Group for Research in Decision Analysis

Bootstrapping methods and semiparametric tests

Bruno Rémillard Professor, Department of Decision Sciences, HEC Montréal, Canada

In most applications, testing for goodness-of-fit or testing other functional hypotheses is complicated by nuisance parameters and the limiting distribution of most test statistics depend on unknown parameters. In this talk I will show that some Monte Carlo methods can be used to estimate P-values for semiparametric test statistics based on empirical processes. Examples will include parametric bootstrap for goodness-of-fit test for copula families and multiplier methods for testing equality between two copulas.