Comparisons of the some estimators for the transcendental logarithmic (translog) model*


ÖRK ÖZEL S., ÇABUK H. A.

COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION, 2021 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Review
  • Publication Date: 2021
  • Doi Number: 10.1080/03610918.2021.1960999
  • Journal Name: COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Applied Science & Technology Source, Business Source Elite, Business Source Premier, CAB Abstracts, Compendex, Computer & Applied Sciences, Veterinary Science Database, zbMATH, Civil Engineering Abstracts
  • Keywords: Multicollinearity, Translog model, Generalized maximum entropy estimator, Restricted generalized maximum entropy estimator, Ridge estimator, RIDGE REGRESSION, DEMAND
  • Çukurova University Affiliated: Yes

Abstract

Flexible functions in economics are functions that do not require a priority restrictor about various substitution elasticities. One of these functions is Christensen et al. proposed by the transcendental logarithmic (translog) function. Translog model suffers from the multicollinearity problem since the squares are added to the model and cross products of variables. Since classical estimators can not be used under multicollinearity, biased estimators can be used to overcome the problem. In this study; ridge estimator, restricted ridge (RRidge) estimator, generalized maximum entropy (GME) estimator, restricted GME (RGME) estimator, ordinary least squares (OLS) estimator and restricted OLS (ROLS) estimator are compared according to the mean squared error (MSE) criteria. We compare aforementioned estimators with Monte Carlo simulation studies and a numerical example. In conclusion, GME and RGME estimators are decided as the most efficient estimators than rest of estimators in terms of MSE criteria when appropriate support matrices and restrictions are selected.