economics without equilibrium

Thought experiment: What would economic theory look like if you weren’t allowed to use the concept of equilibrium? A few thoughts:

  • Programs and routines: Following March and Simon ’58, you could imagine people as rule followers. This version of economics would then be about habits instead of choices. Satsficing instead of optimization.
  • Genetically determined routines:  People are randomly assigned behavioral rules, agents are born with routines. Unsuccessful agents are weeded out, or modify behaviors according to some rule. You might then have equilibria describing populations of actors, thought not their individual behavior. You actually find this in population biology and some kinds of game theory.
  • Autocatalytic behavior: John Padgett has pushing this recently. Assume that agent behavior is triggered by other behavior, just like chemicals (  2H+ O -> H2O). No rationality, just responses to other agents. Then you can actually get self-sustaining reactions that mimick trade patterns.

What other non-equilibrium models might be used to describe economic behavior?

Written by fabiorojas

March 24, 2010 at 6:56 pm

Posted in economics, fabio

10 Responses

Subscribe to comments with RSS.

  1. How can you talk about financial markets without equilibrium? You only have so many securities, and markets have to clear.



    March 24, 2010 at 7:04 pm

  2. What would economic theory look like if you weren’t allowed to use the concept of equilibrium?

    Austrian economics.



    March 24, 2010 at 7:04 pm

  3. You could still have the maximizing behavior without equilibrium. In fact, in macro in the late 1960s and 1970s there was a whole literature on “disequilibrium” behavior–i.e. Keynesian models said that the economy was out of equilibrium, but how did things work when that was true, what dynamics got you to equilibrium?

    I guess this is sort of against the spirit of the question.

    I don’t get why people oppose satisficing behavior and maximizing behavior since if you always see objects in the same order satisficing behavior generates “rational preferences” (that is, it satisfies all the properties that any `legitimate’ preference scheme satisfies).



    March 24, 2010 at 7:05 pm

  4. The comments so far are interesting. Here’s the motivation for the question. In math (I got my undergrad degree in math), we learned that math is based on axioms and that there is much to be learned from adding or dropping assumptions.For example, Euclidean geometry looks much different if you drop a crucial axiom about parallel lines. In modern times, it was found that many crucial results depended on some assumptions that could be dropped.

    So the spirit of the question is: how many economics results require the equilibrium concept? What can be proved without it? How would economic theory look like if you lived without out? Is Austrianism the only equilibrium free theory (obviously not)? Etc.



    March 24, 2010 at 7:12 pm

  5. I like the rationale behind the question. However, if I remember correctly, equilibrium is an outcome not an axiom in itself, so you can’t just drop it. I wonder if there is an axiomatic (micro)economic theory, and if there is, to see how it might change by changing/dropping one assumption at a time.



    March 24, 2010 at 8:29 pm

  6. Well, we have to be careful, equilibrium can be both an outcome or an assumption of economic theory. In game theory, equilibria are definitely outcomes. It follow from utility maximization. But as some of the comments show about it’s an assumption as well: people are trying to compute market clearing prices, or show that the macro-economy is trying to get into equilibrium. Does economics choose axioms that are designed to study equilibrium? What if we choose axioms designed to study something else?



    March 24, 2010 at 8:33 pm

  7. Neoclassical economics models agents as maximizers of a utility function where the argument of the utility function is a vector of commodities. Under some circumstances, aggregate utility is maximized over some “pareto” allocation of these commodities. Neoclassical models have never been too specific as to the particular mechanism for achieving such pareto allocations, however, and they are in fact many.

    One set of non-equilibrium models that would be interesting would be the set that provides an explanation for the mechanism whereby an economy approaches equilibrium. Existing models assume linearity. For inspiration, I think models of nonlinearly coupled oscillators might be useful. One could start with something simple like Ising Models and work your way up. I haven’t seen the Ising Models applied much in social science yet.

    More generally, I haven’t seen much research quantifying the frequency-domain of human (including consumer) behavior. There are characteristic modes to our patterns of behavior that can be mined now thanks to digital technology:

    Click to access temporal-task-footprinting.pdf

    All of your suggestions look good to me too. These are uncharted waters.


    Michael F. Martin

    March 24, 2010 at 9:51 pm

  8. Another slice at answering your question:

    At least part of the reason why the concept of equilibrium made it into economics is because of the influence that thermodynamics had on the early theorists, including Samuelson. Thermodynamic equilibrium, like neoclassical equilibrium, is a good approximation in most situations where key variables aren’t changing fast, or where they are but the equilibrating processes are even faster.

    But thermodynamic equilibrium, like neoclassical equilibrium, is only an approximation. And at the nitty gritty level, not many actual systems are in equilibrium. It’s simply too hard to isolate any system from its environment.

    There’s a great deal of work that’s gone into nonequilibrium statistical mechanics over the past 50 years, some of which has been very productive. You’ve already seen some of the concepts from that field applied to economics — power laws and scaling, self-organization, &c. What is needed is for this to be done starting with the simplest possible models, with indications where appropriate for where the predictions of new theory contradict the predictions of neoclassical economics.

    A problem is that much of neoclassical economics is not falsifiable because of the identification problem. Still, enough evidence is accumulating that traditional linear models (efficient market hypothesis, modern portfolio theory) will eventually get replaced.

    To eliminate the problem with the identification problem, we should give up on measuring supply and demand, and instead model input and output directly. Why should a network of cashflows be harder to model than traffic, for example?


    Michael F. Martin

    March 24, 2010 at 10:11 pm

  9. Thinking about it further: all your suggestions are just different agent heuristics. They don’t tackle the equilibrium assumption so much as the utility maximization bit. But actually you can get some econ results (ie, demand slopes curve down) even with some crazy behavior–as long as markets still clear. Friedman showed you can get efficient markets even if a number of the participants are irrational.

    Clearly with unemployment, inventory management, and certain macro phenomenon you can get a lot of mileage out of weakening (clearly violated) equilibrium assumptions. But note that the canonical non-Neoclassical theory–Keynsian economics–is actually a rational expectations model! While Keynes believed in animal spirits, he phrased much of his work in terms of investor expectations. Investors perceive a bad environment, and that’s proven to be correct. So from a meta-theory point of view; you can get odd results using equilibrium, or get the typical results by breaking a lot of assumptions. It’s tricky.



    March 25, 2010 at 3:07 am

  10. Why, it would look exactly like Autrian economics.


    Troy Camplin

    March 26, 2010 at 7:10 am

Comments are closed.

%d bloggers like this: