Wagner on Agent Based Modeling of Austrian Economics

John Vandivier
I'm now taking my second PhD-level class with Dr. Robert Wagner. This class is Theory of the Market Process II and it focuses on the so-called non-equilibrium theory found in radical subjectivism. Radical subjectivism, typified by Ludwig Lachmann and George Shackle, is regarded as a peculiar offshoot of the Austrian School. Here's the syllabus for more info about the class. For a while I have seen agent based modelling as a rigorous approach to make Austrian economics more computationally rigorous and perhaps more useful as a business tool, without compromising core values. I also think it's a point of possible synthesis with the Neoclassical School and burgeoning behavioral literature. I asked Wagner whether he thought ABM would be viable as a modelling technique for Austrian market process theory and he said it would be suitable if done properly. Keep in mind that Wagner is a radical subjectivist. My line of thinking, perhaps incorrect but I don't think so, is that if a radical subjectivist would ascent to ABM then a typical Austrian would agree all the more. In other words I am considered the radical subjectivist to be a limiting case, a hard case, and a more difficult person to convince compared to someone else such as Coyne or Boettke who I might consider a bit more moderate. Come to think of it, I may have to eat those words as I've yet to ask Coyne or Boettke about their thoughts on radical subjectivism. I know they are both mainline (big tent? contra Salerno) Austrians and at least ordinary subjectivists. Wagner and I also discussed the undecidability problem in particular. The undecidability problem has an Austrian and mainline flavor:
  • Mainline: an individual is indifferent between A and B. Does he consume A, B, a combination, or neither?
    • Typical answer is some of each because we have a taste for variety, but consider lumpy goods such that you only get one or the other. Typical answer is a coin toss.
  • Austrian: an individual is uncertain that either action A or B will bring about desired end C. Does the individual act?
    • One answer is that the individual may have relatively different uncertainties and expected values and may therefore select a superior case, but suppose these factors are even between each choice.
    • Another answer is that an individual does not act unless there is a reason to do so, but this can be rebutted by considering non-action as action B and positing that individuals act by necessity.
    • The last answer I can think of is the one Wagner gave: The individual will eventually choose A or B although we have no reason to think he will prefer either one. I asked if it would be appropriate to use a coin toss simulation as a modelling mechanism for this decision and he said yes, if a problem was undecidable then over the course of time (perhaps modeled by several ticks or iterations of processing logic) a random selection might be fine as a modelling technique.
It was that final point which surprised me and I think is noteworthy. On those grounds there may be room to develop an applied Austrian ABM. More on undecidability in economics is here.