A Quantal Response Model of Firm Competition

The distribution of prot rates in the U.S. economy for 21,714 rms from 1962 - 2012 appears to be highly organized in a Laplace-like distribution. Pos- itive prot rate deviations from the mode appear to be remarkably stationary over time displaying little parametric changes while negative prot rate devi- ations introduce an asymmetry into the distribution that appears to uctuate over time. In this paper I propose a model of \classically" competitive rms facing informational entropy constraints in their decisions to potentially enter or exit markets based on prot rate dierentials. The result is a three parameter logit quantal response distribution for rm entry and exit decisions. Bayesian methods are used for inference into the the distribution of entry and exit deci- sions conditional on prot rate deviations and rm level data from Compustat is used to test these predictions. The model parameters show a uctuating asymmetry in rm exit decisions, an increase in dispersion of negative prot rate dierentials, and a falling general rate of prot.