Report of the Dijon Workshop

by Antoine Godin and Mauro Napoletano

A wide variety of approaches, methodologies and topics were presented during the first workshop, allowing for interesting discussions and exchanges. Clearly, the complementarities of the Stock-Flow Consistent and Agent-Based approaches emerged, even if some were skeptical at first. The influence of finance, fiscal austerity and the construction of the euro zone are at the heart of applied work from both approaches. Furthermore, the topics and issues addressed by papers from both methodologies are similar. On the methodology side, SFC and ABM practitioners share issues regarding estimation or the role of expectations.  On the first issue, the workshop has featured some presentations about methods that could be used to improve the matching between theory and data in both SFC and ABMs. An open issue there, and that could be developed in future workshop, is how far to go in the model calibration and estimation exercises. Indeed, both types of exercises impose strong restrictions on models (e.g. about the ergodicity of the dynamics) which could be limiting in terms of the ability of the model to catch salient features of the reality or that could be hard to detect into some models. Furthermore, the ABM literature has developed robustness checks and stability analysis that need to be further developed in SFC models. This highlights the interest of confronting the two approaches.

This workshop also showed us the importance of getting together and confronting our analysis and results. In particular, the workshop has highlighted the strong complementarities existing between ABM and SFC models. On one hand, SFC models have so far been developed as general aggregative models, i.e. as systems of stock-flow consistent equations describing the laws of motion of the economy at the aggregate level. On the other hand, ABMs provide explicit micro-foundations to macroeconomic relations that, in ABMs, are emergent properties of the disequilibrium interactions occurring among heterogeneous agents. However, the use of the stock-flow consistent approach in ABMs has so far been limited (few exceptions to this are represented by the models of Kinsella et al., 2012 and by Seppecher and Salle, 2012, Raberto et al. 2012). The use of the stock-flow consistent approach in ABMs could thus contribute to improve the rigor of the micro-foundations provided by these models. However, it could also help to micro-found many of the Keynesian dynamics emphasized by SFC models. This is important also because, as it was pointed out in the workshop, SFC models are particularly suitable to study the effects of imbalances at the aggregate level. However, by construction, they cannot study the factors underlying the emergence of those imbalances, such as for example the factors leading to bubble-and-burst dynamics in asset markets. Finally, we should mention the possibility of having some kind of mixed models where some sectors are agent based and others aggregated. Combining ABM and SFC allows thus to offer a wide variety of models with more or less complexity and different levels of aggregation, depending of the subject under scrutiny.

References

  • Stephen Kinsella & Matthias Greiff & Edward J Nell, 2011. "Income Distribution in a Stock-Flow Consistent Model with Education and Technological Change," Eastern Economic Journal, Palgrave Macmillan, vol. 37(1), pages 134-149.
  • Raberto, Marco & Teglio, Andrea & Cincotti, Silvano, 2012. "Debt, deleveraging and business cycles: An agent-based perspective," Economics - The Open-Access, Open-Assessment E-Journal, Kiel Institute for the World Economy, vol. 6(27), pages 1-49.
  • Seppecher, Pascal & Isabelle Salle, 2012. “A Two-Sector Agent-Based Model: Empirical Validation and Prospects” Unpublished.