THE DYNAMICS OF TASK AUTOMATION AND WORKER ADJUSTMENT IN LABOR MARKETS: AN AGENT-BASED APPROACH

Artificial Intelligence and associated technologies are rapidly automating routine and non-routine tasks across industries and can severely disrupt labor markets. This paper presents an agent-based, evolutionary model of labor market dynamics where workers adjust to technology shocks induced by automation. Firms produce a homogeneous service by combining the outputs of tasks performed by workers while stochastically adapting to automation of tasks causing displacement of workers. We develop a model that includes: (i) the description of occupation mobility as a directed graph where nodes represent occupations, and the directed edges represent the mobility pathways along which displaced workers can get retrained and redeployed (ii) explicit microfoundations of the processes of job matching and wage setting between firms and heterogeneously skilled workers. The model focuses on the influence of workers' retraining choices on the employment levels and wage inequality in the labor market. Simulation results indicate distinct tipping points for unemployment and wage inequality with changes in mobility pathways along which workers retrain and redeploy across occupations. An increase in the density of mobility pathways induces a reinforcement effect on employment. Retraining displaced workers without building dense and well-distributed mobility pathways across occupations could widen wage inequality due to excessive crowding of workers around specific tasks. Our work focuses on the finance and insurance industry dataset, where we observe that the reskilling of displaced workers along occupation mobility pathways assisted by a lower retraining cost improves the unemployment levels. Also, if the firms aggressively automate their tasks, an increase in the cost of retraining increases inequality of wages in the labor market.