Efficient estimation of semiparametric copula models for bivariate survival data

JOURNAL OF MULTIVARIATE ANALYSIS

卷:123

Article

Cheng, Guang¹  Zhou, Lan²  Chen, Xiaohong³  Huang, Jianhua Z.² 

[1] Purdue Univ, Dept Stat, W Lafayette, IN 47907 USA

[2] Texas A&M Univ, Dept Stat, College Stn, TX 77843 USA

[3] Yale Univ, Dept Econ, New Haven, CT 06520 USA

关键词: Semiparametric copula model;    Bivariate survival data;    B-spline;    Efficiency;    Consistent covariance estimation;   

DOI  :  10.1016/j.jmva.2013.10.008

来源: Elsevier

【 摘 要 】

A semiparametric copula model for bivariate survival data is characterized by a parametric copula model of dependence and nonparametric models of two marginal survival functions. Efficient estimation for the semiparametric copula model has been recently studied for the complete data case. When the survival data are censored, semiparametric efficient estimation has only been considered for some specific copula models such as the Gaussian copulas. In this paper, we obtain the semiparametric efficiency bound and efficient estimation for general semiparametric copula models for possibly censored data. We construct an approximate maximum likelihood estimator by approximating the log baseline hazard functions with spline functions. We show that our estimates of the copula dependence parameter and the survival functions are asymptotically normal and efficient. Simple consistent covariance estimators are also provided. Numerical results are used to illustrate the finite sample performance of the proposed estimators. (C) 2013 Elsevier Inc. All rights reserved.

【 授权许可】

Free