Agent-based models versus Computable General Equilibrium models and spatial econometrics.

The macroeconomic theory has been developing different models to assess and simulate policies according to the dominant theories in every moment. Currently, many central banks and governments use computable general equilibrium models to predict the effect of their decisions and choose among alternatives. The unpredictability of the crisis by these models and its narrow contribution to the regional impact of policies or the income distribution, make this kind of models a lackluster tool to evaluate and simulate policies, as well as to forecast accurately at local level. To overcome these drawbacks, making use of agent-based models we can learn in depth the relations between different actors, emphasizing the importance of heterogeneity, networks, location and learning by imitation. These models help us understanding not only the expected trend of macroeconomic variables, but its impact on the various actors according to their characteristics.