Abstract
In this paper, a new form of the estimator of unconditional autoregressive model of order one UAR(1) with missing observations has been derived by using unconditional ordinary least squares (UOLS) and unconditional modification weighted symmetric (UMWS) estimators in two cases for unconditional autoregressive model with missing observations when the initial value x(1) similar to N(0, sigma(2)/1-rho(2)) Also, a modification of the formula of weighted symmetric of UAR(p) model with missing observations has been suggested as an extension of Park and Fuller [15], which provides an exact formula for the bias of the parameter estimator of the first order autoregressive process for (UOLS) and (UMWS) estimators. A comparison between (UOLS), (UMWSI) and (UMUSII) estimators using unconditional AR(1) model with missing observation is conducted through a Monte-Carlo simulation at various sample sizes and different proportions of missing observations considering the absolute bias as a criterion of comparison.