INTERACTIVE-AGENT ECONOMIES: AN ELUCIDATIVE FRAMEWORK AND SURVEY OF RESULTS

Abstract

This paper is helpful for understanding interactive-agent economies. It presents a user-friendly framework for exploring the implications of economic interaction, and demonstrates its wide applicability. Agents repeatedly face binary choices and receive i.i.d. shocks; payoffs depend upon one's own action and shocks, as well as upon the actions of a subset of other agents. Simulations explore how properties change as the number of agents, the number of neighbors, and the degree of interaction change. Key results in this framework include the following: The law of large numbers can readily be postponed or defeated, stationarity and ergodicity are trivial to deduce, any equilibrium distribution may be attained in “globally interactive” economies, continuous- and discrete-time analogues generally behave very similarly, dynamics can be sensitive to details of the interaction structure (implying that popular mean-field approximations are not always appropriate), and substantial persistence is typical.