Here. I like it a lot. Especially, you'll find
this most important statement by K. J. Arrow followed by:
Im Hintergrund steht die Prämisse, daß Volkswirtschaftslehre das "wirtschaftliche Verhalten des Menschen" zum Gegenstand habe. Dies aber ist ein Mißverständnis: Volkswirtschaftslehre ist keine Verhaltens-, sondern eine Systemtheorie, die die Funktionsweise einer durch Märkte strukturierten sozialen Ordnung untersucht; und sie gewinnt ihre Eigenständigkeit gegenüber Soziologie und Psychologie gerade dadurch, daß sie ihre künstlichen Modellakteure lediglich mit einer simplen Zielfunktion ausstattet. Damit wird nur die nötige Bewegungsenergie erzeugt, um die Modelle "zum Leben zu erwecken" und ihren Vergleich zu ermöglichen. Qualitätsmerkmal und Fortschrittskriterium ist die Fähigkeit der Modelle, Strukturmerkmale wirtschaftlicher Erfahrung theoretisch nachzubilden.
UPDATE: P. Spahn also refers to the Sonnenschein-Mantel-Debreu (SMD) results. Indeed, the bold statement above (bold in the general and innocent sense) relates to the SMD results. Spahn, after pointing to problems with Walrasian and thus atomistic macro (an oxymoron, indeed), counters the reaction to come up with behaviouristic sophistication. I totally agree that progress in our understanding of the economics system (the impersonal machinery or 'invisible hand' - price system, design of mechanism, competition -guiding action under simple rules of conduct - max U! max P! - and motivated by the most common and reliable incentive, self interest) does not rest on more detailed studies on individual characteristics. This is the very consequence of the SMD results. No matter how many restrictions you impose on individual characteristics, like the assumption of homothetic preferences, or even identical preferences and collinear endowments (Kirman/Koch), it is always possible (with prices bounded away from zero) to find an economy with aggregate excess demand satisfying contituinty, homogeneity, boundedness, and Walras' Law only. To make statements about important properties of our economy - like uniqueness and stability of equilibrium - we need restriction on the aggregate excess demand correspondence (e.g., gross substitutionality for global stability). Thus,
the SMD results imply that general restrictions imposed on agents' characteristics cannot discriminate between economies with different aggregate properties. Atomistic proporties don't yield information about aggregates (beyond B,C,H,W, which suffice for existence only)
! As Hildenbrand and Kriman point out (
Equilibrium Analysis, Amsterdam: North-Holland, 1988):
"Are there still more properties beside the basic four of continuity, homogeneity, satisfying Walras' identity and boundary behaviour? If so, then we could perhaps use these additional properties to start to answer our questions as to uniqueness and stability. - Unfortunately, we will give a clear and unambiguous answer to this question [...]. Our assumptions on individuals yield no more properties than just described and thus we cannot hope to make progress to answering our question if we stick to assumptions on individual characteristics alone. Thus, 'individualistic microeconomics' will not allow us to progress further." (p. 200)
The system is more than just the sum of its components. Note, if "individualistic microeconomics" is a dead end, what about macroeconomics with representative agents (Spahn's Crusoe-economics)? Hildenbrand/Kirman continue:
"What should be absolutely clear is that the conditions that we need on excess demand. that is on our system of equations to give us uniqueness and stability, is mathematically very throughly understood. Thus, the basic problem is not a technical one but rather one of finding economically interpretable properties that will guarantee the mathematical conditions that we need. Now it is obvious from what we have already said, that 'microscopic' or 'índividualistic' assumptions are not going to help. An alternative approach is to give somewhat 'ad hoc' conditions on aggregate excess demand functions themselves which yield the desired results. In doing this, it should be clear to the reader that we are changing standards." (ibid.)
I guess, Prof. Spahn will like that since the results come from within the neo-Walrasian paradigm and thus macroeconomists with a Walrasian fetish have to listen if they care about internal consistency (as they should). Check also
this related statement by F. Hahn.
