Thursday, March 22, 2012, 3pm
Innovation 223

Robert Axtell
Department of Computational Social Science
George Mason University

Modeling Firm Dynamics

A model in which purposive agents self-organize into teams is described and empirically tested against data on U.S. firms. There are increasing returns within teams and agents move between teams or start new teams when it is in their selfinterest. Nash equilibria of the team formation game exist but are unstable for sufficiently large teams. Dynamics are studied using agent-based computing at fullscale with the U.S. private sector (120 million agents). There arise stationary distributions of team sizes, growth rates, lifetimes, output, income, and job tenure, wages that increase with firm size, growth rates that decline with firm age, growth rate variance that falls with firm size, and approximately constant returns to scale at the aggregate level. Job-to-job flows, hiring, unemployment and other labor market phenomena occur for microeconomic reasons, without resort to external shocks. The model quantitatively reproduces many important regularities associated with firms and labor markets.