Which formalization would you argue adequately considers specialization and trade? An Edgeworth box?
This is a good question, because mainstream economists cannot “see” anything that is not in a formal model.
My answer in this case is a definite “No.” The Edgeworth box is an example of two-by-two economic modeling. Other examples include the Ricardian model of comparative advantage and the Heckscher-Olin-Samuelson model of international trade.
The most important aspect of specialization and trade is that we specialize in just a few tasks but we enjoy the products of millions of tasks. This fact was noticed by Adam Smith, but it has not been “formalized” in any useful way that I can think of. So the formal modelers are like drunks who have their preferred lamp posts, but the watch that need to look for is somewhere else.
I wonder if agent-based modelling is making any progress on these fronts.
I know some folks working on such projects, but nothing that has been published. Unfortunately, even if and when those models are working, economists seem very leery of such work. Bordering on distrust and fear, in some cases. The current tools of economists are poorly tuned to understand those sorts of models, it seems. I am optimistic about future generations of economists’ acceptance of such models, however.
One needs to back up to the big picture to get at the nature of the real problem, which isn’t strictly one of scientific understanding in nature.
The problem is when a discipline’s object of study is (1) incredibly complex, (2) cannot be subjected to adequate controlled experiments, (3) sufficiently precise, accurate, and comprehensive data are unavailable (or constructs impossible to observe directly in the first place even in theory, and (4) the ‘results’ of running model simulations are relevant to coalition power politics in teems of justifying controversial major additional government interventions in private affairs.
When those four factors are present, then the incentives, along with typical contemporary academic funding sources and social clique cartel mechanisms, are such that at least a large fraction of the discipline’s scholars will become ‘captured’ and become willing academic hired guns for one partisan agenda.
This doesn’t even have to be corrupt or dishonest at all, because it is in the nature of models of such systems that many variables are needed to reflect the underlying natural complexity, but like von Neumann said, that with enough variables one can ‘prove’ and convince oneself of absolutely anything. And it is easier than ever to exclude critics and contrary voices from the high status discourse debating such issues by portraying them as insufficiently credentialed cranks outside the mainstream consensus. That’s because the consensus is maintained both by the field’s personnel selection and log-rolling tendencies, as well as the fact that there is indeed no shortage of dilletente cranks and crackpots out there.
All of this creates a high risk that the whole output of a field may be nothing more than self-defended bunk, with insiders doing Feynman’s cargo cult science by producing streams of symbolic or verbal manipulations of dubious value or utility.
What is needed, therefore, is not necessarily refinements of these models, which may be impossible in terms of producing frameworks that won’t inevitably be abused in precisely the same ways, but work at the meta or epistemological level about the fundamental limits of our ability to make accurate forecasts or claim knowledge about the effect of interventions on such systems. A science of humility, the major purpose of which is to inspire in all audiences the appropriate presumption of skepticism regarding such claims.
TL; DR version:
Of the most difficult and important intellectual problems in the world today is figuring out when to be skeptical of the expert consensus in any field, and when it is reasonable or unreasonable for one to reject such an argumentum ad verecundiam.
Rob Axtell has an agent-based model that he described to me as “a multi-dimensional Edgeworth box.” It’s the one discussed in “The Complexity of Exchange,” but I think it was originally specified and presented in an earlier conference paper.
Thanks!
I wouldn’t want to give up on the hope of finding some formal models. But ones that can explain specialization and trade are going to be hard to build. Worth looking.
Kling,
Congratulations on becoming a grandparent, I saw that I could not reply on the actual post.
-Long Time Reader
Congrats, Grandpa! And also to Grandma.
Perhaps self-organization and structures in dissipative non-linear systems is an apt zero-order approximation of specialization and trade. The dissipative systems theory – also full of lamp posts and toy problems – at least explains the emergence of sustained large-scale structures in the systems interacting microscopically and locally.