In the context of competing IV econometric models and estimators, we demonstrate a semiparametric Stein-like estimator (SSLE) that, under quadratic loss, has superior risk performance. The method eliminates the need for pretesting to decide whether covariate endogeneity is present and makes use...