Identifiability and estimation of marginal survival functions for dependent competing risks assuming the copula is known

We show that if the copula between the two marginal distributions is known then the competing risks data is sufficient to identify the marginal survival function. An estimator is constructed to estimate the marginal survival functions based on an assumed copula. This estimator is shown to be consistent and to reduce to the Kaplan-Meier estimator when the death and censoring times are independent. Results of a simulation study are presented to compare this estimator to other estimators.