Domenico Colucci

Limited Computational Ability and Approximation of Dynamical Systems

This paper addresses problems due to approximation of dynamical systems. Theapproach presented here is based on the notion of the shadowing property ofa dynamical system and on the use of pseudotrajectories to approximate trueorbits. In particular, specific results in the case of attractors and thenotion of generalized shadowing are reviewed. The evidence described may berelevant for a wide range of economic models, whenever agents that havelimited computational ability, such as agents relying on digital computers,are concerned. The final section presents two sample applications to economicmodels.

Stabilizing inflation in a simple monetary policy model with heterogeneous agents

We study a simple monetary model in which a central bank faces a boundedly rational private sector and has the goal of stabilizing inflation. The systems dynamics is generated by the interaction of the expectations about inflation of the various agents involved. A modest degree of heterogeneity in such expectations is found to have interesting consequences, in particular when the central bank is uncertain about the relevant behavioral parameters. We find that a simple heuristic based on mean and variance of the distribution of behavioural parameters stabilizes the system for a wide parametric region.

Some clues to the determinants of feedback behaviour

We report the results of an experiment designed to investigate the determinants of feedback behaviour in electronic markets. Ratings driven by disconfirmed expectations should in principle reduce the asymmetric information problems of these markets. However, some other motives may influence the decisions of the sellers. In particular, empirical evidence suggests that the economic surplus obtained from the transaction may have some bearing on the way sellers are rated. Our design was meant to test whether and to what extent disconfirmed expectations and/or the transaction surplus play a role in determining the feedback behaviour of buyers in e-marketplaces. The results indicate that both factors affect the ratings, the latter having the stronger effect. One possible empirical implication could be that when an online purchase is a good deal a seller will typically get away with (moderately) exaggerated descriptions of the good on sale, obtaining a positive rating from the buyers.