Abstract. Macro-economic models describe the dynamics of economic quantities. The estimations and forecasts produced by such models play a substantial role for financial and political decisions. In this contribution we describe an approach based on genetic programming and symbolic regression to identify variable interactions in large datasets. In the proposed approach multiple symbolic regression runs are executed for each variable of the dataset to find potentially interesting models. The result is a variable interaction network that describes which variables are most relevant for the approximation of each variable of the dataset. This approach is applied to a macro-economic dataset with monthly observations of important economic indicators in order to identify potentially interesting dependencies of these indicators. The resulting interaction network of macro-economic indicators is briefly discussed and two of the identified models are presented in detail. The two models approximate the help wanted index and the CPI inflation in the US.

It is out of question that economic developments are determined by complex interactions. Modeling such interrelations, the question arises whether the models themselves have to be complex [1] or whether it is sufficient to restrict the model to the essential basic relations and to approximate these structures with linear or linearized functions?In order to answer this question it is helpful to lay down conditions that characterize a certain model as a "complex" one. In this article, models are defined complex if they come up to one or more of the following requirements:• The models exhibit a large number of components and interrelations (e.g., models with a big number of economic agents).• There are interdependencies between the components that make a recursive solution of the problem impossible.• Interactions are nonlinear, even transformations that lead to linear relationships are impossible.• Models are dynamic (the system's state at time t influences future states).The major part of the theoretical and empirical work in economics so far utilize, according to our definition, noncomplex models. They pursue a simplifying approach, using one or more of the following simplification techniques:• Application of exclusively linear functions to enable simple solutions for an even high number of functional dependencies (e.g., input-output analysis)• Reduction of possible interactions by aggregating transactors and transactions (e.g., macroeconomics) • Assuming uniform representative economic agents with identical assumptions about their behavior (e.g., microeconomics or computable general equilibrium models) • Neglecting spatial and temporal relations by utilizing static models without spatial background • Simplification of the model's structure by excluding interdependencies via axiomatic definition of independence (e.g., consumption theory).Each of these simplifications has been removed in isolated attempts. Simulation techniques are the furthestreaching approaches [2]. Nevertheless, there is no class of models so far that avoids all restrictions [3]. This deficit of models becomes even more severe when empirical instead of theoretical approaches are taken into consideration.Whether the lack of the capability of modeling complex structures in economics is a problem or not depends on the question of whether the mentioned simplification techniques do substantially distort or only slightly "disturb" the results. For a long time the latter opinion was taken for

No abstract

No abstract

No abstract