Network data envelopment analysis and an application on railways in Turkey

Data envelopment analysis (DEA) is a non ? parametric method that is used for measuring relative efficiencies of decision making units by using various input and output factors. DEA is usually utilized to evaluate the efficiency of a decision making unit (DMU) which converts inputs to final outputs. Such a DMU is called as a black ? box where it is not necessary to know the processes or the linkages inside the DMU. In this type of a black ? box DEA approach, details of an internal process of the DMU are ignored and therefore the sources of inefficiency can not be identified. Traditional studies about DEA view the system as a whole and ignore the individual processes within the system.DEA researchers have recently developed some network approaches by analyzing the black ? box where the inputs entering the box and outputs coming out of the box is only taking into account. These efforts led to a new methodology that is called as Network Data Envelopment Analysis. Network DEA provides a flexible way to model DEA problems for specific applications. Network DEA provides meaningful links between sub ? DMU?s, helps to measure the efficiencies of internal processes and therefore allows us to look inside a complex DMU with multiple nodes.This study aims to explain this new methodology and introduce some different approaches of it. Traditional DEA and its main models with Network DEA and its various approaches are presented. As an application Turkish Railways? efficiency scores between 1981 ? 2010 are investigated with both Network VZA and traditional VZA methods, and results are compared.