ABSTRACT The article considers the predictive efficiency of some improved estimators of coefficients in linear regression models, constructed with the help of the Stein rule when there are some stochastic linear restrictions on the vector of unknown parameters. Assuming the distribution of disturbances to be not necessarily normal, an attempt is made to compare the performance of improved and mixed regression estimators in terms of their respective risks.
Madhulika Dube hasn't uploaded this paper.
Let Madhulika know you want this paper to be uploaded.