Background

Towards a new modeling paradigm in economics

The financial and banking crisis of 2007-09 and the consequent severe economic recession have caused a crisis of confidence in the science of economics and, in particular, in standard macro-economic modeling. Mainstream dynamic stochastic general equilibrium (DSGE) models characterized by the representative agent paradigm have been unable to forecast timely the advent of the crisis and measure the risks involved. The crisis has indeed highlighted the importance of the interplay among the financial, the credit and the real sectors of the economy, considering both the way productive and real-estate investments are financed and the consequences on the economy of boom and bust cycles in the asset markets. In order to address these issues, we do need models able to take into account the internal financial structure of economic agents, such as balance sheet composition, debt and leverage, as well as the network topology of their credit and ownership relationships, and able to identify counter party risk exposures and shock transmission processes at the systemic level. We need to investigate and establish a new modeling paradigm in economics based on the behavior and interaction of heterogeneous agents.

The agent-based modeling and simulation approach promises to deal with the modeling of the economy as a complex adaptive system and explain the emergence of aggregate economic regularities as a self-organized process from the interactions of heterogeneous individuals. The agent-based approach encompasses the concepts of bounded rationality, heterogeneity, and asymmetric information among economic agents. Furthermore, the agent-based approach naturally allows to investigate the links among credit, financing of investments and real economic activity and their role not just as propagators of exogenous shocks, as in the traditional model settings, but as the main source of endogenous asset bubbles, financial instability and endogenous business cycles, allowing to develop large-scale economic models which result to be far more policy-relevant than standard analytic models based on the representative agent and full rationality hypotheses.