An Agent-Based Model of Worker and Job Matching

This paper proposes and tests an agent-based model of worker and job matching. The model takes residential locations of workers and the locations of employers as exogenous and deals specifically with the interactions between firms and workers in creating a job-worker match and the commute outcomes. It is meant to illustrate that by explicitly modeling the search process and the interactions between firms and individuals, origins and destinations (ODs) can be linked at a disaggregate level that is reasonably true to the actual process. The model is tested on a toy-city and the using Twin Cities are. The toy-city model illustrated that the model leads to reasonable outcomes, with agents selecting the closest work place when wage and skill differentiation is absent. Relaxing these assumptions increases the observed commute. Especially the introduction of wage dispersion in the model increases the the average home to work distance significantly. Using data from Minnesota, the results on aggregate are shown to capture the trends in the observed data, and illustrate that the behavior rules as implemented lead to reasonable patterns. The results and potential future directions are also discussed.