We review the literature on agent-based labor market models by tracing its roots to the microsimulation literature, and surveying a selection of contributions made since the work by Bergmann (1974) and Eliasson (1976). Agent-based models have been applied to explain stylized facts of labor markets as well as for labor market policy evaluations. They also constitute a major part in agent-based macroeconomic models. Besides reviewing the various results achieved, we discuss modeling choices with respect to agents' behavior and the structure of interaction. Our overall assessment is that agent-based labor market models have given us valuable insights into the functioning of labor markets and the consequences of labor market policies, and that they will increasingly become an essential tool of analysis, in particular when the construction of large macro-models is involved.
ABSTRACT
2The labor market is in some respects a very special market, insofar as it interacts with many other economically meaningful domains. It is not just firms and workers meeting together to trade time for money. It is crucial to understand production, on one side, and income, hence consumption and savings, on the other side. As such, it is a key ingredient to any macro model of the economy.In this chapter we provide an original perspective on the agent-based (AB) approach to the modeling of labor markets. 1 We start from a broad definition of the AB computational approach to economic modeling, according to which AB models are characterized by three features: (i) there are a multitude of objects that interact with each other and with the environment, (ii) these objects are autonomous, i.e. there is no central, or "top-down" control over their behavior and more generally on the dynamics of the system, and (iii) the outcome of their interaction is numerically computed (Gallegati and Richiardi, 2009;Richiardi, 2012). To be able to compute the evolution of the system without the resort to external coordination devices, a basic requirement is that the system is specified in a recursive way (Leombruni and Richiardi, 2005;Epstein, 2006). This feature is not only of technical relevance for modeling purposes -as Bergmann (1990) puts it, "The elimination of simultaneous equations allows us to get results from a simulation model without having to go through a process of solution"but bears a substantive belief on how the real systems behave: "The world is essentially recursive: response follows stimulus, however short the lag" (Watts, 1991).