RECURSIVE DIFFERENCING FOR ESTIMATING SEMIPARAMETRIC MODELS

Controlling the bias is central to estimating semiparametric models. Many methods have been developed control in conditional expectations while maintaining a desirable variance order. However, these typically do not perform well at moderate sample sizes. Moreover, and perhaps related their performance, nonoptimal windows are selected with undersmoothing needed ensure appropriate In this paper, we propose recursive differencing estimator for expectations. When method combined targeting derivative of expectation, able obtain asymptotic normality under optimal windows. As suggested by structure recursion, wide variety triple index designs, proposed performs much better sizes than regular or higher-order kernels local polynomials.