Last news Economics Technologies EU Scenarios About us Admin Area
Agent-based Macroeconomics
Home > Agent-based Macroeconomics > A general perspective on EURACE
A general perspective on EURACE

The EURACE economic model compared to existing (agent-based) macro models

In terms of models designed for policy advice the EURACE model follows a fundamentally different approach from the dynamic stochastic general equilibrium models (DSGE) which are currently employed by the European Central Bank (ECB) or the European Commission on Economic and Financial Affairs (ECFIN). These models are typically not as rich in terms of their spatial structure, heterogeneity of firms and households, and institutional details explicitly modelled.  For example, the ECFIN model on which many of the policy recommendations of the European Commission rests has only two countries and three sectors. It lacks a more appropriate spatial dimension and differentiated skills of workers just to name two additional features that the EURACE platform will provide. Thus, for many policy questions, on which we become more explicit in the following section, they are inappropriate tools to come up with satisfactory answers.

The EURACE approach also goes substantially beyond existing macro-economic models  (e.g. Chiarmonte and Dosi (1993,) Delli Gatti et al. (2005), Dosi et al. (2006)). EURACE is by far the largest and most complete agent-based model developed in the world to date. It is the only one explicitly aiming to capture features of the European economy. Its main distinctive and innovative features are:
  • its closure: EURACE is one of the very rare fully-specified agent-based models of a complete economy. EURACE is dynamically complete, that is, it specifies all real and financial stocks and flows and will allow us to aggregate upward from the micro-specifications to the macroeconomic variables of interest;
  • the encompassing types of real and financial markets and economic agents are taken into consideration;
  • the wide use of empirically documented behavioural rules;
  • the different levels of time and space granularity taken into account. Thus, it is possible to investigate the impact of real-life granularity on the economic outcomes, and to analyse the consequences of a modification of this granularity;
  • the treatment of time: asynchronous decision-making across different agents;
  • the explicit spatial structure, allowing to take into account not only regional and land-use aspects, but also more generally the fact that all human activities are localized in geographical space;
  • the evolving social network structure linking the different agents;
  • due to the detailed spatial and time modelling, and to the network structure, the possibility to investigate in finer detail more differentiated policy options;
  • the very large number of agents, possibly allowing to discover emerging phenomena and/or rare events that would not occur with a smaller population;
  • the use and development of innovative software frameworks, code parallelization in order to employ super-computers, allowing very large-scale simulations;
  • the calibration on European economic data and the focus on European policy analysis.