Nonparametric Estimation of Bivariate Mean Residual Life for Right‐Censored Observations

ABSTRACT The bivariate mean residual life (BMRL) is a key characteristic of a pair of lifetimes given that they exceed given values. Its nonparametric estimation is well known for the case of directly observed pairs of lifetimes. For right‐censored observations, typical in survival analysis, the methodology of estimating BMRL is known for the univariate right‐censoring when the two lifetimes are censored by the same censoring variable. The paper considers a general right‐censoring case, and proposes the methodology of a data‐driven estimation of BMRL based on estimation of the conditional residual survival function which is of interest on its own. The supporting simulations are presented.