As this Newsweek article points out, the INET conference challenged much of the previously accepted wisdom in mainstream economics:
Again and again, in different ways, the most perspicacious observers at the Cambridge conference drove this point home: rational models don’t work because there are too many unknowns to justify them. There is no real equilibrium in the real world. The models don’t compute. Literally.
Of course, everyone involved in the conference viewed this as some sort of starting point for a renaissance in the discipline. Preeminent among the participants, I wager, is Joseph Stiglitz, the Nobel laureate who made his marbles by challenging the assumption of perfect information in neoclassical models.
His talk (video here, preliminary pdf here, via Thoma) presumably laid out an agenda for reform. Read it if you have time, or even skim the topic lines; it’s a good primer on the left edge of the mainstream, in which assumptions rationality, representative agents, and even equilibirum are challenged. All of this is good. However, after first reading, I had a nagging feeling that there was something lacking, and I couldn’t put my finger on it. I then went back to the intro, in which he framed the discussion by saying,
Most of the economics profession failed to predict the most important economic event to occur in the history of modern “scientific” economics. If economics is a science, it is presumably to be judged by its ability to predict. It has, by and large, failed that test.
So, as I mention above, he proceeds to dig into the methodological issues explaining why economics failed as a science. There are some really encouraging nuggets at the epistemological level:
This leads one naturally (if one wants to limit oneself to models with high degrees of rationality) to think about models in which there is uncertainty about the model itself, about the adequacy of its description of the world, or in which there is always a residue of fundamental uncertainty: perhaps the world has changed in a way in which this is not a bubble.
But then there are passages like this one:
I do think that policy measures (both macro- and regulatory) have to be addressed with dynamic models, within general equilibrium models and within models in which risk is central. The problem is, in part, that because even “toy” models are complex, there is a need for extensive simplifications, and the simplifications have left out almost everything that is important.
And the thing wraps up without Stiglitz ever really saying, “economics is not a science- we need to approach it as such.” He’s obviously aware of the host of issues explaining why it can’t be, and thus shouldn’t aspire to be, a science. Nevertheless, when it comes to synthesizing all of these issue, his neo-classical, math-addled brain somehow still aims to build better predictive models of the economy. Something doesn’t compute. I probably don’t have anywhere near the math or economics training necessary to make these critiques, but there just seems to be a dissonance withing this type of agenda. Is the issue credibility? A sort of deeply ingrained narrowness? I think we can do better. Of course, I think people like Stiglitz have would have a much better idea of how, but that will require removing their blinders.