Untangling the Gossamer Web: Policy Learning and Informal Networks in Croatia and Serbia

Abstract

Thirty years on from the collapse of Yugoslavia, the paths of the former constituent republics have diverged, with Slovenia and Croatia acceding to the European Union and the rest remaining hopeful Europeans. This article analyses this discrepancy through the lens of policy learning, conceptualising the process of accession as one of learning on the part of the candidate state, and uses social network analysis to identify the impact of informal political networks in parliaments of Croatia and Serbia on their respective EU accession processes. It finds that informal networks in Croatia enabled accession while those in Serbia constrained it.

Why do groups learn? Addressing this question has provided scholars with fertile ground for theorising and helped establish the field of policy learning (Dunlop et al. 2018, p. 2). At its most fundamental level, such learning consists of ‘the updating of beliefs based on lived or witnessed experiences, analysis or social interaction’ (Dunlop & Radaelli 2013, p. 599). Such a definition has allowed the concept to be applied widely,¹ which has, in turn, generated significant attention to the causes and pathologies of policy learning (Dunlop & Radaelli 2018). This article represents a first attempt to quantify one of the often mentioned but little discussed factors in policy learning: informality. Specifically, this article analyses the impact of informal political networks on the European Union (EU) accession processes of Croatia and Serbia, finding that the structure and nature of informal networks in both countries aided learning in hierarchy (LIH) while interfering with learning in epistemic contexts (LEC).

With the European Commission’s recent focus on the rule of law, the time is ripe to investigate the impact of informal networks on polities. ‘Cooperative informal institutions’ (Dunlop & Radaelli 2018, p. 9) and ‘informal’ leadership (Zito & Schout 2009, p. 1117) have been previously recognised as drivers of learning processes whereas other practices can block learning by reducing access and restricting information flows to policymakers (Dunlop et al. 2020). As both Croatia and Serbia are unicameral parliamentary republics, this article uses social network analysis (SNA) to identify the structure of the networks within the Croatian parliament (Hrvatski Sabor, hereafter Sabor) and the Serbian National Assembly (Narodna Skupština, hereafter Skupština) during their respective accession processes.

EU accession was chosen as it provides a standardised formal process for comparison within which Croatia and Serbia shared significant similarities. Both Croatia and Serbia were or are part of the Stabilisation and Association Process (SAP), and both countries had their accession processes delayed due to non-cooperation with the International Criminal Tribunal for the former Yugoslavia (ICTY) (European Commission 2005, 2008), the tribunal set up to try individuals accused of war crimes during the wars that accompanied the dissolution of Yugoslavia. Further, both countries underwent significant constitutional changes in the period immediately preceding their signing of a Stabilisation and Association Agreement (SAA), the core document of the SAP, with the EU (European Commission 2005, 2008). As such, this study examines the countries from the start of the parliamentary period in which they signed their SAAs until accession, in Croatia, or until 2020, in the case of Serbia. This produces a timeline of 2000–2013 for Croatia and 2008–2020 for Serbia, excluding the brief and unstable 2007–2008 Serbian parliamentary period. This allows for the control of variables such as institutional makeup and level of democracy at the start of accession, as both countries were considered sufficiently democratic and institutionally stable to begin the SAP (European Commission 2005, 2008).

This article begins with a presentation of the literature on policy learning and derives from it the conditions for successful learning under the two learning paradigms which apply under the accession conditions: LIH and LEC. The cases are then presented and their selection justified before moving on to a discussion of SNA and the comparative logic behind the choice of cases. This is followed by a discussion of the data, including the collection process. The article concludes with a discussion of how the insights here relate back to the wider literature on policy learning.

Theory: policy learning and informality

As the EU is a complex political, economic and social union, the process of joining requires an equally complex transformation of domestic policy processes. While prior rounds of accession involved forms of negotiation, the accession process of Eastern European states, including those of the former Yugoslavia, was characterised not by negotiation but by adoption (Ágh 1996; Bideleux 1999; Anastasakis 2005). The states of Eastern Europe were, unlike candidates in previous rounds, marked by collapsing economic and welfare systems, and EU membership was considered a way to solve the various crises which they faced during the transition from socialism (Ágh 1994; Bideleux 1999). This situates the EU as an external actor attempting to promote the changes necessary for accession through interaction (Alber et al. 2011). Setting targets, creating a reward and punishment system (conditionality), repeated measuring and observation, and regular meetings and reports were all used to make change in the candidate states ‘irreversible’ (European Commission 2002b, p. 8). Conditionality has been identified as a particularly effective external driver of accession, as rational reward-seeking behaviour on the part of the candidate state drives compliance (Schimmelfennig & Sedelmeier 2005; Dąbrowski 2008; Schimmelfennig & Scholtz 2008).

Alternatively, domestic factors have been highlighted as key intervening factors that explain divergence in the candidate countries (Dimitrova 2002; Schimmelfennig 2005; Perković 2014). Given the violence of the recent past, national narratives and discussions about wartime memory, be they about either of the world wars or more recent conflicts, have both been posited as domestic explanations for this divergence, with national or local leaders mobilising narratives to fuel their political careers (Guzina 2003; Stojanović 2011, 2017; Subotić 2013). Thus, in this framing, the process of joining the EU is one of competing narratives whereby the successful countries have reframed their national myths to support accession (Horvat 2015).

Beyond these cultural aspects, institutional variation has also been posited as an explanation for divergence. This includes explanations such as a ‘goodness of fit’ approach wherein states whose domestic institutional dynamics aligned best with the framework of the EU were best able to accede (Jachtenfuchs 2001; Börzel 2010). The underlying logic is that such institutional arrangements help to drive political interactions in certain states, leading them towards the political and cultural norms of the EU, while the design of others is not as well aligned which results in less Europeanisation, and thus explains the observed divergence in political trajectories (Dimitrova & Dragneva 2009; Iusmen 2014).

While these explanations do offer a convincing framework for understanding divergence between countries on the national level—and this study is only possible because of the groundwork provided by such prior work—they do not address the more granular question of the impact of these factors on the pace of accession. Specifically, under what conditions did informal political networks affect the EU accession process of Croatia and Serbia?

Accession as policy learning

A broad and transformative process such as accession involves changes from judicial reform to sausage labelling and everything in between (Heimbach 2011; Klumbytė 2011). The organisations involved, though, did not simply take up EU norms in a vacuum but altered previous practices through learning as they attempted to adjust their practices to fit within the EU framework. Such learning can be driven, initially, by crisis, wherein previously held ideas and practices are re-evaluated depending on how well they perform (Levy 1994; Kamkhaji & Radaelli 2017). The economic, social and political collapse of the preceding socialist order provided a crisis where previous norms and ideals had been found wanting and were rejected, and new ideas flooded into the region (Csaba 2004).

However, once the initial crisis had passed, how was further learning to take place? For many countries of the postsocialist world, the answer to this question was EU accession. This process promised the alteration of their ‘abnormal’ structures to fit within ‘normal’ EU models (Majstorović 2007), and the completion of a ‘return to Europe’ (Lindstrom 2003; Kostovicova 2004). Within this longue durée process, the type of learning which takes place is dependent on two factors: problem tractability and the ‘certification of actors’ (Dunlop & Radaelli 2018, p. 4). Problem tractability, the inverse of uncertainty, relates to the identifiability of solutions to problems (Dunlop & Radaelli 2013, p. 602). In scenarios where problem tractability is high, that is, where the policy options are known, actors can calculate the pay-offs for different courses of action (Radaelli 1999). However, when problem tractability is low, policymakers need recourse to experts to help them acquire knowledge (Haas 1992; Dunlop & Radaelli 2013, p. 602). The certification of actors, the second factor, refers to the level of expertise associated with the actors who are relied upon to provide information to policymakers (Dunlop & Radaelli 2013, p. 602). The four possible typologies of learning produced by these two factors are given in Figure 1.

FIGURE 1. Typologies of Learning

Source: Dunlop and Radaelli (2018, p. 4).

Due to the structures inherent in the accession process, it is possible to exclude learning in reflexivity (LIR) and learning as by-product of bargaining (LBB) from consideration as the EU and its representatives are, by the nature of the process, experts (Klumbytė 2011). The Commission acts as the arbiter of what must be done and when and if changes are considered sufficient for a chapter to be provisionally closed (European Commission 2005, p. 8; 2013a, p. 1), asserting its expertise. In situations of low problem tractability, EU actors are still, ultimately, experts as they act as gatekeepers or providers of knowledge along with other epistemic experts who have been recruited (Gąsior-Niemiec 2003; Dunlop & Radaelli 2018). Thus, the types of learning in the accession process are reduced to two. Under conditions of high problem tractability, learning takes place in hierarchy (LIH); and under conditions of low problem tractability, learning takes place in an epistemic context (LEC).

But under what conditions does successful learning occur? LIH, as the name implies, is hierarchical: learning is fostered if the rewards for complying outweigh the possible rewards for non-compliance and there is sufficient trust that rewards will be forthcoming and that penalties will be enforced (Dunlop & Radaelli 2018, pp. 11–2). LEC is successful when experts are acknowledged as such, and when they participate in the relevant policymaking bodies (Dunlop & Radaelli 2018, pp. 5, 9). The conditions for successful learning under each typology are as follows. First, for learning in hierarchy, learning is fostered when rewards/punishments outweigh the benefits of non-compliance to the actor (LIH 1), and where the rewards/punishments for compliance/non-compliance are credible (LIH 2). Second, for learning in an epistemic context, learning is fostered when the actors doing the ‘teaching’ are experts (LEC 1), and when the ‘experts’ are inserted into the relevant decision-making bodies (LEC 2).

Informality and networks

However, the actors involved in this process are not isolated beings but socially embedded individuals. Political elites have a variety of connections to other actors, and this wider ‘culture of informality’ pervades formal institutions and helps to regulate conduct within them by enabling or constraining certain types of political action (Lomnitz 1988; Williamson 2009). Relations with linguistic minorities (Trenz 2007), trust (Odera 2019) or history can all shape the culture of informality (Łodziński 2009). Such locally determined cultural practices may be hard to identify (Hooghe & Marks 2009), but they must be considered to fully understand observed political outcomes (Dobovšek & Meško 2008).

By constraining and enabling certain types of political action, the culture of informality shapes the structure of informal political networks which, through their structure, constrain further political action. These networks act as maps of informal practices which pervade formal groups (Bliznakovski et al. 2017). Such ties help to determine informal constraints on political action through mechanisms that are negotiated entirely outside of formal channels (Helmke & Levitsky 2004). These constraints are inherently ambivalent and serve both those seeking power and those who possess it, helping to ensure the replication of informal practices (Ledeneva 2009).

Such informal networks exist widely, not only in formerly socialist countries (Rothstein et al. 1999; Wedel 2003), and they are self-perpetuating as, despite changes in material conditions, individuals become reliant on their functioning as a matter of daily practice (Baez-Camargo & Ledeneva 2017; Ledeneva et al. 2017). Such self-reinforcing socialisation thus provides not only material rewards for individuals but also provides a sense of social stability and purpose beyond the formal practices and institutions in place (Ledeneva 2013).

The existence of these networks, then, alters the necessary and sufficient conditions for policy learning to take place. Specifically, it modifies the previous conditions as follows. For the first combined LIH and informal condition, LIH/I 1, actor learning is fostered under conditions where the rewards/punishments outweigh the cost of compliance as long as these rewards/punishments outweigh contradictory rewards/punishments emanating from within the network. LIH/I 2, the second modified condition, states that when the rewards/punishment for compliance/non-compliance are credible, as long as these rewards/punishments are more credible than those emanating from the veto players in the network, learning will be fostered. For LEC/I 1, the first modified LEC condition, learning is fostered under conditions where the ‘expert’ actors are considered as such by members of the informal networks; and for LEC/I 2 where the ‘experts’ are positioned not just in the correct formal position but also at the correct informal position to transmit knowledge.

Design and data: social network analysis and most similar systems design

To identify and study the impact of informal networks on the learning process, social network analysis (SNA) was used as it provides tools specifically tailored to the study of networks. SNA assumes that networks exist in society and that their structure has an impact on groups (Sørensen & Torfing 2005), and it has been used repeatedly to identify such impact (Scott 1988; Ikeda & Richey 2005; Salvini 2010). In SNA, individuals and the ties between them provide the fundamental building blocks of analysis. Unlike institutions, networks and their ties are not set by rigid rules but are flexible while also constraining and enabling certain types of political action (Wellman 1988; Mizruchi 1994).

To operationalise this, once the characteristics of the individuals and the ties between them have been assembled, standardised values are computed, such as an individual’s centrality to the network or their degree, representative of the number of ties for each individual (Hafner-Burton et al. 2009). These measures can also be computed for the overall network, producing average measures which can be compared across networks. Most importantly, this approach allows for the identification of key veto players and an understanding of how the position of individuals within an overall network can affect outputs.

In addition to SNA, the comparative method was used to draw wider conceptual insights about the impact of these informal political networks. Specifically, the Most Similar Systems Design (MSSD) was employed to identify lines of causation (Lijphart 1971). Although this approach is not ideal in a two-case comparison, it was chosen due to the overarching similarities of the cases in question as the rigorous application of MSSD allows for the identification of key causal differences between the cases. As stated in the introduction, starting the study with the signing of SAAs by Croatia and Serbia sets a baseline for comparison in terms of democracy and in terms of the institutional design of the accession process.

For both countries, the signing represented the end of a period of relative isolation and instability. Both countries were founding members of the Kingdom of Serbs, Croats and Slovenes (Kraljevina Srba, Hrvata i Slovenaca), later the Kingdom of Yugoslavia (Kraljevina Jugoslavija), and became constituent republics of the Federal People’s Republic of Yugoslavia (Federativna Narodna Republika Jugoslavija) and the Socialist Federal Republic of Yugoslavia (Socijalistička Federativna Republika Jugoslavija—SFRJ). With the collapse of the SFRJ in the 1990s amid the wars of Yugoslavia’s dissolution, Croatia and Serbia ended up on opposite sides of the conflict as Croatia was one of the secessionist republics and Serbia, the leading republic in the rump Federal Republic of Yugoslavia, backed ethnic Serbian militias on Croatian territory. As such, until the 1990s, Croatia and Serbia had identical political systems, that of a one-party communist state since 1945, and highly similar economic systems, differing only in levels of development.

Throughout the 1990s Croatia was led by the authoritarian independence leader Franjo Tudjman and vacillated between rhetorical acceptance of EU norms and an authoritarian system which only increased Croatia’s isolation, even after the end of the war in Croatia in 1995 (Sigér 2014, p. 203). This changed with the death of Tudjman in 1999 and the 2000 elections, which resulted in the first handover of power in Croatia, and amendments to the constitution, which changed Croatian from a semi-presidential model with a bicameral parliament to the current unicameral parliamentary system (European Commission 2002a). In the 1990s, Serbia was still ruled by Slobodan Milošević but, following his overthrow in 2001, the country began a pro-EU shift. Negotiations for an SAA began, but it was only signed in 2008, following the dissolution of the State Union of Serbia and Montenegro in 2006, when Serbia passed a new constitution and full cooperation with the ICTY was confirmed (European Commission 2008). As with Croatia, the European Council was optimistic that Serbia’s EU integration could be accelerated given Serbia’s ‘considerable institutional capacity’ (European Commission 2008, pp. 4–5). Table 1 shows further relevant variables for comparison.

TABLE 1 Economic, Political and Demographic Variables in Croatia and Serbia

Variable	Croatia (2000)*	Serbia (2007)^†
Population	4.47 million	7.38 million
GDP per capita	€5,800	€4,002
President	Non-executive	Non-executive
Government type	Unicameral parliamentary republic	Unicameral parliamentary republic
Size of legislature	150–160^‡	250^¶
Electoral system	Proportional Representation (PR), Party List	Proportional Representation (PR), Party List
Electoral districts	12	Single district
Distribution of seats	D’Hondt, 5% threshold (in each district)	D’Hondt, 5% threshold
Minority representation	8 MPs^‖	8 MPs^⸸

Notes: ^‡The number of MPs in the Sabor was not fixed in the period but varied depending on the number of votes cast by Croatians living abroad in the XII electoral district.

^‖ Since the 2002 amendment on national minorities (Ustavni zakon o pravima nacionalnih manjina, 2002, available at: https://narodne-novine.nn.hr/clanci/sluzbeni/2002_12_155_2532.html, accessed 3 June 2021).

^⸸ Although the number of minority MPs is not set by law, minority lists are exempt from the 5% threshold, and the number of such MPs per parliament fluctuated between 8 and 12 between 2008 and 2020.

Sources: *European Commission (2012); ^†European Commission (2009); ^‡Ustav Republike Hrvatske, 1990, l.72, available at: https://zakon.hr/z/94/Ustav-Republike-Hrvatske, accessed 10 March 2023; ^¶Ustav Republike Srbije, 2006, č. 100, available at: http://ustavni.sud.rs/page/view/139-100028/ustav-republike-srbije, accessed 10 March 2023.

As can be seen in Table 1, at the start of the accession process both countries shared significant political and economic similarities. Serbia was poorer with a slightly larger parliament, but both countries seemed, in 2000 and 2008 respectively, to be on similar pro-EU paths. Both countries had emerged from the chaos of the 1990s; both had reformed their constitutional structures; and both had elected pro-EU governments. Indeed, Serbia had one advantage over Croatia in that, by 2008, Serbia had already established a track record of sufficient cooperation with the ICTY for SAA negotiations to conclude, unlike Croatia in 2000.

This is not to ignore, though, that some key differences remained. Croatia, as of 2000, was a unitary parliamentary republic with full control over its territory. Several border disputes with Slovenia led to the stalling of Croatia’s accession process for a period (Geddes & Taylor 2016), but the unitary nature of Croatia was never in doubt. Serbia, on the other hand, has a longstanding dispute over international recognition of its secessionist former province of Kosovo, which declared independence in 2008. Starting the analysis of Serbia in 2008 is intended to control for this discrepancy, as it excludes the one-year (2007–2008) transitional government from study and ensures that each case has four data points. Further, given that the 2007–2008 government was brought down partially by Kosovo’s unilateral declaration of independence (European Commission 2008), this means that in 2008 Serbia had also achieved its current borders.²

It is acknowledged that the Kosovo issue is certainly one of the more difficult points in the Serbian accession negotiations and represents a divergent factor in setting the overall pace of the negotiations but, to date, it has not prevented the opening and closing of other chapters of the acquis (European Commission 2020a). The objective of the Belgrade–Priština dialogue, the intergovernmental EU-sponsored dialogue between Serbia and Kosovo, remains the ‘normalization’ of relations between Serbia and Kosovo (European Commission 2013a, 2019a). While it is, admittedly, unlikely that the form that such ‘normalization’ might take would include Kosovo surrendering its de facto independence, the phrase ‘normalization’ has been sufficiently broad to allow for both Serbia and Kosovo to continue their respective accession processes while dialogue has continued. Indeed, both countries explicitly agreed not to ‘block, or encourage others to block the other side’s progress on their respective EU paths’ (European Commission 2013a, p. 1). As such, for the sake of progress by both countries, the objective of the dialogue remains ambiguous and Serbian accession on other chapters has not been conditioned, to date, on making substantial progress in the dialogue, unlike previous conditions such as Serbia’s cooperation with the ICTY.

Another, perhaps surprising, point of difference is that the Serbian public has been consistently more pro-EU than the Croatian public during the accession process. Given that in the 1990s Serbia, under Milošević, was under international sanctions and was even bombed by NATO in 1999 as part of the war in Kosovo, this runs counter to popularly held belief in Croatia’s relative ‘Europeanness’ as compared to Orthodox and Russophilic Serbia (Larrabee 1990; Vlahutin 2004; Subotić 2013). However, in the six years preceding accession, only 30% of Croatians on average responded that joining the EU would be ‘a good thing’, while 32% responded that it would be ‘a bad thing’ and 36% said it would be ‘neither good nor bad’, compared to 40%, 25% and 30% respectively in Serbia in the period 2014–2020 (European Commission 2010b, 2010c, 2011b, 2014a, 2014b, 2015b, 2015c, 2016b, 2016c, 2017a, 2017b, 2018b, 2018c, 2019b, 2019c, 2020b; Europska Komisija 2007a, 2007b, 2008a, 2008b, 2009, 2011).

Thus, while it is acknowledged that the ongoing dispute with Kosovo and the relative strength of pro-EU sentiment represent divergent factors, along with differences in the size of the parliaments and the relative wealth of the countries, this study will seek to put these issues aside, to the extent that it is possible, and analyse the impact of informal political networks within the parliaments on the accession progress made. This approach positions the structure and nature of the informal networks as the independent variable influencing the dependent variable: the pace of accession. As this section has shown, the two-case comparison controls for a wide variety of possible variables, including economic development, economic history, political history, constitutional design and electoral system, among others. Further, unlike in other new member states, such as those of the 2004 and 2007 enlargement rounds, both Croatia and Serbia were subject to the SAP, and both had their accession processes put on hold due to non-cooperation with the ICTY, the mere existence of which is a unique factor in the accession processes of the Western Balkans.

Data were collected over several periods between 2018 and 2021. The first phase of data collection involved the assembly of a list of members of parliament (MPs) for both the Sabor and Skupština from publicly available information on the respective parliamentary websites.³ This information was entered into GEPHI, a free SNA toolkit (Bastian et al. 2009), and ties were made between the MPs according to their position. Factors such as party affiliation, committee membership and membership in other parliamentary bodies were included to determine the ties for each individual member of parliament (MP). The weighting of the ties is given in Tables 2 and 3. From this information, network maps were generated showing the networks of contacts within the relevant parliaments during each electoral cycle.

TABLE 2 Weighting of Ties in Croatian Networks

Type of tie	Value
Political party (Stranka)	1
Parliamentary club (Klub)	0.5
Committee (Odbor)	0.25
Speaker	0.2
Non-Odbor committee or delegation	0.15
Deputy speaker	0.1
Parliamentary friendship committee	0.01

TABLE 3 Weighting of Ties in Serbian Networks

Type of tie	Value
Political party (Stranka)	1
Working group (Poslanička grupa)	0.5
Committee (Odbor)	0.25
President	0.2
Working group, delegation (Radna tela, Delegacija)	0.15
Deputy speaker	0.1
Parliamentary friendship committee	0.01

Following this first phase, semi-structured interviews were conducted with MPs and key actors involved in the accession process. Nine Serbian interviews were conducted in person during several research trips in 2019–2021 while five Croatian interviews were conducted remotely, due to the travel restrictions imposed by the COVID-19 outbreak. A list of interviews has been provided in Appendix 1. The average length of the interviews was one and a half hours, although they varied in length, from 45 minutes to two and a half hours. For each interview a standardised list of questions was prepared, and the interview was conducted either in English or Serbo-Croatian.⁴ Responses from the interviews were used to corroborate the insights from the data. All interviews are cited anonymously or using the pseudonymous designations chosen by the subjects.

Analysis: Croatian and Serbian accession

Both in Croatia and in Serbia, EU accession was and is not, respectively, a smooth process and the political landscape in both countries shifted throughout the period under analysis. In Croatia, the two-party system ossified between the Croatian Democratic Union (Hrvatska demokratska zajednica—HDZ) and the Social Democratic Party of Croatia (Socijaldemokratska partija Hrvatske—SDP), around which smaller parties orbited. In Serbia, this period was dominated by the rise of the Serbian Progressive Party (Srpska napredna stranka—SNS). In both countries EU integration was declared as a strategic priority by the main parties (European Commission 2005, 2008, 2016a).⁵ Figures 2–4 show the Croatian and Serbian accession timelines individually and superimposed upon each other starting from the signing of the SAA until accession, measured in annual quarters. Key events for Croatia (HR) and Serbia (RS) are highlighted.

FIGURE 2. Croatian Accession Timeline

Source: ‘Croatia’, European Commission, 2016, available at: https://ec.europa.eu/neighbourhood-enlargement/countries/detailed-country-information/croatia_en, accessed 29 September 2020.

FIGURE 3. Serbian Accession Timeline

Source: ‘Serbia’, European Commission, 2016, available at: https://ec.europa.eu/neighbourhood-enlargement/countries/detailed-country-information/serbia_en, accessed 29 September 2020.

FIGURE 4. Croatian and Serbian Accession Timelines (Comparative)

Sources: ‘Croatia’, European Commission, 2016, available at: https://ec.europa.eu/neighbourhood-enlargement/countries/detailed-country-information/croatia_en, accessed 29 September 2020; ‘Serbia’, European Commission, 2016, available at: https://ec.europa.eu/neighbourhood-enlargement/countries/detailed-country-information/serbia_en, accessed 29 September 2020.

As Figures 2–4 show, although both countries applied for membership at almost the same time relative to the signing of their SAAs, with Croatia signing its SAA in October 2001 and Serbia in April 2008, they began to diverge strongly thereafter. By the time Croatia began negotiations, relative to the signing of its AA, Serbia was only being recognised as a candidate, and as Croatia signed its accession treaty, Serbia had only managed to close two chapters of the acquis. Thus, although they began with similar prospects of membership, relative to their signing of the SAA, Croatia proved much more successful in adapting to the requirements of accession and in learning how to become an EU member state.

Networks and accession progress

As mentioned above, during the accession negotiations both Croatia and Serbia underwent four elections. Tables 4 and 5 give the dates and outcomes of each election along with the network values computed. The seat distribution was collected for the end of each parliamentary period to include the formation of committees and any changes in the list of MPs throughout the life cycle of the parliament. As can be seen, both countries shared significant similarities in overall structure of the networks with very similar graph density and clustering coefficients throughout much of the period.

TABLE 4 Croatian Elections and Network Values

Value	2000–2003	2003–2007	2007–2011	2011–2013*
Ruling coalition	SDP, LIBRA, PGS, SBHS, HSS, IDS, LS, SNS	HDZ, HSLS, DC, HSS, HSU, SDSS	HDZ, HSS, HSLS, HSU, SDSS	Kukuriku koalicija (SDP, HNS–LD, IDS, HSU), SDSS
Seats (out of total)	84 (151)	81 (150)	77 (153)	76 (151)
Number of parties in the Sabor	17	14	12	19
Average weighted degree	41.896	60.144	85.064	69.254
Graph density	0.711	1.111	2.161	1.944
Modularity	0.522	0.341	0.41	0.408
Average clustering coefficient	0.684	0.727	0.803	0.8

Note: *This sample was taken at the point of accession even though elections did not take place until 2015.

TABLE 5 Serbian Elections and Network Values

Value	2008–2012	2012–2014	2014–2016	2016–2020
Ruling coalition	DS, G17+, LSV, SDP, SPO, DSHV, SPS, PUPS, JS, PVS, SVM, BDSS, PZDD, DLR, SLPS	SNS, NS, PSSBK, NSS, PS, PVS, PPPS, KUIRS, APPS, BNS, DPM, PVU, RP, SPS, PUPS, JS, PŽK, SDPS, SDA	SNS, SDPS, NS, SPO, PS, PSSBK, DHSS, SPS, PUPS, JS, USS, SVM, PZDD, BNS	SNS, SDPS, PUPS, NS, SPO, PS, SNP, PSSBK, BS, SPS, JS, ZS, KP, USS, NSS, POKS, SVM, PDD, SPP
Seats (out of total)	129 (250)	128 (250)	209 (250)	128 (250)
Number of parties in the Skupština	20	38	23	33
Average weighted degree	84.413	60.791	126.638	91.641
Graph density	0.793	1.208	2.592	2.738
Modularity	0.494	0.438	0.164	0.236
Average clustering coefficient	0.687	0.75	0.886	0.873

Where the countries diverge is in their modularity values. These modularity values, which measure the extent to which subgroups exist within the overall network, are representative of the fracturing of the party system in Serbia in response to the rise of the SNS. In Croatia, although modularity does decrease across time from the high value in the first parliamentary period, it remains above the lows seen in Serbia. This is representative of the fact that in Croatia, while there are several smaller parties, there remain two dominant parties within the Sabor as opposed to the fractured structure in Serbia. Such a configuration of parties in the Sabor allowed learning to take place by creating conditions for top-heavy yet qualified leadership, unlike in Serbia.

Testing learning I: LIH

At the macro level, LIH 1 and 2 are both in evidence in each case. The process of joining the EU entitles candidate states to significant cash transfers, which are to prepare the candidate state for accession, encompassing both physical and bureaucratic infrastructure. Such assistance is also paired with the general financial inflows into a candidate state, such as through trade, which raises the overall value of accession. Table 6 gives a comparison of the state-level financial situations in Croatia and Serbia at relevant periods in the accession process and shows their relative levels of economic integration into the EU.

TABLE 6 Economic Indicators and Incentives

Metric	Croatia*	Serbia^†
GDP/capita	€5,800 (2001)	€4,280 (2007)
	€10,181 (2011)	€6,140 (2018)
GDP/capita (PPP) as percentage of EU average	51% (2001)	36.1% (2007)
	61% (2011)	40.2% (2018)
EU assistance (total, million euros)	€1,255 (2002–2011)	€2,800 (2007–2020)
EU assistance (per person)	€294.6 (2002–2011)	€417.91 (2007–2020)
Share of exports to EU in value of total exports	68.3% (2001)	59.9% (2007)
	62.3% (2011)	67.0% (2018)
Share of imports from EU in value of total imports	71.9% (2001)	56.8% (2007)
	61.8% (2011)	55.9% (2018)

Sources: *European Commission (2012, pp. 51–2); ^†European Commission (2020a, pp. 123–25).

As can be seen from Table 6, both countries received significant financial aid from the EU during their accession process and were also closely tied into the EU economy throughout the process. Both countries, having had their accession process delayed at separate points for non-cooperation with the ICTY, were also able to understand the credibility of EU threats. Thus, at the macro level, LIH 1 and 2 do hold true in both cases, as both countries had experienced the credible rewards of compliance and credible punishments for non-compliance.

However, as stated above, the structure and the character of Croatia’s and Serbia’s informal networks problematise this at the network level. Figures 5 and 6 show the networks in Croatia and Serbia during the first electoral period in both countries when they took their initial steps towards EU integration.⁶

FIGURE 5. Croatia 2000–2003

Note: Key—SDP (red), HDZ (blue), HSS (green), HSLS (yellow), HB (black), HSP (light blue), LS (orange), IDS (light green).

Source: Mesarovich (2023).

FIGURE 6. Serbia 2008–2012

Note: Key—SRS (green), DS (purple), G17 (blue), DSS (red), LDP (dark purple), SPS (light blue), NS (pink), PUPS (light green)

Source: Mesarovich (2023).

As Figures 5 and 6 show, in this early period both parliaments were quite fractured. The Croatian network was split between the HDZ, the ruling party until 2000, and the SDP–LIBRA–HSS coalition. Serbia was governed by the DS coalition, involving a host of other smaller parties, all united against the SRS, the party of convicted war criminal Vojislav Šešelj. As stated in Tables 4 and 5, the modularity scores for Croatia from 2000 to 2003 was 0.522 and in Serbia from 2008 to 2012 it was 0.494, the highest scores, as these were the most fragmented parliaments throughout accession. This contrasts with the situation in the third parliamentary period in each country (2008–2011 in Croatia and 2014–2016 in Serbia), as shown in Figures 7 and 8.

FIGURE 7. Croatia 2008–2011

Note: Key—HDZ (blue), SDP (red), HSS (green), HNS-LD (yellow), HDSSB (purple), SDSS (light blue), IDS (light green), HSP (dark purple).

Source: Mesarovich (2023).

FIGURE 8. Serbia 2014–2016

Note: Key—SNS (red), SPS (blue), DS (green), PUPS (pink), SDPS (brown), JS (purple), SVM (orange), NS (teal).

Source: Mesarovich (2023).

By this period the party constellations in the countries diverged significantly, with the Serbian parliament being almost entirely dominated by the SNS. Serbian accession negotiations began to stall in the 2014–2016 period as well, while Croatia completed its accession negotiations in June 2011. Significantly, the former prime minister of Croatia, Ivo Sanader, was arrested for corruption in 2010, while Serbia failed to implement the recommendations of the Venice Commission regarding rule of law reforms (European Commission 2015a).⁷ This is representative of the impact of the infiltration of informal networks into the political class in Serbia and how it interfered with learning under LIH/I 1 and 2.

In Croatia and Serbia, party structures were hierarchical and advancement was linked to loyalty to the leader rather than individual competence or constituency work.⁸ Yet, in Croatia these parties, which an interviewee described as ‘communist’ style,⁹ acted to enable learning under LIH by ensuring that the accession requirements were enforced from top to bottom.¹⁰ Opposition to policy was not allowed but, informally, if an MP was unwilling to vote against their party, they were often allowed to ‘smoke a cigarette’ or go for a coffee rather than being present for the vote.¹¹ In Serbia, this hierarchical structure resulted in a ‘pyramidal system of psychopaths’¹² wherein, in addition to the ‘carrot’ of political advancement, the ‘stick’ of violence was used to ensure loyalty (Dragoljović 2018; Đurđević 2019).¹³ This reduced the available space for political opposition and helped ensure that the party line was followed in all regards.

Both countries, then, had highly personalised hierarchical party systems but with different accession outcomes. In Serbia, the combination of informal rewards and punishments resulted in the noted phenomenon of preletanje (‘flying over’), where individuals would leave smaller parties to join larger ones, which further reduced the available space to question and critique government policy.¹⁴ In Croatia, instead, there was informal support among opposition MPs to the extent that the Kukuriku koalicija (‘cock-a-doodle-doo’) coalition, which led Croatia into the EU, emerged through an informal meeting of party leaders in a restaurant of that name (Mencinger 2016).¹⁵ While there have been some attempts within Serbia to form such electoral coalitions, none as of the time of writing have been successful in the face of the dominance of the SNS.

In terms of policy, this divergence can be seen clearly in the pace of judicial reforms. Such reform in Serbia has proceeded significantly more slowly than it did in Croatia, with Chapters 23 (Judiciary and Fundamental Rights) and 24 (Justice, Freedom and Democracy) of the acquis opened third and fourth in the succession of opened chapters and yet not closed as of 2020. This is due, in part, to the Commission’s new ‘fundamentals first’ approach, which places rule of law reform at ‘the heart of the enlargement process’ (European Commission 2013b, p. 2). Despite ongoing constitutional reform, progress has been slow, and the acrimonious political environment has impeded the implementation of recommendations from the Venice Commission. There has been a repeated failure by politicians to engage meaningfully with stakeholders through the mandated legislative consultation process, resulting in poor-quality legislation which is not in line with the required standards on judicial independence (European Commission 2018a, p. 14; 2019a, p. 16; 2020a, p. 20).¹⁶ This delay has allowed well-connected individuals to continue using political influence to protect their interests as well as those within their networks (Kmezić 2020).

This contrasts with Croatia, where Chapters 23 and 24 were among the final ones to be opened and were also closed relatively rapidly, both within one year. While, again, these chapters were not opened first, as Croatian accession occurred before the ‘fundamentals first’ approach, it is notable how quickly they closed. This is due to the importance of the hierarchical power structure as these changes were only possible after the retirement of the corrupt Prime Minister Ivo Sanader and the arrest and removal of those associated with him from positions of power by his successor, Jadranka Kosor (Šeperić 2011, p. 468). Once the crucial veto actors were removed, the reformist actors, headed by Kosor, were able to ensure rapid compliance with EU regulations.

Thus, in Croatia the highly personalised and authoritarian party structure was used to ensure compliance with EU rules and enabled learning through repeated compliance, confirming conditions LIH/I 1 and 2 were in place. Conversely, in Serbia, the evidence of hierarchical learning is mixed, with only slow and limited successes in certain chapters of the acquis. While Serbia managed to rapidly close certain specific chapters, others, such as 23 and 24, which run counter to the interests of powerful networks, remain open. All together 18 chapters of the acquis have been opened, and two closed, as the highly credible rewards and threats from the informal networks obstructed learning.

Tested learning: LEC

Turning to the conditions for LEC, which takes place under conditions of lower problem tractability, the conditions for successful learning were partially met by Croatia and not at all by Serbia. As LEC/I 1 states, the expert actors must be considered as experts by the relevant policymakers. In Croatia, EU officials were taken seriously by Croatian actors in their repeated meetings with high-level officials.¹⁷ This held true for Serbia as well.¹⁸

While this meets LEC/I 1, the EU’s representatives were not always interacting with the right people informally in the Sabor or Skupština as per LEC/I 2. To illustrate this, Figures 9 and 10 display the network maps generated for the second parliamentary periods (2003–2007 in Croatia and 2012–2014 in Serbia) when negotiations opened in both Croatia and Serbia. The Croatian political system remained split, with two clear camps: the ruling HDZ and opposition SDP. Party leaders were positioned towards the centre of the map and, generally, had higher betweenness centrality (a measure of the connectedness of individuals in a network) as shown by the larger size of the nodes. In Figure 10 the SNS, which had split from the SRS in the 2008–2012 period, forms a central bloc on the map with the opposition DS situated on the bottom right.

FIGURE 9. Sabor Network 2003–2007

Note: Key—HDZ (blue), SDP (red), HNS-LD (yellow), HSS (brown), HSP (light purple), IDS (light green), SDSS (dark purple), HSU (pink), HSLS (orange), HDSSB (purple).

Source: Mesarovich (2023).

FIGURE 10. Skupština Network 2012–2014

Note: Key—SNS (red), DS (light blue), DSS (green), SPS (light green), LDP (pink), G17 (brown), PUPS (purple).

Source: Mesarovich (2023).

As can be seen from Figure 9, the Croatian political space remained significantly more competitive during the 2003–2007 period. Due to this high level of political competition, parties in Croatia were forced to appoint high-ranking and qualified MPs to positions on important committees.¹⁹ This allowed these committees to summon relevant ministers to give evidence, allowing learning to diffuse throughout the parliament.²⁰ In the period 2012–2014, accession momentum in Serbia was continued from the previous period when, as shown in Figure 6, the fractured political space allowed for learning since, as in Croatia, parliament and its organs continued to be a site of problem-solving and contestation.²¹

As Figures 11 and 12 show, by the fourth parliamentary periods (2011–2013 in Croatia and 2016–2020 in Serbia), the situation in Serbia had deteriorated while Croatia was readying itself for accession. Croatia continued its successful pattern of implementing reforms and acceded to the EU in 2013, as qualified individuals were still appointed to the relevant committees, which continued the trend of the previous parliament.²² By contrast, reforms of Serbia’s judicial system and public administration stalled while debate within the Skupština was curtailed (European Commission 2016a, 2018a, 2019a). This lack of debate was part of the increasing concentration of power in the hands of party leaders and a reliance on more informal decision-making.²³ Such an approach forestalled the learning process by creating alternative, informal sites of decision-making from which EU representatives and pro-EU individuals were excluded.

FIGURE 11. Croatia 2011–2013

Note: Key—SDP (red), HDZ (blue), HNS-LD (yellow), HDSSB (purple), HLSR (brown), HSU (pink), SDSS (grey), IDS (green), HSP-Ante Starčević (black).

Source: Mesarovich (2023).

FIGURE 12. Serbia 2016–2020

Note: Key—SNS (red), SRS (green), SPS (dark green), DS (pink), SDPS (brown), PUPS (blue), Dveri (orange).

Source: Mesarovich (2023).

To return more explicitly to LEC/I 2, in both cases the structure of the informal networks was constructed to block wider learning. Formally, the structures of both parliaments should have helped foster LEC by providing a space for discussion as each law is required to pass the committee stage before going to the plenary session. In Croatia, this worked to a limited extent in key committees. The Committee for European Integration (Odbor za eurposke integracije) and the National Committee (Nacionalni odbor) were at the heart of the accession process. As such, they were staffed with high-ranking and competent individuals who could, and did, summon relevant individuals to give evidence and could, thus, learn from these individuals in the process of scrutinising legislation.²⁴ However, the information contained within these sessions did not lead to wider learning due to informal constraints. The concerns of MPs were frequently overruled, and they were made to toe the party line in some instances.²⁵ While these committees were formally open to all MPs, they were often, informally, closed, even to MPs who wanted just to observe the deliberations.²⁶ As such, the same aspects which enabled learning under LIH, in the meeting of targets and in the observed rapid passage of legislation, constrained wider learning under condition LEC/I 2 in Croatia.

This was repeated in Serbia, where LEC/I 2 was not borne out. With the rise of the SNS, the individuals around President Aleksandar Vučić also became more powerful. Along with the rest of core SNS individuals, they were bonded through their time in opposition as members of the SRS before breaking off to form the SNS in 2008.²⁷ Such loyalty to the centre, combined with relatively weak party institutions, allowed for the election of unqualified MPs, chosen for their loyalty from among friends and associates of party elites rather than by virtue of their expertise on a subject.²⁸ These MPs can obstruct the process of learning as they can actively obstruct the work of committees or passively obstruct the work through a lack of expertise or understanding.²⁹ Such practices interfere with learning under LEC/I conditions, as they can actively or passively interfere with the transfer of knowledge throughout the network.

The impact of informal networks in both countries, then, has acted to impede LEC under condition LEC/I 2. This is not to say that no learning took place, as Croatia did accede to the EU at the end of 2013 and Serbia closed several chapters of the acquis between 2016 and 2020, but the above results do address some concerning trends. Despite the intensity of the accession process, the EU post-accession evaluation concluded that the process did not enable wider ‘whole-of-government’ reform in Croatia (ADE 2019, p. ii). This is corroborated by public rule of law conflicts during the closing months of accession in Croatia, particularly when the government tried to block the extradition of a Yugoslav and subsequently Croatian intelligence officer to stand trial in Germany (Šabić 2019). Incidents such as this, coupled with post-accession backsliding on issues of democracy and corruption (Čepo 2020), highlight the shallowness of reforms rather than the deeper learning which would be expected from successful LEC.

In Serbia, the dominance of the SNS and its associated networks has largely stalled the accession process. Although more chapters have been opened, only two have been closed. Revealingly, the first chapter to be opened, Chapter 32 (Financial Control), remains open. Since it pertains to the effective management of state and EU resources and the implementation of decentralised management principles in public administration, it runs directly counter to the interests of the informal networks and clans which dominate Serbian political parties. Although there are repeated meetings and discussions on the subject in the Skupština, policymaking power is not invested in individual MPs which prevents the knowledge produced within these meetings and discussions from diffusing widely and triggering broader learning. Indeed, MPs are often simply political appointees or individuals seeking financial benefits, who are not particularly interested in public administration reform and who can then serve to disrupt the committee meetings (Keil 2018; Castaldo 2020),³⁰ further hampering the LEC process.

All these difficulties in the Skupština, plus other informal practices like allowing ‘do-over’ votes if the government did not get sufficient support from allied parties, led opposition parties to boycott parliamentary sessions from 2019 and elections in 2020.³¹ While this boycott has been, so far, unsuccessful, it may contain the same seeds of cooperation which helped the Kukuriku koalicija to power in the 2011 election following an informal meeting between Croatian party leaders.

Conclusion

As predicted in both cases, informal political networks have had an ambivalent impact on accession, enabling certain types of action and constraining others. LIH in both cases was fostered, to an extent, as the structure of the accession process aligned with the structure of informal networks within the countries. The overarching hierarchy of the accession process, with the EU acting as a source of rules, created a nested hierarchy where knowledge was passed down from the EU to high-ranking individuals. This allowed for the rapid passage of legislation in both countries through the use of urgent procedures to pass EU-related bills (European Commission 2010a, p. 6; 2011a, p. 5; 2016a, p. 7), and by informally ensuring a minimum of dissent from MPs.

However, the broader impact of informal networks on the learning process is less positive. Top-down control of political parties by a strong executive personality, unsurprisingly, makes for very personal politics. This enabled the formation of the broad Kukuriku koalicija in Croatia but has, so far, hampered the growth of a successful opposition in Serbia beyond the 2020 electoral boycott. While the failure of Serbia to accede to the EU cannot be entirely blamed on informal networks, the current state of the country, with a pliant SNS-dominated legislature after all the main opposition parties boycotted the 2020 election, does highlight the danger that such networks pose to any body politic. Formally Serbia remains committed to accession,³² but it is unclear how the informal networks will continue to affect the process. However, as demonstrated here, it is important to reckon with these networks in order to understand how polities learn and adapt going forwards.

Disclosure statement

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

Additional information

Funding

This work was supported by the University Association for Contemporary European Studies (UACES) under the terms of a UACES-James Madison Charitable Trust (JCMT) Scholarship. The author would like to thank Dr Theofanis Exadaktylos for his comments on an earlier version of this manuscript and the two reviewers for their insightful comments.

Notes on contributors

Alexander Mesarovich

Alexander Mesarovich, School of Social and Political Science, University of Edinburgh, Chrystal Macmillan Building, 15a George Square, Edinburgh EH8 9LD, UK. Email: [email protected]

Notes

1 For a history of policy learning as a discipline, see Dunlop et al. (2018).

2 There have been discussions of further partitions in the region (Rudic 2018), but, as of publication, these ideas have not gained mainstream support.

3 The seat distribution generated from this process can be found in Appendix 2.

4 This is used without prejudice to the naming or existence of the language group.

5 ‘Meeting of the Head of EU Delegation and the President of Serbia’, EU in Serbia, 13 April 2021, available at: http://europa.rs/meeting-of-the-head-of-eu-delegation-and-the-president-of-serbia/?lang=en, accessed 27 April 2021.

6 All SNA maps generated according to the following specifications: Yifan Hu Proportional, Optimal Distance: 100, Relative Strength: 20, Initial Step Size: 20, Step ratio: 0.95.

7 ‘Croatia Jails Former PM Sanader for Party Slush Funds’, Reuters, 13 November 2020, available at: https://www.reuters.com/article/us-croatia-corruption-sanader-idUSKBN27T1BZ, accessed 21 May 2021.

8 Serbia interview 4: former Serbian MP, Belgrade, 18 December 2019; Croatia interview 5: former Croatian MP, conducted via Zoom, 4 March 2021; Serbia interview 6: Serbian politician, Belgrade, 18 February 2020.

9 Croatia interview 1: former Croatian MP, conducted via WhatsApp, 19 May 2020.

10 Croatia interview 5: former Croatian MP, conducted via Zoom, 4 March 2021.

11 Croatia interview 5: former Croatian MP, conducted via Zoom, 4 March 2021.

12 Serbia interview 1: former Serbian MP, Belgrade, 4 June 2019.

13 Serbia interview 2: former Serbian MP, Belgrade, 7 June 2019.

14 Serbia interview 4: former Serbian MP, Belgrade, 18 December 2019.

15 ‘Zašto baš “Kukuriku koalicija”?’, Nova.hr, 10 December 2011, available at: https://web.archive.org/web/20111210223345/http://odluka2011.dnevnik.hr/clanak/zasto-bas-kukuriku-koalicija.html, accessed 19 August 2021.

16 Serbia interview 6: Serbian politician, Belgrade, 18 February 2020.

17 Croatia interview 2: former EU actor in Croatia, conducted via Zoom, 31 July 2020; Croatia interview 3: Croatian politician, conducted via telephone, 20 November 2020.

18 Serbia interview 7: foreign civil servant, Belgrade, 19 February 2020; Serbia interview 9: former Serbian MP, Belgrade, 20 February 2020.

19 Croatia interview 6: former Croatian MP, conducted via Zoom, 12 March 2021.

20 Croatia interview 4: Croatian politician, conducted via email, 23 April 2021.

21 Serbia interview 3: Serbian politician, Belgrade, 17 December 2019; Serbia interview 4: former Serbian MP, Belgrade, 18 December 2019.

22 Croatia interview 5: former Croatian MP, conducted via Zoom, 4 March 2021; Croatia interview 6: former Croatian MP, conducted via Zoom, 12 March 2021.

23 Serbia interview 1: former Serbian MP, Belgrade, 4 June 2019; Serbia interview 4: former Serbian MP, Belgrade, 18 December 2019.

24 Croatia interview 4: Croatian politician, conducted via email, 23 April 2021; Croatia interview 6: former Croatian MP, conducted via Zoom, 12 March 2021.

25 Croatia interview 2: former EU actor in Croatia, conducted via Zoom, 31 July 2020; Croatia interview 6: former Croatian MP, conducted via Zoom, 12 March 2021.

26 Croatia interview 5: former Croatian MP, conducted via Zoom, 4 March 2021.

27 Serbia interview 2: former Serbian MP, Belgrade, 7 June 2019; Serbia interview 8: Serbian MP, Belgrade, 19 February 2020.

28 Serbia interview 1: former Serbian MP, Belgrade, 4 June 2019.

29 Serbia interview 3: Serbian politician, Belgrade, 17 December 2019; Serbia interview 5: former Serbian politician, Belgrade, 18 December 2019.

30 Serbia interview 3: Serbian politician, Belgrade, 17 December 2019; Serbia interview 5: former Serbian politician, Belgrade, 18 December 2019.

31 Serbia interview 2: former Serbian MP, Belgrade, 7 June 2019. See also, ‘Serbian Opposition Says to Boycott Parliament, Demands Snap Election’, Reuters, 11 February 2019, available at: https://www.reuters.com/article/serbia-politics-idINKCN1Q01FZ, accessed 27 April 2021.

32 ‘Meeting of the Head of EU Delegation and the President of Serbia’, EU in Serbia, 13 April 2021, available at: http://europa.rs/meeting-of-the-head-of-eu-delegation-and-the-president-of-serbia/?lang=en, accessed 27 April 2021.

Appendix 1.

List of interview subjects

Croatia interview 1: former Croatian MP, conducted via WhatsApp in English by the author, 19 May 2020.

Croatia interview 2: former EU actor in Croatia, conducted via Zoom in English by the author, 31 July 2020.

Croatia interview 3: Croatian politician, conducted via telephone in English and Croatian by the author, 20 November 2020.

Croatia interview 4: Croatian politician, conducted via email in English by the author, 23 April 2021.

Croatia interview 5: former Croatian MP, conducted via Zoom in English by the author, 4 March 2021.

Croatia interview 6: former Croatian MP, conducted via Zoom in English by the author, 12 March 2021.

Serbia interview 1: former Serbian MP, conducted in Belgrade in English and Serbian by the author, 4 June 2019.

Serbia interview 2: former Serbian MP, conducted in Belgrade in English by the author, 7 June 2019.

Serbia interview 3: Serbian politician, conducted in Belgrade, in Serbian by the author, 17 December 2019.

Serbia interview 4: former Serbian MP, conducted in Belgrade in English and Serbian by the author, 18 December 2019.

Serbia interview 5: former Serbian politician, conducted in Belgrade in Serbian by the author, 18 December 2019.

Serbia interview 6: Serbian politician, conducted in Belgrade in English by the author, 18 February 2020.

Serbia interview 7: foreign civil servant, conducted in Belgrade in English by the author, 19 February 2020.

Serbia interview 8: Serbian MP, conducted in Belgrade in Serbian by the author, 19 February 2020.

Serbia interview 9: former Serbian MP, conducted in Belgrade in Serbian by the author, 20 February 2020.

Appendix 2.

Party and seat distributions in Croatia and Serbia

TABLE A1 Seat Distribution in the Sabor 2000–2003

Party (Gov)	Seats	Party (Opposition)	Seats
SDP (Socijaldemokratska partija Hiratsuka) (A)	44	HDZ (Hrvatska demokratska zajednica)	33
LIBRA (LIBRA—Stranka liberalnih demokrata) (A)	10	HSLS (Hrvatska socijalno-liberalna stranka)	14
SBHS (Slavonsko-baranjska hrvatska stranka) (A)	1	HB (Hrvatski blok)	5
PGS (Primorsko goranski savez) (A)	2	HSP (Hrvatska stranka prava) (C)	4
HSS (Hrvatska seljačka stranka) (B)	16*	HKDU (Hrvatska kršćansko-demokratska unija) (C)	1
LS (Liberalna stranka) (B)	4^†	DC (Demokratski centar)	3
HNS (Hrvatska narodna stranka) (B)	2	HIP (Hrvatski istinski preporod)	3
IDS (Istarski demokratski Sabor) (B)	4	HND (Hrvatski nezavisni demokrati)	1
SNS (Srpska narodna stranka) (M)	1	Nezavisni (unaffiliated)	2
		Nezavisni (unaffiliated) (M)	1
Totals	84		67

Notes: (A) SDP–HSLS–SBHS–PGS Coalition; (B) HSS–LS–HNS–IDS Coalition; (C) HSP–HKDU Coalition; (M) Minority Party. Unaffiliated MPs are not part of any party but group with others. *Includes one minority MP, Zdenka Čunhil, from the Czech and Slovak minority community. ^†Includes two minority MPs, Tibor Santo and Borislav Graljuk, of the Hungarian and other combined minority communities respectively.

TABLE A2 Seat Distribution in the Sabor 2003–2007

Party (Gov)	Seats	Party (Opposition)	Seats
HDZ (Hrvatska demokratska zajednica)	62	SDP (Socijaldemokratska partija Hrvatske) (B)	30
HSLS (Hrvatska socijalno-liberalna stranka) (A)*	3	IDS (Istarski demokratski Sabor) (B)	4
DC (Demokratski centar) (A)	1	HNS–LD (Hrvatska narodna stranka– liberalni demokratni) (C)^†	11
HSS (Hrvatska seljačka stranka)	9	SBHS (Slavonsko-baranjska hrvatska stranka) (C)	1
HSU (Hrvatska stranka umirovljenika)	3	PGS (Primorsko goranski savez) (C)	1
SDSS (Samostalna demokratska srpska stranka) (M)	3	HSP (Hrvatska stranka prava)	5
		HDSSB (Hrvatski demokratski savez Slavonije i Baranje)^‡	3
		Nezavisni (unaffiliated)	13
		SDAH (Stranka demokratske akcije Hrvatske) (M)	1
		Nezavisni (unaffiliated) (M)	3
Totals	81		72

Notes: (A) HSLS–DC Coalition; (B) SDP–IDS–SLD–LS Coalition; (C) HNS–SBHS–PGS Coalition; (M) Minority Party. Unaffiliated MPs are not part of any party but group with others. *Absorbed the LS in 2006. †Formed through the merger of the HNS and LIBRA parties. ‡Formed when three HDZ members formed a new regional party in 2006.

TABLE A3 Seat Distribution in the Sabor 2007–2011

Party (Gov)	Seats	Party (Opposition)	Seats
HDZ (Hrvatska demokratska zajednica)	65	SDP (Socijaldemokratska partija Hrvatske)	53
HSS (Hrvatska seljačka stranka) (A)	6	HNS–LD (Hrvatska narodna stranka–Liberalni demokrati)	5
HSLS/IND (Hrvatska socijalno-liberalna stranka/Nezavisni) (A)	2	HDSSB (Hrvatski demokratski savez Slavonije i Baranje)	4
SDSS (Samostalna demokratska srpska stranka) (M)	3	IDS (Istarski demokratski Sabor):	3
HSU (Hrvatska stranka umirovljenika)	1	HSP (Hrvatska stranka prava)	1
		HLSR (Hrvatski laburisti-Stranka rada)	1
		HSD (Hrvatski socijaldemokrati)	1
		Nezavisni (unaffiliated)	3
		SDAH (Stranka demokratske akcije Hrvatske) (M)	1
		Nezavisni (unaffiliated) (M)	4
Totals	77		76

Notes: (A) HSS–HSLS Coalition; (M) Minority Party. Unaffiliated MPs are not part of any party but group with others.

TABLE A4 Seat Distribution in the Sabor 2011–2013

Party (Gov)	Seats	Party (Opposition)	Seats
SDP (Socijaldemokratska partija Hrvatske) (A)	58	HDZ (Hrvatska demokratska zajednica) (B)	43
HNS–LD (Hrvatska narodna stranka–liberalni demokrati) (A)	14	HGS (Hrvatska građanska stranka) (B)	2
HSU (Hrvatska stranka umirovljenika) (A)	4	HLSR (Hrvatski laburisti–Stranka rada)	6
IDS (Istarski demokratski Sabor) (A)	2	HDSSB (Hrvatski demokratski savez Slavonije i Baranje)	7
SDSS (Samostalna demokratska srpska stranka) (M)	3	HSS (Hrvatska seljačka stranka)	1
		HSP–AS (Hrvatska stranka prava dr. Ante Starčević)	1
		ID (Istarski demokrati)	1
		Nezavisni (unaffiliated)	5
		Bošnjačka demokratska stranka Hrvatske (M)	1
		Nezavisni (unaffiliated) (M)	3
Totals	81		70

Notes: (A) Kukuriku koalicija; (B) HDZ–HGS Coalition; (M) Minority Party. Unaffiliated MPs are not part of any party but group with others.

TABLE A5 Seat Distribution in the Skupština 2008–2012

Party (Gov)	Seats	Party (Opposition)	Seats
DS (Demokratska Stranka) (A)	64	SRS (Srpska radikalna stranka)	77
G17+ (A)	23	DSS (Demokratska stranka Srbije) (C)	20
LSV (Liga socijaldemokrata Vojvodine) (A)	5	NS (Nova Srbija) (C)	9
SPO (Srpski pokret obnove) (A)	4	LDP (Liberalno demokratska partija)	12
SPS (Socijalistička partija Srbije) (B)	11	DHSS (Demohrišćanska Stranka Srbije)	1
PUPS (Partija ujedinjenih penzionera Srbije) (B)	5	SDU (Socijaldemokratska unija)	1
JS (Jedinstvena Srbija) (B)	3	Nezavisni (unaffiliated)	1
SDPS (Socijaldemokratska partija Srbije)	4
DSHV (Demokratski savez Hrvata u Vojvodini)	1
PVS (Pokret veterana Srbije)	1
SVM (Savez vojvođanskih Mađara) (M)	4
BDSS (Bošnjačka demokratska stranka Sandžaka) (M)	1
PZDD (Partija za demokratsko delovanje) (M)	1
DLR (Demokratska levica Roma) (M)	1
SLPS (Socijalno liberalna partija Sandžaka) (M)	1
Totals	129		121

Notes: (A) For a European Serbia (Za Evrposku Srbiju); (B) SPS–PUPS–JS; (C) DSS–NS; (M) Minority Party. Unaffiliated MPs are not part of any party but group with others.

TABLE A6 Seat Distribution in the Skupština 2012–2014

Party (Gov)	Seats	Party (Opposition)	Seats
SNS (Srpska napredna stranka) (A)	53	DS (Demokratska Stranka) (C)	44
NS (Nova Srbija) (A)	7	G17+ (C)	10
APPS (Asocijacija malih i srednjih preduzeća i preduzetnika Srbije) (A)	1	LSV (Liga socijaldemokrata Vojvodine) (C)	3
KUIRS (Koalicija udruženja izbjeglica u Republici Srbiji) (A)	1	ZS (Zeleni Srbije) (C)	1
PSSBK (Pokret Snaga Srbije – BK) (A)	2	DSHV (Demokratski savez Hrvata u Vojvodini) (C)	1
NSS (Narodna Seljačka Stranka) (A)	1	DHSS (Demohrišćanska Stranka Srbije) (C)	1
BNS (Bošnjačka narodna stranka) (A) (M)	1	LDP (Liberalno demokratska partija) (D)	11
DPM (Demokratscka partija Makedoniaca) (A) (M)	1	SPO (Srpski pokret obnove) (D)	4
PUV (Pokred Ujedinjenih Vlaha) (A) (M)	1	SDU (Socijaldemokratska unija) (D)	1
RP (Romani partija) (A) (M)	1	BS (Bogata Srbija) (D)	1
PS (Pokret socijalista) (A)	1	DSS (Demokratska stranka Srbije)	18
PPPS (Pokret privredni preporod Srbije) (A)	1	ZZŠ (Zajedno za Šumadiju)	2
SPS (Socijalistička partija Srbije) (B)	18	NOPO (Nijedan od ponuđenih odgovora)	1
PUPS (Partija ujedinjenih penzionera Srbije) (B)	10	BI	1
JS (Jedinstvena Srbija) (B)	7	UG	1
SDPS (Socijaldemokratska partija Srbije)	9	SVM (Savez vojvođanskih Mađara) (M)	5
PŽK (Pokret Živim za Krajinu)	2	BDZ (Bosanska Demokratska Zajednica) (M)	1
PVS (Pokret veterana Srbije)	1	Nezavisni (unaffiliated)	16
SDA (Stranka demokratske akcije Sandžaka) (M)	2
Nezavisni (unaffiliated)	8
Totals	128		122

Notes: (A) Let’s Get Serbia Moving (Pokrenemo Srbiju); (B) SPS–PUPS–JS; (C) Choice For A Better Life (Izbor za Bolji Život); (D) Overturn (Preokret); (M) Minority Party. Unaffiliated MPs are not part of any party but group with others.

TABLE A7 Seat Distribution in the Skupština 2014–2016

Party (Gov)	Seats	Party (Opposition)	Seats
SNS (Srpska napredna stranka) (A)	119	DS (Demokratska stranka)	15
SDPS (Socijaldemokratska partija Srbije) (A)	10	LSV (Liga socijaldemokrata Vojvodine) (C)	5
NS (Nova Srbija) (A)	6	SDS (Socijaldemokratska stranka) (C)	4
SPO (Srpski pokret obnove) (A)	5	Nova (C)	2
PS (Pokret socijalista) (A)	3	ZS (Zeleni Srbije) (C)	1
SPS (Socijalistička partija Srbije) (B)	25	Preokret (Pokret za preokret)	1
PUPS (Partija ujedinjenih penzionera Srbije) (B)	12	SNP (Srpska narodna partija)	1
JS (Jedinstvena Srbija) (B)	7	ZZŠ (Zajedno za Šumadiju)	1
PSSBK (Pokret Snaga Srbije – BK)	2	SDA (Stranka demokratske akcije Sandžaka) (M)	3
DHSS (Demohrišćanska Stranka Srbije)	1	Nezavisni (unaffiliated)	8
USS (Ujedinjena seljačka stranka)	1
SVM (Savez vojvođanskih Mađara) (M)	6
PZDD (Partija za demokratsko delovanje) (M)	2
BNS (Bošnjačka Narodna Stranka) (M)	1
Nezavisni (unaffiliated)	9
Totals	209		41

Notes: (A) For a Future We Can Believe In (Budučnost u koju verujemo); (B) SPS–PUPS–JS; (C) New Democratic Party–Green (Nova demokratska stranka–Zeleni); (M) Minority Party. Unaffiliated MPs are not part of any party but group with others.

TABLE A8 Seat Distribution in the Skupština 2016–2020

Party (Gov)	Seats	Party (Opposition)	Seats
SNS (Srpska napredna stranka) (A)	92	SRS (Srpska radikalna stranka)	22
SDPS (Socijaldemokratska partija Srbije) (A)	10	DS (Demokratska stranka) (C)	13
PUPS (Partija ujedinjenih penzionera Srbije) (A)	9	ZZS (Zajedno za Srbiju) (C)	2
PS (Pokret socijalista) (A)	3	NS (Nova Srbija) (C)	1
SNP (Srpska narodna partija) (A)	3	LDP (Liberalno demokratska partija) (D)	4
PSSBK (Pokret snaga Srbije – BK) (A)	2	SDS (Socijaldemokratska stranka) (D)	4
SPO (Srpski pokret obnove) (A)	2	LSV (Liga socijaldemokrata Vojvodine) (D)	3
BS (Bolja Srbija) (A)	1	Dveri (Srpski pokret Dveri) (E)	6
NS (Nova Srbija) (A)	1	DSS (Demokratska stranka Srbije) (E)	2
NSS (Narodna seljačka stranka) (A)	1	NARODNA (Narodna stranka)	3
POKS (Pokret obnove Kraljevine Srbije) (A)	1	DSHV (Demokratski savez Hrvata u Vojvodini)	1
USS (Ujedinjena seljačka stranka) (A)	1	Nova (Nova Stranka)	1
SPS (Socijalistička partija Srbije) (B)	20	ZES (Zelena stranka) (M)	1
JS (Jedinstvena Srbija) (B)	6	SDA (Stranka demokratske akcije Sandžaka) (M)	2
KP (Komunistička partija) (B)	1	Nezavisni (unaffiliated)	20
ZS (Zeleni Srbije) (B)	1
SVM (Savez vojvođanskih Mađara) (M)	4
SPP (Stranka pravde i pomirenja) (M)	2
PZDD (Partija za demokratsko delovanje) (M)	1
Nezavisni (unaffiliated)	6
Totals	128		122

Notes: (A) Serbia is Winning (Srbija pobeđuje); (B) SPS–JS; (C) For A Just Serbia (Za pravednu Srbiju); (D) Alliance for a Better Serbia (Savez za bolju Srbiju); (E) Dveri–DSS; (M) Minority Party. Unaffiliated MPs are not part of any party but group with others.