The Capacities and Sustainability of Croatian Cities in Performing Municipal Services

Abstract

The system of local and regional self-governments in the EU is characterized by an extremely large number of small units. The situation in Croatia, a country with only 3.87 million inhabitants, is no different. There are 576 local and regional self-government units. The basic feature of self-government in Croatia is the equal scope of authority in performing tasks, regardless of the number of inhabitants, the area they inhabit, or their ability to pursue sustainability objectives. In this paper, we analyze the capacities of performing municipal services in 127 Croatian cities and examine whether cities differ in their ability to perform public functions. Results show that cities can be grouped into three different clusters. The validity of the results on clusters was confirmed with an ANOVA, where a significant difference was found between all three clusters. Ranking cities according to the capacity for providing municipal services shows that cities with higher capacities are clustered together, implying that cities indeed differ in their ability to perform municipal services. Further, the results indicate that 80 percent of the cities belong to the cluster that consists of cities with a very low capacity for providing public services.

1. Introduction

The system of local and regional self-governments in the countries of the European Union is characterized by an extremely large number of small units [1] measured by the number of inhabitants and the area of the local government unit [2]. The situation in Croatia is no different. There are 576 local and regional self-government units organized in the territory of the Republic of Croatia, of which 556 are local government units (428 municipalities and 127 cities and the City of Zagreb, which has the status of a city and a county) and 20 regional self-government units (counties). The situation is further complicated by the high degree of natural-geographical diversity of Croatian space, which is also reflected in the fact that a significant part of the territory of the Republic of Croatia consists of areas with special development features. In addition to the large number of units, the system of local and regional self-governments is characterized by complexity in the organization of tasks from the self-governing scope. The basic feature of self-governments in Croatia is the equal scope of authority in performing tasks, regardless of municipalities and city sizes, measured by the number of inhabitants and the size of the area they occupy, population density, geographical location, degree of development, ability to pursue sustainability objectives, organization of administrative bodies and public services (organizational capacity), by the number of employees and state employees (human capacity) and income from taxes, aid, own, and dedicated income (financial capacity) for the performance of public services from the self-governing scope. In addition to its advantages, the system of authority defined in this way has numerous disadvantages. One of the most important shortcomings is the impossibility of performing public tasks from the self-governing scope and committing to sustainability objectives at the same level of quality in all municipalities and cities, given the different levels of development, developmental peculiarities, organizational, human and financial capacity, and the size of the local unit itself.

The existing level of quality of services provided by local and regional self-government units in Croatia is low, their capacities are not adequate for the work they are required to perform in accordance with the laws, and the level of transparency in the acquisition of income and spending of budget funds is not adequate [3]. A similar situation can be found in other European countries. Therefore, it does not come as a surprise that fragmentation at the local level of government in European countries is at the center of political attention [4], as well as in academic research [5]. In a number of European countries, reform measures have been taken that not only enabled but also encouraged the merger of local and regional self-government units to achieve their appropriate size and ensure the necessary conditions for them to be able to manage the full range of local self-government functions, i.e., perform tasks within the self-governing scope and at the same time reduce costs [6].

Even though there is an understanding that local self-government units in Croatia do not have equal capabilities to perform municipal services, there is a lack of analytical research that would corroborate this hypothesis and give insight into which units indeed demonstrate higher capabilities. Hence, in this paper, we are analyzing the capacities of performing municipal services in 127 Croatian cities. We are also interested in determining whether cities in Croatia can be grouped according to similar characteristics based on different indicators of providing municipal services. This will enable us to determine how well cities in Croatia are suited to facilitate an equal scope of provided services as well as to indicate how well they are equipped to achieve sustainable development.

In this research, we are focused only on cities. Croatian cities exhibit even more significant differences in terms of size, administrative, human, demographic, economic, and fiscal capacity, and the level of economic development than municipalities. Additionally, there is a (moderate) difference in the self-governing scope between cities and municipalities. Hence, mixing cities and municipalities in the analysis could blur the results. For the same reason, the City of Zagreb, which has the status of a city and a county and hence performs public services from both jurisdictions, is not included in the analysis.

In view of the distribution of cities by their population size, the average population size is 16,010 [7], indicating a dominance of small towns (54.3% of cities have fewer than 10,000 inhabitants) and a lack of medium-sized towns (50,000–100,000 inhabitants). There are only five cities with a population between 50,000 and 100,000 inhabitants. Excluding from the analysis is the city of Zagreb (which has the status of both a city and a county); Split (165,528 inhabitants) and Rijeka (109,533 inhabitants) are the two largest cities with more than 100,000 inhabitants. In contrast, Stara Gradiška is the smallest city in Croatia, having only 875 inhabitants.

The rest of the paper is organized as follows. The next section gives a brief literature review of the provision of public services. The third section gives an overview of the methodology employed and the data used in the analysis. Section 4 presents the results of the analysis. The last section offers conclusions.

2. Overview of Literature

In European cities, there is a trend of increasing population, while in rural areas, the population is decreasing. In terms of population migration, Croatia is no exception. According to Eurostat data [8], 40.2% of the European population lives in predominantly urban regions and 39.2% in the intermediate regions. Projections point to further growth of the population in cities in most European countries, including Croatia. According to United Nations [9], it is expected that 83.7 percent of the European population will live in cities by 2050, and the same source estimates that 71.3 percent of the population will live in cities in Croatia. Therefore, city authorities are faced with a growing demand for infrastructure and services and numerous challenges. After Croatia’s accession to the European Union in 2013, its economic development was largely driven by an increase in exports as companies turned to the large international market. This resulted in higher employment, higher wages, higher private consumption, and an overall increase in the economic development of cities located in the central part of the country (around the capital) and on the coast. However, many Croatian cities, especially medium-sized and small towns in the eastern and peripheral parts of the country, are faced with various problems, such as the migration of the workforce to highly developed countries and various economic and demographic challenges [10].

According to the Law on the Territories of Counties, Towns, and Municipalities in the Republic of Croatia [11], the entire territory of the Republic of Croatia is divided into 556 local government units (127 towns and 429 municipalities) and 21 counties (including the City of Zagreb which is both a town and county). A county is a unit of regional self-government, and it comprises towns and municipalities, while a town is a local self-government unit in which the county seat is situated, as well as every settlement having more than 10,000 inhabitants and representing an urban, historical, natural, economic, and social whole. Exceptionally, a settlement not meeting the conditions may also be determined as a town where the necessary reasons (historical, economic, or geographic) for it exist. In Croatia, the self-governing scope of tasks of local and regional units is determined by the Law on Local and Regional Self-Government [12]. According to this Law, local units (municipalities and cities) have the following responsibilities: organization of settlement and housing, spatial and urban planning, utility services, childcare, social welfare, primary health protection, elementary education, consumer protection, culture, sports and recreation, protection and improvement of environment, fire protection and civil protection, and local traffic in their area [13]. The local units in Croatia are different in terms of size, population, fiscal capacity, and other features, which raises the question of whether they can and how they provide public services to their citizens. For example, only larger cities in Croatia have enough budget resources to achieve positive fiscal capacities [14]. However, the inclusion of citizens in local decision-making is an important mechanism for improving the quality of local governance, including the allocation of budget resources. In that sense, participatory budgeting enables better allocation of budget resources according to citizens’ needs. Still, participatory budgeting in Croatian cities is not well developed; only some cities are involved in citizen participation through participatory budgeting [15]. Similarly, there are relatively large differences between Croatian cities in local budget transparency [16]. However, it has to be noted that some parts of Croatia have, for a long period of time, been losing population, which mostly includes municipalities but also cities [17]. These are mostly areas that have been historically lagging behind in development [18] and, for that reason, have more difficulties in providing public services. Hence, it does not come as a surprise that the quality of public services differs in different cities in Croatia.

Attitudes about who should provide public services have changed over time, starting back in 1956 when Tiebout considered that consumers “vote with their feet”. He believed that economic efficiency is achieved with the provision of local public services, given that consumers can choose where to live, which encourages competition among local communities, and that the choice of living location is a very clear sign of consumer preferences [19]. The existence of multiple levels of government allows authorities to vary the categories and levels of public services across different local units according to citizens’ preferences [20].

Wollmann [21] briefly describes the changes in providing public services in Europe in the past. In the late 19th century, public utilities and social services were, in Western European countries, provided by the state but already, at least in some parts, provided by local authorities, while the energy sector was nationalized. In Central East European countries, public and social services were mostly provided by central state or centrally controlled municipal units. Since the 1980s, various forms of private sector involvement in the provision of services have been noticed. For example, in the UK, the water sector was entirely privatized, and private providers also dominated in providing social services; France outsourced water provision. In Germany, the share of the private sector in providing personal social services increased, while in Sweden, public utilities and personal social services remained under the control of local authorities. Since the 1990s, the different traces in the provision of services across sectors and between European countries have become even wider (for a more detailed overview, see [22]). Due to private sector failure, unsatisfactory reduction of costs, and the need to gain larger control and effective delivery of public services process, the process of decentralization or re-municipalization of service provision has become more often [23]. Croatia, for the last 30 years, also passed through a process of political and socio-economic transition, while the process of decentralization started in mid-2001. Cities are relatively evenly distributed across the country, but there is a high inter- and intra-regional disproportion in the degree of urbanity, level of socio-economic development, and financial strengths of local units. More developed urban areas are mostly near the area of the capital city and along the Adriatic coast.

The recent literature investigates the provision of public services from different aspects. Thus, urban–rural gaps in service delivery [24,25] and inter-urban differences [25,26] are investigated. Liu and He [24] investigate urban–rural differences in the provision of public services and emphasize that the unequal distribution of public services between urban and rural areas is in connection with urban–rural income inequality. Similarly, Pandey et al. [27] notice the similarities between infrastructure and economic inequality. Craig and Halsey [28] analyze local public service efficiency under the assumption of unequal distribution of goods among residents.

Also, there is an increasing number of literature dealing with differences in the quality of life in cities [29,30,31,32]. Węziak-Białowolska [29], based on the literature, groups factors related to the quality of life in cities into five basic groups of factors, namely physical, social, environmental, economic, and institutional factors. Mahendra et al. [33], in their report on necessary transformation for more equitable and sustainable cities, stressed that every third citizen was under-served daily in obtaining core urban services (good-quality housing, transport, water, sanitation, and energy) and the consequences of such inequalities in services were especially visible during the last health crisis caused by the coronavirus pandemic, where the consequences for the health of the population due to inadequate public services, as well as problems with the functioning of entire cities, became more evident.

Hewett and Montgomery [25] noticed the inequality in service provision between larger and smaller cities and that smaller ones were usually under-served compared to larger cities and warned that the decentralization of financial resources to lower levels of government could lead to the consequence of local governments being without financial resources to reach the desired standards in the provision of public services. Holzner and Römisch [34] noticed that there was a correlation between public expenditures for transport, waste and wastewater management, public health services, education (from early childhood to university), social services for families and children, and housing and the expenditures on the culture and leisure of households living in cities and surrounding municipalities. This implies that higher levels of public welfare expenditure increase the quality of life in cities but also in surrounding local government units. Furthermore, the literature indicates that inequalities in the provision of services between local units and differences in their success in raising development are also a consequence of differences in their institutional capacities. Maleković et al. [35] investigated institutional capacity for regional development in Croatia and noticed the need to strengthen institutional capacity both at the regional and local levels. Also, the World Bank [36] noticed that institutional and administrative capacity building is a challenge common to Croatian and other European cities and needs to be improved to boost sustainable urban development through smart solutions in cities.

Pinto et al. [26], in the example of urban green spaces, observed differences in the availability of recreation urban green spaces in the cities of Lithuania and Portugal. Pandey et al. [27] investigated overall infrastructure inequality levels between India and South Africa in the context of the growth of urbanization and noticed the greater inequalities in infrastructure availability in more urbanized regions than in less urbanized regions.

The decentralization index which shows different dimensions of decentralization in European countries on a scale from 0 to 3, indicates that Croatia achieves an average overall level of local decentralization of 1.6, an average level of fiscal decentralization (relative share of overall subnational expenditure compared to total government expenditure is 28 percent), low level of overall administrative decentralization (score 1.2) with the very low level of local autonomy, and a medium level of political decentralization (score 1.5) with very limited ability of local authorities to influence the higher level of government’s legislation and policymaking [37].

3. Methods

The purpose of this analysis is to determine whether cities in Croatia can be grouped according to similar characteristics based on different indicators of providing public services. For this purpose, all 127 cities in Croatia are included in the analysis (excluding the capital city, Zagreb, because Zagreb has, at the same time, the function of a county).

To be able to group the cities, we first collected the data on public services and calculated indicators for 127 cities in Croatia. It was necessary to calculate basic indicators and not use raw data to be able to compare cities that differ in size measured by the number of inhabitants or that differ in other relative terms. In total, we had 19 indicators describing different public functions provided by cities, including general public services, economic affairs, environmental protection, housing and community amenities, health, recreation, culture, education, and social protection (

Table 1. Descriptive statistics of input variables, n = 1271.

Input Variables	Mean	Median	Mode	Std. Dev.	Min.	Max.
Expenditures on public services per capita, in euro	149.5	125.4	17.0	88.1	17.0	622.4
Expenditures on representative and executive bodies per capita, in euro	82.5	59.2	9.3	84.8	0.0	412.7
Local budget expenditures for employees per capita, in euro	227.8	192.8	46.9	126.3	46.9	666.0
Expenditure on public order and safety per capita, in euro	36.3	29.6	0.0	25.8	0.0	128.4
The value of open procurement procedures financed from EU funds as a % of the total value of open public procurement procedures	0.1	0.2	0.0	0.8	0.0	8.4
Environmental protection expenditure per capita, in euro	42.0	26.7	0.0	50.4	0.0	301.3
The rate of separate collection of municipal waste (in %)	17.8	15.0	0.0	14.4	0.0	71.0
Expenditure on waste management per capita, in euro	15.2	9.8	0.0	20.7	0.0	109.0
Expenditure on housing and community amenities per capita, in euro	188.6	160.5	0.8	141.1	0.8	761.4
Utility charges revenues per capita, in euro	85.4	61.7	26.3	84.0	1.2	767.4
City’s contribution revenues per capita, in euro	37.1	11.1	3.9	57.9	0.0	343.7
Health expenditure per capita, in euro	4.4	1.9	0.0	5.9	0.0	31.3
Public expenditure on cultural program activities (total for all cultural activities) per capita, in euro	7.4	1.4	0.0	17.6	0.0	132.4
Expenditure on recreational and sporting services per capita, in euro	36.8	26.5	0.0	31.2	0.0	168.9
Expenditure on primary education per capita, in euro	78.1	7.6	0.0	125.4	0.0	605.1
Expenditure on childcare and early education in euro	114.2	101.2	9.6	63.2	9.6	388.9
Share of the city in the financing of budget beneficiaries in the activity of preschool education (in %)	69.3	70.0	70.0	13.6	0.0	100.0
Child-educator ratio in early childhood education	9.5	9.8	0.0	3.2	0.0	20.8
Allowance to families and households per capita, in euro	6.4	4.2	0.0	7.5	0.0	54.5

Source: authors’ calculations.

). Even though one might argue that some other indicators might also be important for analyzing disparities among cities to provide local public services, we were limited by the availability of data. Secondary data were collected from the Ministry of Finance and the Croatian Bureau of Statistics for the year 2020. Data on revenues and expenditures for cities were collected in accordance with the Budget Act and on the basis of the Guidelines and instructions for the preparation of the state budget, as well as the Instructions for the preparation of the local government budget for each city.

Table 1 shows indicators that served as initial inputs for the factor analysis, with basic descriptive statistics. Already from this table, it can be noticed that some cities spend much larger amounts of money on certain public services than other cities. Additionally, some cities even did not provide certain services to citizens. The larger part of the reason for this is definitely the lack of capacities. However, it has to be stressed that some data might hide the actual developments. For example, some cities are small, and due to demographic structure, they might not have the need for preschool education or have very limited need for such a service. Also, in some cities, there might not be a need for an extended school stay (which is financed from the local budget), which implies that some cities might not offer such a service. Additionally, one might find it curious that average health expenditure per capita was much lower than expenditures on all other public services. The reason behind this is that cities only provide services above national standards or buy some additional equipment. All other expenses for health public services are covered by the county and state budgets. Still, the difference in spending between the highest and lowest-ranked cities on certain services is rather large, and the standard deviation shows a large dispersion in data. However, from this table, we do not know whether some cities are systematically outperforming or underperforming in providing public services or whether there exists similar groups of cities in providing public services.

To group cities according to similar characteristics, we performed a factor analysis. We started with 19 indicators representing different public functions and used a principal component analysis as the extraction method. However, by using all variables, the Kaise–-Mayer–Olkin (KMO) measure of sampling adequacy was below the preferred value of 0.5, and the Bartlett test of sphericity was not statistically significant. Hence, according to the recommendations from the literature [38], variables with significant cross-loadings and/or low loadings were removed from the further analysis. In this way, 13 variables remained, covering the main local public services: expenditures on general public services per capita, expenditures on representative and executive bodies per capita, city’s contribution revenues per capita, health expenditure per capita, utility charges revenues per capita, local budget expenditures for employees per capita, expenditure on primary education per capita, expenditure on recreational and sporting services per capita, share of the city in the financing of budget beneficiaries in the activity of preschool education (in %), the rate of separate collection of municipal waste (in %), the value of open procurement procedures financed from EU funds as a % of the total value of open public procurement procedures, allowance to families and households per capita, and public expenditure on cultural program activities (total for all cultural activities) per capita.

With 13 variables, the KMO measure was 0.68, and the Bartlett test of Sphericity was statistically significant (χ² = 468.073 with 78 df and sig. 0.000), while Cronbach’s Alpha was 0.63, implying that our variables were related and suitable for factor analysis. As stated, we used principal component analysis as an extraction method. The total variance explained with eigenvalues larger than one suggested that we should proceed with five factors. In this way, 66.2 percent of the variance was explained (

Table 2. Total variance explained with eigenvalues.

Component	Initial Eigenvalues			Extraction Sums of Squared Loadings			Rotation Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	3.2	24.3	24.3	3.2	24.3	24.3	2.7	20.9	20.9
2	2.2	16.9	41.2	2.2	16.9	41.2	2.4	18.5	39.4
3	1.2	9.1	50.3	1.2	9.1	50.3	1.3	9.9	49.3
4	1.1	8.2	58.5	1.1	8.2	58.5	1.1	8.8	58.2
5	1.0	7.7	66.2	1.0	7.7	66.2	1.1	8.1	66.2
6	0.9	7.1	73.3
7	0.8	6.0	79.3
8	0.7	5.6	84.9
9	0.6	4.5	89.4
10	0.5	4.1	93.6
11	0.4	3.0	96.6
12	0.3	2.5	99.2
13	0.1	0.8	100.0

Source: authors’ calculations.

). In order to deal with the problem of complex variables, which would make the interpretation of loadings between variables and factors difficult, we used varimax rotation, in which the factors were orthogonal to each other. In this way, we were able to obtain larger loadings on only one factor (in all cases larger than 0.5) (

Table 3. Factor loadings.

		Factor Loadings
Factor 1	Expenditures on general public services per capita	0.826
	Expenditures on representative and executive bodies per capita	0.738
	City’s contribution revenues per capita	0.721
	Health expenditure per capita	0.681
	Utility charges revenues per capita	0.507
Factor 2	Local budget expenditures for employees per capita	0.901
	Expenditure on primary education per capita	0.887
	Expenditure on recreational and sporting services per capita	0.602
Factor 3	Share of the city in the financing of budget beneficiaries in the activity of preschool education (in %)	0.734
	The rate of separate collection of municipal waste (in %)	−0.702
Factor 4	The value of open procurement procedures financed from EU funds as a % of the total value of open public procurement procedures	0.658
	Allowance to families and households per capita	0.678
Factor 5	Public expenditure on cultural program activities (total for all cultural activities) per capita	0.886

Source: authors’ calculations.

).

Next, we used the obtained factor scores to determine the appropriate number of clusters. For this purpose, we used K-means cluster analysis. In this way, we were able to identify a relatively homogenous group of cases, in our case cities, characterized by similar capacities for the provision of local public services [39]. So, input variables are factor scores, which indicate the extent to which each city has a high score on a group of characteristics that have a high loading on a relevant factor. For maximum efficiency, the method for classifying cases was updating cluster centers iteratively. We started with the minimum possible number of clusters (2). However, in this case, univariate F tests from the ANOVA were not statistically significant for all clustering variables, indicating that the difference between the two clusters was not significant. On the other hand, by increasing the number of clusters to three, we were able to obtain significant differences between clusters. Hence, we were able to group cities in our analysis into three different clusters.

Further, we ranked cities according to their capacity for providing public services. Also, we calculated the correlation between the development index and weighted factor points to check whether our analysis of capacities to provide public services corresponds with the level of development of each city.

To rank cities according to the capacity for providing public services, we used individual factor scores for every city and weighted it by the share of explained variance. When we obtained values for all five factors, we summed them for every city and ranked them. A higher rank implies a higher capacity for providing public services. In order to calculate the correlation between the development index and weighted factor points, we used Spearman’s rank-order correlation.

Finally, to check the robustness of the results, an ANOVA was used. An ANOVA was applied to determine the differences in characteristics of socio-economic development across the groups of cities.

4. Results

First, we will focus on identified factor scores. From the factor analysis, we can conclude that public services provided by Croatian cities can be grouped into five factors (Table 3). Factor 1 is related to general public services, housing, community amenities, and health services. It combines the following variables: expenditures on general public services per capita, expenditures on representative and executive bodies per capita, utility charges revenues per capita, municipal contribution revenues per capita, and health expenditures per capita. Factor 2 is related to primary school education and sports and recreation. It comprises local budget expenditures for employees per capita, expenditures on primary education per capita, and expenditures on recreational and sporting services per capita. Factor 3 clusters the public functions related to preschool education and the low rate of separate collection of municipal waste. Factor 4 is related to administrative and social protection capacity. It has a high factor loading on the value of open procurement procedures financed from EU funds as a % of the total value of open public procurement procedures and allowance to families and households per capita. Factor 5 is related to the capacity to promote the cultural sector and has a high positive loading on public expenditure on cultural program activities (total for all cultural activities per capita).

Next, we turn to clustering the cities. Our sample consists of 127 cities in Croatia. We were able to group cities into three different clusters characterized by similar capacities for the provision of local public services. The first cluster is the largest and contains 102 cities; the second contains 20 cities, while the third contains only 5 cities. Figure 1 shows identified clusters in relation to each identified factor from the principal component analysis.

Cluster 1 is characterized by a strong negative relationship with all factors, but most notably with factor 1, which represents developed general public services, housing, community amenities, and health care, and with factor 4, which represents a higher administrative capacity and higher social protection. This group is very large and comprises 102 cities (17,242 inhabitants per city on average). Since this cluster has a negative relationship with all factors, this implies that this group of cities has the lowest capacity to provide public services. It is concerning that the largest group of cities belongs to the group with a low capacity for providing public services.

Cluster 2 comprises 20 cities (14,604.75 inhabitants per city on average) and has a strong positive relationship with all of the factor scores except factor 5. Moreover, this cluster has a more pronounced relationship with all factors (except factor 5) than the other clusters. It is especially characterized by the highest administrative and social protection capacity (related to factor 4) and the most developed functions related to general public services, housing and community amenities, and health care (related to factor 1). On the other hand, this spatial unit is characterized by the pronounced negative relationship with factor 5, which indicates a relatively lower capacity in promoting the cultural sector. Despite the negative relationship with factor 5, it can be concluded that this group of cities has the highest capacities for providing public services. However, it is interesting to note that of 20 cities belonging to this cluster, 16 of them are located in the area of coastal counties. Even though it could be expected that developed coastal cities have high capacities for providing public services due to income generated from tourism activities, it is surprising to find that this cluster comprising of cities from coastal areas has a lower capacity in promoting the cultural sector, which is an important part of touristic activities. However, it has to be noted that (a significant) part of the cultural activities in these cities are financed through local tourist boards. This implies that it is not necessary for those cities to have a low amount of cultural activities but that they are simply financed from different sources. This additionally implies that with these circumstances regarding the financing of cultural activities, cities belonging to this group have more capacities for some other public services.

Cluster 3 comprises only 5 cities (Supetar, Ludbreg, Ozalj, Hvar, and Nin). The predominant feature in this group of cities is the one presented by factor 5: they have by far the strongest capacity to promote the cultural sector. Considering the weak yet positive relationship with factors 1 and 4 to this group of cities, we may also attribute the feature of relatively more developed general public services, housing and community amenities, and health care services, as well as more developed administrative capacity and social protection capacity, compared to Cluster 1 where this problem is most pronounced. However, it has to be noted that cities in this group are small-sized cities with an average size of only 5305 inhabitants. This cluster consists of two island cities, one coastal city, and two cities that are in the continental part of the country.

Ranking cities according to the capacity for providing public services shows that, indeed, cities with higher capacities for providing public services are those from Cluster 2 (

Table 4. City rankings according to the capacity for providing public services.

City	Sum of Weighted Ranks of Factor Scores	Rank	Cluster
Crikvenica	71.1	1	2
Opatija	69.0	2	2
Rovinj-Rovigno	67.9	3	2
Novigrad-Cittanova	67.1	4	2
Umag-Umago	66.9	5	2
Makarska	65.8	6	2
Nin	62.9	7	3
Slunj	62.5	8	2
Obrovac	59.1	9	2
Županja	58.1	10	1
Senj	57.9	11	2
Poreč-Parenzo	57.1	12	2
Dubrovnik	56.5	13	2
Zaprešić	56.5	14	2
Supetar	56.4	15	3
Pazin	56.3	16	1
Biograd na Moru	56.1	17	2
Trogir	55.4	18	2
Vodnjan-Dignano	55.3	19	1
Cres	54.8	20	1
Rijeka	54.5	21	1
Solin	54.1	22	2
Gospić	53.6	23	1
Sisak	53.1	24	2
Vukovar	52.7	25	1
Vodice	52.3	26	1
Korčula	52.2	27	1
Delnice	52.0	28	1
Krk	51.8	29	1
Hvar	51.6	30	3
Mali Lošinj	51.0	31	1
Ivanić-Grad	50.8	32	1
Komiža	50.3	33	2
Otok	49.9	34	1
Labin	49.6	35	1
Bakar	49.4	36	2
Šibenik	49.3	37	1
Pag	49.0	38	1
Varaždin	47.5	39	1
Vrbovsko	47.3	40	1
Buzet	47.2	41	1
Belišće	47.1	42	1
Velika Gorica	47.0	43	1
Đurđevac	46.9	44	1
Čabar	46.8	45	1
Lipik	46.5	46	1
Osijek	46.2	47	1
Kutina	46.1	48	1
Rab	45.9	49	1
Pula-Pola	45.7	50	1
Split	45.7	51	1
Omiš	45.6	52	1
Vrlika	45.4	53	1
Požega	44.8	54	1
Novi Vinodolski	44.6	55	1
Beli Manastir	44.3	56	1
Otočac	44.1	57	1
Ilok	44.1	58	1
Popovača	43.8	59	1
Novalja	43.7	60	2
Buje-Buie	43.3	61	1
Samobor	42.3	62	1
Stari Grad	42.3	63	1
Sinj	41.7	64	1
Virovitica	41.2	65	1
Pakrac	41.1	66	1
Donji Miholjac	41.0	67	1
Hrvatska Kostajnica	41.0	68	1
Bjelovar	40.9	69	1
Zadar	40.7	70	1
Dugo Selo	40.4	71	1
Imotski	40.3	72	1
Karlovac	40.0	73	1
Glina	40.0	74	1
Sveti Ivan Zelina	39.6	75	1
Drniš	39.1	76	1
Koprivnica	38.8	77	1
Novska	38.7	78	1
Vis	38.6	79	1
Ogulin	38.5	80	1
Čakovec	38.3	81	1
Sveta Nedelja	38.2	82	1
Ludbreg	37.9	83	3
Kaštela	37.8	84	1
Ozalj	37.8	85	3
Jastrebarsko	37.6	86	1
Zabok	37.2	87	1
Prelog	37.2	88	1
Ploče	36.9	89	1
Kastav	36.8	90	1
Vrgorac	36.8	91	1
Grubišno Polje	36.5	92	1
Slavonski Brod	36.3	93	1
Križevci	36.2	94	1
Čazma	35.9	95	1
Daruvar	35.7	96	1
Nova Gradiška	35.6	97	1
Knin	34.7	98	1
Garešnica	34.6	99	1
Vinkovci	33.8	100	1
Slatina	33.2	101	1
Kraljevica	33.1	102	1
Opuzen	32.6	103	1
Krapina	29.5	104	1
Kutjevo	29.3	105	1
Metković	28.9	106	1
Našice	28.6	107	1
Benkovac	28.4	108	1
Orahovica	28.1	109	1
Vrbovec	27.5	110	1
Klanjec	27.3	111	1
Valpovo	26.4	112	1
Đakovo	25.6	113	1
Skradin	25.3	114	1
Lepoglava	25.2	115	1
Petrinja	25.2	116	2
Zlatar	25.2	117	1
Donja Stubica	25.0	118	1
Pleternica	25.0	119	1
Varaždinske Toplice	24.3	120	1
Novi Marof	23.7	121	1
Oroslavje	22.6	122	1
Ivanec	21.3	123	1
Mursko Središće	21.1	124	1
Trilj	21.1	125	1
Duga Resa	16.9	126	1
Pregrada	16.4	127	1

Source: authors’ calculations.

). Out of the top 15 cities, 12 are grouped in Cluster 2. Additionally, cities that belong to Cluster 1 are, as expected, mostly ranked in the lower part of the table. Still, there are some evident outliers. For example, Petrinja, the city in the lowest quartile according to the development index, belongs to Cluster 2, which contains the most capacitated cities, while the sum of weighted ranks of factor scores places Petrinja almost at the bottom of the list. Petrinja managed to become a member of Cluster 2 due to a relatively high share of the open procurement procedures financed from EU funds in the total value of open public procurement procedures. Other indicators do not place Petrinja very high. On the other hand, Županja, the city that is almost in the lowest quartile according to the development index, as expected, belongs to Cluster 1, which contains cities with low capacity for providing public services, but at the same time is ranked in the top 15 cities according to the weighted ranks of factor scores. Županja managed to obtain such a result in ranking because of a relatively favorable relationship with fourth and especially third factors.

Detected inequalities between spatial units not only indicate differences in their capacities to provide municipal services, but it can also be argued that cities with low capacities to provide municipal services are least suited to achieve sustainable development. Hence, it seems that the transformation of cities towards a sustainability agenda should start with increasing their capacities to perform municipal services. In this way, cities will not only be able to provide public services at a higher level but will have capacities to commit to sustainable development goals.

To confirm the validity of our results on clusters, especially Cluster 3, which is the smallest, we performed an assessment of characteristics of socio-economic development across clusters of cities using an ANOVA (

Table 5. Assessment of characteristics of socio-economic development across clusters of cities, means, n = 127.

	The total Sample Average	Cluster 1: (n = 102)	Cluster 2: (n = 20)	Cluster 3: (n = 5)	ANOVA
					F	p-Value
Income per capita, euro	3813.7	3745.3	4180.6	3741.3	3.37	0.038
Development index	103.6	102.8	107.4	106.6	8.76	0.000
Budget revenues, euro	383.2	318.2	663.7	585.3	27.71	0.000
Migration ratio	1.06	1.02	1.14	1.38	3.11	0.048
Ageing index	175.1	171.82	191.4	177.6	1.48	0.232
Apartment prices per sqm, euro	972.3	875.8	1398.0	1237.8	11.89	0.000
Population	16,501.1	17,421.7	14,604.8	5304.8	0.78	0.461
Capacity for providing public services—Index	42.4	39.1	57.2	49.3	29.51	0.000

Source: authors’ calculations.

). A significant difference was found between the three clusters in the development index, personal income per capita, budget revenue per capita, migration ratio, and apartment prices per square meter, which confirms the robustness of the conducted cluster analysis. Additionally, the results show that, indeed, Cluster 1 is the weakest. Cities from the first cluster are marked by the lowest average development index (102.75), the lowest budget revenues per capita (EUR 318.23 EUR), and the lowest average migration ratio (1.02). Furthermore, the mean value of the median price of apartments per square meter sold in this cluster was EUR 875.83, significantly lower than the median price of apartments in the second cluster (EUR 1398.02) and in the third cluster (EUR 1237.83). From a spatial point of view, 61 percent of cities from this cluster are in the continental part of the country, while the remaining 39 percent of them are in the area of coastal counties.

It also seems from Table 5 that Cluster 2 is the most advanced. It is characterized by the highest average development index (107.40), the highest income per capita (EUR 4180.58), and the highest budget revenues per capita (EUR 663.72). On average, the median price of an apartment per sqm in this cluster is EUR 1398.02, which means that the apartments in this cluster are significantly more expensive than in the remaining two (p = 0.000). At the same time, apartments in Cluster 2 are less affordable because the average person living in cities from this cluster would have to pay 33.4% of their annual income to buy one square meter of the apartment (2.99 square meters with the whole yearly income). In contrast, in Cluster 1, for an average annual income, an employed person can buy 4.3 square meters of an apartment (23.4% of annual income for 1 square meter), and in Cluster 3, an employed person can buy 3.02 square meters of an apartment (33.1% of annual income for 1 square meter).

Compared to the other two clusters in relation to selected indicators of socio-economic development, Cluster 3 is characterized by the highest value of the migration ratio (1.38, p = 0.048). According to the value of other socio-economic indicators, this cluster is between the first and second clusters, confirming the results of the cluster analysis.

Furthermore, we were interested in relating weighted ranks of factor scores with the development index to check whether, indeed, cities with lower capacities for providing public services are those that are less developed and, hence, less likely to commit to sustainable development. The rank correlation between the development index and weighted ranks of factor scores is statistically significant (

Table 6. Rank correlation between development index and weighted factor points.

			Ranking According to Development Index	Ranking according to Sum of Weighted Ranks of Factor Scores
Spearman’s rho	Ranking according to development index	Correlation Coefficient	1.000	0.460 **
		Sig. (2-tailed)		0.000
		N	127	127
	Ranking according to sum of weighted ranks of factor scores	Correlation Coefficient	0.460 **	1.000
		Sig. (2-tailed)	0.000
		N	127	127

Note: ** significant at p < 0.05. Source: authors’ calculations.

), indicating that our rankings of capacity for providing public services are rooted in reality. It was also interesting to examine the differences between groups of cities considering the capacity for providing public services. To check the significant difference among the three clusters, an ANOVA was applied. The group of cities in Cluster 1 recorded a significantly lower average value of the capacity for providing public services (39.14) compared to the other two clusters (57.24 for Cluster 2 and 49.34 for Cluster 3, p = 0.000).

Finally, we wanted to examine which municipal services dimensions drive the development of cities the most and how. A multiple regression analysis was conducted to determine the relationship. Factor scores related to five characteristics of municipal services were used as independent variables. In contrast, the development index is used as a dependent variable. Regression results are presented in

Table 7. Regression results, n = 127, Development index (dependent variable).

Municipal Service Dimensions, Independent Variables	Coef.	Standard Error
Factor 1: General public services, housing and community amenities, health	1.419 *	0.3613089
Factor 2: Primary school, sport and recreation	2.779 *	0.361309
Factor 3: Kindergardens, waste management	−1.322 *	0.361309
Factor 4: Administrative capacity, social protection	0.331	0.3613089
Factor 5: Culture	0.674 *	0.3613091
Const	103.327 *	0.3598838

Notes: * significant at p < 0.001; R2 = 0.5328, Adjusted R2 = 0.4094; F (5,121) = 18.47; p < 0.000. Source: authors’ calculations.

.

The findings indicate that “Primary school, sport and recreation” and “General public services, housing and community amenities, health” are the most significant factors driving city development. Regression results provided evidence that culture further drove the development of the cities. On the other hand, “Administrative capacities of cities and capacities for the provision of social protection services” have a positive but not statistically significant impact on the development of Croatian cities. It might be puzzling that the coefficient of “Kindergardens, waste management” is negative, but it seems that this is coming from the negative loading from waste management in the rotated component matrix.

5. Discussion and Conclusions

In this paper, we analyzed the capacities of performing municipal services in 127 Croatian cities. Results show that cities in Croatia differ in their ability to perform municipal services, which is a similar result to Sharifi and Khavarian-Garmsir [40], Angelidou [41], and Slack [42]. Croatian cities can be grouped into three different clusters, characterized by similar capacities for the provision of local public services. Cluster 1 is the largest and contains 102 cities, but at the same time, this is the cluster of cities with relatively low capacity for providing municipal services. Cluster 2 contains 20 mostly coastal cities, and this is the group of cities with the highest capacities for providing municipal services. Cluster 3 contains only 5 cities, and this group of cities is characterized by the strongest capacity to promote the cultural sector. Ranking cities according to the capacity for providing municipal services shows that, indeed, cities with higher capacities for providing municipal services are those from Cluster 2.

However, even though we emphasized that 16 out of 20 cities belonging to Cluster 2 are located in the area of coastal counties, we cannot claim that belonging to a coastal county is a necessary condition for having higher capacities for performing municipal services since there are even more cities located in the coastal area that belong to other clusters. Our research just indicated that there are differences between cities according to indicators used to measure capacities to perform municipal services, which is a similar result to Afonso and Fernandes [43]. Still, there are several factors that might affect the quality of performing municipal services, such as budget resources, historical development background, and citizen participation.

Validity of results on cluster analysis was performed with the assessment of characteristics of socio-economic development across clusters of cities using an ANOVA. We found a significant difference between the three clusters in the development index, personal income per capita, budget revenue per capita, migration ratio, and apartment prices per square meter, implying that we correctly identified three different clusters. Further, results show that, indeed, Cluster 1 is the weakest, while cities from Cluster 2 are advanced. This cluster is characterized by the highest average development index, the highest income per capita, and the highest budget revenues per capita compared to Cluster 1 and Cluster 3. An ANOVA also confirmed that cities in Cluster 1 recorded a significantly lower average value of the capacity for providing public services compared to the other two clusters.

Further, a statistically significant rank correlation between the development index and weighted ranks of factor scores indicates that the capacity for providing public services is related to the overall city development. Hence, it is not surprising that results showed the existence of three distinct clusters and that ranking cities according to the capacity for providing municipal services shows that, indeed, cities with higher capacities for providing municipal services are those from Cluster 2, which are more advanced; cities from Cluster 1 occupy the lower part of the list. Additionally, it can be argued that cities with low capacities to provide municipal services are least capable of reaching economic development.

Inequalities between Croatian cities in their capacities to provide municipal services indicate that cities also differ in their abilities to achieve sustainable development goals. Hence, it seems that the transition towards a sustainability agenda should start with improving cities’ capacities to perform municipal services. In this way, cities will be able to provide public services at a more advanced level and will have the capacity to commit to sustainable development goals. So, improving capacities should lead to more sustainable cities that allow their citizens to meet their own needs and improve their well-being without harming the state of natural systems or the living conditions of other people, both in the present and in the future [44], which should ultimately lead to the overall sustainable development of the country.

The results of this research have several implications for policymakers at the local and national levels. The analysis showed which areas need to be improved so that cities can become more developed and sustainable. This means improving public services, housing and community amenities, the cities’ healthcare services and developing primary school education, sports, and recreation, as well as cultural activities. Decision makers at the state level are expected to make changes that will affect the strengthening of the responsibilities in the distribution of revenues and expenditures between the state and local authorities.

Still, there are several limitations of this research. The selection of variables was determined by the quality of the available statistics, which limited the number of variables. Also, collecting data that more precisely measures providing municipal services would provide a more accurate distinction between cities and their capacities. Furthermore, besides the fact that we were able to use data only for one year and that we included only cities in the analysis and not municipalities, the research is only focused on Croatia. Hence, the inclusion of other European countries with similar coverage of municipal services would enable a comparison of cities’ capacities to perform municipal services.