I would like to share some thoughts about the very stimulating discussion on “Achilles’ heels of current economic theories – The debate” (http://www.unifr.ch/econophysics/editorial/show/id/52). I am approaching this as a former physicist (publishing as “Eric Benedict”). I now work in the area of environment and development (publishing as “Eric Kemp-Benedict”), where I have been engaging with economists and economic ideas for over a decade.
I think that there are important research questions in economics where physicists can contribute, and both agent-based and network models are promising avenues of research. But they are not the only possible techniques. I think that a research program should take into account compelling ongoing work in economics (e.g., spatial, behavioral, development, and historical economics), and approaches that were abandoned in past decades because they were out of step with the orthodoxy of efficient markets.
I see two outstanding problems in economics that physicists might be able to help with:
Problem 1 is a re-framing of part of what I’ve read in this editorial and the background papers. I think this re-framing gets at the heart of a central problem. Use of “representative individuals” and “representative firms” derives from an inadequate toolkit for aggregation that is completely unsuited to the insights generated by spatial, behavioral, institutional, and relational economics. Thinking of economies as complex adaptive systems, and studying them using agent-based and network models, is a promising route to constructing effective theories consistent with these insights, but may not be the only approach. For example, Ovchinnikov, 2011, Self-Organized Criticality as Witten-type Topological Field Theory with Spontaneously Broken Becchi-Rouet-Stora-Tyutin Symmetry argues that a suitably complicated field theory with slow and low-amplitude noise compared to internal dynamics is approximately a topological field theory. (An uncomplicated field theory will also be approximately topological, but trivial.) This descriptions seems to fit the assumptions behind CGE models, and, if true, could provide a way to quantify the limits of CGE models, and show where and how an equilibrium/comparative statics approach may be an effective theory in certain limited circumstances. My point here is that by assuming one approach or technique is uniquely suited to this (or any) problem, you can miss insights from alternative approaches. This is an extremely tricky problem that needs broad and creative thinking.
On the second problem, economics has growth accounting/endogenous growth models, price-adjustment mechanisms, and Hamiltonians, but growth accounting models use somewhat ad hoc rules for capital accumulation and production; most economists think of price-adjustment mechanisms as techniques for finding equilibria rather than a dynamic description of real economies; and Hamiltonians in economics look very odd from a physics perspective–they are discounted “snapshots” of the future and are essentially non-dynamical. Agent-based and network models may be very useful for understanding economic dynamics, but again they are not unique. In fact, through the 1990s there was active research into disequilibrium dynamic macro and microeconomics (e.g., Dynamic disequilibrium modeling: theory and applications), including econometric methods to fit such models. In contrast to the claims of some of the background articles, these theories generate instabilities, overshoot, and collapse without treating the economy as a CAS or using agent-based models. As with so many promising research programs, this one died out over the past decades because orthodoxy proclaimed that economies are never out of equilibrium.