Regulatory and Environmental Effects on Public Transit Efficiency. A Mixed Dea-Sfa Approach

Primary tabs

Loading the Internet Archive BookReader, please wait...


Regulatory and Environmental Effects on Public Transit Efficiency. A Mixed Dea-Sfa Approach


WP 13/2006; The aim of this paper is to account for the impact of statistical noise and exogenous regulatory and environmental factors on the efficiency of public transit systems in a DEA-based framework. To this end, we implement a three-stage DEA-SFA mixed approach based on Fried et al. (2002) using a 1993-1999 panel of 42 Italian public transit companies. This allows us to decompose input-specific DEA inefficiency measures into three components: exogenous effects, pure managerial inefficiency, and statistical noise. First, the initial evaluation of producer performance is carried out using conventional variable returns to scale DEA (Banker et al., 1984). Second, a SFA approach (Battese and Coelli, 1992) is used to regress single input slacks on subsidies regulation (cost-plus versus fixed-price contracts) and a set of environmental variables including network speed and user density. Finally, third stage re-runs DEA on inputs purged of both exogenous effects and statistical noise. Results are such that adjusting for the type of regulatory scheme, environmental conditions, and statistical noise increases average efficiency in the industry and reduces dispersion among firms. Furthermore, the implementation of fixed-price subsidies is found to enhance efficiency in the usage of “drivers” and “materials and services” inputs. Such a result sheds some light on the determinants of input-specific efficiency differentials in the industry, improving the existing evidence on mean overall cost efficiency (e.g. Gagnepain e Ivaldi, 2002; Piacenza, 2006). As a policy implication, it is confirmed the relevance of regula tion aimed at replacing cost-plus subsidization mechanisms with high-powered incentive contracts as well as improving operating conditions of public transport networks.


CNR Ceris