Efficiency of Public Expenditures: The Case of Croatian Counties

Public sector efficiency has undergone much scrutiny in literature, especially because of limited public resources for financing public needs at national and sub-national levels. Studies on public sector efficiency usually include country comparisons or analyse the efficiency of public expenditures in a particular sector (e.g. education and health care) (Afonso and Aubyn, 2004 ; Lockheed and Longford, 1989 ; Hollingsworth, et al., 1999). There are a number of studies which look at public expenditure efficiency of local and regional governments (Melkers and Willoughby, 2005 ; Afonso and Fernandes, 2008 ; De Borge and Naper, 2006 ; Afonso and Scaglioni, 2005). The main conclusion from the conducted research is that cost efficiency increases parallel to the size of local government units. The objective of this paper is to explore the efficiency of local government expenditure, with special emphasis on Croatian counties in the period 2005 2014. Croatia consists of twenty counties. Most of the existing public functions in Croatia remain under the jurisdiction of the central state. This is the case even with public functions that are partially decentralised (education, health care and social welfare), since responsibility and financing are divided between the central, regional and local levels of government. The majority of public resources in local budgets is spent on providing a range of public services and ensuring access to education, health care and social welfare to citizens. These three public functions account for almost two thirds of total sub-national government spending. To evaluate the efficiency of public expenditures and the average deviation level in Croatian counties in the period 2006 2015, the Data Envelopment Analysis (DEA) methodology was used to process data. Farnell laid the foundations for the DEA model in 1957, which was later developed by Charnes, et al. (1978). This method identifies the most efficient units in a given set, without assuming any type of functional relationship between inputs and outputs. Our analysis covered twenty Croatian counties (Zagreb, Krapina-Zagorje, Sisak-Moslavina, Karlovac, Varazdin, Koprivnica-Krizevci, Bjelovar-Bilogora, Primorje-Gorski Kotar, Lika- Senj, Virovitica-Podravina, Pozega-Slavonija, Brod-Posavina, Zadar, Osijek-Baranja, Sibenik- Knin, Vukovar-Srijem, Split-Dalmatia, Istria, Dubrovnik-Neretva, Medimurje) and three public functions (education, health care and social welfare). The DEA model inputs comprised total expenditure for education, health care and social welfare of each county, while outputs consisted of population density and gross domestic product per capita in each county. To obtain meaningful results, additional input and output variables were used in this analysis. We expected the DEA model to identify significant differences in the efficiency of expenditure on education, health care and social protection between Croatian counties. This analysis contributes to research evidence that show that larger counties are more efficient in public spending than smaller ones. Finally, this paper identifies some of the main causes of existing inefficiencies in overall public spending at county level, inefficiencies in public expenditure in each of the examined sectors (education, health care, social welfare), and gives recommendations aimed at increasing the efficiency of public spending at county level.