In search of the optimal size for local government: an assessment of economies of scale in local government in Croatia

ABSTRACT

This paper brings an empirical assessment of economies of scale in local government units in Croatia. Using cross-section OLS models, we found statistically significant U-shaped relationship between local per capita expenditures and population size, while controlling for various demographic, socio-economic and institutional factors. The choice of control variables in the paper is based on the existing empirical literature but also includes factors capturing the specificities of Croatian economy and local government institutional setup. Using estimated regression coefficients we calculated the optimal size of local units and showed that population size is below optimal in 72% of cities and 76% of municipalities (based on median results).

KEYWORDS:

• Local government expenditures
• local government
• economies of scale
• territorial fragmentation
• Croatia

Supplemental data

Supplemental data for this article can be accessed here.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. To be precise, there are 128 cities in Croatia but City of Zagreb (capital of Croatia) has a status of both a city and a county. As we explain below, due to this dual status and other specific characteristics (e.g., size) City of Zagreb is excluded from our analysis as it could notably affect our results.

2. Rural index is constructed using the OECD classification of local communities. The index differentiates between rural, intermediate and urban regions, on the basis of the share of rural local communities i.e., communities with less than 150 inhabitants per square kilometres. A region is defined as predominantly rural if more than 50% of the population lives in rural communities; it is considered intermediate if between 15 and 50% of the population lives in urban communities, and it is classified as predominantly urban if less than 15% of the population lives in rural communities (Bartolini 2015).

3. Bikker and Van Der Linde (2016) Breunig and Rocaboy (2008); Drew and Dollery (2016); Drew, Kortt, and Dollery (2012); Hortas-Rico and Ríos (2020); Matějová et al. (2014); Pevcin (2012); Soukopová et al. (2014) and Tavares and Rodrigues (2015).

4. Allers and Geertsema (2016); Blesse and Baskaran (2016); Blom-Hansen et al. (2016); Hanes (2015); Drew, Kortt, and Dollery (2016); Rengewertz (2012); Steiner and Kaiser (2017) and Turley et al. (2018).

5. Our decision to rely on cross-section data rather than panel data stems from the fact that most part of the overall variation in our data sample can be found ‘between’ local units. Variation ‘within’ units (i.e., over time, from one year to the next year) in the sample is very small and not sufficient for analysis. In addition, main variables of interest (population and expenditures) have a high degree of persistency in short time periods. Furthermore, this way the results are comparable to other related studies which also mostly use cross-section techniques, as it is shown in the Appendix 2.

6. To facilitate the comprehension of our empirical strategy and interpretation of results, we provide a brief information on the insitutional framework of local government units in Croatia (structure of revenues, exepnditures, financing etc.) in Appendix 1.

7. We divide the sample of all local government units into cities and municipalities, which differ substantially in the population density. As there are notable differences in population density within each of these two groups, we find it necessary to stratify the sample additionally.

8. Cities and municipalities in Croatia are responsible for covering compensation of employees, current and capital expenditures in pre-school education, while cities are also responsible for covering current and capital expenditures in primary school education.

9. Based on Tax Administration PIT data.

10. Although the reader could assume there could be high level of correlation between tourist activity and taxable income, there is a weak link between taxable income (as a tax base for personal income tax) and tourism in Croatia. Many local units in Croatia with high share of tourism have a relatively low level of employment and activity of local population (this is why some authors see tourism in Croatia as a case of ‘resource curse’). Most workers are only seasonal workers from other counties where they pay taxes. In addition, taxes on short term rental for tourist purposes in Croatia are mostly paid as lump sum taxes (which are paid to the local unit of taxpayer residence, no matter where the property is located – many properties are vacation houses owned by people whose local unit of residence is different from the local unit where tourist activity takes place) and they are not related to taxable income in PIT system.

11. Data on population density and share of older and younger population are obtained from the last official population census in 2011 while data on tourist overnight stays are available only from 2010 due to changes in the methodology.

Additional information

Notes on contributors

Antonija Buljan

Antonija Buljan is a Postdoctoral Researcher at the Faculty of Economics and Business of University of Zagreb, Croatia. Her research interests are public sector economics, fiscal policy and taxation.

Sandra Švaljek

Sandra Švaljek is Deputy Governor of Croatian National Bank in charge for central banking operations and communication. She worked as researcher at the Institute of Economics, Zagreb from 1992 to 2018 and became a senior research fellow. Her research interests include public sector economics, economic policy and climate change.

Milan Deskar-Škrbić

Milan Deskar-Škrbić is a Senior Economic Researcher-Advisor in the Modelling Department of Croatian National Bank and a lecturer at Zagreb School of Economics and Management. His research interests are public sector economics, economic policy, applied macroeconometrics and macroeconomics of small open economies.