| A competitive equilibrium need not exist in an economy with indivisible goods or with other non-convexities. Furthermore, even in convex economies with no externalities, the Walrasian tatonnement can be globally unstable or can lead to undesirable equilibria. Since decentralization using the Walrasian price system is generally infeasible in these 'non-classical' economies, it becomes important to study alternative resource allocation mechanisms which perform well in these situations.;Employing a combination of simulational and analytical techniques, the essays in this thesis explore the potential for decentralization in non-classical economic environments. The starting point for the investigation is the Bidding Process (or B-Process) of Hurwicz, Radner and Reiter (1975), which is a stochastic, dynamic resource allocation process designed for both convex and non-convex economies.;Chapter 2 presents some simulation results on the rate of convergence of the B-Process, while Chapter 3 provides analytical results on the speed of convergence of the Process, as well as a modification of it, termed the Surplus Reallocation (SR) Process. It also demonstrates that the SR Process converges to Pareto optimal allocations in economies with both divisible and indivisible goods. In Chapter 4 another modification of the B-Process--the Free Disposal (FD) Process--is studied. This chapter formally constructs the FD Process as a time-stationary Markov Process on the state space of feasible allocations, and then proves that the FD Process also possesses desirable dynamic properties. Furthermore, analogs of the Welfare Theorems for classical economies are shown to hold. Finally, Chapter 5 employs results from the newly emerging field of 'chaos theory' to provide additional motivation for the research in the earlier chapters. It is proved that even in the two commodity world, the discrete-time tatonnement process exhibits chaos on an open set of economies. As a consequence of this theorem, the case for studying alternative resource allocation processes is greatly strengthened. |