 |
|
|
|
|
|
Centre for Employment Studies
Simulations,
theory and experiments. Marco Novarese * Abstract The analysis starts from a paper of Herbert Simon and Geoffrey Clarkson - titled "Simulation of Individual and Group Behavior", published in 1960 in the American Economic Review - in which they clarify many aspects of simulation, theory and econometric analysis, and of their reciprocal relation. This work seems to have been neglected in the following devel opment of Economics (for exampl e it is quoted just two times in all article available in Jstor and is never quoted in a specialized journal as Jasss). Nevertheless it can be of great interest and usefulness in today's debate on simulations and on their relation with the general methodology of Economics. Clarkson and Simon find three possible kind of simulation analys is : 1) dynamic macr oeconomi c; significant example ar e the model s used in the analysis of the bus i ness cycle: In this realm, simulation is seen as an additional technique for numerical analysis that can be useful because of computer speed and computational power. When a model contain a feedback mechanisms, simulation will probably allow more accurate forecasting then traditional ones; 2) normative model for management science: in this realm the complexity of the environment can be managed more easily and with greater flexibility by simulations than by mathematical techniques as linear programming; 3) economic decision-making: again the usefulness of simulations is related to the degree of complexity they allow to handle. Computer allow also to built agents that manipulate symbols and information, different from numbers (like words or sentences). This characteristic would permit to model situations in which the important factors cannot be represented as real numbers. The last point seems very interesting, as it appears to be one of the main limits of today's simulations, as also Tesfatsion (2002) poses the modelling of individual behavior as one of the critical factors in the ACE researchers agenda. In this area, to investigate the real behavior of agents, the collaboration with Experimental Economics seems necessary and useful for both approaches, as the same Tesfatsion states again. At the moment, there aren't yet realistic models of (individual or social) behavior, empirically founded and accepted by the scientific community. Even experimental economics has generally worked on unrealistic models of situations that can be manipulated with mathematics. The relation under exam will be analysed taking into account some important examples of games analysed through experiment and simulations. Another aspect that will be taken into account, as it's important in Simon's recalled view, is the possible rol e of cases studies as a source of empirical information for understanding and modelling human decision making through simulations. *
Centre for Cognitive Economics, Università del Piemonte Orientale
|
