Title: An algorithm for variable selection in firm valuation models

Authors: Fernando Garcia, Francisco Guijarro, Ismael Moya

Addresses: Facultad de Administracion y Direccion de Empresas, Universidad Politecnica de Valencia, Camino de Vera s/n, 46022 Valencia, Spain. ' Facultad de Administracion y Direccion de Empresas, Universidad Politecnica de Valencia, Camino de Vera s/n, 46022 Valencia, Spain. ' Facultad de Administracion y Direccion de Empresas, Universidad Politecnica de Valencia, Camino de Vera s/n, 46022 Valencia, Spain

Abstract: Firm valuation constitutes a classical topic in finance research. Variable selection in valuation models is a problem addressed by many researchers mainly focused on multivariate analysis. This study proposes a methodology for dealing with the problem of explicative variable selection in financial models for firm valuation. With a view to its eventual automation, it is presented in three consecutive steps. These ensure that the models obtained are parsimonious, and that there is control of their degree of multicollinearity without sacrificing their explicative power. The method combines several statistical techniques, notably simple regression analysis and principal components analysis. To test the proposal, we used a database of stock market and accounting information relating to firms quoted on the Madrid Stock Exchange in 2002 and 2003.

Keywords: firm pricing; business performance modelling; financial statements; multicollinearity; principal components analysis; PCA; regression; variable selection; firm valuation models; Spain.

DOI: 10.1504/IJBPSCM.2009.030639

International Journal of Business Performance and Supply Chain Modelling, 2009 Vol.1 No.2/3, pp.144 - 161

Published online: 29 Dec 2009 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article