Introduction
To explain an event, social scientist may turn to different theories of human action. One of the most prominent is the rational choice theory, which defines the procedure of how rational actors behave – that is, finding the best means to given ends. The ‘pure rational choice theory’ has been contested widely recently, with ‘limited’ (or ‘bounded’) rationality gaining importance in the field. Both of these streams, as well as their problems and application, are dealt with and evaluated by Jon Elster (1989) and James G. March (1994).
Elster – Summary
Elster founds his reasoning in methodological individualism – i.e., the belief that only individual human action can constitute a meaningful fundamental unit in explaining an event. Individuals then, when deciding on what course of action (or non-action) to take, first consider the given opportunity set defined by environmental as well as biological constrains. Then, different plausible actions are compared for the probability of succeeding and the utility they bring, the best choice scoring high on both of the criteria.
In a sense, thus, Elster considers opportunities as a primary determining factor of human action, and preferences as a secondary one, possibly because they are more easily identifiable (‘objective’) to the observer.[1] Nevertheless, people can also deny themselves some opportunities as a response to their weakness of will. The assumptions of a pure rational choice here, however, are perfect foresight and processing capabilities as well as preference consistency[2], which is apparently not the case in the real world. Nevertheless, this simple theory is quite helpful in explaining straightforward actions (e.g., profit maximization) as well as aggregate outcomes.
Bounded rationality, in contrast, retreats on the first and second constrains and allows the actor to optimize on the acquisition of information, given his mental capabilities as such. Here, (almost) any action can be post hoc explained as rational.[3]
March - Summary
March fits rational choice under the ‘logic of consequence’ theories of human action based on individual preferences. From the beginning, the author recognizes the impossibility of a ‘pure’ rational choice because of the necessary prediction of the future state of the world when expectations about consequences of each alternative action are to be evaluated. The resulting uncertainty problem fills the rest of the chapter – with information and risk management strategies of editing, decomposition, heuristics, framing, and others.
When portraying two of these tactics as distinguishable – that is, satisficing and maximizing (though not “likely to be observed in pure form”, p. 18), March does not seem to realize, however, that both of these are in fact maximizing strategies. In addition, in a real world, attention is a scarce resource, and as such has to be traded off against other useful activities. Rational actors thus optimize on the search for information as well as on its procession, or respond to a ‘slack’ (that is, a difference between their actual achievements and aspirations). Risk management depends largely on the personal risk propensity, its correct estimation as well as a social validity of this estimation. Here, March seems to depart (relatively to Elster) from methodological individualism, because he does not explain the functionality or nature of the ‘social validity’ concept.
Similarities
Both authors are regarding actor preferences more or less as given. This way, the procedure of rational choice described in both can support the saying that ‘reason is the slave of the passions’ (David Hume). In addition, however, both texts endorse limited rationality as better at explaining everyday human decision-making than relying on unrealistic assumptions of the pure rational choice (e.g., March, p. 4: “pure versions of rational choice are hard to accept as credible portraits of actual individual or organizational actors”).
Second, neither Elster nor March is really clear on whether rational choice is a positive or a normative theory. This would not constitute a serious problem unless we try to decide on whether an action followed rational decision-making or not (as below).
Neither of the authors, in addition, touches value-freely upon the actor’s time preference. Elster goes as far in this direction as to claim that utility should be maximized over “many” periods (p. 28), and ‘impulsive’ actions (i.e., those that prefer utility now over utility in the future) thus have to be ‘irrational’ (p. 40).[4] Neither one of the authors can deal with preference inconsistency: March only notes that people cannot really compare all alternatives simultaneously, but rather do it sequentially, which can lead to contradiction (pp. 8-9).
Both texts have thus problem with defining what actually an ‘irrational’ action is. If judging an action post hoc from the ‘pure rationality’ point of view, neither well-equipped scientist can judge whether actor made a rational decision, because neither he possesses perfect foresight and computational capabilities. Limited rationality, on the other hand, provides for numerous maximizing strategies in place at the same time (e.g., March), making judgment over (ir-)rationality of an action a matter of reading the actor’s mind.[5]
Differences
Whereas Elster grants the opportunities (constraints of the environment) a significant role in shaping an actor’s decision-making process, March touches this issue only lightly. Moreover, for Elster, when maximizing utility, even the beliefs (springing from desires) have to be optimal. This goes a bit against his other conclusion that “passions are not subject to rational assessment” (p. 31). March does not make similarly controversial claims about the rationality of desires.
In the second text, on the contrary, March goes much deeper into the bounded rationality issue (that is the title of his whole first chapter of the book), leaving for pure rational choice only the first very few pages. Rather immediately, the author turns to explain the numerous strategies to optimize at decision-making under the ‘harsh’ conditions of reality – he thoroughly describes attention allocation, information search, as well as risk management.
Conclusion
The two views on rational choice of Elster and March are compatible, though not identical. Where Elster tends to prefer more some kind of ‘modified’ pure rational choice, March explicitly focuses on bounded rationality theories instead. Both authors however defend a counter-intuitive ‘rational ignorance’ as a rational practice.
It seems to me though, that the more positive (explanatory) the rational choice theory tries to be, the more actions of the real world it can justify, the more parameters come there into play, and the less applicable it is for a social scientist to explain an event post hoc or predict ex ante. The more normative it is (i.e., distinguishing between rational and irrational actions), the less rules and variables it is based on, on the other hand, the less it reflects of what the people’s decision-making is about in the real world, however.
Annotation:
1. Elster, Jon (1989). Nuts and Bolts for the Social Sciences. Cambridge: Cambridge University Press, pp. 13-41.
2. March, James G. (1994). A Primer on Decision Making. How Decisions Happen. New York: The Free Press, pp. 57-102.
[1] Although the procedure can start with preferences and come to the same outcome – rejecting preferred outcomes that cannot be reached due to the constrains
[2] Consistency here means a) stability with time as well as b) transitivity (i.e., A>B and B>C => A>C).
[3] Similar to: Mises, Ludwig von (1947/1996). Human Action. Auburn: Ludwig von Mises Institute, pp. 19-22.
[4] For more on time preference in economics see: Mises, Ludwig von (1947/1996). Human Action. Auburn: Ludwig von Mises Institute, pp. 479-537.
[5] To know which maximizing strategy and limitations actually were in place.
