Invited Speaker

An agent-based simulation approach to evaluate and design policies

Jean-Daniel Kant (Sorbonne Université, LIP6, Paris, France)

In this talk, I will present an agent-based simulation methodology to evaluate and design policies: public policies, economic policies, firm policies, … First, we could build an agent-based model to simulate the ex-ante effects of the policy, in order to improve its design. I will illustrate this idea in the case of labor policies, with the WorkSim ABM of labor markets, a comprehensive and calibrated model of the French Labor Market. We used WorkSim to evaluate a wide range of labor policies. The validation process, based on calibration, is crucial in order to set the model on real data.

Another – and complementary – approach is to use optimization algorithms to seek optimal policies. To do so, I will process the OptiPol methodology, a
generic method we applied to labor policies.

I will conclude by discussing the limits of all these approaches, and how the
could contribute to improve decision-aid for complex systems and policy
design.