In this talk, we introduce the WorkSim model, a novel agent-based framework to study labour markets. The first and empirical objective of the model is to reproduce the gross flows between the important states: employment, unemployment and inactivity, and the ratios of individuals in these states. The distinction between fixed-term contracts and open-ended contracts is a main feature of these labour markets, so we build two different sub-states within employment, including the necessary additional flows. The model also formalizes the most relevant legal rules of the labour market precisely as they have written in the French law.
A novelty of the model is that it simulates the flows on the basis of the rational decisions of individual heterogeneous agents, within the constraints of this institutional system. This is done in an intellectual framework that is based on the search concept but with bounded rationality and an emphasis on firms' anticipations formation on their own demand in a stochastic environment, rather than the standard independent productivity shocks on individual jobs. This leads to a second and theoretical objective, the development of a novel micro-based formalization of the endogenous choice by each employer between opening a open-ended or a fixed-term contract when faced to a favourable demand.
The model is then calibrated by a CMA-ES algorithm to set 60 parameters and fit 63 aggregate variables in a steady state. Once the model is calibrated (in this paper on the French labour market in 2014), the third objective is to do economic analysis, starting by a characterization of the nature of the labour market under study. In the talk, we start by examining the patterns of gross flows and stocks at the aggregate level and at the levels of different categories of labour, including sensitivity analyses for some exogenous parameters and variables. Finally the calibrated model is a tool for experimenting labour market policies. We show some evaluations here, including recent changes in the labour law in France.
Gérard Ballot is Emeritus Professor of Economics at the University Panthéon-Paris 2 and a member of the CRED (Research center in Law and Economics) at Paris2. He has been the founder and director of ERMES, a CNRS affiliated Research Laboratory in this university (1987-2000) focusing on personnel economics and labor markets. He also created and directed an undergraduate and graduate program in managerial and industrial economics (1998-2010).
He has built an agent-based model of the French labor market, ARTEMIS as early as 1980, winning the French Economic Association Prize. He then worked with analytical models and econometric methods on personnel economics and later on the effect of firms’training on their innovation and performance. The largest part of his research since 1990 is on the development of agent-based models, first a macro and Schumpeterian-Keynesian-Wicksellian model, MOSES, second a new model of the French labor market, WorkSim, with an emphasis on institutions and employers’ decisions over contracts types under uncertainty. He is also works with the same methodology on models of competence based R&D alliances which are aimed at quality and new goods innovations in a competition framework with consequences for growth, and market and employment structure.
Jean-Daniel Kant is currently an associate professor with Habilitation (HDR) at the University Pierre et Marie Curie (UPMC) in Paris, ranked first science University in France. He works at its Computer Science Laboratory (LIP6), one of the leaders in computer science research in France, within the SMA team, devoted to Multi-Agent Systems (MAS).
He is interested in modelling and simulation of complex systems, and more particularly to the contributions of Information Technologies in the Humanities and Social Sciences. He is currently developing research projects of multi-agent simulation, especially in economics (agent-based computational economics) to model the French labour market or the labour organization in firms. He also applied multi-agent simulation to sociology (job satisfaction, social networks, dynamics of opinions and attitudes), diffusion of innovation, political science (democratic insitutions) or supply chains. His approach is multidisciplinary and relies heavily on modelling of individual and social behaviours, from theories derived from economics, sociology, cognitive and social psychology, artificial intelligence.
He participates in several journal and conference reviews and committees, and also the cofounder and co-leader of MAGECO, the first French network of research on Agent-Based Computational Economics.