Herbert Dawid, économiste de l'innovation

C'est avec grand plaisir que nous accueillons le 17 novembre 2022, Herbert Dawid, économiste de l'innovation, spécialisé en Agent-based Computational Economics, Industry Dynamics et Policy Design in Complex Environments, à l'université de Bielefeld.

Herbert Dawid nous présentera en séminaire le papier intitulé "Implications of Algorithmic Wage Setting on Online Labor Platforms: A Simulation-Based Analysis".

Résumé :
We study how the use of machine-learning based algorithms for the determination of wage offers affects wages emerging on online labor platforms. We consider scenarios where firms compete by posting tasks and associated wage information on an online labor platform, and heterogeneous workers send applications based on the posted information. Firms use Q-learning to update over time the rules determining their wage offers. We analyze properties of the firms’ emerging rules and the resulting wages for different assumptions about the design of the labor platform as well as about the type of Q-learning algorithm used by employers. We show that the platform design has no substantial impact on the level of the emerging wages, but that a specific property of the learning algorithm, called experience replay, is crucial for determining whether wages converge to the (one-shot) equilibrium wages or to a much lower level close to the collusion outcome.
 

Le séminaire a lieu à 13h30 en salle de séminaires 227