Conflict” argues that ethnic conflicts arise when four conditions hold: first, the disintegration of a state presents a minority ethnic group with a choice between joining a new state ruled by an ethnic majority (with which the minority once peacefully co-existed) or fighting for independence. Second, the minority must calculate its expected costs from fighting and winning to be smaller than the expected costs of suffering indefinite exploitation at the hands of the majority in the new state. Third, the minority must feel that its chances of winning a war are better today than they would be in the future. Finally, the ethnic majority must be unable to commit to not exploiting the minority in a new, majority-ruled state. When these four conditions hold, Fearon argues, minority ethnic groups will reason that it is in their best interest to fight for independence, because the majority cannot commit to not exploiting them in the future. Moreover, it is best to fight immediately, because the minority expects the odds of success to decline over time. Fearon employs this model to explain the revolt by the Serbian minority in the newly independent state of Croatia in 1991.

Fearon’s analysis suffers from a number of problems identified in Green and Shapiro’s book. First, it is not clear what Fearon’s null hypothesis is. Against what alternative possibility is he testing his explanation? Indeed, as Green and Shapiro would predict, some potential null hypotheses of the theory seem absurd: for example, the proposition that the Serbs thought war in the future would be better than war today but still decided to fight today. The most reasonable null hypothesis is that the Croats were able to credibly commit to not exploiting minorities, but hatred or some other mechanism drove the Serbs to war nevertheless. But Fearon offers no description of what a credible commitment would look like, other than one enforced by an external third party. As a result, his hypothesis is not falsifiable – there is no hypothetical evidence that could disconfirm his theory. As argued below, he explains away cases of non-conflict by asserting (with no evidence) that one of the conditions of his model must not have held.

A related observation is that Fearon fails to consider alternative explanations for ethnic conflict. The only alternative perspective offered in his article is a straw-man reference to the “standard view” among journalists that ancient animosities explain ethnic warfare, although no citation is provided to verify that anybody actually subscribes to this view.[11] Without explaining the inadequacy of alternative theoretical perspectives, it is difficult to evaluate the relative merits of his theory.

Second, Fearon “projects” his evidence rather than conducting a systematic empirical study. Perhaps for Fearon the Croatian commitment to protect Serb rights was not credible, but there is no evidence provided to suggest that the Serbs themselves shared this perspective. Were the Serbs really worried about exploitation, and if so, why? The strength of the credible-commitment explanation relies on this assumption, but there is no empirical support given for it. This raises a larger question with implications for the predictive power of the model: under what conditions will a minority be worried about exploitation by a majority during the formation of a new state? Simply asserting that minorities always worry about exploitation, as Fearon seems to do, is not a compelling defense of the theory.

Fearon’s explanations of confirming cases of ethnic conflict also suffer from the projection problem. In citing the war over Nagorno Karbakh as evidence for his credible-commitment explanation, he merely states that the conflict resulted from the inability of the Azerbaijani government to commit to protecting the rights of its Armenian minority.[12] No evidence whatsoever is provided for this assertion – the reader is expected to take this description at face value. This is precisely what Green and Shapiro would identify as “projecting evidence from theory” – claiming by assertion that reality mirrors the model at issue.

Finally, Fearon’s predictions are what Green and Shapiro would term “slippery.” There is no hard standard against which to measure the success or failure of his explanation, and there are no guidelines with which to evaluate future evidence about ethnic conflicts. How can we tell if a credible commitment exists? How do we know whether a minority viewed war today as better than war tomorrow? Fearon’s only mechanism for doing this is to evaluate the outcome and infer from revealed preferences: if war occurred, then a credible commitment was lacking, and war today must have been seen as preferable to war tomorrow. The result is a tautological claim – using this standard of evidence, a war that occurred despite the availability of a credible commitment is not even hypothetically possible. Furthermore, it is impossible to predict future conflicts with Fearon’s model, because he fails to explain how we can identify the conditions for his model beforehand. This is the curse of post-hoc theory development.

An example of such a tautological claim is Fearon’s explanation for non-conflict in Kosovo. Fearon argues that the minority Kosovars “did not fight en masse to secede because they lacked any serious prospect of success at a reasonable cost – it was better to accept greater oppression (and probably a lower ability to fight for secession in the future) than to fight for secession in the present.”[13] Yet Fearon does not offer a single piece of evidence for this claim. His only “proof” that the Kosovars assigned a higher expected utility to acquiescence is that they did not go to war. This is tautological rational choice theory at its worst – the conditions for the theory are defined by the outcome. There is no evidence that could disconfirm Fearon’s explanation, because the independent and dependent variable are defined by the same empirical result.

The appeal of Fearon’s argument lies in the fact that his assertions usually match empirical outcomes. But in the case of Kosovo, they did not. The rebellion and war that erupted after Fearon wrote his article would seem to confound all of Fearon’s explanations: both that Kosovars viewed oppression as preferable to war, and that the appeal of war for Kosovo declined over time. This empirical fact should cast serious doubt on Fearon’s method, if not his model altogether.

The accusations of Green and Shapiro thus appear to apply well to Fearon’s explanation for ethnic conflict. Fearon advances unfalsifiable hypotheses, asserts the applicability of his evidence, neglects alternative explanations, and defines his independent variable in terms of the dependent variable. This is not to say that his theory is false, or that a rational choice approach to ethnic conflict is hopelessly flawed. Rather, it appears that greater attention must be paid within the rational choice camp to the methodological problems that are endemic to such explanations. Reformulation of Fearon’s theory is thus both necessary and possible.

-----------------------

[1] Donald Green and Ian Shapiro, Pathologies of Rational Choice Theory : A Critique of Applications in Political Science (New Haven: Yale University Press, 1994), p. 35.

[2] Ibid., pp. 36-7.

[3] Ibid., p. 37.

[4] Ibid., p. 39.

[5] Ibid., p. 41.

[6] Ibid., p. 43.

[7] Ibid., p. 44.

[8] Ibid., pp. 43-44.

[9] Ibid., pp. 44-46.

[10] James Fearon, “Commitment Problems and the Spread of Ethnic Conflict,” in David Lake and David Rothchild, eds., The International Spread of Ethnic Conflict: Fear, Diffusion, and Escalation (Princeton: Princeton University Press, 1998), pp. 107-126.

[11] Ibid., pp. 108 and 115.

[12] Ibid., p 120.

[13] Ibid., p. 122.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download