Complex dynamics and adaptive fuzzy rule-based expectations - economic simulations with GENEFER

On last yearÌs conference in Barcelona the authors presented an innovative expectation formation hypothesis. The assumption of fully rational agents is rejected and replaced by a bounded rationality approach that is modelled by means of a fuzzy rule-base. These rules as well as their components (antecedents and consequents) dynamically adapt to a changing economic environment. The technical realization is done by the software GENEFER (Genetic Neural Fuzzy Explorer). An early version of this software was presented last year. Since then the authors have significantly extended its abilities and implemented a COM interface, so that it is applicable to any simulation in any other programming language. Our recent research concentrated on economic applications of GENEFER to demonstrate its expectations modelling and learning abilities: 1. Modelling inflation expectations within a macroeconomic business cycle model 2. Modelling agents' expectations in an artificial multiple agent foreign exchange market 3. Computing business cycle indicators with real world data The latter is on our current research agenda and we plan to obtain first results within the near future. These results will give first insights into GENEFER's forecasting abilities. All these current research projects along with a brief software documentation can be found on our website www.genefer.de. We would be happy to present the most interesting application on this yearÌs conference in Yale.