Abstract: A computer intensive resampling technique called the method of Pao-Zhuan Yin-Yu is systematically developed as an alternative to the bootstrap and the jackknife. The method is sequential parametric resampling scheme which searches for an optimal estimator (in the minimum variance unbiased estimator sense) and provides an estimate for the variance of the optimal estimator. Several numerical examples are given, including inference for a coefficient in an autoregressive model where the observations are dependent. Numerical results show that the method performs extremely well in almost all cases.
Key words and phrases: Pao-Zhuan Yin-Yu, resampling, contourization, empirical conditional expectation, Rao-Blackwell theorem, bootstrap.