Semiotic Tools For Economic Model Building

Scientific researchers, when faced with real world data, try to detect the hidden relations and laws that are not readily apparent. This is the basic motivation for what is called "model building". Several techniques were developed in order to facilitate that work. Statistics provided ways to build efficient models by using the minimal amount of contextual information. Computational intelligence continued with this trend of automating the construction of models for restricted domains. In this paper instead, we claim that model building requires the active participation of researchers and their previous knowledge and intuitions. Contextual information must be taken into account to faithfully represent the real world phenomena. To improve this task, we need more expressive instruments. Semiotics, a discipline highly concerned with iconic reasoning tools, is the basis on which we will build the desired procedures for model building. The method we introduce here, with its great expressiveness, is extremely useful for economic model building. This is because in Economics (especially in microeconomics) the heterogeneity of data and the different statistical methods applied lead to very different models. With more expressive methods, these differences will disappear or at least will become easy to detect where they come from. The method that will be applied here is a kind of semiotic data-mining technique for generating models. This semiotic engineering will be applied to the analysis of the degree of convergence among economies. This last issue has been a source of discussion for economic growth theory in the last years. Since several factors are involved, it seems clear that more than a mere statistical analysis is required to detect the relations between sources of growth and the rate of growth. A semiotic approach will be useful on this issue.