CFE 2020: Start Registration
View Submission - CMStatistics
B0504
Title: Model selection and goodness of fit tests for conditional copula models Authors:  Jacco Wielaard - University of Twente (Netherlands) [presenting]
Alexis Derumigny - University of Twente (Netherlands)
Abstract: Conditional copulas, also known as copula models with covariates, are models that describe the dependence between several random variables of interest, conditionally to some known explanatory variables. It is often assumed that these conditional copulas belong to a fixed parametric family, with a (conditional) parameter depending on the explanatory variables. We propose several goodness-of-fit tests for the assumption of good specification of a parametric conditional copula model, without any constraint on the conditional margins. Two such tests that use different bootstrap resampling procedures are compared in a simulation study. Finally, these tests are applied to a dataset of financial returns.