NONPARAMETRIC IDENTIFICATION OF THE MIXED HAZARD MODEL USING MARTINGALE-BASED MOMENTS
证明了混合风险模型的非参数可识别性，利用鞅结构构造矩条件，提出广义矩估计方法，适用于含时变协变量和删失事件的数据。
Nonparametric identification of the Mixed Hazard model is shown. The setup allows for covariates that are random, time-varying, satisfy a rich path structure and are censored by events. For each set of model parameters, an observed process is constructed. The process corresponding to the true model parameters is a martingale, the ones corresponding to incorrect model parameters are not. The unique martingale structure yields a family of moment conditions that only the true parameters can satisfy. These moments identify the model and suggest a GMM estimation approach. The moments do not require use of the hazard function.