Enterprise Hall, Room 318
November 08, 2009, 07:00 PM to 07:00 PM
The purpose of this project is to examine the benefits of basing economic analysis of accident law on the notion that individuals form a spontaneous order as they respond to the institution of tort law, each other, and their environment. Furthermore, I examine the efficacy of agent-based modeling as an analytical tool that exploits this realization. I discuss the theoretical implications of the neoclassical analytical perspective and offer an alternative based upon the insights of spontaneous order economics.
I develop an artificial society in which virtual agents pursue productive, though inherently accident prone, activity. The accidents that occur are chance encounters and agents bear a private cost for engaging in behavior that reduces their likelihood. First, I employ empirical techniques based on neoclassical theory and fit regression models to simulation data in order to determine the wealth maximizing behavior under various liability rules. I confirm some of the major theoretical conclusions but I also identify several common simplifying assumptions that may adversely affect the accuracy of conclusions under certain circumstances.
I contrast the mainstream mode of analysis with an evolutionary approach that uses a computational model comprised of heterogeneous agents that implement satisficing algorithms to select strategies on the basis of their individual experience. The power of this perspective is not simply that it assists in developing a genetic-causal explanation as a means for evaluating the desirability of various liability rules, but it enables the detailed exploration of population dynamics and a close examination of out of equilibrium behavior. I find that for the artificial society under consideration, system level steady-state does not necessarily imply agent equilibrium, agents often elect to be careful even when the neoclassical theory predicts otherwise, and negligence rules differ in their ability to rid society of negligent behavior.