Federalism,policy change,and social security in Belgium: Explaining the decentralization of family allowances in the Sixth State Reform

Introduction

Much has been written in the last two decades about social policy change (e.g. Gilbert, 2002; Humpage, 2010; Streeck and Thelen, 2005). Historical institutionalist scholars like Paul Pierson (1994, 2001) have emphasized continuity and path dependence stemming from the embedded nature of social programmes and their constituencies. Reacting to this literature, authors such as Jacob Hacker (2004) and Kathleen Thelen (2004; Mahoney and Thelen, 2010; Streeck and Thelen, 2005) have stressed the impact of incremental yet transformative changes that alter the social policy landscape over time. Yet other researchers have claimed that path-departing social policy change is so widespread that it is the rule rather than the exception (Gilbert, 2002).

In federal states, the debate on social policy change is not only about the evolving nature of social programmes but also about whether and how their territorial organization may change over time (on federalism and social policy, see Obinger et al., 2005). The territorial issue is central in Belgium, a prototypical case of ‘holding-together’ federalism (Stepan, 1999). In fact, over the last 45 years or so, Belgium has experienced a major yet incremental transformation from its original centralized unitary structures to a decentralized federal state ¹ exposed to strong centrifugal tendencies (Deschouwer, 2012; Erk, 2003; Poirier, 2011; Popelier and Cantillon, 2013; see also Cantillon et al., 2011). The main force behind these centrifugal tendencies is Flemish nationalism (e.g. Erk, 2003; Poirier and Vansteenkiste, 2000), which has successfully pushed for the decentralization of a substantial number of public policies in the country (e.g. culture, education, economic development, employment policy, local powers and telecommunications).

In Flanders, the decentralization of social programmes has been an objective not only of the explicitly nationalist parties (Nieuw-Vlaamse Alliantie (N-VA) and Vlaams Belang/Vlaams Blok) but also of the Christian Democrats, the traditionally dominant actors in Flemish politics who have long been associated with the Flemish Movement (Béland and Lecours, 2005; Poirier and Vansteenkiste, 2000). Prior to 2011, this objective was basically unfulfilled. In fact, the social insurance system known as ‘social security’ at the centre of the Belgian social protection system had remained territorially centralized. The decentralization of a system that was created long before the shift from a unitary to a federal state is an unlikely outcome for two main reasons: the attachment of most social partners to country-wide social insurance and, what is more important, the traditional opposition of Francophone politicians in an institutional context where support from a majority of French-speaking parliamentarians is needed for change (Béland and Lecours, 2005). In 2011, however, after record-long (18 months) negotiations between political parties, the agreement over the formation of the new Di Rupo coalition government led to a new state reform which, among other things, featured the decentralization of family allowances. If we consider that family allowances involve nearly €6 billion in annual benefits, they are a significant component of social security in Belgium (Rocour, 2013). ² A complex package featuring provisions on issues ranging from fiscal policy to the Senate, this Sixth State Reform is a large-scale and multifaceted endeavour, with important institutional, political and policy significance (Dumont, 2015b; Goossens and Cannoot, 2015; Palsterman, 2012; Popelier and Cantillon, 2013; Reuchamps, 2013). In this article, we focus on the family allowance provisions of the Sixth State Reform because this policy area is the first component of social security to be entirely decentralized in Belgium.

The puzzle at the centre of this article is the decentralization of a programme that is part of Belgium’s social security system. Decentralization appeared quite unlikely considering existing institutional settings and legacies, but it is now taking place as part of the Sixth State Reform, whose implementation began in 2014. Using an historical institutionalist approach that takes into account nationalist pressures on existing policy legacies, we ask why the decentralization of family allowances became possible when the institutional obstacles that had so far prevented the territorial re-organization of any part of social security were still in place. Our argument is that increasing political pressure generated by the surge of N-VA in the context of the government formation crises altered how some Francophone parties assessed the national unity situation and their own position regarding social security decentralization. This logic led to negotiated ‘quid pro quos’ (Bonoli, 2000) that helped bring about a partial decentralization of social security in Belgium.

Methodologically, the article is based on a qualitative, process-tracing approach (Bennett and Checkel, 2015) that uses the available primary and secondary evidence to follow the development of particular institutional and political dynamics over time. The research uses relevant scholarly works, media stories and official documents, such as party platforms and government publications. Moreover, in the summer of 2015, the co-authors conducted 16 semi-structured interviews with policy experts, government officials and party representatives from Flanders and the French-speaking community. These interviews were used to obtain information about the origin and the content of the Sixth State Reform, as well as the political dynamics that led to it. In 2004, the authors had conducted a similar series of interviews, which provided them with insight about the actors and the institutional configurations at the centre of the debate on the decentralization of social security in Belgium (Béland and Lecours, 2005). Although these earlier interviews are not used directly in this article, they helped the co-authors to assess the nature of the political changes that have, over the last years or so, led to the Sixth State Reform.

This article is divided into four main sections. In the first section, we explain why and how, starting in the 1980s, Flemish nationalism became concerned with social security and began to advocate its decentralization. In the second section, using insight from historical institutionalism, we discuss the factors that accounted for the absence of any territorial re-organization of social security despite strong Flemish pressures and a federalization process involving significant decentralization in many policy fields. The section ends with a brief overview of the ‘quid pro quos’ (Bonoli, 2000) institutionalist approach we apply to our case and combine with the claim that nationalistic pressures can impact policy change (Béland and Lecours, 2008). The third section develops an explanation for the decentralization of family allowances in the context of the Sixth State Reform. In the final section, we discuss this policy area and the larger political implications of the constitutional transfer of the power to legislate and administer family allowances from the federal government to the communities.

Flemish nationalism and the push to decentralize social security

From the creation of Belgium in 1830 until the 1980s, the primary focus of the Flemish Movement was the protection and promotion of the Dutch language and Flemish culture, first in Belgium as a whole, and then in Flanders specifically. The Flemish Movement’s struggle on language and culture first produced important language legislations (McRae, 1986), and then led to the federalization of the state beginning in 1970 (Deschouwer, 2012). From the Flemish perspective, Belgian federalism was always meant to be dynamic, that is, each constitutional and institutional change beginning with the first state reform of 1970 was viewed as a step in a longer process.

The Flemish push to decentralize social security emerged in the 1980s at a time when the Flemish economy had clearly and significantly surpassed that of Wallonia, which had been the economic powerhouse of the country for most of its existence. ³ Studies on the territorial and fiscal dimension of social security conducted by Flemish researchers considered social security revenues and expenditures in Wallonia and Flanders, and came to the conclusion that the centralization of all social programmes benefit Wallonia and cost Flanders (Dethée, 1984). Unsurprisingly, the Flemish nationalist parties Vlaams Blok (the former Vlaams Belang) and Volksunie were the first to explicitly denounce the seemingly negative fiscal and economic consequences of the centrally administered social security system for Flanders, and to demand its decentralization. Their political arguments were also the most straightforward: Flemings and Francophones ⁴ belong to two different nations, and social solidarity should be confined to each of these national communities (see Vlaams Blok, 2003). In the early 1990s, the Flemish Liberals (Vlaamse Liberalen en Democraten, VLD) and Christian-Democrats (Christen-Democratisch en Vlaams, CD&V) began favouring the partial decentralization of social security (Béland and Lecours, 2008: 173–4). The CD&V specifically targeted family allowances, and above all health insurance, as a first step in decentralization. The positioning of CD&V on this issue was particularly significant because it was the dominant party in Flanders and a habitual member of federal coalition governments. For the CD&V, the idea of a partial decentralization of social security (initially with a focus on child benefits and health insurance) went perfectly with its developing discourse on confederalism. In the context of this discourse, the Flemish Christian Democrats formulated a technical argument in support of the partial decentralization of social security: the idea of a need for homogeneity or the congruence of policy-making powers (Poirier and Vansteenkiste, 2000). This argument nevertheless came with the nationalist subtext of community solidarity, where Flemings were typically presented as more industrious and more resource conscious than Francophones. By the late 1990s and early 2000s, all Flemish political parties, except for Greens (first called Agalev then Groen!), supported the partial decentralization of social security. As for the Flemish Socialists, they were ambiguous but, caught in nationalist outbidding, they started to request the regionalization of significant elements of employment policy (Dumont, 2015a: 195). In 1999, the Flemish Parliament adopted a motion supporting the decentralization of health insurance and family allowances. ⁵

Francophone parties opposed the notion of decentralizing any part of social security as soon as the issue came up in Flanders (Béland and Lecours, 2008). They appealed to pan-Belgian solidarity, reminding Flemings that Wallonia was once the economic engine of the country and had therefore economically supported the North of the country. They refuted the notion propagated by some Flemish nationalists that Walloons were lazy and wasteful, ⁶ pointing out that structural conditions linked to the decline of Wallonia heavy industry were responsible for the greater costs of healthcare and employment insurance in the South than in the North. For Wallonia, the perspective of even a partial decentralization of social security was seen as a major threat, as it could place the region in a difficult fiscal position if it chose to maintain existing levels of social benefits. ⁷ More important, however, was the idea that social security was one of the last remaining pan-Belgian programmes, and that its decentralization could mean the end of Belgium as a political community. In fact, the late 1990s marked the beginning of a period of great angst about the future of the country (Martiniello and Swyngedouw, 1998), ⁸ and a centrally administered social security became, from a Francophone perspective, emblematic of Belgian national unity. For all these reasons, Francophone parties stated they would stand firm and refuse any decentralization of social security. ⁹ Furthermore, the governance structure of social security and the nature of the Belgian political system made change in the territorial structuring of social programmes difficult.

Institutional obstacles to the decentralization of social security in Belgium

One of the most debated perspectives on social policy stability and change is historical institutionalism, a theoretical school that emphasizes the impact of existing policy arrangements and political institutions on welfare state development and restructuring (Campbell, 2004; Lecours, 2005; Steinmo et al., 1992; Weaver and Rockman, 1993). Two contributions of historical institutionalism to policy analysis are particularly useful for understanding the resilience of the central management of all social security programmes up to the 2011 constitutional reform.

First, historical institutionalist scholars such as Paul Pierson (1994, 2001) have shown how historically constructed policy legacies both constrain and create opportunities for reformers. Reform agents seeking to bring about path-departing change work in a policy environment populated by existing programmes and the constituencies they have created over time. For instance, existing social programmes create large groups of beneficiaries that may, in turn, mobilize to expand or at least protect these programmes (Campbell, 2003; Pierson, 1996). Simultaneously, such programmes can shape the perceived interests and the institutional position of business and labour organizations involved in the management of such programmes. This is certainly the case for social insurance programmes, especially when these social partners are involved in their management and gain symbolic and material benefits from this institutional configuration (Béland, 2001).

In Belgium, the centralized institutional legacies of social security, including payroll contributions tied to work rather than one’s place of residence, and the position of social partners (employers and trade unions) within its governance structures are considered significant factors in the debate over the decentralization of this system (Béland and Lecours, 2005). To understand the role of the social partners, it is important to note that modern social security in Belgium was created long before the federalization of the country, which began in 1970. An extension and rationalization of the social insurance system that emerged before the Second World War, modern social security is comprised of seven main elements: family allowances, health insurance, disability insurance, old-age insurance, unemployment insurance, industrial accidents and social assistance (for an historical perspective, see Vanthemsche, 1994). National employers and labour organizations play a central role in the governance of this system, and their political legitimacy is derived from this institutional position, and from the fact that they represent workers and employers who pay the social insurance contributions at the centre of social security. In this context, one would expect these national organizations to oppose the decentralization of social security, as such a change could deprive them of much of their prestige and policy influence, as well as their control over massive social insurance contributions paid by their members (Béland and Lecours, 2005).

A survey conducted in 2010 among national business and labour organizations confirms this basic supposition (Bouteca et al., 2013). Support for the maintenance of existing territorial centralization proved especially strong among the three main national labour organizations: the Confederation of Christian Trade Unions (ACV/CSC), which is the largest in the country; the General Federation of Belgian Labour (ABVV/FGTB), in which French speakers are overrepresented; and the General Confederation of Liberal Trade Unions of Belgium (ACLVB/CGSLB), which has the smaller membership. The 2010 poll confirmed that these organizations opposed the devolution of health insurance as well as ‘Social Security initiatives at [the] state level’ (Bouteca et al., 2013: 305). On the business side, the Federation of Enterprises in Belgium (VBO/FEB) also opposes decentralization, although the level of opposition is not as high, on average, as it is among labour organizations. This polling data confirm earlier qualitative research based on interviews conducted in 2004, which pointed to the clear opposition of national business and labour organizations to any significant change in the territorial governance of social security in Belgium (Béland and Lecours, 2008). Considering the central institutional position of this national business organization and the high union density in Belgium, ¹⁰ this opposition from ‘social partners’ is meaningful and consequential. On the employer side, however, it is worth noting that Flanders’ Chamber of Commerce and Industry (VOKA), the leading Flemish business organization, advocates social security decentralization and is very much in line with the demands of the N-VA. In contrast, the key Walloon organization is less supportive of decentralization, something confirmed by the 2010 poll discussed above (Bouteca et al., 2013: 305).

Second, historical institutionalism explores how veto points inherent to political institutions may empower certain actors to shape or block reform (Immergut, 1992). The number and nature of veto points vary from one political system to the next, and the identity of the actors who hold such powerful institutional positions has direct consequences for social policy reform (Bonoli, 2000; Immergut et al., 2006; Marier, 2008). Veto points play a direct role in both unitary and federal states. In fact, federal institutions may feature a multiplicity of veto points, which can empower reform opponents and make meaningful policy change less likely (Immergut, 1992; Obinger et al., 2005).

Although Belgium is a federal state, it is the consociational nature of the political system (Deschouwer, 2012) that provides Francophones with a veto point when it comes to constitutional reform. Consociational democracy (Lijphart, 1977) is a structure whose logic is based on the various communities (or ‘segments’) that constitute the system and have the right to say ‘no’, so that no policy or legislation that could affect them is adopted without their consent. In the case of Belgium, as specified in article 99 of the Constitution, federal cabinets have to feature an equal number of French-speaking and Flemish ministers. Moreover, the decentralization of components of social security can only be implemented through ‘special laws’, which require support from a majority of parliamentarians in each language group as well as from an overall supra-majority (two-thirds) in Parliament. The Belgian political system thus gave Francophone parties an absolute veto point regarding the decentralization of social security (Béland and Lecours, 2005). Their assurances through most of the 2000s that such a change would never occur stemmed largely from a combination of political and institutional obstacles: Francophone parties appeared to be united in staunch defence of the status quo on the territorial structuring of social security, and the consociational structures of Belgian federal politics clearly gave them the tools to perpetuate this status quo if they all so wished.

Yet, in January 2014, the special laws executing, among other things, the transfer of constitutional responsibility for family allowances from the federal to community governments was promulgated. Why did this happen? As suggested in the next section, the surge of the N-VA and the political crisis resulting from long and difficult government formation negotiations in 2010–2011 featured the expression of growing dissatisfaction for Belgium in Flanders, which put tremendous pressure on French-speaking politicians to find a solution that could calm things down and fight off the notion that Flanders would be better off with some kind of confederal arrangement with the rest of the country. To end a political stalemate that seemed to undermine the political future of a country they want to keep, Francophone parties finally agreed to what they felt was a necessary yet limited concession: the decentralization of family allowances.

To shed light on this key episode, we draw on the work of Giuliano Bonoli (2000) on ‘quid pro quos’ political actors’ bargain over and agree on to bring about policy change amidst strong institutional obstacles (on coalition building and social policy, see also Häusermann, 2010; Palier, 2005). As our institutionalist analysis shows, in Belgium, negotiations over government formation are ‘critical junctures’ (Pierson, 2004) that may create a window of opportunity for social policy change when partisan actors are forced to compromise through ‘quid pro quos’ that are likely to involve the territorial question in the context of ongoing nationalist pressures in favour of decentralization emanating from Flanders. In the following analysis, we show how ‘quid pro quos’ over government formation led to the decentralization of family allowances as part of what Bruno Palier (2005) calls an ‘ambiguous agreement’, in which specific political actors find common ground to bring about policy change ground for very different reasons. To sum up, our argument is that nationalist pressures to decentralize parts of social security in Belgium (Béland and Lecours, 2008) finally brought about significant policy change through ‘quid pro quos’ (Bonoli, 2000) necessary to reach an ‘ambiguous agreement’ (Palier, 2005) over the formation of a new government and the advent of the related Sixth State Reform in 2011.

How family allowances came to become decentralized

Flamish nationalist parties have played a significant role in shaping contemporary Belgian politics. The Flemish nationalist party Volksunie was instrumental in placing federalism onto the political agenda in the 1960s and 1970s, and it was the first political party to argue for the decentralization of social security. As its popularity began to wane in the 1990s, its radical right-wing offshoot, Vlaams Blok, picked up steam with a clearer secessionist stance (as well as an anti-immigration platform), reaching 12 percent of the overall vote in the 2007 federal elections and making it the second strongest party in Flanders. The Vlaams Blok was the subject of a ‘containment’ policy (cordon sanitaire) on the part of the traditional parties, which identified it as ‘non-democratic’ (or ‘fascist’) because of its anti-immigration and anti-diversity positions. This isolation meant that the Vlaams Blok’s nationalist discourse generated little ‘outbidding’ (Horowitz, 1985) or ‘keeping up’ on the part of the other Flemish parties; its highly controversial social positions meant there was some form of natural limitation to its growth potential.

The implosion of Volksunie in 2001, primarily as a result of an internal conflict between its left-wing and right-wing factions, was tremendously consequential for Belgian politics. The demise of Volksunie led to the creation of two new parties. The first one was Spirit, a left-wing party with some nationalist tendencies whose disintegration in 2009, after a short cartel with Socialistische Partij Anders (SP.A), pushed its members to go to the traditional parties. As a result, the Flemish Socialists, Christian Democrats and Liberals all received an infusion of nationalist members. The second party coming out of the implosion of Volksunie was N-VA. N-VA emerged as a more clearly nationalist party than its predecessor. On the political future of Flanders, N-VA seeks independence, which would be reached by a gradual transfer of powers from the federal to the Flemish government and would involve a special status for Brussels and, potentially, confederal links with Wallonia. ¹¹ Ideologically, N-VA is clearly a right-wing party when it comes to economic redistribution and market regulation. ¹² Thus, the decentralization of social security could allow the N-VA to achieve part of its nationalist objective (by increasing the powers of the Flemish government) as well as its policy and socio-economic objectives because the control of some aspects of social security by the Flemish government could presumably translate into a more liberal approach to social protection.

The N-VA’s effect on Belgian politics was first felt in the federal elections of 2007. After a poor initial performance in the 2003 elections (where it won only one seat), N-VA changed its leader and formed an electoral cartel with CD&V. This 2007 election and its aftermath represented the real launch of the N-VA as a driving force in Flemish and Belgian politics. The CD&V/N-VA cartel easily won a plurality of sets (the N-VA won five seats) after having campaigned on the promise of significant state reforms. The decentralization of components of social security was part of the cartel’s state reform proposals, although this issue was overshadowed by the question of the Brussels–Halle–Vilvoorde (BHV) electoral district. ¹³ Benefitting from the charismatic leadership of Bart De Wever, the N-VA’s alliance with the traditional powerhouse of Flemish politics augmented its profile and raised the question of the party’s potential participation in a federal government for the first time. Government formation negotiations lasted about 6 months and exacerbated the widely shared perception that Belgium was a country whose political future seemed constantly in doubt. ¹⁴ The N-VA was eventually excluded from the CD&V-led federal executive, although it initially supported this government on the condition that the state reform it considered adequate be implemented. However, a few months into the life of the new government, the nationalist party voiced its dissatisfaction with the government’s approach to state reform and withdrew its support. The state reform and BHV issues continued to fester, and the second, Yves Leterme’s CD&V-led government fell in April 2010.

The governments created after the 2007 elections did not produce any constitutional change. At that point, it looked like Francophone parties would be keeping their vow to block any decentralization of social security. However, the political dynamics of this first government formation crisis would dramatically reduce their ability to do just that. In fact, the Flemish parties that were featured in the executives that governed Belgium between 2008 and 2010, the Liberals and Christian Democrats, had made state reform an important part of their respective platforms (especially CD&V). Failure to deliver state reform represented a blow to their credibility. In contrast, the N-VA profited immensely from the absence of state reform. Its leader, Bart De Wever, could argue that the other Flemish parties had been unable to deliver for Flemings, and that Francophone parties again showed their commitment to the defence of the status quo. De Wever could also ramp up his denunciations of Belgium’s partitocracy, which he argued prevented Flemings from achieving the type of constitutional and institutional change they wanted.

Additional pressure for state reform and the decentralization of important components of social security came in the form of the Flemish government’s so-called Octopus note. This document, which was presented in the Belgian Senate on 1 February 2008, built on the 1999 Flemish Parliament resolutions and stated in quite a lot of details how the Flemish government saw the future of the country and the next state reform (Werkgroep ‘Staatshervorming’, 2008). In the note, then Flemish Minister – President Kris Peeters expressed Flanders’ desire for further decentralization in a wide array of policy matters (from labour market to innovation policy) in the name of greater efficiency. On social security, the note states,

An efficient and effective distribution of Social Security resources to a level as accessible as possible, and without a negative impact on the quality of care, is only possible if the current fragmentation of competencies is remedied. In 1999, the Flemish parliament thus demanded power over the formulation of norms as well as the implementation and the financing of all health and family policies, including health insurance and family allowances. (Werkgroep ‘Staatshervorming’, 2008: 9) ¹⁵

The political importance of the Octopus note is that it was supported by two Flemish coalition governments and, therefore, bound the four major Flemish parties together (N-VA, CD&V, Open VLD and SPA). For these parties, participating in a federal executive without a commitment from Francophone parties to proceed with a state reform, including a partial decentralization of social security, would be a nearly impossible political proposition.

The first government formation crisis in 2007 and the absence of a constitutional reform between 2008 and 2010 led to N-VA’s strong performance at the June 2010 federal elections: it won 27 seats, the most of any party, not only in Flanders but also in Belgium as a whole. In French-speaking Belgium, the Parti socialiste (PS) won a plurality of seats with 26. These results produced a political situation that would make government formation extremely difficult. The N-VA certainly had a legitimate claim to be part of a government, but a party ultimately seeking the independence of Flanders governing Belgium was a hard pill to swallow for Francophone parties. More concretely, the N-VA would be a particularly difficult partner for Francophone parties because it was the Flemish party least likely to compromise on state reform and the decentralization of social security. The other Flemish parties, more specifically CD&V, had a strong interest in having N-VA in government. ¹⁶ Indeed, a repeat scenario of 2007–2010, where the N-VA stood on the side lines and criticized other Flemish parties for their inaction, was quite unattractive to them. As ‘winner’ on the Francophone side and the habitual party of federal coalitions, the PS fully expected to be part of the new government. However, the PS was a really bad fit to form a government with N-VA. Ideologically and from a public policy perspective, the PS’ social-democratic nature clashed with the N-VA’s right of centre thinking and programme. For the N-VA, the PS represented the ‘old Belgium’, the political force of the constitutional, institutional and socio-economic status quo. ¹⁷ For the PS, the N-VA was the ultimate threat to Belgian solidarity and redistributive social policies.

The government formation process primarily involved seven political parties (N-VA and the traditional parties in each community), and it featured several different politicians assuming various formal roles, as specified by the King, in the hope of finding some agreement towards a governmental programme. Bart De Wever was one of these politicians, as he was initially the informateur and later the clarificateur. ¹⁸ The major challenge was to find common ground between the N-VA, which stuck very much to the content of the Octopus note, ¹⁹ and the PS. De Wever’s clarificateur report was rejected by all Francophone parties. As the country was heading towards establishing an unofficial world record on the length of government formation, pressure on the traditional parties, especially Francophone parties, to form a government grew. In fact, this second government formation crisis in a few years seemed to give credence to the N-VA’s argument that Belgium was fundamentally inefficient and perhaps irremediably broken. In Flanders, public opinion support for N-VA was growing (De Standaard, 2011). The incentive structure for Francophone parties, and for the PS more specifically, began to change. ‘Keeping Belgium together’ now seemed to require a compromise on the partial decentralization of social security rather than a staunch refusal to budge on this issue. Moreover, as Bart De Wever could not possibly become the Prime Minister of Belgium (an inconceivable thought for most and something he was seemingly uninterested in), there was a chance that PS leader Elio Di Rupo could be the first Francophone to occupy this position in decades when he was named formateur in May 2011. Compromising on the decentralization of some component of social security was now a more attractive option for Francophone parties, especially for the PS.

Family allowances emerged as the component of social security most likely to be the decentralized. For Flemish parties, family allowances had the advantage of involving large sums of money. In the context of intense partisan competition within Flemish politics, it was crucial for the traditional Flemish parties, especially the CD&V, to be in a position to say they delivered a significant state reform without the N-VA. To make this claim plausible, it was important to be able to be more specific about the size of the reform (Dumont et al., 2015), and family allowances fulfilled this essential condition as they helped to put a hard number on the fiscal impact of the reform (Goossens and Cannoot, 2015). For Francophone parties, family allowances could be depicted as less central to social security than other parts of the system like unemployment insurance. In reality, the programme was described as essentially self-contained and involving no spill over into the more central aspects of social security, such as health insurance and unemployment insurance. ²⁰ Finally, the fact that ‘family policy’ as a field had long been devolved and that family allowances are clearly distinct from employment-related matters further reinforced the case for decentralization. ²¹

In the end, Elio Di Rupo’s proposal for a Sixth State Reform was rejected as insufficient by the N-VA but was endorsed by the five other traditional parties, which agreed to join the (Francophone) Socialists in a new government. ²² The N-VA was once again on the government’s side-lines but, this time, Flemish parties were in a position to say they had adopted the most important state reform ever, one that created more ‘homogeneous blocs of competencies’ and that moved ‘the centre of gravity’ of the federation from the federal government to the communities and regions (Dumont et al., 2015: 14, 19). Francophone parties could tell their constituents they had found a way out of a crisis that seemed to threaten the very existence of their country without sacrificing the integrity of core components of social security such as unemployment insurance. For MR, the decentralization of family allowances was not a problem from a policy point of view because family policy had never been a major component of their programme. ²³ CDH was historically much more connected to the notion of family but, finishing a distant third in French-speaking Belgium, it did not have much leverage. The PS would gain in prestige and visibility and its leader, Elio Di Rupo, became Prime Minister.

The advent of the Sixth State Reform during the bargaining leading to the formation of the Di Rupo government is a typical example of how ‘quid pro quos’ (Bonoli, 2000) can help bring about policy change amidst powerful institutional obstacles. Indeed, in order to secure the consent of the traditional Flemish parties to participate in a new government, French-speaking parties had to ‘give something away’ as a response to ongoing Flemish territorial demands. The decentralization of family allowances became a key aspect of an ‘ambiguous agreement’ (Palier, 2005) that the different partners involved in the institutional deal could depict as a ‘victory’ to their supporters.

The Sixth State Reform: policy and political consequences

The Belgian state’s Sixth State Reform has important policy and political implications. This reform is complex and multifaceted. It includes a reform of the Senate; splits the controversial BHV/Bruxelles–Hal–Vilvorde electoral district located at the edge of the capital region in two; involves a significant increase in regional fiscal autonomy; and, finally, features a new transfer of powers to the constituent units of the Belgian federation (Goossens and Cannoot, 2015). For the first time since the federalization process began in 1970, this transfer of powers directly affects social security (Dumont, 2015a: 175, 187–9). The Sixth State Reform directly affects three main social policy areas: labour market policy, healthcare and family allowances, a component of social security previously under the exclusive control of the federal state. ²⁴

The section of the government agreement on the Sixth State Reform pertaining to the transfer of family allowances to the communities is less than a page long and offers relatively few details about implementation. Interestingly, the first thing mentioned in this section of the agreement regarding the transfer of responsibility is that ‘The right to family allowances will be established in the Constitution’. ²⁵ This provision simply means that community governments will not be able to dismantle family allowances altogether, even as they become solely responsible for this element of social security. ²⁶

According to the agreement, the Flemish, French- and German-speaking communities would take over the management and provision of family allowances outside of Brussels. In the Brussels-Capital Region, however, the French and the Flemish communities would be pushed aside and the Common Community Commission (COCOM) put in charge of running the family allowance system for all the inhabitants of the region regardless of the language they speak. ²⁷ This transfer is meaningful from an institutional standpoint because it

marked the first time in Belgian institutional history that community powers were allocated to the COCOM. This transfer of powers is remarkable, as the bilingual character of Brussels and the lack of an own autonomous culture are traditionally invoked as arguments against granting community powers to Brussels. (Goossens and Cannoot, 2015: 47)

The reason for the decision to transfer the responsibility for family allowances to COCOM in Brussels has to do with the preoccupation (especially of Francophone parties) that no ‘sub-nationalities’ be created in Brussels (i.e. no permanent linguistic label, for more details see Dumont, 2015a: 202–8; Poirier, 2011: 362). In fact, the fear was that the distribution of different family allowances in Brussels could produce such sub-nationalities ‘and differential treatment which might be incompatible with the constitutional principle of equality’ (Goossens and Cannoot, 2015: 47; on the constitutional issue, see Dumont and Van Drooghenbroeck, 2011).

The challenges associated with the implementation of the transfer of responsibility for families are significant. These challenges are only alluded to in the short discussion about family allowances in the agreement document mentioned above, which allows for a ‘transition period’ during which time the communities and COCOM (Brussels) can use existing federal institutions to manage and pay out family allowances until they are in a position to assume full control over this policy area. Based on expert interviews conducted in Belgium in June 2015, it is clear that the ongoing process of decentralization has proven to be a demanding task that is likely to extend for several more years in Brussels and in the communities. In addition to determining who should receive benefits from which government, managerial issues are a concern. For instance, federal civil servants in charge of the existing system are to be transferred to the constituent units, which cannot fire them as their employment is protected. The organizational culture of these former federal civil servants could potentially clash with the one prevailing at the constituent unit level. Mentioned by several of the policy experts we interviewed, this situation is only one of the many issues complicating the implementation of the transfer process. Ongoing discussions between federal and sub-state actors have been lengthy and, among the civil servants and policy experts involved, the implementation process is typically seen as complex and demanding.

The implementation process is rather technical in nature, and many ordinary citizens are not even aware of it. For instance, according to a national poll conducted in the late summer 2013, 43 percent of respondents claimed to only have a ‘superficial’ knowledge of the decentralization of family allowances. At the same time, 27 percent of respondents declared that they knew nothing or virtually nothing about this process, and 12 percent of people claimed that they did not even know such a process was taking place. Only 19 percent of those surveyed knew this issue ‘very well’ or ‘relatively well’. At the same time, 49 percent of respondents in Flanders strongly believed that it would lead to greater administrative efficiency compared to 40 percent in Brussels and Wallonia. These results ‘seem to confirm that Francophones are more worried about state reform’ and its impact than Flemings are (Rocour, 2013). This difference in perceptions between the two language communities is hardly surprising; the decentralization of social security has been associated with Flemish nationalism since the 1980s and Francophone parties did not ask for yet another state reform. Overall, decentralization remains a more popular idea in Flanders than in French-speaking Belgium, and it is likely that the most recent state reform alleviated Francophone anxiety over the future of the country only for the short term.

The main political consequence of the Sixth State Reform of the state was that the ‘institutional question’ could be removed from the agenda for the campaign and the government formation negotiations that followed the May 2014 federal elections. In fact, persistent economic challenges, combined with the ongoing implementation of the Sixth State Reform, meant that the political focus was on the country’s socio-economic problems. Because its leader Bart De Wever had signalled his willingness not to press his institutional programme, the N-VA, which had won 26 seats, formed part of the federal executive for the first time. Thus, the new federal government that emerged from the elections was made up of the N-VA, CD&V, Open VLD and the Francophone liberal party (Mouvement réformateur, MR). MR’s leader, Charles Michel, became Prime Minister.

The composition of the Michel centre-right coalition government is novel in at least two ways. First, the N-VA is part of a coalition government for the first time at the federal level. Interestingly, despite the focus of the party on struggling for more autonomy for Flanders, the N-VA entered a government that explicitly stated that it was not going to enact further state reforms. Second, and also for the first time, this coalition is deeply asymmetrical insofar as it features only one Francophone party and three Flemish parties. Most notably absent from the coalition is the PS, which had been part of every single federal government in the previous quarter century.

The Sixth State Reform is unlikely to be the last. In Flanders, most parties view Belgium as evolving towards even more decentralization. As suggested above, the N-VA seeks independence in the long-term, while the CD&V typically speaks of a confederal model for the country’s communities, although its position over the issue has somewhat fluctuated over time. For its part, Open VLD has also supported a confederal model in the past. The decentralization of elements of social security is a central aspect of these alternative models to the current federal system. Hence, we can expect that Flemish parties will press for transfers of powers from the federal level to constituent units on issues, such as healthcare and unemployment insurance. In addition, the country’s fragmented social protection system faces serious governance issues that create the sense that more reforms are needed to sort out what remains a highly complex and fragmented institutional setting (Popelier and Cantillon, 2013). On the Francophone side, the belief that the full decentralization of social security would lead to the end of Belgium as a meaningful political community is still very present, and that Francophone parties will attempt to avoid any further steps towards the territorialization of social security (Béland and Lecours, 2008). However, as we just saw, sometimes the consequences for Francophone parties of exercising their collective vetoes on the decentralization of parts of social security are detrimental to their overarching objective of stability and some form of unity for the country.

Conclusion

Over the last two decades, much had been written about the issue of social policy change. Although powerful institutional factors do increase the likelihood of inertia, change is much more widespread and significant than the early scholarship on welfare state restructuring, such as the work of Paul Pierson (1994), suggested (Hacker, 2004; Thelen, 2004). Even in Belgium, for the longest time a textbook case of relative stability due to powerful institutional logics (Béland and Lecours, 2005), significant change to the territorial organization of social security is now becoming a reality in the aftermath of the Sixth State Reform, as the decentralization of family allowances is now underway. Despite the opposition of powerful constituencies and the collective veto points of Francophone parties, this new reform challenges the claim that the country-level social insurance system created before the advent of federalism in Belgium is unlikely to change as a result of the above-stated institutional factors (Béland and Lecours, 2005).

How did this change become possible? The answer here first lies in increased Flemish nationalist pressures to decentralize parts of social security generated by the surge of the N-VA and the resulting government formation crises, which altered how some Francophone parties assessed the national unity situation as well as their own perceived political interests. Negotiations leading to the formation of a government were unusually long in 2007 and 2010–2011 in large part because of disagreement between Flemish and Francophone parties over state reform. This context creates strong pressures on Francophone parties to compromise in order to show that Belgium can work. At the same time, Francophone parties have to try to shield more vital segments of social security, such as health insurance and unemployment insurance, from decentralization. In this context, it is clear that the strong nationalist mobilization in Flanders associated with the rise of the N-VA and the need for Francophone parties to satisfy at least some Flemish demands in the context of long government negotiations created a ‘critical juncture’ for potential territorial social policy change in Belgium. This ‘critical juncture’ led to policy change in the territorial organization of social security because the issue is central to the ‘quid pro quos’ (Bonoli, 2000) leading to the formation of the Di Rupo government. This logic of quid pro quos according to which Francophone parties agreed to support the decentralization of family allowances to help end the crisis over government formation led to an ‘ambiguous agreement’ (Palier, 2005) through which both Flemish and French-speaking traditional parties could claim ‘victory’ to their respective electorates. Ultimately, the partisan logic inherent to the long negotiations leading to government formation reduced the influence of social partners, which are not formally involved in these negotiations. For these social partners, especially the trade unions, the political crisis provoked by the stalemate over government formation was probably seen as a bigger threat to the unity of a country they wanted to preserve than the decentralization of a limited component of social security.

Only time will tell if Francophone parties will agree to accept more social security decentralization in future government formation negotiations, but Francophone observer Daniel Dumont (2015a) claims that the seeds of a Seventh State Reform are clearly planted in the Sixth State Reform, which would make that next reform very likely, in the future. The divide between the Flemish and the Francophone communities over state reform and the territorial structuring of social security could lead to the emergence of a new political crisis that will force Francophone parties to make some concessions.

The short- and medium-term stability of Belgium seems to very much hinge of the strength and political positioning of the N-VA. As long as the N-VA is the dominant party in Flanders, government formation is likely to be a difficult process. The current environment of austerity policies and concerns with security favours the N-VA, which puts forward a strong message of financial responsibility and law and order. In fact, the N-VA is a nationalist party, but it also gets votes on the basis of its right-wing ideological and public policy agenda. This gives the party a certain flexibility in its position in the context of the government formation negotiations since it can choose to place its state reform agenda on the backburner and find like-minded parties on socio-economic issues, as happened in 2014. It remains to be seen, however, how often the party can accept to be part of government coalitions without implementing any part of its state reform agenda. Party politics will drive future reforms as it has shaped previous ones. ²⁸

Belgium presents a unique political system, which means parallels with other states or generalizations are challenging. Still, at a broad level, this research contributes to the study of territorial policy change by stressing the interaction between nationalist pressures and ‘quid pro quos’ (Bonoli, 2000) involved in government formation negotiations, which can directly impact key decisions about federalism and decentralization. Simultaneously, our analysis suggests that such interactions can, at least in part, break the logic of institutional inertia that is so central in the current literature on policy stability and change. Over time, in the context of powerful political pressures for change and a ‘critical juncture’ (Pierson, 2004), represented in our case by negotiations over government formation, significant policy change is possible, as even traditional institutional obstacles such as existing constituencies and veto points prove insufficient to secure the status quo.