On the local convergence of economic mechanisms

For certain informationally decentralized economic models, it is known that the associated dynamical process need not be locally stable. In this paper, the causes of the conflict between the local stability of the process and the decentralization of information are identified. This motivates a resolution for this problem. The method describes how approximate, local information about the economy can be used to modify a decentralized, dynamical procedure to achieve local stability. Explicit procedures are given for constructing the decentralized systems with which the agents communicate this approximate information. As an illustration, this approach provides a partial response to the well-known convergence problems with the tâtonnement model.