The total successive raise procedure: analysis of estimators obtained from the canonical and original model.

J. García Pérez, C. García García, R. Salmerón Gómez

Different methods have been developed to avoid the instability of estimates derived from collinearity. The ridge regression (RR) introduces a same factor k for each explanatory variable even when the model presents only one variable with a variance inflator factor higher than the established threshold. Alternatively, the raise procedure is more versatile and flexible than the ridge regression and may mitigate collinearity only raising one vector. Garcia et al. (2019) present the total successive raise (TSuR) and show that it leads to the generalized ridge regression (GRR) if all the columns of matrix X are raised from the canonical model. The estimator obtained raising all the columns of matrix X from the original model is analyzed in this work. The contribution is illustrated with an empirical application.

Palabras clave: collinearity raise regression

D. R. Santos Peñate, C. Florido de la Nuez, R. Suárez Vega

J. P. Ramos Requena, M. A. Sánchez Granero, K. Balladares, J. E. Trinidad Segovia