Another Puzzle of Subnational Variation: the Decentralisation Project in Ukraine

ANOTHER PUZZLE OF SUBNATIONAL VARIATION:

THE DECENTRALISATION PROJECT IN UKRAINE

Author: Alexander van der Wusten Supervisor: Imke Harbers

Second reader: Joost Berkhout Date: 22.06.2020

Course: Master Research Project: States and State-building Word count: 15.944

TABLE OF CONTENTS

I. INTRODUCTION

3

II. THEORY

6

III. DECENTRALISATION IN UKRAINE

16

A) HISTORICAL LEGACIES 16

a) Regional divisions 16

b) Administrative-territorial structure of Ukraine 18

B) THE 2014 DECENTRALISATION REFORM 19

a) Political context 20

b) Content of the reform 22

C) HYPOTHESES REVISITED 23

a) Decentralised parties 23

b) Industrial assets 26

c) Local participation 27

d) Ethnic heterogeneity 30

IV. EMPIRICAL ANALYSIS

32

A) METHODS 32

B) DATA AND VARIABLES 33

a) DEPENDENT VARIABLE: ATC CONCENTRATION 36

b) INDEPENDENT VARIABLES: EXPLAINING (UN)SUCCESSFUL

AMALGAMATION FOR OBLASTS, RAYONS, AND VILLAGES 36

c) CONTROL VARIABLES 1: PUBLIC GOODS PROVISION 41

d) CONTROL VARIABLES 2: INDUSTRIAL LEGACIES 44

C) OBLAST-LEVEL ANALYSIS 45

D) RAYON-LEVEL ANALYSIS 49

V. CONCLUSION

53

VI. ANNEX: DESCRIPTIVE STATISTICS

56

I. INTRODUCTION

Decentralisation shows up in all sorts of organisations, both in the public and private sectors (universities, armies, companies). Numerous definitions of the term have been developed in the literature, but in its most basic form, decentralisation involves the shifting of authority away from a top layer in charge of the entire organisation to smaller units at lower tiers or even to dependent outsiders (Falleti 2005, Manor 1999, Schneider 2003). In the private sector, for example, this can mean that certain programmes or projects are outsourced or delegated to a third party. In the public sector this can mean that lower administrative levels receive more budgetary autonomy - they are able to decide independently on what kind of projects to spend taxpayers’ money (Manor 1999: pp.4-5).

We can currently observe the latter in Ukraine. Since 2014, the Ukrainian government has embarked on an ambitious reform path: apart from the privatisation of state companies and professionalisation of the civil service, one of the central reforms includes the transferral of responsibilities and resources to newly formed administrative units, conditional upon the merger of smaller administrative units into larger ones (a restructuring of the country’s territorial-administrative makeup) (Romanova and Umland 2019: pp.3-5). A central component of the reform is that the smallest units can decide themselves whether they want to merge and form a new unit, a decision that is encouraged by the central government through various fiscal incentives and support programmes (ibid.). However, as the reform was introduced, it could be observed in the years that followed that some regions witnessed multiple mergers, while other regions did not (ibid.: p.9). My central research question then is what explains subnational variation in the merger of the smallest administrative units as well as the subsequent decentralising transferral of administrative responsibilities and resources in Ukraine.

Decentralisation is seen as improving the efficiency and quality of government (Schneider 2003: pp.33-34). The topic of decentralisation is thus a relevant

one, because the question of how the functioning of government can be improved remains pertinent for both citizens and governments across all tiers alike (ibid.). With regard to Ukraine specifically, decentralisation reform is a relevant topic for two main reasons. Firstly, serious (sometimes unrealistic) hopes have been vested on this reform - it is hoped that it will bring both transparency and efficiency to local government administration in Ukraine and that in this way Ukraine will also approach European standards (Solodkyy et al. 2020). Secondly, structural reform of this kind has so far been somewhat under-researched for post-USSR countries, while extensive research has been done on decentralisation for Central and Eastern European states, such as Poland (Levitas and Herczyński 2003; Levitas 2017). How do explanatory factors that seem to explain subnational decentralisation in other parts of the world (predominantly Latin America, a central focus in the literature) affect the decentralisation process in the post-Soviet space? Similar to many Latin American countries, Ukraine has a clientelistic political environment where relationships are characterised by personal ties and kinship networks (Faust and Harbers 2012; Lankina and Libman 2019). While this is not a comparative research with Latin American cases, it has to be recognised that these clientelistic networks and, consequently, decentralisation reform in Ukraine could well work out differently from Latin American cases as they are part of a different historical past.

My findings suggest that Ukraine’s historical past still has a significant influence on the country today. Subnational variation in decentralisation outcomes seems to be largely explained by the presence or absence of Soviet industrial assets that have created a particular set of attitudes, deeply impacting social and political life and with it the incentives of subnational governments to support or oppose the voluntary creation of new administrative units. In addition, the formation of new units seems to be impacted, to some extent, by other factors, such as the ability of the central government to provide public goods across its territory, which seems a basic condition for the creation of well-functioning local governments (Romanova and Umland 2019).

The structure of my thesis is as follows. Firstly, I will clarify some definitions of the decentralisation concepts and discuss the findings of the literature on subnational variation in decentralisation outcomes, proposing four preliminary hypotheses. Secondly, I will outline the historical and administrative-territorial context of Ukraine, before proceeding with an explanation of the current decentralisation reform in Ukraine and its political dimension. In light of these findings, I will revisit my four initial hypotheses. Thirdly, I will proceed with an empirical analysis, explaining methods and data, before continuing with correlation and regression analyses for different administrative levels. My interpretation of the analyses’ results attempts to answer my main research question. Finally, I will conclude by summarising my findings, pointing towards avenues for further research, and making some general remarks.

II. THEORY

Before I set out and outline the puzzle I hope to solve, I want to clarify the definitions. For the domain of government, decentralisation can be characterised as ‘a process of state reform composed by a set of public policies that transfer responsibilities, resources and authority from higher to lower levels of government in the context of a specific type of state’ (Falleti 2005: p. 328). Since I am considering a case where decentralisation takes place in government, I am limiting myself to the public sector. This means that I will not discuss privatisation (the handover of tasks formerly performed by state agencies to the private sector) and particular types of delegation, when they involve the transfer of tasks outside political systems (Manor 1999: pp.4-5). Furthermore, I do not want to elaborate on conceptualisation issues in this research; I will limit myself to a couple of main points on the type of

responsibilities, resources and state authority that is reallocated downwards as

well as the context of a specific type of state in which decentralisation takes place.

The type of responsibilities transferred to lower levels of government mainly concern public services provision - e.g. education, health care, social welfare and housing (so-called administrative decentralisation) (Falleti 2005: p.329). Furthermore, lower levels of government can receive tools or resources through which they increase their revenues - e.g. throughthe creation of new subnational taxes or the strengthening of their taxation capacities (fiscal decentralisation) (ibid.). Finally, the political authority of subnational government (or local interest groups) can be increased when they are politically independent to make certain decisions (political decentralisation) (ibid.). In this way, decentralisation is measured along three dimensions: administrative, fiscal and political (Falleti 2005; Schneider 2003). These three dimensions are distinct, but there are overlaps (Schneider 2003: p.41). In order for decentralisation to bring about any change for the lowest tiers of

government, it must be a mixture of these three dimensions (Manor 1999: p. 7).

Also, this definition sees the content of decentralisation policies and their interactions with the broader political and economic systems as highly determined by the type of state in which they are taking place (Faletti 2005: p. 329). In order to fairly compare decentralisation policies across countries, policies must take place within the same type of state (ibid.). Federal states are by the nature of their constitution provided with definite chances for decentralisation. But there are always spaces for struggles between central and at least subnational actors of some importance over the distribution of power and resources. These struggles continue over time and may repeatedly readjust relations between the center and subnational actors. This has thus an impact on the institutional outcome of the state in different stages resulting in different degrees of decentralisation (Gonzalez 2008: p.217). Other explanations involve unitary states, which are characterised by weak subnational actors and where decisions about the intergovernmental transfer of responsibilities, resources and authority are in the end made almost exclusively by the central government (ibid.).

Finally, there are more macro-sociological perspectives where it is argued that not only formal political institutions, but also embedded patterns of social organisation determine whether institutions are relatively centralised or decentralised (Boone 2003: p.20). Patterns of social organisation are partly comprised of local communal structures, a key variable in sociological analysis, which comes down to the extent to which control (over persons, resources, access to markets) is concentrated in the hands of a ruling elite or dispersed among many (ibid.: p.21). Concentrated control pools political resources or assets into the hands of a narrow set of actors that hereby gains political clout vis-à-vis other groups (ibid.: pp.22-23). Depending on the extent to which these elites can be co-opted by a central state, a more centralised or decentralised state structure will be created (ibid.: pp.28-33).

B) PUZZLES

From the 1980s onwards, in developing countries, decentralisation in government was seen as a solution to the problem of inefficient central bureaucracies (Schneider 2003: pp.33-34). The idea was that increasing the

responsibility of local governments over public goods and services as well as

the financial resources at their disposal to take care of these would result in a better match of public goods supply with local demand. In the end, the offering of public goods and services would be improved through a more “business-like” organisational structure that makes use of managerial and accounting techniques normally associated with the private sector (Cheema et al. 1983: pp.19-20).

While most of the research on public goods provision has been focused on the national level, one of the most important questions is what explains varying outcomes in levels of public goods provision within a country, also referred to as the puzzle of subnational variation (Giraudy et al. 2019: pp.11-13; Faust and Harbers 2012: pp.54-55). Some argue that subnational variation is explained by a variety of local factors, which are overlooked in national-level research. Among others, these are the emergence of clientelist networks (also involving political parties), the presence of social capital and ethnically heterogenous communities (Giraudy et al. 2019; Hopkin 2003; Gervasoni 2010; Olson 1993; Alesina et al. 1999).

Faust and Harbers focus on one factor that explains subnational variation in public goods provision (social capital) in order to explain subnational decentralisation outcomes in Ecuador (Faust and Harbers 2012: pp.52-55). They consider an ‘optional decentralisation framework’ in which local governments could apply for additional administrative responsibilities (ibid.: p. 55). Their findings suggest that the presence of social capital on the subnational level makes local governments apply for additional administrative responsibilities (ibid.: p.54). More broadly, they argue that the study of optional decentralisation frameworks in this way is a possibility to observe

‘subnational heterogeneity’ in action through the decisions subnational governments make about whether to apply for additional responsibilities or not (ibid.: p.55).

The first puzzle that emerges from these results and suggestions is whether the approach of Faust and Harbers, employing factors that initially explain subnational variation in public goods provision (social capital), will hold also in other contexts (a similar type of state that also employs an ‘optional decentralisation framework’). This is the puzzle I want to focus on in this research.

The other puzzle that emerges is whether the existing infrastructure for the provision of public goods and services impacts decentralisation outcomes in the first place. In developing countries, local governments may take over responsibilities in public services provision if central governments are unable or unwilling to provide these services (Faust and Harbers 2014: p.55). Conversely, it could well be possible that local governments themselves are unwilling or unable to take over responsibilities when the necessary infrastructure, implemented by the central state, for local public goods and services provision is lacking (Herbst 2014: pp.23-24). The question underlying this puzzle is thus whether the level of public goods and services provision negatively or positively impacts the willingness of subnational governments to take up additional administrative responsibilities. In this research, I want to control for this puzzle.

Looking at the figure below, Faust and Harbers in their research only consider relations C and, to some extent, A. I want to build on Faust and Harbers’ research, firstly, by selecting three other local explanatory factors other than social capital, and apply them to a similar context; and, secondly, by including the level of public goods provision (indicated as an intermediate variable in the figure below) as a control - put differently, by adding relationship B.

Fig.1. Theoretical framework

I want to consider the 2014 decentralisation reform in Ukraine. I will elaborate on the precise content of Ukraine’s decentralisation reform in the following chapter. For now, I will only mention two things. Firstly, the 2014 decentralisation reform in Ukraine employs something similar to an optional decentralisation framework (Romanova and Umland 2019: p.4). Secondly, Ukraine is a unitary state that has undergone a tremendous societal and economic transition in the past decades, a trajectory, in a way, similar to Ecuador (Sutela 2012; Faust and Harbers 2012). The similarities are, of course, by no means perfect - Eastern Europe after the Cold War is a very different political context from Latin America in the 1990s. Other success stories in terms of decentralisation reform have emerged in Eastern Europe - most notably, Poland’s decentralisation reform after 1989 (Levitas and Herczyński 2003; Levitas 2017). This means that, although we are considering the same type of state, Ukraine requires a different kind of focus, something I will also elaborate on in the following chapter.

My research question therefore is: what explains subnational variation in the

current decentralisation process in Ukraine?

In the next section I will consider four possible explanations of subnational decentralisation within an optional decentralisation framework and formulate hypotheses on the basis of these explanations.

Social capital

Level of public goods provision

Decentralisation outcomes B

A

C) HYPOTHESES

Garman et al. consider a fiscal decentralisation process that applies evenly throughout the territory of a federal state. They argue that the internal structure of political parties that hold offices on the subnational level may explain the eventual degree of fiscal decentralisation throughout the country (Garman et al. 2001). More specifically, they consider ‘the lines of authority and accountability’ within a party and whether these lie closer to subnational or national party figures (Garman et al. 2001: p.212). There are several ways in which to measure these ‘lines’: for example, by looking at internal party procedures, such as control over candidate nominations for national legislative office; or by focusing on the nature of career opportunities for politicians once they leave elected office (ibid.). I want to focus instead on more informal ways of decision-making within political parties in the context of unitary states with a strong clientelistic tradition, such as Ecuador and Ukraine. In the context of an optional decentralisation framework, as already argued by Faust and Harbers (2012: p.54), the type of party in power has an impact on the eventual decision of local government whether to apply for additional administrative responsibilities or not.

As outlined in the previous section, the macro-sociological perspective on the type of context in which decentralisation takes place suggests that social structure is determined by the extent to which control (over persons, resources, access to markets) is concentrated in the hands of a ruling elite or dispersed among many (Boone 2003: p.21). In areas where the state’s central apparatus is weak and there is a strong clientelistic tradition, regional economic elites have a need to secure their interests (Hopkin 2003: p.231). For this purpose, they may join a political party, providing the party with ‘packages of votes’, and in return receiving privileges by their representatives in (sub)national government (often in the form of contractual or job privileges) (ibid.: p.232). This means that political parties are organised around regional elites who are in control of ‘packages of votes’ on which the electoral success of the party as a whole depends (ibid.).

These political parties are often founded by ‘diffusion’ - when regional party organisations that were initially separate decide to unify (ibid.: p.233). The informal influence of regional elites (from these formerly independent party organisations) on decision-making within the united party as a whole is dependent on their ability to transfer votes from one party to the next at any time they please - for example, if they are promised a more sizeable reward by a rival party (ibid.: p.232). Consequently, those elites in control of larger ‘packages’ have relatively more influence within the party’s decision-making, primarily in areas where procedures are not officially documented in protocols and statutes, such as campaign strategy, party discourse, or programmatic proposals (so-called ‘zones of uncertainty’) (ibid.: p.231). Such parties, in which the ‘lines of authority and accountability’ are thus in the hands of (sub)regional elites, could be considered to have decentralised party structures (ibid.: p.233).

Decentralisation reforms that provide regional tiers of government with greater (financial) autonomy could enhance the ability of regional party elites to better control their own ‘packages of votes’ (ibid.: p.232). They can also claim responsibility for spending on public infrastructure, increasing local support for regional party elites and in turn their leverage within the party as a whole (ibid.). In addition, the opening up of new political positions as a result of decentralisation means that more party representatives could participate in the country’s political arena, even if they lose elections for the national presidency or legislature (Jenkins 1997 as cited by Manor 1999: p.49).

In this way, intracountry variation in decentralisation outcomes (whether local governments are more likely to take on additional administrative responsibilities within an optional decentralisation framework) could be explained by the presence or absence of political parties with relatively decentralised (decision-making) structures on the levels of subnational government. My first hypothesis is therefore:

H1: The presence of political parties with relatively decentralised (informal) decision-making structures makes local governments more likely to take on additional administrative responsibilities.

As discussed in the explanation to H1, regional elites play an important role in the context of weak central states - these elites control valuable economic resources, making them powerful vis-à-vis the central state (Boone 2003: p. 21). When the central state is heavily intertwined with such regional elites (through, for example, political parties), the state enjoys relatively easy access to significant economic resources that are independent from broad domestic taxation. This means that central and subnational governments who collect resource rents from valuable economic assets enjoy the political benefits of spending without facing the political costs of taxation (Gervasoni 2010: pp. 302-309). These rents may be used to award hefty procurement contracts, finance extensive clientelism, dominate the advertisement market, all of which decrease possible mechanisms of transparency and accountability by social actors (ibid.).

This means that any efforts to undermine these mechanisms (for example, through decentralisation reforms that involve increased control of subnational governments over local budget and taxation) would be opposed by both powerful economic elites as well as subnational governments who are largely dependent on resource rents, since this may negatively affect the economic privileges of either side. It is noteworthy in this regard that decentralisation has not occurred in most of the rentier states that mainly derive their revenues from oil exports (Manor 1999: p.41). My second hypothesis is therefore:

H2: The presence of valuable economic assets (that could be a source of resource rents) makes local governments less likely to take on additional administrative responsibilities.

The incentives of local governments to provide public goods (or, for that matter, take on additional policy responsibilities) may be reinforced by the active presence of civil society groups or social movements. The public choice literature argues that such groups or movements create social capital - mechanisms of accountability as a result of which local administrators feel more obliged to meet local demand for public goods (Putnam 1993). This is suggested by the findings of Faust and Harbers (2012: p.54), who considered the process of (optional) administrative decentralisation in Ecuador in the period 1998-2008: their findings suggest that on subnational territories where parties connected to local interest and civil society groups are represented (and enjoy broad popular support) local governments have an incentive to opt for additional administrative responsibilities, even in the absence of fiscal incentives, because such parties provide mechanisms for accountability and transparency. On those territories where there are no such interest groups, local governments do not have an incentive to provide public goods (ibid.: p. 58). My third hypothesis is therefore:

H3: The presence of social capital makes local governments more likely to take on additional administrative responsibilities.

The public goods literature suggests that ethnic heterogeneity within an administrative unit (be this a metropolitan or rural area) negatively impacts the share of local public resources that are spent on ‘productive public goods’, such as education, sewage and sanitation systems, and road infrastructure (Alesina et al.: pp.1243-1244). The idea is that different ethnic groups have preferences for different public goods: ‘representatives of interest groups with an ethnic base are likely to value only the benefits of public good that accrue to their groups, and discount the benefits for other groups’ (ibid.: p.1244). This means that whatever public good is eventually chosen, it will be the less preferred option for the majority of the ethnically heterogeneous community. As a result, the population will undervalue the provision of public goods, prefer

to keep taxes low, and devote more resources to private consumption rather than public consumption (ibid.: p.1251).

Alesina et al. bring up Tiebout’s solution for the heterogeneity problem that comes down to the freedom of ethnic groups to separate from heterogenous communities and establish their own ethnic communities that are then able to provide the public goods they prefer. This solution has several problems, such as the multidimensional nature of public goods, limitations to mobility and serious legal constraints. This means that, despite the possibility of sorting ethnicities into separate communities, heterogeneous preferences will remain (ibid.: pp.1246-1247). My fourth hypothesis is therefore:

H4: The presence of ethnic heterogeneity within an administrative unit makes local governments less likely to take on additional administrative responsibilities.

Finally, the importance of historical legacies and how they might influence the decentralisation process, creating opportunities and problems, should not be underestimated. Historical legacies explain the type of state under consideration, the kind of decentralisation reform that takes place within a state as well as the more local context. In this way, centralised authoritarian regimes in Latin America attempted to repress any decentralisation project and also civil society which in turn created problems for decentralisation at the local level (Manor 1999: pp.58-59). Similarly, in Ukraine, the Soviet communist legacy impeded the development of well-functioning party systems, social capital and civil society (Kvartiuk 2016). In the next section I will elaborate on the four hypotheses in the context of its past and the 2014 decentralisation reform.

III. DECENTRALISATION IN UKRAINE

Before I discuss the decentralisation reform of 2014, I will outline two historical legacies that are very important for fully understanding the Ukrainian context after the country’s independence in 1991. These are the regional divisions in Ukraine and the territorial-administrative structure of the country. I will discuss these two legacies in turn.

a) Regional divisions

There are regional differences that have their roots in the way the Ukrainian state has historically been shaped. What is today known as Ukraine used to be part of the Polish-Lithuanian Commonwealth, the largest European medieval state by landmass. In 1772-1793, the territory was divided between two of Europe’s dominant powers at the time, Russia and Austria. Up until the collapse of both empires in respectively 1917 and 1918, different imperial policies shaped Ukrainians on both sides of the border: on the Austrian side, state-sponsored schools and periodicals as well as the right of ethnic Ukrainians to form political parties emerged; on the Russian side, the territory became underdeveloped with no proliferation of public institutions and with ethnic Ukrainians prohibited to hold public office (Peisakhin 2015: pp.25-28). Only in 1939 with the partition of Poland between Nazi Germany and the Soviet Union did the Ukrainian territory become united again, now under Soviet rule (ibid.: p.27).

The different experiences of both territories under different empires had a lasting impact on Ukrainians: Peisakhin’s study on identity and attitudes of Ukrainians on both sides of the former imperial border finds that on what used to be the Russian side, Russian identity and language are predominant, while in the former Austrian parts, Ukrainian (and anti-Russian) attitudes and language prevail (Peisakhin 2015). It is therefore not surprising that,

compared to central and western Ukraine, a relatively small amount of the population in Ukraine’s eastern and southern regions voted for Ukraine’s independence from the USSR in 1991 (Solchanyk 1994: p.48). Feeling historically closer aligned to Russia, it has housed considerable regionalist (and even separatist) movements, such as in the eastern Donbas region and the Crimea (Kuzio 2015: pp.174-175; Romanova 2011: pp.323-325, 331-332). Furthermore, the eastern regions have been industrialised in Soviet times, while the Western part of the country, largely joining the USSR only in 1939, didn’t go through this industrialisation trajectory (Lankina and Libman 2019: pp.141-142). As a result, the eastern regions house valuable mega-industries in coal and steel production, metals- and machine-building, piping and petrochemicals, and military production (ibid.: p.133). These major industries had a central place in the regions’ social life in Soviet times. Lankina and Libman consider the powerful informal networks of the industrial plants’ managers and central planners in the Soviet Union’s planning economy: the only way to meet unrealistically high production targets was to extend bribes to party executives and various planning bodies; in addition, to maintain the loyalty of the workforce, managers had to provide support in the form of tourist vouchers, scarce consumer goods and the like. In this way, major industrial facilities in Soviet times were first and foremost providers of security, support and rewards, meted out through hierarchical political networks (ibid.: pp.130-131).

Considering these industries’ important positions, they were ‘well placed to consolidate control over enterprises and their workforce after communism collapsed’ (ibid.). In the 1990s, the so-called ‘red director’ elite (former managers of the industrial plants) took advantage of the low coercive power of the state at the time to enrich themselves by trading in subsidised exports and exploiting price controls (Wilson 2014: p.42). In this way, they quickly became regional elites, having a dominant influence on regional political and economic life. Their influence worked according to Soviet-era mechanisms: Lankina and

Libman (2019: p.130) consider the power of major Soviet-era industrial facilities to rapidly mobilise the workforce for communist party rallies in the 1970s and 1980s. In the same way, regional elites from the 1990s onwards could rely on the largely loyal industrial workforce for political ends: during the 2015 local elections, heads of factory shops were to ensure that workers showed up at the polling stations and voted for the regional party supported by the plant’s owner (ibid.: pp.131-132).

In sum, Lankina and Libman (2019) argue that Soviet industrialisation brought with it a particular set of attitudes, way of doing things, management tools, and social codes. This legacy still has a visible effect on the political landscape of eastern Ukraine today, while the absence of major industrial assets in central and western Ukraine explain the peculiarities of the political landscape in those parts of the country.

b) Administrative-territorial structure of Ukraine

The current administrative-territorial structure of Ukraine stems from the constitutions of the Ukrainian Soviet Socialist Republic of 1929, 1937 and 1978. These constitutions divided the Ukrainian territory (from highest to lowest levels) in oblasts, rayons, and cities, towns, and villages (Tkachuk et al. 2012: pp.32-33, 37-38). Subnational governments’ responsibilities were outlined in rather broad terms: it included the provision of services in the sectors of education, public health, housing and public utilities (OECD 2018: p. 257). There was no clear separation of functions between different levels: on the overall, the system was based on rigid vertical power in which upper-level executives could annul the decisions of lower-level executives, which indicates that actual authority lied with the higher levels (Tkachuk et al. 2012: p.43). All power at the subnational level seemed to be concentrated in the hands of directly elected oblast assemblies, which carried out the functions of self-government and state power (ibid.: pp.42-44). Oblast assemblies received the total subnational budget and redistributed it among lower levels of subnational

government. This meant that lower levels were very much dependent on allocations from oblast budgets (OECD 2018: pp.177-189). This “trickle-down” budgeting system or matryoshka (Russian doll) budgetary model lead to lobbying and informal networks (ibid.: p.178). The post-independence 1996 Constitution significantly increased the political autonomy of lower-level executives (Romanova 2011: pp.325-327). However, no changes in the overall system of administrative-territorial structure took place - this system was still largely based on the old Soviet model. This meant that the dependence on higher levels (with lobbying and informal networks) remained as well as the system’s main inefficiencies, discussed in the next paragraph (Tkachuk et al. 2012: pp.44-45).

This system remained highly inefficient due to several reasons. Firstly, Ukraine is split into 27,000 villages with a highly unequal amount of residents: 24% of villages have a population of 100 residents or less, while other villages have around 5800 residents. This means that villages with relatively fewer inhabitants have a hard time providing basic services and infrastructure to their residents (ibid.: pp.57-58). This is reminiscent of the previously outlined discussion by Herbst (2014: pp.23-34) on the lack of necessary infrastructure, implemented by the central state, to provide local public goods and services in rural areas. In addition, rural areas are becoming depopulated - the rural population is falling with 1% every year. In practice, this means that, among other things, the decrease in number of children in village schools translates into decaying social infrastructure (Tkachuk et al. 2012: pp.55-56). Secondly, administrative borders often overlap: for example, city borders encompass wide areas of farmland. This complicates local taxation as well as running an effective network of municipal organisations (ibid.: p.60).

B) THE 2014 DECENTRALISATION REFORM

The described inefficiencies of Ukraine’s administrative-territorial system were the primary reason why from 1998 onwards several failed attempts have been undertaken to change Ukraine’s territorial-administrative structure. These

reforms had two components that are central in the subsequent 2014 decentralisation reform I want to focus on in this research: the merger of smaller administrative units into larger ones, hereby reducing the excessive number of villages; the transferral of responsibilities and resources to lower levels of subnational government (Tkachuk et al. 2012: pp.62-69; Levitas and Djikic 2017: p.16). I will explain how in the 2014 decentralisation reform the transferral of responsibilities and resources is conditional upon the merger of smaller administrative units. But first I will outline the political context of the 2014 decentralisation reform.

a) Political context

Already in 2005-2008, under the leadership of then deputy prime minister Roman Beszmertnyi, a plan was developed, in partnership with the Association of Ukrainian cities, to consolidate 12,000 villages and small municipalities into 1200 larger political units, while transferring responsibilities from rayon and oblast levels to these new consolidated villages (Levitas and Djikic 2017: p. 16). In addition, the proposal included the introduction of a presidential representative on oblast level, with the idea of hereby making subregional units less susceptible to regional or oblast separatism. The proposal was supposed to take the form of a constitutional amendment. However, by the time the ruling coalition attempted to pass the proposal through parliament in 2009, it failed, since it had already lost overall credibility and parliamentary support (Levitas and Djikic 2017: p.17; Romanova 2011: pp.332-335). Viktor Yanukovich was elected as President in 2010 - he proceeded with consolidating his presidential powers, suppressing active opposition, and effectively recentralising the country by making oblast assemblies and administrators dependent on him personally for financial resources and security (Romanova 2011: pp.333-334; Kuzio 2015: p.182). Yanukovich’s decision not to sign an Association Agreement with the EU in 2013 sparked a series of mass protests, eventually leading to his ousting and to the election of a progressive coalition of forces to the presidency and legislature in 2014, centred around the figure

of Petro Poroshenko (Kuzio 2015: p.183; Sydorchuk 2016: pp.444-446) Soon after his election, a new administrative-territorial or decentralisation reform was initiated, very similar in content to Beszmertnyi’s 2009 proposal (Levitas and Djikic 2017: p.17).

Administrative-territorial reform became relevant again in 2014 with the new president and coalition government due to the following three reasons. Firstly, the reappearance of the 2009 plan was associated with the unrealised hopes of the 2005-2009 government that the new 2014 progressive coalition was again expected to realise (ibid.). Secondly, as a reaction to Yanukovich regime’s ousting in 2013, Russia annexed the Crimea in 2014 and started sponsoring violence in the eastern Donbas region in the same year, with Russia and Russia-affiliated forces within the country subsequently calling for the far-going federalisation of Ukraine (ibid.). The 2009 decentralisation plan was seen as a potential remedy for the secessionist regional movements (as was also the intention of the 2009 proposal) that appeared in 2013-2014, also preventing the conflict with Russia of getting out of control (Dudley 2019: pp.4-5). Thirdly, the existing administrative-territorial system (with the dominating oblast level and the matryoshka system of subnational fiscal budgeting) was and remained largely inefficient, as described in the previous section (OECD 2018: pp. 177-189; Tkachuk et al. 2012: pp.44-45).

Soon after Poroshenko’s election, the newly formed coalition government started developing proposals for territorial-administrative or decentralisation reform (that in fact also had many of the same people working on it as for the 2009 proposal) (Levitas and Djikic 2017: p.17). The reform would be taking the form of a constitutional amendment and a budget code revision (Sasse 2016). Similar to the 2009 proposal, villages would be able to merge on a voluntary basis and presidential representatives would be installed in the oblasts (ibid.). However, in the end, similar to the 2005-2009 coalition, Poroshenko’s government was not able pass the constitutional amendment through parliament, due to opposition within parliament and mass protests.

Although the coalition government eventually obtained the required 2/3 majority, it had to abandon the proposal for a constitutional amendment by 2015, finding it too politically sensitive (Roberts and Fisun 2014: p.10; Kozlov 2015: pp.1-3; Dudley 2015: pp. 14-16). Still, the major difference with the 2009 attempt was that the Poroshenko government proceeded with the decentralisation process anyway, without a constitutional basis (Dudley 2019: p.5). Up until this day, even under the new Zelensky presidency since 2019, the reforms still have not been grounded in the country’s Constitution (ibid.: pp.15-16).

b) Content of the reform

Apart from the constitutional amendment, two concepts were developed in 2014, outlining the general strategy for the subsequent decentralisation reform: the “Concept of Reforming Local Self-Government and Territorial Structure of Power”, approved by the cabinet of ministers in April 2014, and the “State Strategy for Regional Development 2015 – 2020”, approved in August 2014 (ibid.: p.5). The reform created the possibility for voluntary amalgamation of villages into so-called Amalgamated Territorial Communities (ATCs), improving the state’s capacity to deliver public services, and redesigning the country’s administrative structure and territorial divisions (Romanova and Umland 2019: p.6).

Apart from the villages that initiate their merger into larger units, the central government, rayons and oblasts are all part of the ATC formation process: the national government provides official guidelines on establishing sustainable ATCs capable of providing basic services; oblast administrations design ‘prospective plans’ for the formation of ATCs in each oblast, in accordance with the criteria set by the national government; elected rayon councils have to approve these plans, which in turn need approval from the central government. Thus if a group of villages decides to amalgamate, this has to be approved first by the oblast, rayon and, finally, the national level. When villages manage to

pass these stages and finally merge into an ATC, elections are held for the new ATC assembly and the ATC mayor (ibid.: pp.7-8).

Newly-formed ATCs would receive additional policy responsibilities, primarily in the areas of health care, primary and secondary education (ibid.: pp.8-9). They would also receive increased administrative functions in the management of municipal land by ATCs - land that used to belong to rayons is transferred to newly formed ATCs (Lindegaard and Webster 2018: 53). At the same time, parliament amended the budget and tax codes in 2015. As a result, villages would receive fiscal support directly from the national government, hereby circumventing possible negotiations with the oblast and rayon administrations within the matryoshka system. At the same time the idea is that ATCs would have increased control over local taxes, so that they would not be dependent on subsidies only (OECD 2018: pp.198-199).

C) HYPOTHESES REVISITED

In light of the specific Ukrainian context, I revisit the four hypotheses I proposed in the theory chapter. My first two revisited hypotheses focus on explaining the (un)successful consolidation of villages into ATCs for the oblast

and rayon levels by considering the impact of political parties and industrial

assets on ATC formation. My third and fourth revisited hypotheses focus on explaining the (un)successful consolidation of villages into ATCs for villages by considering the impact of local participation and the ethnic composition of villages on the provision of public goods and, related to that, ATC formation.

a) Decentralised parties

In 2015, local elections took place in which villages, oblasts and rayons elected their representatives for the local councils. These local councils, in particular village councils, would have a central say in whether a village amalgamates into an ATC with surrounding villages in the years to come. Also, oblast and rayon councils would have a serious role to play: as discussed in the previous

section, oblast and rayon levels could block the merger of villages into ATCs and thus the transferral of corresponding functions and responsibilities to these ATCs (Romanova and Umland 2019: pp.7-8) While the electoral procedures under which the 2015 local elections took place are important to consider, I am not going to do that here. Instead, I will focus on the internal structures and organisations of those political parties that participated, further building on my hypothesis from the theory section. I will focus on two major types of parties that ran on opposing platforms during the presidential and parliamentary elections of 2014 and actively participated again in the 2015 local elections. On the one hand, there was the Opposition Bloc, a party formed in 2014. It was (for the most part) a reconfiguration of the Party of Regions, a political party formed in 2002 as a coalition of regional elites in eastern Ukraine to protect their economic interests through the unification of five regional parties (Kuzio 2015: p.176). In line with Hopkin’s (2003: p.233) argument in the theory section, this would be a party that was formed by ‘diffusion’ - separately formed regional parties eventually unified into one single party, the Party of Regions. As a result, there were (more or less) equally strong interests at play within the party. Therefore, the Party was not characterised by some sort of central control, but instead adopted decisions through ‘consensus’ (Kuzio 2015: p.181).

Viktor Yanukovich, the presidential candidate of the Party, was elected as President in 2010, but after the 2013 Euromaidan events the Party was banned from participating in the subsequent presidential and parliamentary elections of 2014. As a result, many of its former members joined the newly-formed Opposition Bloc (Whitmore 2014: p.8). The Bloc took over many of the existing structures of the Party and, consequently, it had similar decision-making procedures as the Party of Regions as well as relatively well-developed regional and local party structures (Sydorchuk 2016: p.455). In addition, in the context of the decentralisation reform, the Opposition Bloc was in favour of the transferral of 75% of the corporate tax to local governments. This would,

however, exclusively benefit the eastern regions, such as Dnipropetrovsk, where current corporate tax revenue is relatively high compared to western regions (Decentralization 2019).

On the other hand, there was the government coalition that emerged after the 2014 parliamentary elections, consisting of largely two parties: Petro Poroshenko’s Solidarity party (or the Petro Poroshenko Bloc) and Prime Minister Yatseniuk’s People’s Front (Sydorchuk 2016: pp.445-447). These parties accounted for a total of 221 of the 226 needed to form a parliamentary majority - the remaining seats were made up of smaller parties (Goncharova 2016). Both of these parties (as well as the supporting, affiliated ones) that entered parliament in 2014 seemed to be ‘virtual parties’ (Whitmore 2014: pp. 5-6). In the words of one Ukrainian observer: ‘The very party of Poroshenko does not exist. There is no site, the telephone or address is not known to which one can refer to become a member of Solidarity’ (Leshchenko 2014 as cited by Kuzio 2014: p.319). Similar to the governing coalition in the 2005-2009 period, the parties lacked a proper organisation at the local level and effective decision-making structures (Sydorchuk 2016). As a result, the 2014 coalition’s support base was unstable (mostly spread out over the central and western regions) and mobilisation at the grass roots level was complicated (Whitmore 2014: pp.5-6; Kuzio 2014: p.312).

As discussed in the theory section, Hopkin (2003: p.232) argued that decentralised party structures would have a positive impact on the decentralisation process. Following his argument, the Opposition Bloc, a party with relatively decentralised decision-making structure that is supported by powerful regional economic elites, would see an opportunity in the 2014 decentralisation reform (that, among other things, provides regional tiers of government with greater financial resources through the revision of the tax and budget codes) to expand their political power in the region and hereby further secure their economic interests. It could therefore be expected that in those regions where decentralised parties (thus, eastern-based parties) won

majorities in (sub)regional assemblies, there would be increased amalgamation of villages into ATCs.

This leads me to the following specification of my first hypothesis:

H1: Regions where parties with a relatively decentralised party structure have a majority in the elected assembly are more likely to have a relatively greater part of their territory covered by ATCs than those regions with a coalition government party majority in the elected assembly.

b) Industrial assets

As discussed in the theory section, social structure is determined by the extent to which control (over persons, resources, access to markets) is concentrated in the hands of a ruling elite or dispersed among many (Boone 2003: p.21). When the central state is heavily intertwined with such regional elites (through, for example, political parties), the state enjoys relatively easy access to significant economic resources that are independent from broad domestic taxation (Gervasoni 2010: pp.302-309).

As outlined in the previous sections, in Ukraine control over valuable industrial assets is mainly in the hands of regional elites based in the eastern regions (Lankina and Libman 2019). In the explanation to the first hypothesis it was shown how they would see the creation of ATCs as an opportunity to expand their power through political parties (Kuzio 2015). However, the creation of ATCs (with the transferral of responsibilities etc.) could, using another argument, just as well be perceived as a threat by these elites.

The coalition government envisaged that ATCs would receive increased control over land as well as income from land taxes. With regard to control over land, the idea was that jurisdiction over land plots outside city/town/locality limits would be transferred to ATCs (Lindegaard and Webster 2018: p.38). This would initiate land surveys by ATC authorities to clarify boundaries and ownership for

purposes of taxation (ibid.: p.53). Valuable industrial assets are usually located on such land plots (Nizalov et al. 2018: pp. 88-91). With regard to increased control over income from land taxes, one of the main aims of the 2014 decentralisation reform was to increase the villages’ fiscal autonomy - apart from the central government’s direct fiscal transfers and grant in order to help newly formed ATCs (Levitas and Djikic 2017), ATCs would also have greater freedom to set rates on the new local taxes that were introduced in the 2015 Tax Code and establish exemptions (OECD 2018: pp.198-199). This would imply that, in the case of non-co-opted ATC leadership (i.e. ATC leadership not somehow connected to the Opposition Block and other reconfigurations of the Party of Region), industrial assets falling within the boundaries of an ATC could be more heavily taxed.

From this it seems clear how eastern industrial elites would perceive the formation of ATCs as a threat to their interests. Such elites would be able to oppose the formation of ATCs through blocking the formation process on the oblast and rayon levels, which they would be able to do through their influence on (sub)regional administrators and assemblies: oblasts and rayons are dependent on these elites, since they are a valuable source for rent-extraction (Dudley 2019: p.27). For example, oblasts and rayons are able to secure bribes in exchange for lease contracts with companies seeking to develop or work the land (ibid.).

This brings me to the specification of my second hypothesis:

H2: Regions with valuable industrial assets would be more likely to have a relatively smaller part of their territory covered by ATCs than those regions with relatively fewer / less valuable industrial assets.

c) Local participation

Parallel to the ‘top-down’ government initiatives to improve the provision of public goods and services by local governments, citizen initiatives were already

attempting to do this since the early 2000s (Kvartiuk 2016: p.1130). There are two ‘bottom-up’ mechanisms incentivising the state to provide public goods (Putnam 1993). The first mechanism is ‘electoral reprisal’ - local voters can sanction their government with non-election in case public goods are not sufficiently provided (Kvartiuk 2016: p.1125). Since this is rather complicated in Ukraine with the non-transparent electoral mechanisms (Lankina and Libman 2019: p.134), hope has been vested in local initiatives with civic-minded individuals that through their extensive social networks would be able to put pressure on local governments, thereby decreasing their rent-seeking activities and improving the quality of public services (Kvartiuk 2016: pp. 1126-1127).

Such initiatives appeared in Ukraine in the following way. After independence, local public goods that were provided by collective farms in Soviet times started to be gradually transferred to local governments and (largely) state-owned communal enterprises (Kvartiuk 2016: pp.1129-1130). Because local governments were often unable to sufficiently finance this infrastructure, the local private sector was pressured into co-financing these assets (ibid.). The inability of local governments to provide basic public needs led to initiatives of Community-Based Organisations (CBOs) since the early 2000s that, with significant help of international donors, contributed to the development of basic infrastructure in their communities, such as water supply and drainage systems (ibid.: pp.1130-1131).

One such major initiative was the ‘Community Based Approach to Local Development in Ukraine’ by the United Nations Development Programme (UNDP) (ibid.: p.1130). The aim of this project was to strengthen participatory governance and encourage community-based initiatives in Ukraine, working on projects to refurbish local health posts, set up sewage and drainage systems, improve water supplies and more. The programme’s funding was conditional upon the active involvement of community members in the CBO’s activities (ibid.). The project ran from 2007 until 2017 and was implemented in three

phases, with the first and second phases taking place respectively in the periods 2007-2011 and 2011-2015 (UNDP 2015).

After gaining independence, town hall meetings, a particular informal institution from the Soviet past, would serve as a platform where local inhabitants could get involved in local public policy and exert pressure on the government with respect to the issues of their interest (Kvartiuk 2016: p. 1131). These more or less informal ways in which locals participated were formalised through community organisation (CBOs) that were created with the help of the UNDP, among other international organisations (ibid.: p.1129). These CBOs improved the effectiveness of the ‘participation - governance’ relationship and also did largely succeed in improving the quality of public goods, because they matched the preferences of local inhabitants better than state companies did (ibid.).

In this way, CBOs arguably contributed to creating social capital and some sort of civic culture in the villages they were operating in (Kvartiuk 2016). Following this line of argument, CBOs can be seen as having laid the groundwork for the creation of ATCs years later: villages with a civic culture and formalised ways in which to put pressure on the local government are more likely to take on additional administrative functions, primarily in health care and education, by becoming part of an ATC. In other words, if a village houses a CBO that was established there before or during 2015, such a village would be more likely to have become part of an ATC in later years, since the activities of a CBO indicate the presence of social capital.

This leads me to the specification of my third hypothesis:

H3: Regions with a relatively higher number of CBOs would be more likely to have a relatively greater part of their territory covered by ATCs than those regions with a relatively lower number of CBOs.

d) Ethnic heterogeneity

There are indications in the literature that the heterogenous ethnic composition of some regions, such as southern Odessa and Kherson oblasts, has sparked the creation of paramilitary groups and inter-ethnic conflict in the wake of the 2013 Euromaidan events (Nagai 2019; De Waal and Jarábik 2018). The conditions for conflict have mainly been created by poor local administration - a combination of poor minority integration policies and high levels of corruption (Nagai 2019: pp.8-9). In Kherson oblast, this has led to the creation of dominant agrarian cartels and limited opportunities for minority groups (Crimean tatars) to integrate in the region’s economic life (ibid.). In Bessarabia, in the south of Odessa oblast, this is less of a problem: there is a lot of multiculturalism, cohesion, shared history between the different ethnicities (ibid.: pp.7-8).

Still, although there seems to be much more cohesion in a region such as Bessarabia, the heterogeneous ethnic composition of the territory complicates the decentralisation process in a different way. If two villages populated by different ethnic groups consolidate into an ATC, one village with one particular dominating ethnic group may become the new regional centre and, as a result, own the regional school and health clinic (De Waal and Jarábik 2018). As argued by De Waal and Jarabik (2018), in line with Alesina et al.’s (1999) reasoning in the theory chapter, this may trigger resentment from different ethnic groups and thus impede the formation of ATCs on the territory.

In the western regions, Zakarpattia oblast with its significant Hungarian minority is facing similar problems (Nagai 2019: p.10). In the more eastern regions, such as Kharkiv and Dnipropetrovsk as well as the Donbas region, the minority groups are composed of mainly ethnic Russians (Solchanyk 1994: pp. 47-50). To an even greater extent than in Bessarabia, the Russians have been living on these territories for hundreds of years and are integrated in the regions’ economic and social life (ibid.). Still, the threat of separatism has proved to be serious in these regions in 2013-2014, and both ethnic Ukrainians

and Russians appeared to have widely opposed political views and public goods preferences, which may in turn also complicate the formation of ATCs (Solchanyk 1994; Arel 2002; Giuliano 2018: p.167).

This brings me to the specification of my fourth hypothesis:

H4: Regions with an ethnically heterogenous population are more likely to have a relatively smaller part of their territory covered by ATCs than those regions with a ethnically more homogenous population (composed of predominantly ethnic Ukrainians).

IV. EMPIRICAL ANALYSIS

The aim of this research is to estimate the impact of local-level political, participatory and fiscal variables on ATC formation. The following relationship will be examined:

ATC formation = _{f (Political, Participatory, Fiscal, Controls)}

Formation of ATCs is the function of local-level political, economic, fiscal and control variables.

The idea is to conduct an analysis on oblast and rayon levels. The point of the oblast-level analysis is to find broad explanations for the country as a whole. For the rayon-level analysis, a selection of eastern oblasts (and their rayons) is made.

In addition to the oblast-level analysis, a selected rayon-level analysis is conducted for two reasons. First of all, the rayons are important players in the proposed hypotheses. The empirical analysis at the level of oblasts shows the oblasts on the one hand as players, but on the other hand they serve also as collections of rayons with scores in which, it must be admitted, a lot of important internal variation gets lost. The analysis at the rayon level makes an effort to account for this variation to some extent. Secondly, the country-wide number of rayons is very large and within the time constraints a selection has to be made.

Following the empirical analysis at the oblast level, we make a selection of oblasts for the second part, the rayon-level analysis. The selection is based on the results of the Electoral Quality (EQ) score outlined in the findings of Lankina and Libman (2019) in their research on Soviet developmental legacy in

Ukraine (included as a control variable in the rayon-level analysis). The EQ score shows relatively lower scores for six eastern oblasts in Ukraine - Dnipropetrovsk, Zaporozhie, Mikolaev, Odessa, Kharkiv, Kherson (ibid.: pp. 149-150). These six oblasts (as well as the Donetsk and Luhansk oblasts) will 1

be selected for the rayon-level analysis. I expect my independent variables 2

(especially on political parties and industrial assets) to matter in the eastern regions because they address certain informal, clientelistic dynamics that, as discussed in the previous chapters, are characteristic of the Ukrainian eastern regions in particular. Therefore, the selection of these eight oblasts seems justified.

For the oblast-level analysis, my number of cases (N) varies between 20 and 22 oblasts, which is too small for a regression analysis. I will therefore run Pearson r bivariate correlations, since both my dependent and independent variables (apart from the political party variable) are continuous. For my rayon-level regression, where I consider between 106 and 115 cases, I will conduct a multiple linear regression analysis. The results are displayed as robust variance estimates using within-cluster estimation (Rogers 1993). The choice for robust variance estimates is explained by my use of oblast-level variables in the rayon-level regression analysis. By using robust variance estimates, I control for intra-group correlation of rayons within oblasts (ibid.). 3

B) DATA AND VARIABLES

What follows is an outline of the operationalised variables and the corresponding data for both my oblast-level and rayon-level analyses. A table

Since the EQ score is based on the 2015 local election results, Donetsk and Luhansk oblasts are

1

not taken into account - on oblast-level, no elections took place in 2015 due to the armed conflict in these two regions (Lankina and Libman 2019: p.146).

In only a selection of rayons (and villages) of the Donetsk and Luhansk oblasts that are still

2

under central control of the Ukrainian state did elections take place (Lankina and Libman 2019: p. 146; Central Election Commission of Ukraine 2015). In my research, these rayons will be taken into account.

The robust variance estimates for SPSS have been computed using a source recommended by

3

of these variables with more specific information on descriptive statistics can be found in the Annex. In addition to my four independent variables, I am using three control variables for my rayon-level analyses, also outlined in the following subsections.

My independent variables are measured on different levels of subnational government, but they also focus on different levels. Building on the arguments advanced in the previous chapter, the political party and industrial assets variables focus on explaining the (un)successful consolidation of villages into ATCs for the oblast and rayon levels. Put differently, it focuses on the reasons why oblasts and rayons may actively pursue or oppose the formation process of ATCs. The local participation and ethnicity variables, on the other hand, focus on explaining the (un)successful consolidation of villages into ATCs for the village level or, in other words, the reasons why villages may actively pursue or oppose consolidating into ATC with their neighbours.

My two control variables are focused on the impact of the level of public goods provision on decentralisation outcomes. In the theory chapter I deliberately chose not to make these factors the main focus of my research, but I recognised that they have an impact on ATC formation (see figure 1). It therefore seems useful to include them as controls. Population density seems to be a condition (albeit not a necessary one) for effectively providing public goods and services, while the eventual quality of local institutions seems to be based on the eventual effectiveness of public goods provision. The third control variable is Lankina and Libman’s Electoral Quality (EQ) score, the historical legacy of Soviet industrialisation that, as discussed in the previous chapter, seems to be related to some of my independent variables.

Fig.2. Status of the voluntary amalgamation of communities into ATCs in each oblast as of July 2019

a) DEPENDENT VARIABLE: ATC CONCENTRATION

Figure 2 above clearly outlines the patterns of amalgamation since the start of the decentralisation reform in 2014 and the elections in the first ATCs a year later. The Zaporizka and Dnipropetrovska oblasts in the east took an early lead in the process, which can be read off the map, looking at the relatively high concentration of ATC territory in these regions (Romanova and Umland 2019: pp.9-10). While the time aspect (the exact yearly growth of ATCs) is in itself important to consider when researching decentralisation in Ukraine, it is not the focus of this research.

ATCs are instances of post-2014 decentralisation outcomes in Ukraine. On oblast level, I measure the percentage of total territory (in ha) covered by ATCs up until the present day. I do not focus on the number of formed ATCs per oblast, because the unequal number of ATCs per region often seems to be explained by varying population density or other related factors (Romanova and Umland 2019: p.5), while size (percentage of territory) seems to say something about the political preferences of subnational governments. In addition, the percentage of territory provides some control for the varying size of different oblasts. For both the oblast and rayon levels, the percentage of territory covered by ATCs (with the final ATCs formed in December 2019) therefore seems a more appropriate measure.

b) INDEPENDENT VARIABLES: EXPLAINING (UN)SUCCESSFUL AMALGAMATION FOR OBLASTS, RAYONS, AND VILLAGES Political parties

The 2015 local elections took place for all levels of subnational government (Central Election Commission of Ukraine 2015). Most of the national parties that participated in the 2014 parliamentary elections also took part in the 2015 elections (ibid.). The 2015 local election results are largely similar for the oblast and rayon councils - four, five major parties took the lead (ibid.).