Computation and Complexity in Economic Behavior and Organization

From MaRDI portal





This book considers a general class of models of information and information processing, or computation, (over the reals) for economics. The basic class of models F-networks, which represent computation over finite directed graphs, is presented in Chapter 2, along with a proof that it suffices to consider fan-ins for the directed graphs in many cases. Chapter 3 deals with the basic mathematical results needed, and a survey of other approaches, in terms of representing functions of several variables in terms of compositions of fewer numbers of variables. The most important tool used is Leontief's theorem, on when such a local representation is possible for differentiable functions, in terms of the partial derivatives. This is applied to game theory. In Chapter 4 the model is extended so that it can apply to human agents, and is applied to an Edgeworth box economy to show that the equilibrium message function takes at least 2 stages to compute. In Chapter 5 as a solution to bargaining games, the Nash solution is in general informationally more complex than the Kalai-Smorodinsky solution. In Chapter 6 theoretical results are presented in terms of ranks of bordered Hessian matrices as to when functions of a product of manifolds can be represented in terms of fewer coordinates on each manifold. Chapter 7 applies this to questions of market equilibrium and incentive-compatible mechanisms. In one example, the market channel cannot achieve an efficient coordination over time, but a direct channel can. In a second example, the direct channel requires less information transfer in order to achieve efficiency than a market channel.











This page was built for publication: Computation and Complexity in Economic Behavior and Organization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4781176)