Entrepreneurship: Getting Nonnormality Right

Abstract:     
Decision problems in environments with iid shocks are typically modeled as involving payoffs that follow a martingale process. The paper formalizes entrepreneurial skills as the ability to design optimal investment strategies for problems, in which payoffs do not follow a martingale. It develops a new model of states of nature, such that Gaussian states (finite variances and finite expectations) are as likely as non-Gaussian states (infinite variances and infinite expectations). To determine a true state, an "entrepreneur" performs a nonstandard Bayesian inference with a Markov Chain Monte Carlo algorithm. Her competitor uses a standard Bayesian inference and 96% of the time concludes that a true state is Gaussian even when, in reality, it is non-Gaussian. By properly applying strategies optimal for non-Gaussian states, and strategies optimal for Gaussian states, the "entrepreneur" finds and benefits from profit opportunities that are missed by her competitor.

Keywords: Information and Uncertainty, States of Nature, Games with Nature, Markov Decision Problem, Entrepreneurship, Nonnormal Probability Distributions, Extreme Volatility, Bayesian Inference, Bayes Factors, Markov Chain Monte Carlo Algorithm

JEL Classifications: D8, C1

Working Paper Series