Asymptotic aspects of ordinary ridge regression

Ordinary ridge regression is considered as an alternative to least squares regression for the analysis of multicollinear data. When the ridge factor is estimated from the data, the authors obtain the asymptotic distribution of several ridge factors, and propose a new ridge factor based on asymptotic considerations. The mean squared errors of operational ordinary ridge regression estimators are evaluated asymptotically, and they propose several statistics useful for the selection of ridge factors.