Estimation of Zellner-Revankar Production Function Revisited
Sudhanshu K. Mishra
affiliation not provided to SSRN
Arnold Zellner and Nagesh Revankar in their well-known paper "Generalized Production Functions" [The Review of Economic Studies, 36(2), pp. 241-250, 1969] introduced a new generalized production function, which was illustrated by an example of fitting the generalized Cobb-Douglas function to the U.S. data for Transportation Equipment Industry. For estimating the parameters of their production function, they used a method in which one of the parameters (theta) is chosen at the trial basis and other parameters relating to elasticity and returns to scale are estimated so as to maximize the likelihood function. Repeated trials are made with different values of theta so as to obtain the global maximum of the likelihood function.
In this paper we show that the method suggested and used by Zellner and Revankar (ZR) may easily be caught into a local optimum trap. We also show that the estimated parameters reported by them are grossly sub-optimal.
Using the Differential Evolution (DE) and the Repulsive Particle Swarm (RPS) methods of global optimization, the present paper re-estimates the parameters of the ZR production function with the U.S. data used by ZR. We find that the DE and the RPS estimates of parameters are significantly different from (but much better than) those estimated by ZR. We also find that the returns to scale do not vary with the size of output as reported by ZR. A Fortran program for estimation of the ZR function with RPS and DE has also been appended
Number of Pages in PDF File: 20
Keywords: Zellner-Revankar production function, maximum likelihood, global optimization, Repulsive Particle Swarm, Differential Evolution, U.S. Data, Transport Equipment Industry, variable Returns to scale, sub-optimality, Fortran Program
JEL Classification: C13, C15, C16, C39
Date posted: December 11, 2006
© 2015 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo6 in 0.375 seconds