CFE 2019: Start Registration
View Submission - CMStatistics
B0636
Title: The identifiability of the copula competing risks model under exclusion restrictions Authors:  Ralf Wilke - Copenhagen Business School (Denmark) [presenting]
Ming Sum Simon Lo - City University of Hong Kong (Hong Kong)
Abstract: Competing risks duration models are routinely applied in many disciplines such as biostatistics, mechanical engineering, economics and other social sciences. A convenient and general way to describe risk dependencies is to model the joint distribution by a copula that links the marginal distributions of latent competing durations. While the non-identifiability of the competing risks model complicates informative empirical analysis, a series of contributions has obtained identification results under different sets of restrictions. We develop a new identification result under exclusion restrictions. We show that the existence of exclusion restrictions ensures model identification under mild additional restrictions. While exclusion restrictions are common in econometric models, previous identification results for competing risks models do not rely on them.