A1485
Title: Identifiability and estimation of the competing risks model under exclusion restrictions
Authors: Ralf Wilke - Copenhagen Business School (Denmark)
Munir Hiabu - University of Copenhagen (Denmark)
Ming Sum Simon Lo - United Arab Emirates University (United Arab Emirates) [presenting]
Abstract: The non-identifiability of the competing risks model requires researchers to work with restrictions on the model to obtain informative results. A new identifiability solution is presented based on an exclusion restriction. Exclusion restrictions are popular in many areas of applied research, and it appears natural to use them for the identifiability of competing risk models. By imposing the exclusion restriction coupled with an Archimedean copula, any parametric restriction is avoided on the marginal distributions. A semiparametric estimation approach is introduced for the nonparametric marginals and the parametric copula. The simulation results demonstrate the usefulness of exclusion restrictions in survival or duration analysis, as the degree of risk dependence can be estimated without parametric restrictions on the marginal distributions.
