Semiparametric Estimation of Censored Transformation Models.

Many widely used models, including proportional hazards models with un- observed heterogeneity, can be written in the form (Y ) = min[ 0 X + U; C], where is an unknown increasing function, the error term U has unknown distribution function and is independent of X, C is a random censoring threshold, and U and C are conditionally independent given X. Thispaper develops new n 1=2 -consistent and asymptotically normal semiparametric esti- mators of and which are easier to use than existing estimators. Moreover, Monte Carlo results suggest that the mean integrated squared error of predic- tions based on the new estimators is lower than for existing estimators.