What We Assume When We Assume Rationality

This term, I’m sitting in on a Master’s level microeconomics course. I’ve read a lot of micro – a couple textbooks, tons of blog posts, and some journal articles here and there, plus some historical pieces through a history of economic thought course – but I’ve never taken a legit microeconomic theory course before. I think the class is going to be equal parts frustrating and fascinating, but will likely inspire more than a couple blog posts.

Today, in addition to course preliminaries, we covered two things. The second was a quick review of basic calculus. This followed from the first substantive discussion: the “what is economics?” debate. The professor presented a common definition of the field, the Robbins/scarcity definition: economics is “the science which studies human behavior as a relationship between ends and scarce means which have alternative uses.” (See Backhouse and Medema’s work on the history of this definition.) The prof then went on to disagree with this definition, arguing that economics should not be defined by its object but rather by its method. She then went on to give a quick greatest hits of Gary Becker’s career, and assigned his Nobel Prize lecture for the next course. This method consists of two principles: an equilibrium principle (supply and demand meet up) and an optimization principle, that is, the idea that people and firm maximize over some utility function.

This framework, twice described as “beautiful”, leads naturally to a discussion of differential calculus and Lagrangians. It also justifies Becker’s expansion of economics into crime, the family, and so on. But what do economists really assume when we assume optimization of preferences as the essence of the economic method?

I think the core of the model is not actually self-interest*, or even consistency of preferences, but stability and unity of the individual. That is, for rationality to make sense as a model, individuals (or firms) need to be unified actors whose preferences are relatively invariant across contexts and at least short periods of time.

I think we tend to take this basic model of individuals for granted in our daily lives, which is part of the appeal of individual rationality as a methodological principle. I also think it’s false, which is both a precondition for a lot of successful interactions, and a source of problems when we make the assumption that individuals are continuous processes. Think, for example, of Goffman’s** work on the presentation of self, frames, and all that. The point is not just that our presentation is strategic, but in fact that the self is interactional and social – who we are is a function of the environment we are in. Or think of Foucault’s work on the history of the subject, Althusser on interpellation, Butler on performance, etc. Of, for an economist’s take on it, think of Schelling’s fantastic essay on Self command in practice, in theory and in a theory of rational choice, which discusses impossible behavior where you act against your future self (for example, having a friend hide your cigarettes, or asking a loved one to stop you from killing yourself). The point is, the “individual” is a historical construct, enforced through all sorts of practices, and to some extent an aim we strive for (“to thine own self be true”) but not a necessary and always present feature of the world.

And that’s a big part of why I’m not an economist. And I think it’s a really helpful scope condition – in cases where it makes the most sense to think of an individual (or, importantly, a firm) as a unified actor with stable preferences, economics has a lot of insight. But I think it can also lead to either frustration (why are these people so irrational?!) or prescriptive rather than descriptive findings (here’s how they should be rational!). In other words, economists, and economic thought, try to make the world more like the one they assume it to be by helping individuals be true to themselves, and by ignoring how much individuals usually aren’t.

*As I love to quote, economist John Quiggin notes that the definition of rationality does not require self-interest at all, but that economists tend to jump straight to self-interest after defining rationality. Thus, they tend to present one definition which is “vacuous” and one which is “false”, and when challenged tend to oscillate between these definitions as “sin (1/x) as x approaches 0.”

Advertisements

11 Comments

  1. afinetheorem

     /  September 7, 2011

    Dan,
    Glad to see you’re taking some theory! I would be really careful when talking about what economists assume about the real world. You could argue (and I certainly would) that methodologically economists assume *nothing* about the real world. Rather, we assume things in the context of a model, and hope the model can give us insight into particular situations where those features are most relevant. For example, beyond the stability paradigm you mentioned that is common in many models, there is also a huge literature in micro theory on temptation from a decision-theoretic perspective that assumes no such thing (i.e., Pesendorfer-Gul, or Dekel-Lipman-Rustichini).

    If economics is unique among social sciences in terms of methodology, it is not because of RCT or unity of self or anything like that, but rather the use of parsimonious mathematical modeling of social situations. “Toy model” is a compliment and a benefit, not a negative.

    • Every field, I think, has within it folks who do things that challenge key premises of the underlying model. But I do think it’s a unifying feature of rational choice approaches that is little discussed, and is very distinct from a lot of research in sociology. We tend to critique the structure of preferences (other-regarding, etc.) or knowledge (assumptions of perfect, or boundedly imperfect knowledge as opposed to Knightian uncertainty, etc. see also Keynes 1937 and the post-Keynesians), but those critiques are in some ways easy to address without challenging some of these stability issues (e.g. boundedly-rational behavior is just optimizing in a world where processing and information gathering are costly). And maybe there are ways to patch in unstable preferences without disturbing much, but I’m not sure. It gets close to an attack on methodological individualism and some really basic ontological issues.

      But in the long run, you are probably right that the really crucial feature is the faith that toy models produce useful models for advancing the science, and, I would argue, for insight about real world problems (even if there is always a heroic leap made to get to specific applications). At the moment though, those toy models mostly rely on relatively stable and unified individuals.

  2. John

     /  September 7, 2011

    Terrific post! Thanks for sharing. Many of your insights here started becoming clear to me when one of my professors (I’m an econ major) attempted to offer some unsolicited career/grad school advice. She basically plugged me into a constrained utility function and delivered the results, without bothering to consult me about my preferences or my perception of the constraints involved.

    I balked, among other reasons because it seemed like a foolish way of making personal decisions. I remember telling her something like “I don’t know ‘myself’ and am not even sure such a thing is possible.” Thinking about it since then, I realized that what I was getting at was that in reality preferences are unstable and constraints aren’t fully knowable. I should revisit my literary theory textbook (Are Foucault/Butler/Althusser read in sociology as well?)

    Also, I think defining any discipline in terms of a methodology is wrongheaded and, in the case of economics, exactly what’s produced the current sorry state of the field. My own intuition leans toward methodological eclecticism — a discipline is like an algebra: a set of questions coupled with some methods for approaching those questions. Both can change over time, but it’s the questions that are primary, not the methods.

  3. John

     /  September 7, 2011

    Also:
    “This framework, twice described as ‘beautiful’…”

    Is that a note of skepticism I detect?

    I’ve often wondered about this “sexy proof” trope—it seems like a counterproductive shibboleth to me.

    I really wish this page had a ‘Criticism’ section: http://en.wikipedia.org/wiki/Mathematical_beauty

    • It’s not skepticism exactly, it’s more fascination. Beauty or elegance are really important to how higher-level math judges its progress – elegant solutions are praised over brute force hacks, etc.* It’s interesting how the logic translates to economics. The “beauty” here is how a relatively simple formalism can be applied to different domains, much as people got excited about inverse square laws when they realized they covered so many physical situations, or the like. What’s missing is actually Friedman’s stuff – the notion that economic science is justified by its predictive validity, not the elegance of its formalism (or its accuracy in modeling human behavior). According to a historian of the Chicago school, Friedman actively disliked Becker’s approach – Friedman did not like economics should become rational choice theory, and thought the rational self-interest assumptions were crucial to understanding market behavior, but did not want to extend them to everything.

      * I was a math major in undergrad and got into soc in part because of my interest in how numbers were mobilized in the rhetoric of the social sciences.

      • John

         /  September 7, 2011

        Interesting. My ignorance here is vast, but I’d stop short of equating ‘beauty’ and ‘elegance’. If we think of the latter as nothing more than the privileging of minimalism, then its utility—both substantive (Occam) and ‘rhetorical’ (given its audience, it’s easier to parse)—is clear to me.

        But ‘beauty’ seems to be used to denote something else—an aesthetic enthusiasm for something ineffable (Erdős: “If you don’t see why, someone can’t tell you). And this, at least to my jaundiced eye, doesn’t properly have a place in science, if for no other reason than because it’s often counterproductive.*

        *For this reason:
        “In some cases, natural philosophers and other scientists who have made extensive use of mathematics have made leaps of inference between beauty and physical truth in ways that turned out to be erroneous. ”
        http://en.wikipedia.org/wiki/Mathematical_beauty

  4. Simon Halliday

     /  September 7, 2011

    For an interesting comment on and more nuanced approach to rationality, I’d recommend Herb Gintis’s book ‘The Bounds of Reason’. He uses rationality in a way that appears much more reasonable to me – he tries to introduce perspectives from anthropology, psychology and sociology to come up with a better understanding of rationality and its applicability (his aim being a ‘unified behavioural sciences’ which I think is unlikely, but at least he’s making a go of it).

  5. John

     /  September 10, 2011

    Paul Krugman sez:
    “What we really need is a change in the destructive social dynamics that brought us to this point. And I wish I knew how to do that. But my problem is obvious: I’m an economist, and it seems that we need some kind of sociologist to solve our profession’s problems.”

    Fix it, Dan! 🙂

  6. Spot on the mark. Note also that this is not the only fanciful assumption in economic models. Stuff such as perfect information and free entry and exit to markets are just as fictitious, and a mere device to justify the use of mathematical models. In effect, this may be Economic fighting to make sure its status as a science is not threatened; the problem is as much epistemological as ontological.

  1. Roeper is… an Endless Meeting « A (Budding) Sociologist’s Commonplace Book
  2. Weekly List Bookmarks (weekly) | Eccentric Eclectica @ ToddSuomela.com