Resumo

LOPEZ GARCIA, María del Rosario; RAMIREZ VALVERDE, Gustavo; RAMIREZ VALVERDE, Benito  e  TERRAZAS GONZALEZ, Gerardo H.. Shrinkage estimators in simultaneous equation models for the analysis of the beef market in México. EconoQuantum [online]. 2019, vol.16, n.1, pp.103-123. ISSN 2007-9869.  https://doi.org/10.18381/eq.v16i1.7157.

The main objective in this work was to use LASSO regression (Least Absolute Shrinkage and Selection Operator) as a method of instrument selection in the estimation of two-stage least squares (MC2E) in a system of simultaneous equations proposed to perform an econometric analysis of the market of beef in Mexico in the period 1972-2011. A determining factor in the performance of the estimators is the degree of correlation of the instruments with the endogenous variables in the first stage. When the instruments are weak, the MC2E estimators are inconsistent, biased, and with large variances; Asymptotic results fail even with large samples. The results show that using LASSO can select relevant instruments, and have better estimators that result in a better policy design.

Palavras-chave : Weak instruments; LASSO regression; two stage least squares; C30; C61; C13.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )