Title: Multiple local optima for seemingly unrelated regression models
Authors: Norman K. Womer
Addresses: College of Business Administration, The University of Missouri – St Louis, One University Blvd, St Louis, MO 63121, USA
Abstract: This paper provides an example of multiple local maxima to the likelihood function for a seemingly unrelated regression (SUR) model with a cross equation restriction. Since this is the least complex of the various structural equations models (SEM) and since maximum likelihood is often the preferred technique for SEM the problem of multiple local maxima is expected to be pervasive.
Keywords: point estimation; maximum likelihood; structural equation modelling; SEM; cross equation restrictions; specification test; multiple local optima; seemingly unrelated regression models.
DOI: 10.1504/IJCEE.2016.073343
International Journal of Computational Economics and Econometrics, 2016 Vol.6 No.1, pp.44 - 55
Received: 21 May 2014
Accepted: 21 Feb 2015
Published online: 30 Nov 2015 *