Home
Comparing bayesian and corrected least squares

Fco. Javier Ortega Irizo
Departamento de Economía Aplicada I
Universidad de Sevilla
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it

Jesús Basulto Santos
Departamento de Economía Aplicada I
Universidad de Sevilla
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it

José Antonio Camúñez Ruíz
Departamento de Economía Aplicada I
Universidad de Sevilla
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it

  • Abstract
    In the econometric approach to deterministic frontier production models, the use of maximum likelihood estimation has major problems, because the model violates the usual regularity conditions that allow to establish the desirable asymptotic properties of the estimators. To avoid this diffi- culty, we may use other approachs. One of the methods most widely used is the corrected least-squares estimator. Alternatively, we can perform the Bayesian estimation using Gibbs sampling. In this paper, we make a comparative study of both approachs using simulation methods. We will conclude that the bayesian estimator has better properties in terms of bias and mean squared error, especially for the intercept term; this fact can result very important to estimate the individual efficiencies, which is one of the main objectives of these models.
  • Keywords: Production models, Bayesian Inference, Gibbs sampling, Corrected least-squares estimator.
  • AMS Subject classifications: 90B30,62F15,62F30,62J99.
  • PDF PDF (511.16 KB)