ABSTRACT
[T]he fact that any satisfactory theory must be grounded in the theory of individual behavior has further consequences for some current work and for the way in which we view certain forms of analysis. To begin with, the stability problem is not satisfactorily solved by showing that there exist some adjustment processes which converge. However interesting certain adjustment processes may be, unless there is a reason to believe that they arise from the optimizing behavior of agents, they cannot be regarded as providing more than a computational algorithm for finding equilibria. Indeed, the situation here is worse than that involved in the ad hoc specialization of excess demand functions to achieve a stability proof. We know that such specialization can obtain under special circumstances. We often do not know that particular convergent processes are ever consistent with a sensible story about the behavior of individual agents.
