Bootstrapped insights into empirical applications of stochastic dominance

Bootstrapping, a very versatile statistical technique, significantly amplifies the understanding and success of empirical applications of stochastic dominance. Its ability to calculate the standard deviations of order statistics reveals the uncertainty of the critical estimates of the tails of cumulative density functions. Understanding this uncertainty reveals why a wide variety of tail shapes all cause a notable loss in power for stochastic dominance tests. Simulations show that the smoothing inherent in bootstrapping can significantly increase the power of tests when dominance exists in the population.