The Nature of Equilibrium with Semiordered Preferences: A Correction
HANS KEIDING of the University of Copenhagen has pointed out that Theorem (3.1) of our recent paper [1] is false as stated. In particular he has illustrated that when n = 2, using the notation of [1], one can construct a proper (1) correspondence Z: S2 -> P(R2) satisfying Definition (3.3) and such that Z(p) c 4th quadrant, for each p E S2. By Walras' Law this means Z'(p) 0 only when p = (1, 0). Thus, if S c Sn satisfies Z(p) s 0 for all p E S, then S = {(0, 1)}, which is not a subset of full dimensionality in Sn. We remedy this by adding to Definition (3.3) the economically innocuous assumption that if a good's price be zero the excess demand for it will be positive: