Method of construction of contently interpreted recreational models in conditions of multicollinearity

Abstract

Annotation: the construction of linear regression models without distorting the meaningful effects of multicollinearity.

Discussion: the issue of econometric modeling in a situation where there is multicollinearity continues to be relevant, despite a rather large arsenal of recommendations. Practically all recommendations are oriented towards solving computational problems related to the degeneracy of the matrix of normal equations. We can assume that problems of a computational nature are basically solved. But the problem related to the meaningful meaning of the final version of the econometric model is not considered in any of the methods and, as before, remains unresolved. Results: An original way of constructing linear regression models is proposed for cases when the effects of multicollinearity are manifested by distortions of the meaningful meaning of the factor influence on the modeled index. The conducted empirical studies confirmed the effectiveness of the proposed method.