Matching dynamics and optima in a multi-agents labor market setting, 2015.

One of the greatest difficulties meet on the labor market is that it ensures that all the people who want to work can find a job. But it is not the only one. A second difficulty consists to find a social arrangement which approaches as much as possible the social optimum without forgetting that the optimum can be reached with the rejection of certain candidacies. It seems impossible to solve a social optimum problem of the size of the french labor market because this one is furthermore constituted of 26 million job-seekers and that no computer is capable of managing the matching problem which, surprisingly is not of an excessive complexity-it is of order n 3-, but which quickly collides with the computational capacity of machines, however powerful they may be. It's the reason why it is essential to study restricted size markets, that is markets of less than 10000 job-seekers, to observe how practical solutions which can be organized as, for example, a geographical progressive balkanization either geographic or by skill level, leads to a departure from the Pareto optimum. In this context, this paper suggests to compare the level of efficiency of a decentralized actually practicable solution with the optimum which would be if only it was possible to collect all necessary information to implement and compute it