An empirical analysis of airport efficiency: the Croatian case

Abstract

The main objective of this paper is to present an empirical application of the DEA method, by proposing an adequate model for determining the relative efficiencies of seven Croatian airports, and by analysing the obtained scores. To highlight different crucial aspects of airport performance in Croatia, four indicators are used for the period 2009-2014: personnel costs/airport throughput unit (ATU), total expenditures excluding personnel costs/ATU and total assets/ATU as input variables, while the output variable is total revenue/ATU.
To overcome certain limitations associated with the relationship between the number of the observed entities and the number of employed variables, and to provide dynamic efficiency results that usually reflect reality better than static ones, window analysis is used as an extension of basic input-oriented DEA models. In general, the findings indicate that, over the observed period, performance rankings change, and, with the exception of the last observed year, the relative performance of Croatian airports is gradually declining. Consequently, the airports of Split, Pula and Zadar were found to be efficient (“best practice” airports) in the four years, and the airports of Zagreb and Osijek in one single year. Based on the efficiency score averaged across the observed period, Split turned out to be most efficient whilst Osijek appeared to be least efficient. Total assets per ATU are identified as the most significant source of inefficiency.
Using both constant and variable returns to scale assumptions, this paper is the first to decompose the technical efficiency of Croatian airports into two components – pure technical efficiency, which reflects the ability of an airport to obtain maximal outputs at an optimal scale, and scale efficiency, which reflects the distance of an observed airport from the most productive scale size. The significance of these results is in the fact that they offer the possibility of directly identifying inefficiency causes, and can serve as a basis for an a posteriori correction of previously made disadvantageous decisions.