Title: A two-stage DEA with partial least squares regression model for performance analysis in healthcare in Algeria

Authors: Hassiba Djema; Mohamed Djerdjouri

Addresses: Ecole des Hautes Etudes Commerciales, 11 Chemin Doudou Mokhtar, Ben Aknoun, 16170 Alger, Algeria. ' School of Business and Economics, State University of New York, 101 Broad Street, Plattsburgh, New York 12901, USA

Abstract: Governments in many developing countries are overwhelmed by the inefficiencies of their healthcare facilities. This is the main cause of increasing healthcare costs. A major challenge faced by policy makers and administrators is how to assess and identify these inefficiencies. In this paper, we discuss measuring operational efficiency of public hospitals in Algeria using a two-stage DEA and partial least squares regression model. The paper will present results of a pilot study to evaluate the technical and scale efficiency of a sample of 174 hospitals. First, the Kohonen algorithm is used for the classification of the dataset. Then, a two-stage DEA has been employed. In the first stage the model calculates an efficiency score for each hospital. This helps to identify efficient as well as inefficient hospitals in the set. The information provided includes technical and scale efficiency levels, a measure of possible input reductions and output improvements, and identification of appropriate benchmarks. In the second stage the efficiency scores are used as the dependent variable to investigate the determinants of efficiency. This is done using the partial least squares (PLS) method to analyse factors which explain efficiency in the healthcare sector.

Keywords: partial least squares regression; data envelopment analysis; DEA; Kohonen algorithm; linear programming; operational efficiency; public hospitals; healthcare costs; performance evaluation; Algeria.

DOI: 10.1504/IJADS.2012.046506

International Journal of Applied Decision Sciences, 2012 Vol.5 No.2, pp.118 - 141

Published online: 09 Aug 2014 *

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