Eficiency of Water Utilities: Does Local Public Ownership Matter? Evidence from Italy

In the last decades the governance of local public services has been the focus of administrative reforms to pursue an improvement in the eciency and productivity levels coping with increasing constrains in nancial re- sources. In this scenario, changes in governance and ownership of local public services shave occurred with a shift from public to private owner- ship (Pollitt and Bouckaert, 2011). Our paper focuses on the eciency analysis of Italian water utilities. The attention toward this type of utili- ties is motivated by two reasons: rstly, in recent years the Italian water industry has been at the centre of debate about the possibility of priva- tization; secondly in 2011 the change in the regulation conferred powers to the Italian regions to decide on which administrative body should be the provider of local water services. In addition, the law does not exclude that regions themselves could exercise this power. Thus, the evaluation of the eciency and productivity analysis could have an important policy implication for regions and other water industry regulators. Moreover, the debate about the determinants of the performance of public service organization is a long lasting one, especially with regard to the dichotomy between public and private ownership (Andrews et al. 2011). On the bases of this background, the paper addresses the following research questions: Is the ownership (public, mixed, or private) a signi cant determinant of the eciency level of water utilities? Can size and geographical location be regarded as moderators of the relationship between ownership and ef- ciency? The method applied in the paper combines Data Envelopment Analysis (DEA) with cluster analysis. This method has the advantage to rank water utilities on bases of their eciency score without requir- ing any assumption on the distribution function of the data and to work well with small samples (Coelli et al 2005). The cluster analysis allows classifying the utilities ex-post instead of ex-ante, thus it provides more rigours results (O'Donnell et al. 2008, Balaguer-Coll et al. 2013). The results show that the ownership has not a signi cant e ect on eciency per se, however the combination of size and geographical location provides interesting insights on the di erence observed in the performance.