My colleague Alexandre Chirat and I have just published a short piece in ProMarket, a publication of the Stigler Institute, about the relevance of Anthony Downs’s economic analysis of democracy to account for the rise of populism in contemporary liberal democracies. This piece is based on a working paper that hopefully will be published in the future. Please read the ProMarket piece to have an idea what the paper is about but the basic point is relatively simple. It might be tempting to assume that the rise of populism is linked to various forms of irrationality. Irrationality from politicians who argue for bad economic policies and disseminate false beliefs. Irrationality from voters who adopt these false beliefs. We contend however that the kind of rational-choice analysis of democracy that Downs has developed is also insightful to understand populist strategies and votes in a democracy.
I don’t want to develop the Downs-inspired analysis of populism further here. The publication of this piece provides a good opportunity however to reflect on the analytical power of rational choice analysis. Non-economists have often been very critical of it and the way it is used to study a wide range of phenomena. There are countless publications on this theme.[1] What is relatively new is that economists themselves have become increasingly skeptical of the rationality assumption and therefore of the value of models based on it. We are constantly reminded that people don’t understand anything about probabilities, that they are messing up their intertemporal choices, that they overreact to losses, that their judgments are biased by the way a choice problem is presented to them, and so on.
A standard defense that is still sometimes used has been to argue that the value of the rationality principle and of the rational choice approach in general lies in its ability to make predictions about the aggregate behavior of agents in a market or elsewhere. Milton Friedman famously made this claim in a 1953 paper, a paper that is otherwise largely forgettable – its crude positivism has definitely not aged well. More sophisticated versions of this argument were provided by Armen Alchian, Gary Becker, and more recently Vernon Smith and other experimental economists. As Becker notes himself, however, this is not so much a defense of rational choice theory as an argument that the fundamental results of economic theory (e.g., that the demand curve is downward slopping) do not depend on the rationality assumption.[2]
Can we imagine a more positive defense? Let’s be clear first on what is meant by “rationality.” In standard economics, an agent is rational if and only if their preferences are well-ordered, i.e., they are reflexive and transitive. That means that an agent cannot strictly prefer a state of the world to itself (reflexivity) and that if they weakly prefer x to y and y to z, then they weakly prefer x to z. There is nothing else in the simplest certainty case. In cases of risk/uncertainty, the ordering condition is extended to lotteries (probabilistic distributions of outcomes) and we need some additional and arguably strong conditions (independence axiom plus a bunch of conditions making sure that agents’ beliefs can be represented by probabilistic functions). The point that is often misunderstood is that the rationality principle has nothing to do with an assumption about what people want or should want, for instance, that they are self-interested and motivated by monetary gains. What it postulates however is that people’s choices can be meaningfully related to their preferences, the latter being a theoretical construct that captures intentional states such as desires and beliefs. On this version of the rationality principle, to assume that an agent is rational is to assume that her behavior is not erratic and can be meaningfully interpreted in terms of what that agent wants and believes.
Well before the axiomatization of decision and choice theory, this understanding of the rationality principle was provided and defended by Max Weber. Weber makes it clear that what makes the rationality ideal-type relevant for the social scientific inquiry is that it is a useful device to interpret individuals’ behavior, to give meaning to it. Weber then got things messy by distinguishing between “means-ends rationality” and “value-rationality” but the point remains: by assuming that people are rational, you’re able to decipher what may have motivated their behavior. Or at least, you can make assumptions. If you give up the rationality principle, you can indulge in all more or less plausible fantasies to explain their behavior.
Rational choice theory makes far more sense if you understand it as a module of Daniel Dennett’s intentional system theory. Dennett argues that humans are able to interpret, explain, and predict others’ behavior by taking the intentional stance. We routinely ascribe to others intentional states like intentions, desires, and beliefs. On Dennett’s externalist account of intentional states, a desire is not so much an internal property of the agent who is supposed to be moved as a hypothesis that accounts for the agent’s behavior. The ascription of intentional states to an agent is always relative to a context and to an underlying assumption regarding the rationality of the agent’s behavior. Even more importantly, the intentional stance is not an option to make sense of individuals’ behavior. Consider indeed hypothetical Martians who are observing Humans and are trying to predict our future on the basis of superhuman abilities making them equivalent to Laplacean super-physicists: “Our imagined Martians might be able to predict the future of the human race by Laplacean methods, but if they did not also see us as intentional systems, they would be missing something perfectly objective: the patterns in human behavior that are describable from the intentional stance, and only from that stance, and that support particular generalizations and predictions.[3] Dennett’s point is that the intentional stance is not merely instrumental; it is the only way to observe real behavioral patterns.
As Dennett himself points out, decision theory and game theory are powerful theoretical devices to ascribe intentional states to agents. They rely on a rationality principle that is very helpful to articulate individuals’ intentional states and behavior with social phenomena that result from them. If Dennett is right about the evolutionary roots of the intentional stance, that means that rational choice theory is not the random product of scientific inquiry. It is in some sense a necessary device to analyze social phenomena.
In the case of populist politics, while it is surely true that part of the populist vote is fueled by cognitive biases, it is also the case that it results from purposeful and intentional behavior. Postulating that people are rational is relevant because it forces us to uncover the reasons that people have to adopt this behavior. It may happen that a fully rational choice account of populism is impossible or limited, in which case we may have to invoke behavioral explanations that rely on specific and highly contextual psychological assumptions. However, the default assumption of social sciences must be the rationality principle. Rational choice theory will therefore never die.
[1] Among the most sophisticated and lucid critiques, see the writings of Jon Elster, for instance Jon Elster, Explaining Social Behavior: More Nuts and Bolts for the Social Sciences (Cambridge University Press, 2007). In political science, for a mostly empirically-based criticism, see Donald Green and Ian Shapiro, Pathologies of Rational Choice Theory: A Critique of Applications in Political Science (New Haven, Conn.: Yale University Press, 1996).
[2] Gary S. Becker, “Irrational Behavior and Economic Theory,” Journal of Political Economy 70, no. 1 (February 1, 1962): 1–13, https://doi.org/10.1086/258584.
[3] Daniel C. Dennett, The Intentional Stance (MIT Press, 1989), p. 25, emphasis in original.
Thank you for your comment. I agree that as soon as we are dealing with choices under risk, transitivity and reflexivity are not enough. But as you say, RDU is ultimately a generalization of EU and therefore of the rationality principle. My view is that the rationality principle is the default hypothesis to study social behavior. In some cases, you may need to use some of its generalizations to provide a fuller account of the phenomenon at stake. And maybe in other (rarer) cases, you may need a completely different theory of "nonrational" behavior.
Rationality is a more problematic concept than you imply here. You can't get far with just transitivity and reflexivity. The first big difficulty comes up with the independence axiom under uncertainty. Lots of people violate it, and this violation can be accommodated by various generalizations of EU, like rank-dependent utility (my baby!). Several responses are possible
(i) Even though RDU lacks appealing properties like dynamic consistency, RDU choices are rational
(ii) RDU choices are mistakes, and careful reflection would lead RDU people to follow EU
(iii) RDU people are irrational and that's too bad
The same set of response arise in relation to any choices that violate some proposed rationality norm. It's always possible to model such choices as the maximization of a (possibly stochastic function), so some analog to RDU is always available.