Other way round, I think

Cosma writes:

1. Obviously, macroeconomic phenomena are the aggregated (or, if you like, the emergent) consequences of microeconomic interactions. What else could they be? Analogously, the macroscopic physical properties of condensed matter all ultimately emerge from molecular interactions.

2. Macroeconomic theories which do not derive such phenomena from microscopic interactions are thus incomplete, and intellectually unsatisfying. Analogously, theories of condensed matter which do not derive the phenomena from molecular interactions are incomplete.

So: the true and complete theory of macroeconomics must emerge from the true and complete theory of microeconomics.

to which I don't wholly object ... or maybe I do. I agree with 1. And I agree with the lemma after "So:". But I don't really agree with 2. Not so much in what it says, as the way it says it.

My argument would be that what is it that we really care about when we do economics? The big questions, the ones of price, production, employment and trade. The stuff about individual transactions and agents can be ferociously interesting, and even sometimes useful, once in a blue moon (if you want to buy a second hand car for example), but the point of studying economics is to get answers that are useful at the aggregate, policy level.

So I'd turn this on its head. It isn't about "Does macroeconomics need microfoundations?". It's about "Does microeconomics need macroconsequences?". Leaving the stuff about condensed matter out, I'd rewrite 2 as:

2: Theories of microscopic interactions which do not imply accurate predictions about macroeconomic phenomena are thus incomplete, and intellectually unsatisfying.

And then the rest of the argument goes through mutatis mutandis, right through to the analogies with particle physics. Price theory, the expected utility model of choice theory and Nash/von Neumann/Morgenstern game theory have done very well in their fields - they even quite clearly describe certain aspects of the underlying truth about how economic actors behave. But they've failed the most crucial test - they don't help us generalise out to a theory of the behaviour of the whole economy, taken as a whole. They don't even (as the statements made by half the Chicago faculty over the last three years demonstrate more than adequately) help us to avoid saying things about the macroeconomy which fail to respect adding-up and stock/flow-consistency constraints, or well-established empirical relationships.

The New Keynesian project worked in this sort of spirit - recognising that we basically only had one family of macroeconomic models which even slightly worked, and attempting to work back into microeconomics to justify it. This approach, unfortunately, led us to DGSE and all that; it didn't work. I think that the conclusion from that ought to be that you can't get the Keynesian (or rather, the correct - the eventual macro model might not be Keynesian I suppose) conclusions simply from price theory and game theory[1], and so microeconomics needs to be re-established on some other basis that preserves the insights of what Phil Mirowski called the "cyborg science" era but moves on. In some unspecified way that I don't care about because I don't really agree with Cosma that there's anything particularly unsatisfying about macroeconomics without microfoundations[2].

[1] Or at least without explicitly modelling such a complicated and combinatorially huge nexus of principal/agent interactions at every management level of every firm, giving you a map potentially several times larger than the territory.

[2] This might be because my brain is smaller than his and thus more easily repleted.

Edit: reading back, I don't think I've emphasised enough how much I agree with Cosma's actual argument - I just think it's a damnable impudence on the part of microeconomists to have ever claimed otherwise! I also tend toward the view (also, I think, held by JW Mason) that any "correct theory of microeconomics" which has enough relevant and accurate macroimplication, is going to have imported a hell of a lot of specific and contingent institutional detail, rather than looking anything like a general theory of decision-making (there's another subject where all that stuff belongs, and it's called "decision theory", not "economics", but I think that demarcation dispute is for another time).