Model averaging based on weighted generalized method of moments with missing responses

Model averaging based on the least squares estimator or the maximum likelihood estimator has been widely followed, while model averaging based on the generalized method of moments is almost rarely addressed. This paper is concerned with a model averaging method based on the weighted generalized method of moments for missing responses problem. The weight vector for model averaging is obtained via minimizing the leave-one-out cross validation criterion. With some mild conditions, the asymptotic optimality of the proposed method in the sense that it can achieve the lowest squared error asymptotically is proved. Some numerical experiments are conducted to evaluate the proposed method with the existing related ones, and the results suggest that the proposed method performs relatively well.