Many disagreements implicating economic analysis arise from terminological misunderstandings. For example, a challenge when teaching Law and Economics is persuading students not to fight theorems or games. Unfamiliar students sometimes treat theorems or games as if they were hypotheses or theories. A similar phenomenon explains sometimes profound disagreements between scholars relying upon neoclassical economic theory, including first generation law and economics scholars, on one side, and those influenced by behavioral economics, on the other.
Game theory rests on specified games. These include, among others, the prisoner's dilemma, the driving game, and the battle of the sexes. Social choice introduces various social welfare theorems, including, most famously, Arrow’s Impossibility Theorem (or simply Arrow’s Theorem). Games and theorems can be expressed as proofs, demonstrating their internal consistency. From stated premises, the conclusions follow. If we assume the conditions of a prisoner's dilemma, the dominant outcome (or pure Nash equilibrium) is mutual defection. If we assume the conditions for a cycle, then for any proffered outcome, an alternative has majority support. Failing to acknowledge these results produces analytical confusion.
The problem is partly terminological and partly analytical, and the distinction is important. “Theory” and “theorem” sound similar but have different meanings. A theorem is an internally consistent proof; the conclusions inexorably follow from the premises. By contrast, a theory is an assertion intended to explain an observed phenomenon. Some theories are robust; others are bogus. Either way, theories posit an explanation about something in the world. (One source of confusion: “Game Theory” and “Social Choice Theory” would each be more descriptive if they were identified, instead, as “Games” and “Social Choice.”) In the sciences, including economics, a social science, meaningful theories are subject to falsification. Otherwise, theories rest on faith, which is generally appropriate for matters of purely personal belief, such as religion, but less so for theories advanced as scientific.
A central claim of neoclassical economics is that models built upon the simplifying premise of individual rationality provide the basis for more robust accounts of observed phenomena than models resting on alternative premises. First-generation law and economics scholars advance the related claim that rationality-based models provide more robust accounts of common law doctrine than alternative explanations. (I have previously written on how economists define "rationality."). The latter claim has generated substantial disagreements among legal scholars.
To be fair, economists sometimes use a shorthand and simply state that individuals are rational, a premise that behavioralists challenge. My sense is that fewer behavioralists find problematic the modest claim that rationality provides the basis for more robust models of human behavior than models built on other premises.
Here are a few prominent examples from the behavioral literature challenging rationality. (For interested readers, here are some prominent sources).
(1) the endowment effect: some people appear to value personal possessions above fair market value, potentially inhibiting social welfare enhancing exchange;
(2) zero pricing effect: people sometimes choose options that are nominally free over alternatives even when the marginal benefit of an alternative exceeds the marginal cost (above zero), thereby diminishing personal utility;
(3) default rule effect: people tend to favor default positions, for example contributions or portfolio allocations for retirement savings, thereby foregoing potentially superior strategies; and
(4) laboratory gaming effects: participants in structured games (typically with low stakes) sometimes thwart predicted strategies based strictly on formal payoffs.
Sometimes these effects implicate a kind of heuristic bias, meaning a method of processing information that renders us human beings, not computers. Other times, they reflect a difference between nominal and actual costs or payoffs. What primarily interests me is not the merit of each separate claim as the basis for refuting rationality. There is a large literature on these points, with behavioral economists emphasizing the importance of each observed phenomenon (see earlier links), and with traditional economists claiming to reconcile such observations with neoclassical theory. (Some examples here).
Over the years, I have followed the literature, and I have personally observed the resulting arguments. Neoclassical economists typically posit alternative, rationality, accounts of the observed phenomena, and behavioralists typically respond that such accounts are circular, too neat, or both. The conversations are interesting, and, yes, I have my views. My immediate point, however, is that the resulting disagreements often arise as much from conflating or confusing terminology as from differing substantive views of the world. Among the conflated terms are premises, theories, and models.
Although economists sometimes express insights with imprecise phrasing, economic theory does not depend on the universality of individual rationality. Economic theory is able to sustain itself even if we accept that some individuals are generally irrational and that most individuals are sometimes irrational. Economic theory rests on the more modest claim that the rationality premise generates more robust models than models resting on alternative premises. The individual rationality premise might be true or false, but either way, specific observations seemingly in tension with individual rationality do not undermine the economist’s central posit. To refute that claim, behavioralists would need to offer up alternative and more robust models, resting on alternative premises.
But this is almost never the manner in which the conversations between the two camps take place. The fault is mutual. Neoclassical scholars try to refute each observed instance of a heuristic bias with an alternative rationality account. Behavioralists try to refute the claimed robustness of economic theory by identifying specific instances of observed behaviors that reflect heuristic biases. In fact, both perspectives can coexist without contradiction; economic theory can remain robust even if individual rationality is not universal.
I previously distinguished a theory from a theorem. It is also important to distinguish premises and models. A model is a simplified construct, as with a map. Models omit unnecessary detail, for example, the tendency of persons to process information based on heuristic biases or shorthands. Models are not designed for descriptive accuracy; they are constructed from simplifying premises (something posited, but not proved by the model itself) to allow the testing of hypotheses. Models allow hypothesis testing by isolating variables. The process is dynamic, with newer models incorporating features left out of earlier versions that generated failed hypotheses. Theories are right or wrong; models are not. But models can be more or less helpful in developing a theory.
Economic hypotheses rely on models or theorems, but they are not themselves models or theorems. The prisoners’ dilemma is not a theory. We might observe that Congress tends toward excess spending, ballooning the national debt; society tends toward excess carbon emissions, exacerbating global climate change; and fishermen tend toward excess fishing, risking stock depletion. In each instance, we might posit that the affected actors (congresspersons; individuals, firms, and nation states; and fishermen) are situated in a prisoners’ dilemma with respect to each other given the observed activity (spending, carbon emissions, or fishing). The model in each instance applies the prisoners’ dilemma to the relevant context. The hypothesis is that the model provides the basis for a more robust account of the observed phenomenon than other theories. An effective refutation cannot seek to discredit the prisoners’ dilemma. Instead, it requires an alternative model, perhaps resting on a different game, or a more nuanced version of the same game, that provides a more robust account.
If behavioral insights in tension with at least some understandings of rationality are not viewed as refuting economic theory, both sides gain. Such insights potentially serve as a bridge to other fields, such as psychology (especially evolutionary psychology), sociology, and anthropology. Critics sometimes maintain that economic theory is hegemonic, extending simplified models into myriad complex fields. Economists tend to find such claims overblown, observing that the models assume human motivations or preferences, without seeking to explain them. For such explanations, we must look to other disciplines. Behavioral insights offer a possible bridge from economics to these other disciplines. When traversing that bridge, however, we might all learn more if we respect each other’s lanes.
I welcome your comments.