The decentralised provision of social services raises concerns about availability of services in different geographical areas, particularly in low- and middle-income countries with weak governance and fiscal redistributive capacities. Yet the interconnection of different decentralisation regimes and territorial inequalities in the provision of social services remains underexplored. This article engages with one aspect of this puzzle, the implications of the fiscal conditions on exacerbating (or overcoming) territorial inequalities in services provision. Using the Croatian system of early childhood education and care (data for the 2005-2018 period) as an empirical lens, the article shows that in the absence of a well-established policy and fiscal framework sensitive to regional inequalities in administrative and fiscal capacities, decentralised systems can only institutionalise territorial inequalities in services provision. Next to the legal entitlement to a certain service, inter-territorial fiscal equalisation policies are crucial in overcoming fragmentation in social rights along territorial lines.
Keywords: early childhood education and care; decentralisation; devolution; local welfare systems; territorial inequalities; fiscal decentralisation
JEL: H75, H39, I22
Following a mainstream subsidiarity discourse that “going local […] is the best solution for all problems” (Kazepov and Barberis, 2017: 307), European welfare systems have undergone reforms aimed at strengthening local autonomy in the provision of social services (see, e.g., Ruano and Profiroiu, 2017). Reforms were typically guided by the idea that decentralisation, that is, the transfer of responsibility for defining, financing and providing services to lower levels of government, will bring more efficient services that better respond to diverse regional preferences and needs (Darby, Muscatelli and Roy, 2003; Neuman, 2005; Bartlett, Maleković and Monastiriotis, 2013; Hlepas, 2016), as well as greater accountability and transparency of governance at the local level (Czike, Krémer and Tausz, 2002). Nevertheless, there are certain challenges inherent to decentralisation, such as the risk of “institutionalising” territorial disparities in service provision and increasing inequalities in access to social services (Kazepov, 2008; Andreotti, Mingione and Polizzi, 2012). Concerns about equality have been particularly raised in relation to low- and middle-income countries and countries with poor governance and weak mechanisms devoted to equalization of local government fiscal capacities (Rodríguez-Pose and Ezcurra, 2010; Kyriacou, Muinelo-Gallo and Roca-Sagalés, 2015; Kazepov and Barberis, 2017; Liu, Martinez-Vazquez and Wu, 2017).
Yet the intersection of different decentralisation regimes and territorial inequalities in the provision of social services remains underexplored (cf. Kazepov and Barberis, 2017
; Costa-Font, 2010
; Costa-Font and Turati, 2018
) and asks for further elaboration. This article aims to contribute to this debate, in particular with respect to the implications of the different fiscal mechanisms and conditions involved in decentralisation on the exacerbation (or alleviation) of territorial inequalities in the provision of social services (see Kazepov and Barberis, 2017
). Using the Croatian case as an empirical lens, the article explores the patterns and dynamics of early childhood education and care (ECEC) services expansion within a strongly devolved system marked by high territorial fragmentation and the absence of inter-territorial fiscal equalisation mechanisms. It shows that in the absence of a well-established fiscal framework sensitive to regional inequalities in administrative and fiscal capacities, decentralised services can only institutionalise territorial inequalities in the provision of social services.
The contribution to the existing literature is twofold. First, the article contributes to the literature on fiscal decentralisation (e.g., Liu, Martinez-Vazquez and Wu, 2017
; Martinez-Vazquez, Lago-Penas and Sacchi, 2017
) by pointing out the importance of fiscal equalisation mechanisms for equalising the provision of social services within decentralised systems (cf. Kazepov and Barberis, 2017
). By building an original database covering 556 Croatian local government units (LGUs)1
over the 2005- 2018 period and exploring the patterns of ECEC provision, the article adds to the evidence on trends and dynamics of territorial inequalities in service provision in systems that did not establish fiscal equalisation mechanisms and whose financing largely depends on the local tax base determined by fiscal policy set at the national level. While the ECEC institutional framework in Croatia was stable in the last two decades (Baran, Dobrotić and Matković, 2011
; Dobrotić, Matković and Menger, 2018
), fiscal policy was subject to frequent reforms which affected the disposable revenue of LGUs. That allowed exploring whether the change in LGUs’ revenues itself translated into (dis)investments in service delivery and the reduction of regional inequalities in ECEC provision, or whether there is a need to reform the fiscal policy mechanisms involved in decentralisation in the light of the literature that points to the importance of inter-territorial fiscal equalization (e.g., Liu, Martinez-Vazquez and Wu, 2017
). Focusing on a single country is beneficial here as it allows for the control of the variations in an institutional context and limits the variations in structural and cultural factors to a minimum, overcoming one of the main challenges in cross-country studies on decentralisation (Faguet and Sánchez, 2014
). At the same time, it retains ample variation in observations from several fiscal policy reforms in the 2005-2018 period, the 2009-2014 economic crisis and a large number of heterogeneous LGUs. Besides, in this way, the article also adds a new evidence base to earlier studies published in Croatia, which problematised inequalities in social rights provision produced by the ill-advised policy and fiscal framework involved in decentralisation (for recent findings see, e.g. Dobrotić, 2016
; Dobrotić, Matković and Menger, 2018
; Babić, 2018
; Berc, Blažeka Kokorić and Opačić, 2019
; Babić and Šućur, 2022
Yet this is the first study built on panel data and explicitly connecting funding mechanisms, that is, the fiscal conditions involved in the decentralisation, and territorial inequalities in service provision. Second, the article contributes to the literature on ECEC systems and developments. It provides necessary empirical evidence and insight into the governance-related challenges European countries are facing in setting up ECEC entitlements for all children, the main goal of an ongoing “global agenda of childcare” (Rostgaard, 2018
: 101). It points out how reaching the European Union’s ECEC targets3
may be particularly challenging in countries with highly decentralised systems operating within the context of large territorial economic disparities, particularly if coupled with vague ECEC framework law and a lack of budget transfer mechanisms dedicated to securing funds for the ECEC function (cf. OECD, 2001
; Neuman, 2005
; Moss, 2007
; Kazepov, 2008
The article starts with a discussion of the effects of decentralisation on territorial disparities in the provision of social services, including an overview of the main characteristics of the Croatian ECEC system and the policy context within which it operates. Based on these insights, an analytical framework is set out, which is then applied in the analytical section utilising the empirical evidence on territorial patterns of ECEC services development in Croatia to examine the dynamic and limits of ECEC expansion within a strongly devolved system characterised by high local discretion in ECEC provision and the absence of earmarked mechanisms of inter-territorial fiscal equalisation, coupled with large regional inequalities in the local units’ fiscal and administrative capacity. It concludes by emphasizing the importance of a well-established policy framework and fiscal system (including fiscal equalisation mechanisms) behind decentralised services to ensure equal rights for all citizens and overcome the fragmentation of social rights along territorial lines.
2 Decentralisation, territorial inequalities in the provision of social services and the ECEC system in Croatia
Kazepov and Barberis (2017) point to the interconnection of different “territorial regimes” and the extent of institutionalised territorial disparities in public service provision, with underlying regulative mechanisms playing an important role in palliating or exacerbating these inequalities. The literature indicates that the absence of legal entitlements to certain public services and of inter-territorial fiscal equalisation policies, as well as weak governance (particularly inadequate coordination between central and local bodies), tend to yield a fragmentation in citizenship and social rights along territorial lines (Andreotti, Mingione and Polizzi, 2012; Kazepov and Barberies, 2017; Liu, Martinez-Vazquez and Wu, 2017). The inequalities in social rights may be additionally exacerbated in the context of high territorial fragmentation, that is, “a structure of many independent units of government with very small populations, and limited public resources and management capacity” (Hortas-Rico and Rios, 2020: 963). This raises the risk of having many LGUs with limited administrative and fiscal capacities, which struggle to establish and provide services due to both the small economy of scale and administrative inefficiencies (e.g., staff costs and overheads multiply across many local administrations; Faguet and Sánchez, 2014).
All these features are to be found in the Croatian system of ECEC governance and may pose a significant implementation barrier in service provision to local communities with fewer financial resources (see, e.g. Brennan et al., 2016
) and weak administrative capacities (Bartlett, Maleković and Monastiriotis, 2013
). Namely, the ECEC system in Croatia has been devolved since its establishment in the early socialist period. Functional decentralisation was followed by financial decentralisation in 1959 when the ECEC funding was fully transferred to the municipal level (Iris, 1984), while a legal entitlement to ECEC or fiscal equalisation mechanisms were never established (Baran, Dobrotić and Matković, 2011
). Weak central state involvement paved the way for the institutionalisation of territorial disparities in ECEC accessibility and affordability (Dobrotić, Matković and Menger, 2018
), which only intensified with the transition from a socialist to a capitalist regime. The policy context has changed profoundly since 1990 as local finances were reduced, and the territorial organisation became even more fragmented. The number of LGUs (towns and municipalities) quintupled from 109 to 556. The majority (80%) of LGUs have fewer than 5,000 inhabitants (Koprić, Musa and Đulabić, 2016
and therefore poor fiscal and administrative capacities are a commonplace. As the financing of operational costs and the development of new ECEC infrastructure remained exclusively dependent on LGUs and their fiscal capacities, large territorial disparities in ECEC provision persisted (Dobrotić, Matković and Menger, 2018
Although the decentralisation literature argues for fiscal autonomy “to ensure that financing and expenditure responsibilities are linked at the margin, so that local politicians can bear the costs of their decisions” (Darby, Muscatelli and Roy, 2003
: 8), it also argues for some limits to be imposed on fiscal autonomy when it comes to equality of access to public services. Fiscal equalisation mechanisms are seen as one of the instruments that may equalise service provision within the decentralised systems (Kazepov and Barberis, 2017
). That is particularly important as not all local communities have the same revenue base at their disposal. Wealthier communities tend to have better access to resources (including more skilled workers), thus, challenging the basic assumptions on which decentralization is based (Rodríguez-Pose and Gill, 2004
; Lessmann, 2012
). Hence, although fiscal autonomy can contribute to the accountability of local policymakers, a guarantee of equal access to quality public services also requires some degree of inter-territorial redistribution to provide adequate resources to less developed communities (Darby, Muscatelli and Roy, 2003
; Rodríguez-Pose and Gill, 2004
; Liu, Martinez-Vazquez and Wu, 2017
). The same is stressed in the ECEC literature, which emphasizes that the entitlement to ECEC must be coupled with a funding system that enables all children equal access to quality ECEC (Moss, 2007
). Governance is seen as a critical component of the ECEC system, which may importantly affect the availability, affordability and quality of ECEC services as well as their efficiency in achieving equity goals (Kagan and Cohen, 1997, cited in Neuman, 2005
). At the EU level, such a framing of both governance and funding issues was integrated into the 2019 Council Recommendation on High-Quality ECEC systems (Council of the European Union, 2019
Fiscal equalisation mechanisms have never been introduced into the Croatian ECEC system. The 2001 decentralisation reform in Croatia established local responsibilities for four public services: primary and secondary education, healthcare, social assistance and fire protection. It was coupled with a fiscal package aimed at increasing local revenues by: 1) an increase in the share of personal income tax assigned to LGUs; 2) an increase in the fiscal autonomy of LGUs by allowing them to collect additional local taxes and surtax;5
3) an earmarked part of the personal income tax revenue for the newly decentralised functions;6
and 4) an equalization grant for decentralised functions. Moreover, the wage costs for most of the newly decentralised functions continued to be further paid from the central budget. As the reform covered only the newly decentralised functions and the already decentralised ECEC system was not part of the reform package, neither a funding mechanism nor an obligation for LGUs to provide ECEC was introduced. Still, the ECEC system could have benefited from a higher share of the personal income tax being transferred to LGUs or from the autonomy given to LGUs to introduce new local taxes (Act on the financing of regional and local government units, No. 33/2000; 59/2001). However, the LGUs’ own-tax revenues continued to form a small proportion of their total revenues (Jurlina Alibegović, 2013
). Most of the local revenue came from taxes from the resident population and businesses (defined at the state level), with more developed areas therefore performing better. As a result, LGUs retained low fiscal autonomy and revenueraising authority, and the share of local budgets in the consolidated budget of the central government and GDP continued to be small (table 1). In the two decades following the 2001 decentralization, the government has implemented several reforms affecting the disposable revenues of LGUs (table 1). Repeated personal income tax reforms at the national level (2003, 2005, 2008, 2010, 2013, 2015, 2017) that aimed to decrease the tax burden on labour and increase the disposable income for citizens had a secondary effect of unevenly reducing this revenue stream for LGUs. Therefore, personal income tax reforms have usually been followed by reforms pertaining to the financing of LGUs (in 2007, 2012, 2015, 2017 and 2018), which resulted in a short-term increase in the LGUs’ revenues. Finally, the 2009-2014 crisis severely affected LGU budgets, as employment levels (and personal income tax revenues) nosedived (Vukšić, 2014
). Consequently, the contraction of local revenues was much stronger than the GDP change in 2010 and 2011 (table 1).
Therefore, while the legal framework of and governance setting for a strongly devolved ECEC system in Croatia remained stable in the last twenty years, the LGUs’ capabilities to improve ECEC provision remained highly dependent on the national fiscal policy, where a steady procession of minor reforms affected the level of revenues at the disposal of each LGU in an unpredictable manner. Such exogenous interventions, together with the 2009-2014 crisis, provide enough variation in local revenues to explore the patterns and limits of ECEC expansion within a devolved system underlined by the weak autonomy of LGUs, their heterogeneous fiscal and administrative capacities and the absence of a fiscal equalisation mechanism.
LGUs revenues and fiscal reform effects on disposable LGU revenue, the share of local government revenues in the consolidated governmental revenues and GDP
3 The analytical framework
This article aims to point out that there are the limits of ECEC development within strongly devolved ECEC systems characterised by the absence of a legal entitlement to a regular ECEC program, which operates within: 1) the context of pronounced territorial fragmentation creating unequal fiscal and administrative capacities of LGUs; and 2) a fiscal policy framework that does not provide budget transfers from the central to the local level dedicated to the ECEC function (i.e. lacks a fiscal equalisation mechanism). In particular, it analyses whether and to what extent changes in LGUs’ disposable revenue caused by the central state-led fiscal reforms affected LGUs’ investments in ECEC. Other ECEC policy elements being stable, frequent changes in taxation and financing rules during the observed period have affected LGUs’ abilities to tax and spend differently, depending on their economic activity, wage structure and sources of income and therefore exogenously introducing revenue variation both among and within LGUs. In other words, the article examines whether changes (constraints or expansions) in LGUs’ fiscal capacity may be associated with changes in ECEC enrolment rates within a system that does not oblige LGUs to provide this service and the context of pronounced inequalities in LGU’s fiscal and administrative capacities, which particularly affects underdeveloped areas. The investigated period also allows assessment of the effects of external shocks (i.e. the economic crisis) on resilience/ retrenchment of the local ECEC provision in such a system.
Towns and municipalities (LGUs) are used as a basic unit of analysis. The dataset was compiled for the purposes of this article from the institution-level Croatian Bureau of Statistics’ education and population reporting and the Ministry of Finance’s local budget series. It is organised as a panel covering the 2005-2018 period for the entire population of 556 LGUs. For each LGU, it contains detailed annual information on revenues, budget and consolidated public expenditure for ECEC function7
, number of children and educators in ECEC, and an estimate of the population. As for time-invariant attributes, the official categorisation of mountain (85) and island (47) LGUs is applied, as well as the Eurostat degree of urbanisation (DEGURBA) category for all the LGUs. Based on our exposition, the following variables are considered in the analysis:
The overall ECEC enrolment rate (0-6) is a key service provision outcome of interest. Although enrolment rate indicators are usually monitored separately for the nursery (0-2) and kindergarten (3-6) level, the pooled indicator is used in this analysis as Croatia has a unified ECEC system, and budgetary expenditures on those two programmes are not formally separated and can spill over.
Total local revenues collected
(in fixed 2018 value) is used as an indicator of the LGUs’ fiscal capacity at any given year.8
The effect of both the level and change in revenues is explored.
Share of locally-sourced revenues
is applied as a fiscal autonomy indicator of LGUs (see Akai and Sakata, 2002
), and it excludes state-level equalisation transfers. In addition, the level of surtax on personal income
that LGUs are free to adjust indicates the commitment of local governments to increasing their revenues that may be spent on the ECEC function.
Share of ECEC expenditure in the LGUs’ budget
is used as an indicator of LGUs’ commitment to investing in ECEC, alongside which is an indicator showing whether a town has fiscal responsibility for primary education
(34 such towns in Croatia) as those might have a greater capacity to govern educational systems, including the ECEC system (see, e.g., Rodríguez-Pose and Gill, 2004
The model controls for the degree of urbanisation, remote island and mountain location
as ECEC services are more challenging to establish in areas with a more dispersed population, and for the LGUs population size
, as efficiency loss might emerge due to smaller economies of scale or administrative inefficiencies (see Faguet and Sánchez, 2014
). Also, the impact of change in the size of the preschool population
is assessed as capacities might increase due to demographic pressures.
The analytical strategy is as follows: First, territorial inequalities in the overall ECEC enrolment rates, including their changes over the 2006-2018 period, are described. Then, the existing relationship between ECEC enrolment rates and budgetary commitment to ECEC is plotted against the fiscal capacity of the LGUs (per capita). This being established, a pooled OLS regression model is estimated using the most recent 2016-2018 data and a full set of fiscal and spatial indicators to identify those contributing to the established ECEC enrolment rates. Penultimately, to fully utilise the panel form, a first difference regression is applied with 2006-2018 data, with an eye towards the contribution that the changes in revenue (in the prior year) and demographic pressure may have had on the change in ECEC enrolment rates. In order to check for the resilience of once-established ECEC services to austerity, the effects of revenue growth and cuts are estimated separately. Finally, several first difference regressions are deployed to explore the dynamics of ECEC investments; that is, how general budget change translates into the ECEC commitment, ECEC budget (per capita and child), and a pupil-educator ratio.
Some of the data collected were not used in the analysis. Two municipalities with fewer than ten children of preschool age are excluded. For the first difference analysis of the panel data, LGUs without ECEC capacities in a given year (zero children attending) are not included – a total of 2,395 observations (155-204 LGUs per year), neither are LGUs that established a provision in a given year (total of 127 observations). As for the revenue indicator, there were 275 occasions on which reported revenues peaked or collapsed in one year by more than 50%, only to reverse to the baseline a year after. Observations involving such outlier occurrences were omitted from the regression analysis.
4.1 Regional differences in ECECenrolment rates and the LGUs’ fiscal capacity
The expansion of the ECEC system in Croatia was slower than in other former socialist countries of Central Eastern Europe (cf. figure 1; Stropnik, 1989; Zrinščak, 2002; TransMonee, 2019). In the 1990s, there was only modest growth in ECEC enrolment rates, which mostly emerged due to negative demographic developments (Matković and Dobrotić, 2013). A more noticeable increase in ECEC capacities reappeared in the 2000s (figure 1), and the overall ECEC enrolment rate increased from 31.0% to 46.1% between 2005 and 2018.9 During this period, the ECEC population size remained rather stable, but public investment in ECEC increased from 0.46% to 0.60% of GDP between 2005 and 2009 and levelled off afterwards.
There is a large variation in ECEC enrolment rates both within the NUTS3 regions and particularly between them (cf. Dobrotić, Matković and Menger, 2018
). As indicated in figure 2, several large areas in northeast and central Croatia have no or very sporadic ECEC provision. At the same time, higher ECEC enrolment rates are a characteristic of coastal areas and large urban centres (cf. Dobrotić, Matković and Menger, 2018
for 0-2 and 3-6 breakdown). ECEC enrolment rates plot (figure 2) closely resembles the wealth disparities between LGUs (cf. figure 2).
However, territorial disparities in ECEC enrolment rates, while still high, seem to be narrowing, as the population-weighted coefficient of variation steadily declined from 0,616 in 2005 to 0,414 in 2018 (figure 1).10
Over the 2005-2018 period, new capacities were established in 52 LGUs, and ECEC enrolment rates in several hitherto underprovided areas were improved, while the provision declined or remained stagnant in several coastal areas (figure 2).
Figure 3 (left pane) demonstrates a moderate correlation (r = -0,34) between the total local revenues collected per capita and ECEC enrolment rates in 2016-2018. This association is not evident in high-revenue LGUs, collecting above HRK 5,000 per capita (r = 0,02, n.s.). Also, almost all LGUs without the ECEC services belong to the group of low-revenue LGUs. Yet, there seems to be no association between revenues collected and LGUs’ commitment to ECEC (r = -0,05, n.s.), and there is considerable variation in LGUs’ ECEC expenditure at all revenue levels, particularly among the LGUs with low fiscal capacity (figure 3, right pane). On the one hand, 40 towns and municipalities (containing 9.2% of the preschool population) invest more than 15% of their budget in the construction and operation of ECEC facilities. On the other hand, 215 small towns and municipalities (accounting for 12.4% of the preschool population) invest less than 5% of their budget in ECEC. However, most LGUs with moderate-to-high fiscal capacity invest between 6 and 14 per cent of their budget in ECEC.
ECEC enrolment and financing indicators 2005-2018
Regional differences in ECEC enrolment rates (0-6), LGUs' total revenues per capita (2018 prices), change in ECEC enrolment rates (0-6) between the 2005-2007 and 2016-2018, and change in preschool population size (0-6)
ECEC enrolment rate and fiscal capacity, and commitment to ECEC and fiscal capacity (average, 2016-2018)
4.2 Cross-sectional evidence on fiscal and spatial indicators contributing to the ecec enrolment rates
The pooled linear regression model based on the 2016-2018 data shows that after controlling for the size and density of the population as well as fiscal commitment indicators, the ECEC enrolment rates are still strongly associated with the LGUs’ fiscal capacity – each HRK 1,000 per capita in local revenues is associated with a 3 percentage point higher ECEC enrolment rate. Therefore, the difference from the 2nd to 9th revenue decile (1.91 to 7.25) could account for a 16.1 percentage point difference in ECEC enrolment rates. Moreover, greater LGU fiscal autonomy contributes to higher enrolment rates: the difference from the 2nd to 9th revenue decile (29% to 77%) could account for a 6.7 percentage point difference in ECEC enrolment rates.
Apart from the revenues, LGU commitment to investing in ECEC is firmly associated with ECEC enrolment rates, as a ten percentage point difference in the share of ECEC expenditure in an LGU budget (e.g., from 5% to 15%) accounts for a 13.7 percentage points difference in ECEC enrolment rates. Finally, having authority for running the primary education (currently, most of LGUs with a population size over 30,000, and about a third of those with a population size 10-30,000) spill over to 7.3 percentage points higher ECEC enrolment rates.
Population size and density
do matter for provision. Fiscal indicators being controlled for, LGUs with too few children to organise an efficient-sized ECEC facility are still likely to have a lower enrolment rate. Following the model, LGUs with fewer than 2,000 residents are likely to have an 8.7 percentage point lower enrolment rate than those with over 5,000.11
Also, rural areas have lower ECEC enrolment rates than more densely populated LGUs. However, municipalities officially categorised as mountainous do not have lower ECEC capacities, and on islands, ECEC enrolment rates are currently higher than in similar LGUs on the mainland (cf. figure 2).
Pooled regression model: ECEC enrolment rate (0-6), 2016-2018
4.3 Estimating the effect of changes in LGU revenues on ECEC enrolment rates
Having a panel of fourteen years of data, externally introduced variation in the level of LGU revenues caused by frequent changes in local government income and taxation rules and the 2009-2014 economic crisis allows for a closer look at changes in ECEC enrolment rates. More precisely, it enables us to explore when and under which circumstances ECEC enrolment rates change. For that purpose, the analysis starts with descriptives of the first differences in ECEC enrolment rates. On average, ECEC enrolment rates in LGUs increased by 1.19 percentage points per year. Some growth was present in times of both declining and rising revenues. However, the increase in ECEC enrolment rates was considerably slower after years of considerable reductions in LGUs’ revenue per capita, and strongest after years of moderate revenue growth or stability (table 3).
Average change in ECEC enrolment rates, by annual change in LGU revenues per capita (fixed prices), 2006-2018
The net effect of the budgetary change was isolated using a first-difference OLS regression (observations clustered by LGUs), allowing the assessment of the effect of the budgetary change on change in ECEC enrolment while also controlling for pressures related to demographic change (table 4). The first-difference regression indicates that an increase in an LGU’s revenues of HRK 1.12 million (about 150,000 EUR) led to one child newly enrolled in ECEC. In addition, the reductions in LGU revenues, many of which occurred in the 2009-2014 crisis period, were not associated with a subsequent reduction in ECEC provision. A similar pattern was also identified concerning the indicators related to demographic trends. An increase in the number of children was followed by an increase in ECEC capacities (6.9 places per 10 children increase). However, the decline in population size did not lead to a reduction in ECEC enrolment.
First-difference regression on change in ECEC enrolment, 2006-2018
4.4 Estimating the effects of change in LGU budgets on ECEC investments
Apart from the ECEC enrolment rate, an increase in an LGU disposable budget might have led to an increase in other kinds of ECEC investments such as expenditure per child or staffing. The mechanism of such developments was explored via four auxiliary first difference specifications, deploying different dependent variables (table 5). The change in the ECEC budget per capita (1) is responsive to changes in the general LGU budget to a similar extent when the LGU budget is both expanding and contracting, with an additional HRK 33 being allocated for a HRK 1,000 budgetary increase, and HRK 36 being divested for HRK 1,000 of budgetary contraction. However, if the share of ECEC expenditure in the local budget is already high, the change in the ECEC budget per capita is suppressed downwards. The effect of demographic change on the change of the ECEC budget per capita is below the threshold of statistical significance, in cases of both demographic contraction and expansion. The change in the share of the budget dedicated to ECEC (2) indicates the stickiness of ECEC spending. When the LGU budget is contracting, the share of an LGU’s ECEC budget grows. However, the share of the budget devoted to ECEC usually declines in years of budgetary expansion. These findings (1 and 2) indicate that in good fiscal years the ECEC budget grows slower than the LGUs’ budget in general, yet, in bad fiscal years the cuts to ECEC services funding are limited. Also, population pressure slightly contributes to the change in the proportion of the budget dedicated to ECEC.
First-difference regression on change in ECEC investment, 2006-2018
Both in times of budget expansion and, in particular, budget contraction, a considerable adjustment happens with respect to public expenditure per enrolled child (3). According to the model, one HRK change in the LGU budget per capita is associated with about HRK 1.47 increase (if budget is growing) or a HRK 1.82 decline (if budget is decreasing) in the ECEC public expenditure per enrolled child. Moreover, expansion in ECEC enrolment rates contributes to lower investment per child, about HRK 436 per one percentage point increase in ECEC enrolment rate, if the change in ECEC capacity is not followed by a proportional increase in investment. However, the change in investment per child is not related to the hiring (or dismissal) of ECEC staff, and neither is the LGUs’ commitment to ECEC, as neither contributes to the change in a pupil-educator ratio (4). Here the change occurs only due to a change in the ECEC enrolment rates (increased capacity). This finding points to trade-offs between ECEC quality and accessibility.
5 Concluding discussion
This article explored the patterns and dynamics of the development of devolved social services provided under the conditions of strong local disparities in the fiscal and administrative capacity, coupled with a lack of inter-territorial fiscal equalisation mechanisms. Using territorial inequalities in the provision of ECEC services in Croatia as an empirical lens – a case within which the local governments have a high level of discretion in service provision, no service mandate and no earmarked transfers from the central budget – the article shows that the inadequately established policy framework and above all the lack of a fiscal system including fiscal equalization mechanisms that underpin decentralised services may be an important obstacle to providing equal rights to all citizens in the country. In other words, reaching the EU’s ECEC targets that ask for high ECEC coverage (especially for older age groups) may be particularly challenging in countries with highly decentralised systems coupled with vague policy frameworks, prone to the institutionalisation of territorial inequalities in the provision of social services. Fourteen years of observational data have provided ample evidence of higher ECEC enrolment rates in localities with higher total revenues and greater fiscal autonomy (i.e. wealthier communities less dependent on central state transfers), as well as in those that are more committed to the ECEC function or have the administrative capacity to govern primary education. The analysis revealed a slow but steady increase in ECEC enrolment rates, followed by a reduction in disparities in ECEC provision among LGUs over the 2005-2018 period (in part due to saturation in prosperous regions). However, the ECEC expansion is very gradual, territorial disparities are still large, and the EU’s ECEC targets are far from being achieved at the national level.
Concerning the role of fiscal policy developments, a robust pattern of ECEC enrolment rates growth was identified in the years following the increase in LGU disposable revenue. At the same time, already existent ECEC capacities seem to be resilient to the budget cuts. However, this growth mechanism does not seem very efficient, as in general, only a minor part of the budget growth was invested in ECEC. Consequently, overall revenue growth reduced the share of ECEC expenditures in the local budgets. Moreover, in the periods of budget expansion, the ECEC expenditure per child was prone to fast growth while the teacher-pupil ratio (seen as a proxy for ECEC quality) was not being improved. Conversely, the ECEC system has proved to be rather resilient to revenue decline, avoiding a reduction in ECEC enrolment rates via two mechanisms: (a) increasing the share of ECEC expenditures in the local budget, and (b) taking a cost-cutting route, that is, decreasing the ECEC expenditure per child and potentially affecting the quality of services being provided. The latter might have been instrumental in containing the growth of local ECEC expenditures, although those increased from 0.45% to 0.63% GDP in the 2005-2018 period.
While demonstrating a certain resistance of the already established system to external shocks, these findings point at the limits of ECEC expansion within a policy and institutional setting such as that found in Croatia. First, a sizeable post1990 territorial fragmentation gravely limited the administrative and fiscal capacities of many LGUs, particularly in less developed areas of Croatia (see, e.g. Koprić and Đulabić, 2018
). This is an essential obstacle to higher investments in new ECEC capacities with provision being weaker in the smallest municipalities and towns, which do not have enough population to organise an efficiently sized ECEC facility on their own, indicating the importance of economies of scale (cf. Faguet and Sánchez, 2014
). At the same time, inter-municipal collaboration in joint ECEC services provision in Croatia is rare and, in most cases, inherited from the pre-fragmentation period (see Dobrotić, Matković and Menger, 2018
). Second, the fiscally decentralised ECEC system in Croatia never introduced mechanisms to address the weak fiscal capacity of many LGUs, especially after the 1990s when the number of LGUs quintupled, resulting in many of them not being able to serve basic local needs (see Koprić, Musa and Đulabić, 2016
). Equalisation funds were primarily established to serve newly decentralised functions (primary and secondary education, healthcare, social assistance and fire protection), while ECEC has never become part of this package. Such a setup severely limited the development of the ECEC system, particularly in the absence of an obligation to dedicate a part of the budget to the ECEC function. Third, the vague ECEC framework law inherited from the socialist period has never been the subject of comprehensive reform, providing a lot of local discretion in services provision. Among other things, a child’s entitlement to the regular ECEC program has never been established, leaving this function low among the priorities of many LGUs.
These findings indicate that it is not decentralisation per se, but the inadequate underlying mechanisms and conditions that are likely to contribute to the persistence of regional inequalities in ECEC provision. They particularly indicate the importance of the regulatory capacity of the state to provide equal rights to all citizens. As shown, in highly devolved systems, it is crucial to have a well-established fiscal system to achieve territorial equality and efficiency in service provision. As pointed out by Rodriguez-Pose and Ezcurra (2010
: 639), “positive effects of political decentralisation on cohesion will be easily counterbalanced by the unequal capacity of regions in the core and in the periphery of these countries to make the most of decentralised resources, especially in the absence of well-established territorially progressive fiscal systems”. Thus, “the regulatory capacity of the nation-state plays a crucial role in […] guaranteeing citizens’ rights independently of the local conditions in which a person is embedded” (Andreotti, Mingione and Polizzi, 2012
Administrative data on ECEC attendance and financing at the LGU level used in this article were collected within the project “Analiza pristupačnosti, kvalitete, kapaciteta i financiranja sustava ranoga i predškolskog odgoja i obrazovanja u Republici Hrvatskoj” commissioned in 2017 by the Croatian Ministry for Demography, Family, Youth and Social Policy, and further updated within the “Study on financing and governance in pre-primary education” commissioned in 2019 by UNICEF Zagreb Office.
* The authors would like to thank the two anonymous reviewers for their very useful comments and suggestions.
1 Croatia is administratively divided in 20 counties and the City of Zagreb (regional level), and 127 towns and 428 municipalities (local government units).
2 Discussing the development of eldercare services in Croatia, Dobrotić (2016: 30) stressed that “’limited decentralisation’, that is, the fact that decentralisation of the eldercare function was not accompanied by adequate fiscal decentralisation, contributes to the maintenance of regional inequalities in service coverage”. Similar findings in later studies also showed that social benefits and service provision tend to be lower in less developed areas (Dobrotić, Matković and Menger, 2018; Babić, 2018; Berc, Blažeka Kokorić and Opačić, 2019; Babić and Šućur, 2022).
3 Barcelona goals target an ECEC enrolment rate of 33% of nursery-age children (0-2) and 90% of kindergarten-age children (from 3 to primary school; European Council, 2002), and the educational target of the Europe 2020 strategy required an ECEC enrolment rate of 95% of children aged four or older (until they enter primary school; European Commission, 2011).
4 There is a great variety in population size of LGUs, from 137 to 803,900 residents as of the end of 2017, the median standing at 2,826.
5 New local taxes (e.g., taxes on the use of land or properties) were permitted and the introduction of different levels of surtax on income taxes (max 10 to 30%, depending on the size of the municipality/city). LGUs have the autonomy to determine their level, however, with maximum tax ceiling rate for each purpose set at the national level (Act on the financing of regional and local government units, No. 33/2000; 59/2001). They can also charge various fees (e.g., fees for ECEC services).
6 An additional share of personal income tax was transferred for decentralised functions: 2.9% for primary education, 2% for higher education, 2% for social assistance, 2.5% for healthcare and 1% for fire protection (Decision on decentralised functions, No. 75/2001). The share increased in 2007 at 3.1%, 2.2%, 2.2%, 3.2% and 1.3% respectively (Decision on decentralised functions, No. 143/2006), however, it decreased again in 2014 at 1.9%, 1.3%, 0.8%, 1% and 1% respectively (Decision on decentralised functions, No. 33/2016).
7 As the functional budget breakdown of local government is published in consolidated fashion (including publicly owned ECEC facilities) only up to 2015, such consolidated LGU data for the ECEC function for the 2016-2018 period was provided by the Ministry of Finance.
8 Only the EU transfers to LGUs are not included as they are mostly awarded for highly specific projects at the time of their implementation, and none were invested into the ECEC sector until 2018.
9 In 2018, nursery-age children (0-2) enrolment rate reached 23.7% and kindergarten-age children (3-6) enrolment rate (regular programs) stood at 61.3%. That still places Croatia among the EU countries with the lowest ECEC enrolment rates, particularly for kindergarten-aged children (cf. OECD, 2019).
10 With the exception of the 2009-2012 crisis period when the variation coefficient stagnated at about 0.54-0.55.
11 The size of the preschool population at average stands at about 7% of total population, and the optimal ECEC facility size, stipulated by the national pedagogical standard, is 340-400.
No potential conflict of interest was reported by the authors.
Akai, N. and Sakata, M., 2002. Fiscal decentralization contributes to economic growth: evidence from state-level cross-section data for the United States. Journal of Urban Economics,
52, pp. 93 - 108 [CrossRef
Andreotti, A., Mingione, E. and Polizzi, E., 2012. Local Welfare Systems: A Challenge for Social Cohesion. Urban Studies,
49(9), pp. 925 – 1940 [CrossRef
Babić, Z. and Šućur, Z., 2022. The Matthew Effect in Local Welfare Policy in Croatia. Croatian and Comparative Public Administration,
22(1), pp. 159 – 186 [CrossRef
Babić, Z., 2018. Decentralizacija socijalne skrbi i socijalne nejednakosti: slučaj Hrvatske. Revija za socijalnu politiku,
25(1), pp. 25 – 48 [CrossRef
Baran, J., Dobrotić, I. and Matković, T., 2011. Razvoj institucionaliziranog predškolskog odgoja u Hrvatskoj: promjene paradigme ili ovisnost o prijeđenom putu? Napredak: časopis za pedagogijsku teoriju i praksu, 152(3 – 4), pp. 521 – 539.
Bartlett, W., Maleković, S. and Monastiriotis, V., 2013. Decentralization Paradigms in South East Europe: An Introduction. In: W. Bartlett, S. Maleković and V. Monastiriotis, eds. Decentralization and Local Development in South East Europe.
Cham: Palgrave McMillian, pp. 3 – 28 [CrossRef
Berc, G., Blažeka Kokorić, S. and Opačić, A., 2020. Dostupnost socijalnih prava i usluga za obitelji u općinama u ruralnim područjima Hrvatske. Revija za socijalnu politiku,
27(2), pp. 113 – 136 [CrossRef
Brennan, C. [et al.], 2016. How can states ensure access to personal assistance when service delivery is decentralised? A multilevel analysis of Iceland, Norway and Sweden. Scandinavian Journal of Disability Research
, 19(4), pp. 334 – 346 [CrossRef
Costa-Font, J. and Turati, G., 2018. Regional healthcare decentralization in unitary states: equal spending, equal satisfaction? Regional Studies,
52(7), pp. 974 – 985 [CrossRef
Costa-Font, J., 2010. Does devolution lead to regional inequalities in welfare activity? Environment and Planning: Government and Policy,
28, pp. 435 – 449 [CrossRef
Czike, K., Krémer, B. and Tausz, K., 2002. The Impact of Decentralization on Social Policy in Hungary, Latvia and Ukraine. In: K. Tausz, ed. The Impact of Decentralization on Social Policy. Budapest: Open Society Initiative, pp.1 – 88.
Darby, J., Muscatelli, A. V. and Roy, G., 2003. Fiscal decentralisation in Europe: a review of recent experience. European Research in Regional Science, 13, pp. 1 – 32.
Dobrotić, I., 2016. Razvoj i poteškoće sustava skrbi za starije osobe u Republici Hrvatskoj. Društvena istraživanja,
25(1), pp. 21 – 42 [CrossRef
Dobrotić, I., Matković, T. and Menger, V., 2018. Analiza pristupačnosti, kvalitete, kapaciteta i financiranja sustava ranoga i predškolskog odgoja i obrazovanja u Republici Hrvatskoj. Zagreb: Ministarstvo za demografiju, obitelj, mlade i socijalnu politiku.
Faguet, J. P. and Sánchez, F., 2014. Decentralization and access to social services in Colombia. Public Choice,
160, pp. 227 – 249 [CrossRef
Hlepas, N. K., 2016. Is it the twilight of decentralization? Testing the limits of functional reforms in the era of austerity. International Review of Administrative Sciences
, 82(2), pp. 1 – 18 [CrossRef
Hortas-Rico, M. and Rios, V., 2020. Is there an optimal size for local governments? A spatial panel data morel approach. Regional Studies
, 54(7), pp. 958 – 973 [CrossRef
Iris, G., 1984. Društvena briga o djeci predškolskog uzrasta. In: M. Škrbić [et al.], eds. Socijalna zaštita. Zagreb: JUMENA, pp. 318-325.
Jurlina Alibegović, D., 2013. Less is More: Decentralization in Croatia and its Impact on Regional Development. In: W. Bartlett, S. Maleković and V. Monastiriotis, eds. Decentralization and Local Development in South East Europe
. Cham: Palgrave McMillian, pp. 51-66 [CrossRef
Kazepov, Y. and Barberis, E., 2017. The territorial dimension of social policies and the new role of cities. In: P. Kennett and N. Lendvai-Bainton, eds. Handbook of Europan Social Policy.
Cheltenham: Edward Elgar, pp. 302 – 319 [CrossRef
Kazepov, Y., 2008. The subsidiarization of social policies: actors, processes and impacts: Some reflections on the Italian case from a European perspective. European Societies
, 10(2), pp. 247 – 273 [CrossRef
Koprić, I. and Đulabić, V., 2018. Evaluation of the Decentralization Programme in Croatia: Expectations, Problems and Results. In: I. Koprić, H. Wollmann and G. Marcou, eds. Evaluating reforms of local public and social services in Europe
. Cham: Palgrave Macmillan, pp. 243 – 260 [CrossRef
Koprić, I., Musa, A. and Đulabić, V., 2016. Local Government and Local Public Services in Croatia. In: W. Hellmut, I. Koprić and G. Marcou, eds. Public and Social Services in Europe
. Cham: Palgrave Macmillan, pp. 201 - 216 [CrossRef
Kyriacou, A. P., Muinelo‐Gallo, L. and Roca‐Sagalés, O., 2015. Fiscal decentralization and regional disparities: The importance of good governance. Papers in Regional Science,
94(1), pp. 89 – 107 [CrossRef
Lessmann, C., 2012. Regional inequality and decentralization: an empirical analysis. Environment and Planning,
44(6), pp. 1363 – 1388 [CrossRef
Liu, Y., Martinez-Vazquez, J. and Wu, A. M., 2017. Fiscal decentralization, equalization, and intra-provincial inequality in China. International Tax and Public Finance,
24(2), pp. 248 – 281 [CrossRef
Martinez-Vazquez, J., Lago-Penas, S. and Sacchi, A., 2017. The impact of fiscal decentralization: a survey. Journal of Economic Surveys
, 31(4), pp. 1095 – 1129 [CrossRef
Matković, T. and Dobrotić, I., 2013. Promjene u obuhvatu programima predškolskog odgoja i obrazovanja u Hrvatskoj na nacionalnoj i županijskoj razini između 1990. i 2012. godine. Revija za socijalnu politiku
, 20(1), pp. 65 – 73 [CrossRef
Moss, P., 2007. Bringing politics into the nursery: early childhood education as a democratic practice. European Early Childhood Education Research Journal
, 15(1), pp. 5 – 20 [CrossRef
Neuman, M. J., 2005. Governance of early childhood education and care: recent developments in OECD countries. Early Years
, 25(2), pp. 129 – 141 [CrossRef
Rodríguez-Pose, A. and Ezcurra, R., 2010. Does decentralisation matter for regional disparities? A cross-country analysis. Journal of Economic Geography
, 10(5), pp. 619 – 44 [CrossRef
Rodríguez-Pose, A. and Gill, N., 2004. Is there a global link between regional disparities and devolution? Environment and Planning
, 36, pp. 2097 – 211 [CrossRef
Rostgaard, T., 2018. Childcare as a global policy agenda. In G. B. Eydal and T. Rostgaard, eds. Handbook of Family Policy
. Cheltenham: Edward Elgar Publishing, pp. 96 – 110 [CrossRef
Ruano, J. M. and Profiroiu, M. (eds.), 2017. The Palgrave Handbook of Decentralisation in Europe
. Cham: Palgrave Macmillan [CrossRef
Stropnik, N., 1989. Družina – Vir Kvalitete Življenja. Ljubljana: Institut za sociologijo.
Vukšić, G., 2014. Employment and employment conditions in the current economic crisis in Croatia. Financial Theory and Practice
, 38(2), pp. 103 – 138 [CrossRef
Zrinščak, S., 2002. Socijalna i obiteljska politika u tranzicji. Revija za socijalnu politiku
, 9(1), pp. 107 – 115 [CrossRef