Shrinkage Estimators for Prediction Out-of-Sample: Conditional Performance

We find that, in a linear model, the James–Stein estimator, which dominates the maximum-likelihood estimator in terms of its in-sample prediction error, can perform poorly compared to the maximum-likelihood estimator in out-of-sample prediction. We give a detailed analysis of this phenomenon and discuss its implications. When evaluating the predictive performance of estimators, we treat the regressor matrix in the training data as fixed, i.e., we condition on the design variables. Our findings contrast those obtained by Baranchik (1973 Baranchik , A. J. ( 1973 ). Inadmissibility of maximum likelihood estimators in some multiple regression problems with three or more independent variables . Ann. Statist. 1 ( 2 ): 312 – 321 .[Crossref], [Web of Science ®] , [Google Scholar]) and, more recently, by Dicker (2012 Dicker , L. ( 2012 ). Dense signals, linear estimators, and out-of-sample prediction for high-dimensional linear models. arXiv:1102.2952 [math.ST].  [Google Scholar]) in an unconditional performance evaluation.