We consider two sequential problems: minimum risk problem and bounded risk problem under linex loss function. We shall show that the least square estimate in linear regression model is improved by another estimator asymptotically.