• 1 Introduction

  Interest groups are important players in democracies, as they provide crucial linkages between the state and society. Basic questions on interest groups concern who gets mobilised, who is politically active and who enjoys access to the policy process and/or gains media attention. These questions all pertain to whether systems of interest representation are biased or rather diverse. Interest group scholars usually conceive of a biased system of interest representation as lacking diversity and where access and influence are skewed towards a small number of well-resourced interests, especially economic interests (Lowery et al., 2015). Many domestic systems, as well as the European system of interest representation, have been found to be characterised by bias in various policymaking arenas and the news media (Binderkrantz, 2012; Binderkrantz et al., 2015; Bunea, 2017; De Bruycker & Beyers, 2015; Lowery & Gray, 2004; Rasmussen & Gross, 2015; Schlozman et al., 2012). Similarly, in Belgium, scholars have observed a bias towards a limited number of privileged, mostly economic, interests that gain regular access to the policy process and receive media attention, while many interest groups enjoy no or only limitedly access. In short, interest representation in Belgium is characterised by a strong core-periphery dynamic (Beyers et al., 2014a; Fraussen et al., 2015; Fraussen & Wouters, 2015; Verschuere & De Corte, 2014).
  Although representational bias in Belgium is often linked to neo-corporatism and consociationalism, various recent developments, including federalisation, the politicisation of domains such as migration and the environment, and growing contestation of elitist neo-corporatist practices, have challenged traditional patterns of interest representation (Beyers et al., 2014a; Fraussen & Beyers, 2015; Fraussen et al., 2017; Hooghe, 1995, 1998; Van Den Bulck, 1992). These developments may have resulted in a larger, more diverse, fragmented and competitive interest group system. Hence, it is doubtful whether traditional concepts of Belgian interest representation, such as consociationalism and neo-corporatism, still adequately characterise the overall pattern of state-society relations. After all, similar developments have also affected the Belgian party-political landscape and the overall political-administrative system (De Winter et al., 2006; Deschouwer, 2012; van der Meer et al., 2019; Van Haute & Wauters, 2019).
  Studying the size and diversity of the Belgian interest group community allows us to assess the extent to which neo-corporatist patterns – such as the privileged status of economic interest organisations as core policy insiders compared to the more peripheral role of citizen groups – are still prevalent. Although the size and diversity of interest group systems at the national as well as European and international levels have been long-time concerns in the literature (for an overview, see Halpin & Jordan, 2012), we know relatively little about the overall system of interest representation in Belgium. In Belgium, efforts mainly focused on Flanders, while systematic research on national and Francophone organised interests has been limited (Fraussen & Beyers, 2015; Verschuere & De Corte, 2014). Moreover, while providing empirical depth and rigor, analyzing a small set of well-known organised interests – or ‘usual suspects’ – in one specific arena or region does not capture the overall nature of Belgian interest group politics (Bouteca et al., 2013).
  In the first section of this article, we highlight three macro-concepts characterising Belgian politics and its system of interest representation, namely neo-corporatism, consociationalism and federalism. We also discuss two factors – the politicisation of policy domains such as the environment, migration and the growing contestation of neo-corporatist practices – challenging traditional patterns of interest representation. Next, we elaborate conceptual and methodological issues concerning the mapping of interest group populations and present our datasets on Belgian organised interests. We combine data sources on mobilisation in four arenas: parliament, the executive branch, advisory councils and the news media. The third part provides a first analysis of the size and diversity of the Belgian interest group system. We find that: (1) the national government level continues to be an important venue for interest mobilisation, despite the continuous transfer of competences to the subnational and European levels; (2) neo-corporatism remains a persistent feature of interest representation, despite substantial interest group diversity; and (3) mobilisation patterns differ across government levels and political-administrative arenas. The concluding section offers some general reflections on how the systematic combination of different data sources delivers important insights into the nature of the Belgian system of interest representation.

  1.1 Interest Representation in Belgium

  The Belgian system of interest representation is traditionally characterised as neo-corporatist with a consociational legacy (Siaroff, 1999; Van Den Bulck, 1992). Belgian neo-corporatism entails extensive institutionalised concertation processes (i.e. social dialogue and advisory councils) between the government and a few business associations, labour unions and/or institutional associations (e.g. schools, hospitals, health insurance providers). Since the 1970s, neo-corporatist practices have spilled over from socio-economic policies to other policy domains such as environmental protection (Hooghe, 1995, 1998; Kriesi et al., 1995). These domains are also characterised by privileged access for a limited number of prominent interest groups, albeit not in the classic tripartite way of business associations and labour unions as government interlocutors (Fraussen, 2014). In other domains such as justice, foreign affairs or migration and even within the aforementioned domains, various arrangements for interest representation exist that are not or quasi-neo-corporatist in nature (van den Bulck, 1992). Hence, it might not be appropriate to characterise an entire system as neo-corporatist; instead, we need to analyze sectorial/policy domain variation in interest representation and go beyond the traditional areas of welfare state policies.
  Moreover, the neo-corporatist nature of the Belgian system of interest representation is inseparable from consociationalism (Beyers et al., 2014a; Fraussen & Beyers, 2015; Van Den Bulck, 1992). This implies a cultural/religious and socio-economic ideological segmentation into so-called ‘pillars’, pacifying Belgium’s main political cleavages. The Christian, socialist and liberal pillars represent(ed) dense organisational networks, with strong ties to their respective political parties. The combination of neo-corporatism and consociationalism often involves reaching consensus on policies between the various peak associations, each tied to a pillar. This pillarisation has coincided with extensive government patronage, since pillar organisations were and still are strongly involved in the formulation and implementation of welfare state policies (i.e. providing unemployment benefits and health care reimbursements).
  However, consociational practices gradually declined since the 1990s due to decreasing representativeness of peak associations – because of declining membership, internal heterogeneity – and/or the delaying effect on public policymaking that extensive consultation of peak associations produces. Whereas in the heyday of corporatism, interest intermediation relied on the peak associations’ ability to align and appease their members in exchange for political concessions and/or funding, this has shifted as members of peak associations increasingly bypass their organisation and lobby the government directly (Grote et al., 2008; Kriesi et al., 2006). Moreover, multiple more specialised interest organisations which focus on issues more closely tied to narrow constituencies have been established. In sum, the tension between acting upon the membership interests and reaching political compromises is nowadays much more prevalent; this has put corporatist and consociational practices under strain.
  Currently, consociationalism is predominantly applied to pacify the language cleavage, for instance through language parity requirements and the devolution of competences from the national to the subnational entities. Devolution has resulted in substantial interest group communities at different government levels (Fraussen, 2014; Keating & Wilson, 2014). On the one hand, the growing scope and volume of regional government activities triggered organised interests to mobilise at the subnational level. On the other hand, regional governments themselves actively stimulated a system of interest representation by subsidising regionally based organisations and establishing a system of advisory councils. The devolution dynamic also resulted in organised interests splitting up along linguistic lines and/or creating separate ‘branches’ in each subnational entity (Celis et al., 2012; Fobé et al., 2010; Fraussen, 2014; Heylen & Willems, 2019; Keating & Wilson, 2014; Verschuere & De Corte, 2014). At the same time, neo-corporatism and consociationalism have impacted subnational interest representation profoundly; neo-corporatist practices, initially applied at the national level, were mimicked at the regional level, making peak associations to still play a prominent role at the regional level.
  However, traditional modes of interest representation are increasingly put under pressure. More specifically, the ‘permanent conflicts of interest’ frequently deadlock concertation at the national level, especially in a context of social policy retrenchment (Arcq et al., 2010; Van Gyes et al., 2017). This has fueled, as in other countries, political contestation over presumably ‘elitist’ (neo-corporatist and consociational) practices; populist discourses using an anti-establishment rhetoric found its way into Belgian party politics (Jagers & Walgrave, 2007; Pauwels, 2011; van den Berg et al., 2014). The de-politicisation of (old socio-economic, linguistic and religious) cleavages by institutionalising them into the political system and applying principles of ‘grand coalitions’ and power sharing ultimately fed the emergence of populist criticism (Deschouwer, 2012). Regarding interest representation, this has induced a shift from behind-the-door corporatism to the ‘primacy of politics’. In recent years, for instance, parties at all government levels have tried to decrease the involvement of organised interests in various policy domains, sought to limit the proliferation of advisory councils and implemented budget cuts in various subsidy programs (Fobé et al., 2013; Heylen & Willems, 2019). Moreover, the ‘mediatisation’ of public policymaking constrains corporatist interlocutors to negotiate and produce compromises behind ‘closed doors’ (Häusermann et al., 2004; Kriesi, 2006).
  Finally, due to the politicisation – i.e. increased public salience and intensified party–political conflict – and the widening scope of interest mobilisation tied to policy domains such as migration, justice and the environment, patterns of interest representation have changed (Fraussen, 2014; Hooghe, 1998; Kriesi et al., 1995). As in other European countries, multiple citizen groups became mobilised on ‘new politics’ issues not covered by the traditional corporatist interlocutors (see also Binderkrantz et al., 2016; Kriesi et al., 1995). This has resulted in a more diverse and fragmented set of organised interests seeking access to policymaking processes and the news media. Hence, business associations and labour unions, the principal interlocutors of governments in neo-corporatist systems, might no longer be each other’s sole competitors to gain access and influence.
  In short, as in other small European neo-corporatist countries, corporatist and consociational patterns of interest representation are increasingly put under pressure (Binderkrantz & Christiansen, 2015; Christiansen et al., 2018; Häusermann et al., 2004; Rommetvedt et al., 2013). However, at this moment we lack systematic data on the extent to which the composition of the Belgian interest group population still corresponds with these traditional patterns of interest representation, or that, due to federalisation, societal and political constraints (such as anti-elitist attitudes) or opportunities (such as growing politicisation of certain policy areas), the overall pattern of interest representation has become more diverse and fragmented.

  1.2 Defining and Mapping Interest Group Populations

  One important challenge for answering these questions concerns the conceptualisation of interest groups and its implications for mapping group populations. A commonly used definition of organised interests includes three criteria: (1) being organised, (2) aiming to influence public policy and (3) achieving political goals through informal and formal political engagements outside the electoral arena (Beyers et al., 2008). The latter component sets interest groups apart from parties; typically, interest groups do not seek office through elections like parties but try to achieve their goals through formal (e.g. advisory councils) or informal engagements with policymakers.
  Next, ‘organised’ refers to a minimal level of structural association, thus excluding broad societal movements and waves of public opinion. Some scholars emphasise the membership component or collective, constituency-based features interest organisations must have (Jordan et al., 2004). Hence, organised interests include organisations with formal members – individuals or other organisations such as firms or institutions – as well as organisations with more informal constituencies – donors or supporters (Jordan & Maloney, 2007). These organisations advocate for enfranchised (e.g. the self-interest of affiliates such as companies or professional groups) and disenfranchised (e.g. the poor, the environment, animal rights, child protection) constituencies (Halpin, 2006).
  Finally, interest organisations should show some level of political activity and articulate a collective interest; they potentially aim to influence public policies (Jordan et al., 2004). However, this criterion entails that many civil society organisations, often labelled ‘service/non-profit organisations’, would not be characterised as interest organisations simply because they demonstrate limited or no political activities. For instance, non-governmental organisations (NGOs) primarily focusing on development aid or national sports federations, becoming only occasionally politically active, would not be included (Halpin, 2006; Jordan et al., 2004). Yet, these organisations can play a key role in the policy process because they deliver public services (e.g. social welfare, youth work). As such, these latter organisations are often involved in less visible instances of advocacy, while social movement organisations (SMOs), labour unions, business associations and citizen groups are often pursuing policy influence through more visible advocacy tactics.
  Depending on which conceptual component is emphasised, interest group scholars tend to apply two kinds of data collection strategies for mapping interest group populations, focusing either on behavioral (advocacy or lobbying for policy influence) or on organisational aspects (mobilising a constituency) (Berkhout et al., 2018). First, a frequently used approach by scholars emphasising organisational aspects is called ‘bottom-up mapping’. These scholars are mostly interested in varying levels of collective action, the density and diversity of interest group communities and how organisational entities are established. Typical data sources are directories and encyclopedia of organisations. This approach has been used for mapping transnational advocacy, as well as interest group communities at the national level (Berkhout et al., 2015, 2017; Fraussen & Halpin, 2016; Hanegraaff et al., 2011, 2015; Wonka et al., 2010, 2018). The extent to which groups are politically active is not a central criterion for mapping a community. Irrespective of their involvement (and interest) in policymaking processes, all organisations having a collective supporter or membership component are included in the mapping effort. This inclusive and broad mapping is occasionally followed by a survey among the identified interest groups focusing on organisational characteristics and general tendencies in advocacy strategies and/or influence (Hanegraaff et al., 2016; Heylen et al., 2018).
  Second, studies focusing on the behavioral component of interest representation tend to prefer a top-down mapping strategy. Interest organisations are identified through their participation in specific policymaking processes. Examples are studies using the US state lobby registration rolls or all interest groups registered at the German Bundestag (Klüver & Zeidler, 2019; Lowery & Gray, 1995, 2004), lists of organisations attending political events such as global diplomatic conferences (Hanegraaff et al., 2015), organisations participating in public consultations, parliamentary hearings, advisory bodies (Bunea, 2017; Fraussen et al., 2015; Pedersen et al., 2015; Rasmussen, 2015) and organisations appearing in the news media (Binderkrantz, 2012; Binderkrantz et al., 2017a; De Bruycker & Beyers, 2015). These data sources are particularly suited to study advocacy strategies and influence tied to specific policy dossiers (Berkhout et al., 2018; Beyers et al., 2014b). Compared to bottom-up mapping, the threshold for inclusion is relatively high, as organisations only weakly or not involved in policymaking processes, or engaged with policymakers through other or less visible venues and channels, are usually filtered out. Hence, a top-down mapping mostly identifies organisations visibly advocating on specific policies, but it does not necessarily lead to a valid and comprehensive estimate of the extent to which particular societal segments have been able to establish interest organisations and overcome their collective action problems.
  This distinction affects the conceptual boundaries of interest group populations. A behavioral focus might underestimate the size of an interest group population, as scholars mostly focus on groups demonstrating significant political activities. An organisational approach, by comparison, focusing on the mobilisation of constituencies, also has some limits, as it is not immediately clear to what extent organisations effectively seek to influence public policy. Hence, the emphasised component of the definition – political activities or organisational constituencies – strongly shapes the nature of the studied organisational population.
  Our mapping of the Belgian interest group community relies on an extensive scrutiny and triangulation of multiple data sources. By combining a top-down and a bottom-up approach we seek to account for the potential limitations each specific method entails. We relied on a bottom-up registration of organisations and a survey implemented among high-level representatives (such as the director, chair, president or secretary-general) of these organisations (see Appendix). A distinction was made between concentrated groups representing the self-interests of well-circumscribed constituencies and diffuse groups representing broader societal segments. The former set of organisations includes professional associations (e.g. lawyers), business associations (e.g. the chemical industry) and associations representing institutions and (semi-)public authorities (e.g. hospitals), while the latter set of organisations includes citizen groups such as cause groups (e.g. consumer rights, environmental protection) and identity groups (e.g. youth, patients, the LGBT community, migrants), as well as constituency-based service/non-profit organisations (e.g. social welfare) (Baroni et al., 2014; Beyers et al., 2008, 2014b; Binderkrantz et al., 2015).

  Table 1 Overview of datasets

  Dataset	Sampling approach	Outcome	Period	Data repository
  Registered interest organisations in the KBO	Bottom-up	1,678 organisations	2015	www.cigsurvey.eu
  Survey of interest organisations	Bottom-up	771 survey responses	2015	www.cigsurvey.eu
  Interest organisations as members of advisory councils	Top-down	616 advisory councils with 1,154 organisations	2016-2017	www.ibias.eu
  Interest organisations in the news media	Top-down	110 policy issues with 247 organisations	2014-2018	www.ibias.eu

  These data are combined with two top-down maps. First, we identified interest groups gaining news coverage through a content analysis related to 110 policy issues included in a 2014 voter survey (see also Appendix). Second, we identified groups with access to 616 advisory councils at the national and subnational levels. The latter two datasets thus include groups demonstrating a substantial level of political activity. Table 1 gives an overview of the data sources, and each of them will be discussed in the following sections in relation to some key findings. Combined, these datasets assess interest group mobilisation, involvement (i.e. seeking contact), access (i.e. granted contact) and prominence (or pre-eminence) across different political arenas and government levels (for a detailed conceptual discussion, see Halpin & Fraussen, 2017).

• 2 Density and Diversity of the Belgian System of Interest Representation

  2.1 A Bottom-Up Census of Belgian Interest Groups

  First, we describe the demography of the Belgian interest group population based on a bottom-up mapping. The bottom-up census was primarily drawn from the Kruispuntbank voor Ondernemingen (Crossroads Bank for Enterprises, CBE), the official federal government register documenting the legal statuses of enterprises and organisations in Belgium. Through multiple semi-automated processes based on the NACE classification code S94 and manual operations aimed at grouping organisational conglomerates, we identified 1,461 Belgian interest organisations.‍¹x NACE is the abbreviation of the French Nomenclature statistique des activités économiques dans la Communauté européenne. This European industry classification system consists of a six-digit code and is systematically used in most national statistical data systems (see http://goo.gl/8NLquM). The full definition of S94 reads as follows: This division includes activities of organisations representing interests of special groups or promoting ideas to the general public. These organisations usually have a constituency of members, but their activities may involve and benefit non-members as well. The primary breakdown of this division is determined by the purpose that these organisations serve, namely interests of employers, self-employed individuals and the scientific community (group 94.1), interests of employees (group 94.2) or promotion of religious, political, cultural, educational or recreational ideas and activities (group 94.9). We supplemented this list with organisations identified through SectorLink and Filantropie.be.²x SectorLink (currently www.bsae.be) provides an overview of Belgian professional associations, industry groups and business federations. It includes organisations recognized as professional associations by the ‘Hoge Raad van de Middenstand’, as well as the member organisations of the main peak business associations. Filantropie.be (currently www.goededoelen.be) is a voluntary register with mostly non-profit organisations and encompasses organisations active at the national, subnational and local level (n=2,904 on 15 December 2014). It is an online platform developed through a cooperation between the Koning Boudewijnstichting and the National Bank of Belgium. This resulted in a set of 1,678 interest organisations. Table 2 presents this demography by group type across government levels.
  First, the census delineates not one but three distinct systems of interest representation, namely at the national level, the Flemish level and the Walloon/Francophone government level.³x Also, the Brussels Capital Region and the German-speaking community have their own interest group community. Due to the strongly locally based nature of the latter and the considerable overlap with the Flemish and Francophone/Walloon interest group communities, we decided not to include these smaller communities in our bottom-up mapping. Some 41% of groups are mobilised nationwide, while 35% and 24% of the groups limit their activities, respectively, to the Flemish and Walloon/Francophone government level. Second, considering the distribution of group types, an obvious observation is not only the prominence of economic interests, but also the considerable presence of non-business interests. Although business and professional groups account for 50% of the entire Belgian interest group community, a considerable share of 30% are cause groups and identity groups. The distinctiveness of the interest group communities at each government level is substantial. As Table 2 demonstrates, business interests are strongly mobilised at the national level (42% of identified groups represent business interests), while the prevalence of business is less outspoken at the subnational level (18% of Flemish organisations and 13% of Walloon/Francophone groups represent business interests). Vice versa, cause groups and identity groups are especially mobilised at the subnational levels. Respectively, 18% and 15% of the organisations active in Flanders, and 18% and 24% of Francophone groups represent an identity or cause group. Hence, when looking systematically at a wide range of groups, the enormous diversity of interest groups across government levels is remarkable, which reflects the division of policy competencies in a federal setting (Fraussen, 2014; Heylen & Willems, 2019; Keating & Wilson, 2014).
  To explore this diversity further, Table 3 reports the mean year of foundation, the mean staff size and the median level of financial resources across group types. The figures corroborate that the Belgian system of interest representation is characterised by substantial diversity between and within group types. For instance, labour unions are few in number (less than 5% of the population), but they mobilise a huge number of individuals (more than three million Belgian citizens are labour union members), and they trump all other group types in terms of staff and financial resources. By contrast, while business associations make up the largest share of the population and have mostly corporate members, they are characterised by relatively lower staffing levels. Interestingly, although business groups have a reputation of being well-resourced (Dür & Mateo, 2013), we also observe non-business interests possessing substantial resources. Compared to business associations, cause groups and identity groups have on average the same or an even larger capacity in terms of financial resources and staff.
  To summarise, traditional neo-corporatist organisations such as business associations and labour unions are still prominent, but the contemporary Belgian interest group community also exhibits substantial diversity and signs of a more pluralist system of interest representation – i.e. having many different interest groups competing to get their voices heard by policymakers.
  Figure 1 takes a closer look at the founding dates of the organised interests that exist today. This overview allows us to tentatively discuss the impact of major institutional and societal changes on the contemporary interest group community.⁴x There are two important aspects to be aware of when considering Figure 1. First, the evidence only concerns founding dates of groups that currently exist, which are all survivors. It tells us little about the composition of the groups’ system in previous eras and the dynamics associated with organisational mortality and survival in the past. Second, the strong decline in recent founding rates (4) should be dealt with cautiously. Although a possible explanation for this is the financial crisis starting in 2008 and government austerity suppressing organisational establishment and survival (Heylen et al., 2018), an entry lag in public directories for several years must be taken into account (Bevan et al., 2013; Fraussen & Halpin, 2016). A first peak in the establishment of organisations is situated in the post-war period (1). This is the time the welfare state was established, incentivising the founding and growth of socio-economic interest groups playing a key role in developing and implementing welfare state policies (Deschouwer, 2012). The founding dates by group type – shown in the Appendix (Figure 2A) – confirm that especially business groups and professional associations were established during the post-war period, together with labour unions.
  Further growth peaks manifest in the late 1970s, the period after the enactment of the first state reform (2), and in the late 1990s, marked by the continuous devolution of competences from the national to the subnational level (3). First, devolution stimulated existing organisations to split up their nationwide structure into Flemish and Francophone branches. For instance, one of the largest environmental associations in Belgium (i.e. Bond Beter Leefmilieu/Inter-Environnement) was first established as a nationwide organisation in 1971, consisting of four regional branches (Fraussen, 2014). However, due to the increasingly outspoken claims of the nationwide association on nuclear energy, several important private sponsors withdrew their funds, and this incentivised the association to foster more structural ties with policymakers. As a consequence of these intensified interactions with policymakers, the association had to deal with growing cultural-linguistic tensions between its Flemish and Francophone strands. This eventually led to the disbandment of the association along subnational lines in 1979. Moreover, instead of adopting a nationwide structure, many new organisations established themselves immediately at the subnational level. The Flemish and Francophone governments increasingly provided financial resources and policy access, especially in areas of important competencies, creating incentives for organisations to have a clear subnational territorial focus (Celis et al., 2012; Keating & Wilson, 2014).

  Table 2 Demography by group type across government levels

  	National (%)	Flemish (%)	Francophone (%)	Total (%)
  Business	42	18	13	26
  Professional	25	25	21	24
  Labour	3	1	2	2
  Identity	8	18	18	14
  Cause	12	15	24	16
  Leisure	8	19	16	14
  Associations of institutions	2	4	6	4
  Total (%)	41	35	24	100

  Note: Percentages based on N = 1,678 from bottom-up mapping

  Table 3 Basic organisational features by organisation type

  Group type	Number of groups	Foundation (mean)	Staff (mean FTE)	Budget (median category)
  Business associations	235	1973	9	€100,000-500,000
  Professional associations	199	1970	44	€50,000-100,000
  Labour unions	20	1947	58	€5,000,000-10,000,000
  Identity groups	144	1976	26	€100,000-500,000
  Cause groups	205	1985	15	€100,000-500,000
  Leisure associations	145	1974	7	€100,000-500,000
  Associations of institutions	36	1986	8	€100,000-500,000
  Number of observations	n = 984	n = 947	n = 768	n = 851

  Note: Numbers based on survey responses

  Founding dates of Belgian interest organisations
  Notes: 1 = post-war period; 2 = late 1970s; 3 = 1990s; 4 = recent era

  The growing number of new interest organisations should also be seen in the context of post-materialist issues supplementing – from the late 1960s/early 1970s onwards – the left–right socio-economic cleavage (2 and 3). An assessment of the disaggregated founding dates by group type confirms that the founding of cause and identity groups exploded from the 1970s onwards (Figure 2A in the Appendix). Many of these social movements are today well-established organisations, as illustrated by their formal recognition as members of several advisory bodies (Defourny et al., 2005; Dewachter, 1995; Fobé et al., 2010). This rise of citizen groups thus reflects growing public concerns with topics such as the environment, climate change and human rights (Hooghe, 1995, 1998; Kriesi et al., 1995).
  Overall, the devolution of policy competences and the territorial fragmentation along linguistic lines has resulted in distinct interest group communities with little interaction between them. Many socio-economic policies remain the prerogative of the national level and this is reflected in the prominence of traditional neo-corporatist associations at the national level, while most social movements/citizen groups active in the field of the environment, transportation or culture operate at the subnational level. These interest group communities have developed separately, and few groups have incentives to organise at the national level. Only those groups – the so-called ‘social partners’ consisting of the peak business associations and labour unions – for which key policy interests are still determined by the national government maintain their nationwide organisational structure and resist the devolution of competences in areas such as social security and labour market policy (Bouteca et al., 2013).

  2.2 Bottom-Up Mapping of Seeking and Gaining Access to Policymakers

  In this section, we map the extent to which different group types are insiders to the policymaking process. We rely on two data sources, the survey data and the dataset on advisory councils, to assess who is seeking and gaining access to the policy process. First, to assess inside tactics – the seeking of direct contact to policymakers – we use the survey data (Beyers et al., 2016). These contacts are initiated at the discretion of interest organisations – respondents in the survey – themselves and do not warrant an invitation by policymakers (Halpin & Fraussen, 2017). Of all interest organisations in our sample, 41% at least once per month or more directly contact a policymaker, 30% do so at least once every three months and 23% do so at least once a year. These observations signify the importance of inside advocacy tactics for Belgian interest organisations.
  When it comes to advocacy strategies, most interest groups prefer inside strategies – directly contacting policymakers – over outside strategies – reaching out to the broader public and members – to affect public policies (Figure 2). When engaging in outside activities, Belgian interest organisations mainly use media-oriented strategies such as contacting journalists and organising press conferences. Activities involving members such as signing petitions and staging protests are less frequently used. Compared to these media-oriented tactics, the least used outside tactics – such as developing research reports, publishing opinions online and organising stakeholder meetings – are also strategies with a smaller target audience.

  Inside and outside advocacy tactics (n = 641)
  

  Interest groups can choose to directly address policymakers located within various political arenas – ranging from the legislative to the executive and administrative branches of government – when they seek to influence public policy. The survey included questions probing the frequency of contacts with government officials initiated with the purpose to ‘influence public policy’. Figure 3 compares the prevalence of contacts initiated across government levels with the executive branch of government (ministers and cabinets), the administration (civil servants within ministerial departments and agencies) and the parliament.
  We can draw three conclusions by comparing these results. First, in each jurisdiction, the parliament is least contacted compared to the administration and the executive branch. When we consider weekly and monthly contacts together, we observe, depending on the jurisdiction, less than 30% of interest groups developing regular contacts with parliamentarians. While some longitudinal research conducted in other neo-corporatist European countries found that the parliament as a lobbying target has gained substantial importance since the heyday of corporatism (Gava et al., 2017; Rommetvedt et al., 2013), this seems to be less markedly the case in Belgium. Belgium is still characterised by a weak parliament, ‘politicised’ government administrations and large personal cabinets of ministers (van den Berg et al., 2014; van der Meer et al., 2019). Second, most contacts are initiated with the administration, especially at the subnational level. While the joint ‘weekly’, ‘monthly’ and ‘once every three months’ contacts with the national administration sum to 62%, this comes to 76% for the Walloon/Francophone administration and 69% for the Flemish administration. One reason for the differences between the subnational and national levels could be the extent to which groups depend on subnational subsidies (Celis et al., 2012; Heylen & Willems, 2019). Third, the executive branch in Flanders and Wallonia is contacted on a more regular basis by interest groups – respectively 33% and 29% of groups seek contacts at least once every three months or more frequently – compared to groups seeking contact with the national government – at this level only 19% seek contacts ‘at least once every three months’ or more frequently with ministers and cabinet members. In short, while interest mobilisation differs across government levels, also the variation in lobbying strategies reflects the multi-level structure of the Belgian polity. Moreover, the clear prominence of the administration as a lobbying target seems to be a persistent feature of neo-corporatist and consensual policymaking in Belgium (see also van den Berg et al., 2014; van der Meer et al., 2019). This finding matches the assessment of ‘corporatist resilience’ observed in other small European countries (Binderkrantz & Christiansen, 2015; Binderkrantz et al., 2015; Christiansen et al., 2018).

  Seeking access to different governmental branches by government level
  

  2.3 Top-Down Mapping of Gaining Access to Advisory Councils

  Advisory councils, and their composition, are one of the foremost formal and institutional expressions of neo-corporatist practices (Christiansen et al., 2010). Moreover, access, more specifically obtaining seats in advisory councils, can be seen as a lobbying success; it reflects the effectiveness of an interest organisation in passing a certain threshold that is beyond their own discretion and gaining recognition by policymakers (Binderkrantz et al., 2017b; Halpin & Fraussen, 2017). We mapped interest group membership in 616 advisory councils at the national (n = 290), Flemish (n = 116) and Walloon/Francophone (n = 167) government levels. Three information sources were used: councils’ own websites and dedicated government webpages, Politiek Zakboekje/Mémento Politique 2016 and Moniteur Belge (see also Appendix). This allowed us to check the following criteria: (1) sufficient information availability, (2) dealing with policy formulation or implementation, not individual administrative acts or management tasks such as hiring and selection or awarding project funding, (3) permanently established and active during the legislature 2014-2019 and (4) at least one member is a non-governmental stakeholder. Finally, we conducted a detailed coding of these 616 advisory councils and mapped a total of 1,154 interest groups being council members.
  Of the entire interest group population, 41% of national organisations, 38% of Flemish organisations and 45% of Francophone organisations have access to at least one of these 616 advisory councils. The somewhat higher percentage of Francophone organisations having access – compared to Flemish organisations – might be due to the overall higher number of Francophone advisory councils. Across group types enjoying access (i.e. light grey bars in Figure 4), business and professional associations make up the largest category. Respectively 23% and 26% of groups having access are business and professional associations, illustrating a clear prominence compared to other group types. Associations of public authorities and institutions (12%), cause groups (14%) and identity groups (11%) also make up quite a substantial portion in the system of advisory councils. Compared to other group types, labour unions (3%) constitute only a small portion of all groups having access. Economic interests are core policy insiders compared to the more peripheral position of citizen groups.
  To assess this representational bias further, we compared access across group types relative to their total share in the population (i.e. dark grey bars in Figure 4). If certain group types dominate the population, it would be no surprise that these types enjoy higher levels of access. Or by contrast, if some group types, for instance labour unions, are less numerous, this could affect their access. A total of 85% of all associations of (semi-)public authorities and institutions and 82% of all labour unions have access to at least one advisory council. While the overall portion of labour unions is small compared to the total number of interest groups having access to at least one council, the overwhelming majority of Belgian labour unions does gain access. Also, business associations, professional associations, identity groups and cause groups enjoy substantial access. Respectively, between 48% and 60% of all these groups have access to at least one advisory council.

  Table 4 List of top 20 advisers

  1. ACV/CSC	2. FGTB/ABVV	3. ACLVB/CGSLB	4. Landsbond der Christelijke Mutualiteiten Alliance Nationale des Mutualités Chrétiennes
  422 seats Christian labour union National	373 seats Socialist labour union National	201 seats Liberal labour union National	190 seats Health care association National
  5. Union des Classes Moyeness	6. Union Nationale de Mutualités Socialistes Landsbond der Socialistische Mutualiteiten	7. Belgische Vereniging van Artsensyndicaten Association Belge des Syndicates Médicaux	8. Unie van Zelfstandige Ondernemers
  170 seats Business association: small and medium enterprises Francophone	164 seats Health care association National	161 seats Professional association: doctors National	157 seats Business association: small and medium enterprises Flemish
  9. Vlaams Netwerk van Ondernemingen	10. Brussels Enterprise and Commerce	11. Union Wallon des Entreprises	12. Zorgnet-Icuro
  126 seats Business association Flemish	114 seats Business association Brussels	111 seats Business association Francophone	95 seats Association of health care facilities Flemish
  13. Verbond van Belgische Ondernemingen Fédération des Entreprises de Belgique	14. Fédération Wallonne de l’Agriculture	15. Landsbond van de Onafhankelijke Ziekenfondsen Union nationale des Mutualités Libres	16. Union des Villes et Communes de Wallonie
  90 seats Business association National	86 seats Professional association: farmers Francophone	78 seats Health care association National	78 seats Association of municipalities and cities Francophone
  17. Boerenbond	18. ACOD/CGSP	19. AXXON Physical Therapy	20. Landsbond der Liberale Mutualiteiten Union Nationale des Mutualités Libérales
  69 seats Professional association: farmers Flemish	67 seats Socialist labour union: public sector National	67 seats Professional association of physiotherapists National	62 seats Health care association National

  However, we need to be careful as these conclusions concern access to at least one advisory council and not the absolute number of seats these organisations hold across multiple councils. In this regard, we can clearly detect a core-periphery dynamic. Few groups have seats in a high number of advisory councils. Of all groups with access (n = 1,154), the overwhelming majority (n = 719 or 58%) has access to only one council, and most of these (n = 375 or 52%) have only one seat per council.
  Table 4, presenting the top 20 organisations with the most seats, illustrates this skewed access pattern. This list consists mostly of labour unions, peak business associations and (professional) associations in the health care sector. No identity groups (e.g. youth, patients, gender and migrants) or cause groups (e.g. environment, human rights, traffic safety) are among this set of core insiders. The centralisation of the interest group system around a few business, labour and professional groups is a typical feature of consociationalism and neo-corporatist systems (Grote et al., 2008; Kriesi et al., 2006). Policymaking and implementation – especially in welfare state domains – is still a matter of concertation among organisations representing key socio-economic segments (Beyers et al., 2014a; Deschouwer, 2012; Van Den Bulck, 1992). These top 20 interest organisations, for instance, have a strong presence in influential socio-economic advisory councils (i.e. the National Labour Council and the Central Economic Council, at the national level, SERV in Flanders and CESE in Francophone Belgium). Moreover, the reach of these core insiders is much wider than traditional welfare state domains. The fact that they also enjoy substantial access to advisory councils in other domains such as environment, transport and cultural policy clearly demonstrates the prominence of these actors among policymakers. It illustrates the pre-eminence or ‘taken-for-grantedness’ of these groups and their viewpoints among policymakers, despite the presence of other groups that represent similar constituencies (Halpin & Fraussen, 2017).
  However, in these domains also citizen groups gain substantial access. Citizen groups’ rise in numbers following the increased politicisation of certain policy domains is to a certain extent matched by these organisations’ successful entry into the system of advisory councils. Although neo-corporatist patterns of interest representation still rule the system of advisory councils, the years since the economic and financial crisis are characterised by increasing political and public contestation for these ‘elitist’ closed-door decision-making structures and more frequent deadlocks of social dialogue because of retrenchment (Van Gyes et al., 2017). Neo-corporatist practices have increasingly been put under pressure and the traditional interlocutors of government face more competition from other types of interest groups to gain access to the system of advisory councils.

  Comparing access of different group types to advisory councils List of top 20 advisers
  

  2.4 Top-Down Mapping of Interest Groups in the News Media

  For assessing media attention, we rely on a content analysis of various news media outlets for a sample of 110 specific policy issues included in a 2014 voter survey, comprising 37 federal issues, 34 Flemish issues and 39 Walloon/Francophone issues (see also Appendix). First, the relevant media coverage from June 2013 to December 2017 in four media outlets was automatically scraped from GoPress.⁵x GoPress is the online press database and monitoring service for all Belgian newspapers and magazine publishers (www.gopress.academic.be). In Flanders, the news media outlets selected were De Standaard (715,100 daily readers) and De Morgen (448,500 daily readers). In Wallonia, the media outlets were Le Soir (639,400 daily readers) and La Libre Belgique (339,700 daily readers). For more information, see https://www.cim.be/nl/pers/bereik-resultaten. To identify relevant articles, we applied a computer-automated Boolean search with up to six keywords – in both Dutch and French for national issues – closely related to the policy issues.⁶x Keywords were carefully selected based on the name of the policy issue in the online voter survey and extensive desk research including legislative initiatives on the policy issue. The saturation point for identifying key words was inductively determined by checking the number of (new) relevant articles that could be found by entering a new keyword in the GoPress search tool. This resulted in 26,512 unique newspaper articles. Next, we automatically identified interest organisations active on these issues based on a curated dictionary containing 2,340 organisation names and abbreviations.⁷x This list of 2,340 organisations resulted from a combination of identified organised interests through the bottom-up mapping and the mapping of advisory councils’ members. Additional coding was done so that different variations on the name and acronym of an interest group could be stored in the curated dictionary. The advantage of a curated dictionary is that it allowed us to quickly sift through a substantial number of newspaper articles. However, to account for the limitations of the computer-automated identification, coders manually added organisations making relevant claims in the selected articles and excluded those newspaper articles containing irrelevant claims. A manual coding was opted for because the claims interest groups made in the news are often complex and multi-faceted. In total, we sampled 2,740 newspaper articles in which interest groups were identified (for an overview, see Appendix).
  Although outside strategies are usually deployed by groups to gain news coverage, media coverage could also be due to the fact that journalists (or policy advocates) disclose hidden lobbying activities and/or publicly challenge some organised interests (i.e. some groups cannot escape media attention due to being a policy insider and/or the need for counteractive lobbying). The overall media attention groups gain is rather limited and also in the media arena considerable bias is present. Across the 110 policy issues, we identified 247 unique interest organisations making relevant claims on these issues in the sampled newspaper articles, which is only 11% of all mapped organisations. Across all group types appearing in news coverage (see Figure 5), business associations (26%) and cause groups (24%) make up the largest categories (i.e. light grey bars in Figure 5). Also, identity groups (15%), associations of (semi-)public authorities and institutions (15%) and labour unions (11%) make up quite substantial portions. Professional associations (9%) and leisure associations (4%) gain comparatively less attention.
  However, when comparing media attention across group types relative to their total share in the population, a different picture emerges. Of all labour unions, 56% appears at least once in the news, making them the group type with the most attention. Cause groups also enjoy substantial levels of attention; 18% of these groups appears in the news – which is slightly more than business associations (13%). Nonetheless, compared to the population – except for the labour unions – the vast majority of groups does not appear in the news. The scarce media attention for interest groups might be due to the overall harsh competition to gain news coverage, not only among interest groups themselves, but especially with parties and politicians (Tresch & Fischer, 2015).
  Similar to the access interest groups enjoy to advisory councils, media attention also displays a profound core-periphery dynamic (Fraussen & Wouters, 2015). Of all groups attracting media attention, a high number (n = 95 or 40%) appears in only one article. Looking at the distribution across group types (Table 5), business associations attract more media attention – exemplified by their higher mean and maximum values – compared to identity groups and cause groups. For instance, while the top 25% of business associations appears 7.25 times in the news (Q4), the top 25% of identity groups appears only four times in the news. Cause groups are more on par with business associations; the top 25% of them appeared six times in newspaper articles related to one of the 110 policy issues. Again, labour unions are successful when it comes to media attention, as the median labour union appears in six newspaper articles on the same policy issue and the top 25% of them appears in 15.5 articles (Q4).
  The clear prominence of business associations and labour unions in the news indicates that – although every arena has its own logic – some characteristics, such as the core-periphery structure of interest representation in advisory councils, are also reflected in the media arena. As some research demonstrates, the news media often attribute news value to powerful political insiders and pay less attention to outsiders. This has led authors to characterise the – Belgian as well as Scandinavian – news media as an arena of ‘privileged pluralism’ (Binderkrantz et al., 2015; Fraussen & Wouters, 2015; Tresch & Fischer, 2015). However, some groups with less inside access do seem to be able, at least to some extent, to make up for this through the news media. For instance, many cause groups gain substantial news coverage.

  Comparing media attention for different group types The distribution of media attention by group type, at least one media hit
  
  Table 5 The distribution of media attention by group type, at least one media hit

  Number of media hits	n	Min.	Mean	SD	Q1	Median	Q4	Max.
  Business	64	1	7.62	13.7	1	3	7.25	88
  Professionals	22	1	2.59	2.77	1	2	2.75	13
  Labour unions	28	1	16.64	26.48	1.75	6	15.5	106
  Identity groups	38	1	4.21	5.53	1	2	4	24
  Cause groups	60	1	4.53	4.58	1	3	6	21
  Leisure associations	10	1	2.9	2.9	1	1	1.75	11
  Associations of institutions	25	1	4.32	6.64	1	2	4	27
  Total	247	1						106

  Note: Numbers based on the media data

• 3 Conclusion

  The descriptive analyses presented in this article give some tentative insights into some key features of Belgian interest representation. We find that: (1) neo-corporatist core-periphery structures continue to be a persistent feature, (2) the national government level remains an important venue for interest groups, despite the continuous transfer of policy competences to the subnational and European levels, but (3) patterns of interest representation vary across government levels and policymaking arenas. First, our descriptive overview demonstrates that neo-corporatist mobilisation patterns are quite persistent. As in other European countries, the traditional neo-corporatist interlocutors of government – labour unions, peak business and professional associations – tend to dominate in absolute numbers (Dür & Mateo, 2013; Wonka et al., 2010). However, while in absolute numbers labour unions and business associations are achieving more political voice, cause groups and identity groups may shout as loudly – for instance because of their resource endowment and substantial media attention – and therefore have considerable chances to influence public policymaking.
  Still, the (peak) business associations and labour unions enjoy more access to traditional neo-corporatist venues such as advisory councils compared to citizen groups – an outspoken core-periphery dynamic is present. Although these traditional neo-corporatist practices are mostly present at the national level, they also prevail at the subnational level, exemplified by the multitude of advisory councils established by subnational governments and these councils’ composition, in which the traditional peak business associations and labour unions are also prominent (Fobé et al., 2013; Fraussen & Beyers, 2015; Keating & Wilson, 2014). A rather small set of groups – mostly business associations and labour unions – has access to a large number of councils and gains substantial media attention. Hence, the media arena resonates the political power of insider groups, and, in the case of Belgium, this perpetuates traditional neo-corporatist patterns of interest representation.
  Second, one important consequence of the Belgian federal state structure is the presence of different interest group communities at the national and subnational levels. Drivers for these diverging patterns of mobilisation and the emergence of a multi-layered interest group system are multiple. On the one hand, many organised interests have – confronted with the continuous devolution of competencies – rescaled their organisational structure and activities towards the subnational level and new interest organisations are mostly established at the subnational level (Fraussen, 2014; Keating & Wilson, 2014). This is especially the case for identity groups, cause groups and associations of institutions and public authorities – all are predominantly mobilised at the subnational level. By contrast, business interests, professional associations and labour unions are still primarely mobilised at the national level. On the other hand, the Belgian governments have themselves actively developed distinct systems of interest intermediation, for example, by awarding subsidies or setting up consultation venues (Celis et al., 2012; Fraussen, 2014; Heylen & Willems, 2019). As a result, Belgium offers an excellent case to assess how multi-layered political institutions shape the mobilisation of societal interests and their interaction with public authorities. It enables an analysis of how institutions and political elites can provide incentives for the formation of subnational interest communities with distinct features and dynamics, as well as (possibly) constrain the establishment or maintenance of nationwide groups bridging territorial interests. In short, devolution has created an incentive to ‘abandon’ the center, but it has, in the case of Belgium, not resulted in the hollowing out of the national interest group community.
  This brings us to the third conclusion, namely that interest mobilisation varies profoundly across political arenas and branches of government. For instance, while the media arena resonates the core-periphery structure of the political–administrative arena of advisory councils, it does not provide a perfect mirror. In this regard, the media arena is somewhat more inclusive of citizen groups (compared to advisory councils). This can be framed in the context of increased public attention and interest mobilisation on issues such as the environment and human rights (see Binderkrantz, 2012). In addition, while the traditional neo-corporatist actors have maintained their core position in advisory councils dealing with welfare state policies, in other ‘new’ domains also citizen groups have gained substantial access (Willems, 2020). The growing number of citizen groups is to a certain extent matched with these organisations’ successful entry into the system of advisory councils. At the same time, the system of advisory councils and social dialogue is increasingly criticized and contested by political and public actors (Van Gyes et al., 2017). In addition, substantial variation can be observed across government branches. The evidence reveals that, compared to the administration and the executive branch, Belgian organised interests least contact parliamentarians.
  An important limitation of our characterisation of interest representation in Belgium is that we only focused on organised interests, while excluding other actors such as companies and semi-public authorities such as universities or hospitals – often referred to as ‘pressure participants’ – from our analyses (Jordan et al., 2004). These entities have no intermediary function; they do not represent a constituency or membership, and their potential political activities are usually a by-product of their core business – if their interests are threatened, they mobilise politically. The expertise these actors have at their disposal, as well as their economic significance (in terms of employment and/or investments) means that they play a crucial role in any political system (Salisbury, 1984; see also Lowery, 2007). Some results – such as the prominence of business interests and the importance of inside lobbying – might even be more pronounced if we would have included these actors in our analyses (see for instance Aizenberg & Hanegraaff, 2020); follow-up research could investigate the role of these ‘pressure participants’ more closely.
  Overall, the evidence suggests that the federalisation of Belgium, the politicisation of issues such as migration and the environment and growing political contestation towards elitist neo-corporatist practices put pressure on traditional patterns of neo-corporatism. These developments have nourished a more competitive interest group system, larger in size as well as more diverse and fragmented. Hence, traditional concepts of Belgian interest representation, such as consociationalism or neo-corporatism, can no longer adequately characterise the overall pattern of state–society relations in Belgium. The Belgian system of interest representation has become considerably segmented and characterised by distinct constellations of organised interests at each government level and distinct mobilisation patterns across political and public arenas. Nonetheless, the neo-corporatist legacy has proven to be resilient, as the prevailing prominence of labour unions and peak business associations demonstrates. In essence, when it comes to interest mobilisation, we observe on top of the persistent neo-corporatist patterns substantial ingredients of a more pluralist system of interest representation.

References

• Aizenberg, E., & Hanegraaff, M. (2020). Is politics under increasing corporate sway? A longitudinal study on the drivers of corporate access. West European Politics, 43(1), 181-202.

• Arcq, E., Capron, M., & Léonard, E. (2010). Dynamiques de la concertation sociale. Brussels: CRISP.

• Baroni, L., Carroll, B. J., Chalmers, A. W., Marquez, L. M. M., & Rasmussen, A. (2014). Defining and classifying interest groups. Interest Groups & Advocacy, 3(2), 141-159.

• Berkhout, J., Beyers, J., Braun, C., Hanegraaff, M., & Lowery, D. (2018). Making inference across mobilisation and influence research: Comparing top-down and bottom-up mapping of interest systems. Political Studies, 66(1), 43-62.

• Berkhout, J., Carroll, B. J., Braun, C., Chalmers, A. W., Destrooper, T., Lowery, D., Otjes, S., & Rasmussen, A. (2015). Interest organizations across economic sectors: Explaining interest group density in the European Union. Journal of European Public Policy, 22(4), 462-480.

• Berkhout, J., Hanegraaff, M., & Braun, C. (2017). Is the EU different? Comparing the diversity of national and EU-level systems of interest organisations. West European Politics, 40(5), 1109-1131.

• Bevan, S., Baumgartner, F. R., Johnson, E. W., & McCarthy, J. D. (2013). Understanding selection bias, time-lags and measurement bias in secondary data sources: Putting the encyclopedia of associations database in broader context. Social Science Research, 42(6), 1750-1764.

• Beyers, J., Bernhagen, P., Boräng, F., Braun, C., Fink-Hafner, D., Heylen, F., Maloney, W., Naurin, D., & Pakull, D. (2016). Comparative Interest Group Survey Questionnaire (www.cigsurvey.eu).

• Beyers, J., Braun, C., & Haverland, M. (2014a). Plus Ça Change, Plus C’est Pareil. European integration and interest group politics in the low countries. In H. Vollaard, J. Beyers, & P. Dumont (Eds.), European Integration and Consensus Politics in the Low Countries (pp. 134-154). London: Routledge.

• Beyers, J., Dür, A., Marshall, D., & Wonka, A. (2014b). Policy-centred sampling in interest group research: Lessons from the INTEREURO project. Interest Groups & Advocacy, 3(2), 160-173.

• Beyers, J., Eising, R., & Maloney, W. (2008). Researching interest group politics in Europe and elsewhere: Much we study, little we know? West European Politics, 31(6), 1103-1128.

• Binderkrantz, A. (2012). Interest groups in the media: Bias and diversity over time. European Journal of Political Research, 51(1), 117-139.

• Binderkrantz, A., Chaqués-Bonafont, L., & Halpin, D. (2017a). Diversity in the news? A study of interest groups in the media in the UK, Spain and Denmark. British Journal of Political Science, 47(2), 313-328.

• Binderkrantz, A., & Christiansen, P. (2015). From classic to modern corporatism. Interest group representation in Danish public committees in 1975 and 2010. Journal of European Public Policy, 22(7), 1022-1039.

• Binderkrantz, A., Christiansen, P., & Pedersen, H. (2015). Interest group access to the bureaucracy, parliament, and the media. Governance, 28(1), 95-112.

• Binderkrantz, A., Fisker, M., & Pedersen, H. (2016). The rise of citizen groups? The mobilization and representation of Danish interest groups, 1975–2010. Scandinavian Political Studies, 39(4), 291-311.

• Binderkrantz, A., Pedersen, H., & Beyers, J. (2017b). What is access? A discussion of the definition and measurement of interest group access. European Political Science, 16(3), 306-321.

• Bouteca, N., Devos, C., & Mus, M. (2013). The future of Belgian federalism as seen through the eyes of the social partners: A continuing obstacle to social policy decentralization? Regional & Federal Studies, 23(3), 293-309.

• Bunea, A. (2017). Designing stakeholder consultations: Reinforcing or alleviating bias in the European union system of governance? European Journal of Political Research, 56(1), 46-69.

• Celis, K., Mackay, F., & Meier, P. (2012). Social movement organizations and changing state architectures: Comparing women’s movement organizing in Flanders and Scotland. Publius: The Journal of Federalism, 43(1), 44-67.

• Christiansen, P. M., Mach, A., & Varone, F. (2018). How corporatist institutions shape the access of citizen groups to policy-makers: Evidence from Denmark and Switzerland. Journal of European Public Policy, 25(4), 526-545.

• Christiansen, P. M., Nørgaard, A. S., Rommetvedt, H., Svensson, T., Thesen, G., & Öberg, P. (2010). Varieties of democracy: Interest groups and corporatist committees in Scandinavian policy making. Voluntas: International Journal of Voluntary and Nonprofit Organizations, 21(1), 22-40.

• De Bruycker, I., & Beyers, J. (2015). Balanced or biased? Interest groups and legislative lobbying in the European news media. Political Communication, 32(3), 453-474.

• De Winter, L., Swyngedouw, M., & Dumont, P. (2006). Party system(s) and electoral behaviour in Belgium: From stability to balkanisation. West European Politics, 29(5), 933-956.

• Defourny, J., Develtere, P., Marée, M., Meireman, K., Mertens, S., & Raymaekers, P. (2005). Het Verenigingsleven in België. Een Kwantitatieve en Kwalitatieve Analyse. Brussel: Koning Boudewijnstichting.

• Deschouwer, K. (2012). The Politics of Belgium: Governing a Divided Society. Houndmills: Palgrave Macmillan.

• Dewachter, W. (1995). Adviesraden als Stap naar Institutionalisering en Medebeslissing. Res Publica, 37(1), 67-78.

• Dür, A., & Mateo, G. (2013). Gaining access or going public? Interest group strategies in five European countries. European Journal of Political Research, 52(5), 660-686.

• Fobé, E., Brans, M., Vancoppenolle, D., & Van Damme, J. (2010). De Werking van de Vlaamse Strategische Adviesraden Verkend: Een (Zelf)Analyse. Tijdschrift voor Bestuurskunde en Bestuursrecht, 6(2), 115-130.

• Fobé, E., Brans, M., Vancoppenolle, D., & Van Damme, J. (2013). Institutionalized advisory systems: An analysis of member satisfaction of advice production and use across 9 strategic advisory councils in Flanders (Belgium). Policy and Society, 32(3), 225-240.

• Fraussen, B. (2014). The visible hand of the state: On the organizational development of interest groups. Public Administration, 92(2), 406-421.

• Fraussen, B., & Beyers, J. (2015). Who’s in and who’s out?: Explaining access to policymakers in Belgium. Acta Politica, 51(2), 214-236.

• Fraussen, B., Beyers, J., & Donas, T. (2015). The expanding core and varying degrees of insiderness: Institutionalised interest group access to advisory councils. Political Studies, 63(3), 569-588.

• Fraussen, B., Bossens, N., Wilson, A., & Keating, M. (2017). Interest groups and policy analysis in Belgium: Examining the policy-analytical capacities and practices of prominent citizen and economic groups. In. M. Brans & D. Aubin (Eds.), Policy Analysis in Belgium (pp. 193-211). Bristol: Policy Press.

• Fraussen, B., & Halpin, D. (2016). Assessing the composition and diversity of the Australian interest group system. Australian Journal of Public Administration, 75(4), 476-494.

• Fraussen, B., & Wouters, R. (2015). Aandacht Trekken of Advies Verstrekken? De Aanwezigheid van Middenveldorganisaties in Adviesraad-en Beeldbuispolitiek. Res Publica, 57(2), 159-183.

• Gava, R., Varone, F., Mach, A., Eichenberger, S., Christe, J., & Chao-Blanco, C. (2017). Interests groups in parliament: Exploring MPs’ interest affiliations (2000-2011). Swiss Political Science Review, 23(1), 77-94.

• Grote, J., Lang, A., & Schneider, V. (2008). Organized Business Interests in Changing Environments: The Complexity of Adaptation. Houndsmills: Palgrave Macmillen.

• Halpin, D. R. (2006). The participatory and democratic potential and practice of interest groups: Between solidarity and representation. Public Administration, 84(4), 919-940.

• Halpin, D. R., & Fraussen, B. (2017). Conceptualising the policy engagement of interest groups: Involvement, access and prominence. European Journal of Political Research, 76(3), 823-732.

• Halpin, D. R., & Jordan, G. (2012). Estimating group and associational populations: Endemic problems but encouraging progress. In D. Halpin & G. Jordan (Eds.), The Scale of Interest Organization in Democratic Politics (pp. 1-22). Houndsmills: Palgrave Macmillan.

• Hanegraaff, M., Beyers, J., & Braun, C. (2011). Open the door to more of the same? The development of interest group representation at the WTO. World Trade Review, 10(4), 447-472.

• Hanegraaff, M., Beyers, J., & De Bruycker, I. (2016). Balancing inside and outside lobbying: The political strategies of lobbyists at global diplomatic conferences. European Journal of Political Research, 55(3), 568-588.

• Hanegraaff, M., Braun, C., De Bièvre, D., & Beyers, J. (2015). The domestic and global origins of transnational advocacy: Explaining lobbying presence during WTO ministerial conferences. Comparative Political Studies, 48(12), 1591-1621.

• Häusermann, S., Mach, A., & Papadopoulos, Y. (2004). From corporatism to partisan politics: Social policy making under strain in Switzerland. Swiss Political Science Review, 10(2), 33-59.

• Heylen, F., Fraussen, B., & Beyers, J. (2018). Live to fight another day? Organizational maintenance and mortality anxiety of civil society organizations. Nonprofit and Voluntary Sector Quarterly, 47(6), 1249–1270.

• Heylen, F., & Willems, E. (2019). Writing blank checks? How government funding affects interest organisations’ advocacy behaviour in a multi-layered context. Journal of European Public Policy, 26(6), 863-882.

• Hooghe, M. (1995). De Invloed van de Milieubeweging op het Milieubeleid: Neo-corporatisme of Culturele Vernieuwing? Res Publica, 37(1), 93-108.

• Hooghe, M. (1998). Selectieve Uitsluiting in het Belgisch Politiek Systeem. Innovatie en Protest door Nieuwe Sociale Bewegingen. Res Publica, 40(1), 3-21.

• Jagers, J., & Walgrave, S. (2007). Populism as political communication style: An empirical study of political parties’ discourse in Belgium. European Journal of Political Research, 46(3), 319-345.

• Jordan, G., Halpin, D., & Maloney, W. (2004). Defining interests: Disambiguation and the need for new distinctions? The British Journal of Politics & International Relations, 6(2), 195-212.

• Jordan, G., & Maloney, W. (2007). Democracy and Interest Groups: Enhancing Participation? Houndsmills: Palgrave Macmillan.

• Keating, M., & Wilson, A. (2014). Regions with regionalism? The rescaling of interest groups in six European states. European Journal of Political Research, 53(4), 840-857.

• Klüver, H., & Zeidler, E. (2019). Explaining interest group density across economic sectors: Evidence from Germany. Political Studies, 67(2), 459-478.

• Kriesi, H., Adam, S. & Jochum, M. 2006. Comparative analysis of policy networks in Western Europe. Journal of European Public Policy, 13(3), 341-361.

• Kriesi, H., Koopmans, R., Duyvendak, J.-W., & Giugni, M. G. (1995). New Social Movements in Western Europe. A Comparative Analysis. London: University College London.

• Lowery, D. (2007). Why do organized interests lobby? A multi-goal, multi-context theory of lobbying. Polity, 39(1), 29-54.

• Lowery, D., Baumgartner, F., Berkhout, J., Berry, J., Halpin, D., Hojnacki, M., Klüver, H., Kohler-Koch, B., Richardson, J., & Schlozman, K. L. (2015). Images of an unbiased interest system. Journal of European Public Policy, 22(8), 1212-1231.

• Lowery, D., & Gray, V. (1995). The population ecology of Gucci Gulch, or the natural regulation of interest group numbers in the American states. American Journal of Political Science, 39(1), 1–29.

• Lowery, D., & Gray, V. (2004). Bias in the heavenly chorus: Interests in society and before government. Journal of Theoretical Politics, 16(1), 5-29.

• Pauwels, T. (2011). Measuring populism: A quantitative text analysis of party literature in Belgium. Journal of Elections, Public Opinion and Parties, 21(1), 97-119.

• Pedersen, H., Halpin, D., & Rasmussen, A. (2015). Who gives evidence to parliamentary committees? A comparative investigation of parliamentary committees and their constituencies. The Journal of Legislative Studies, 21(3), 408-427.

• Rasmussen, A. (2015). Participation in written government consultations in Denmark and the UK: System and actor-level effects. Government and Opposition, 50(2), 271-299.

• Rasmussen, A., & Gross, V. (2015). Biased access? Exploring selection to advisory committees. European Political Science Review, 7(3), 343-372.

• Rommetvedt, H., Thesen, G., Christiansen, P. M., & Nørgaard, A. S. (2013). Coping with corporatism in decline and the revival of parliament: Interest group lobbyism in Denmark and Norway, 1980–2005. Comparative Political Studies, 46(4), 457-485.

• Salisbury, R. H. (1984). Interest representation: The dominance of institutions. American Political Science Review, 78(1), 64-76.

• Schlozman, K. L., Verba, S., & Brady, H. E. (2012). The Unheavenly Chorus: Unequal Political Voice and the Broken Promise of American Democracy. Princeton, NJ: Princeton University Press.

• Siaroff, A. (1999). Corporatism in 24 industrial democracies: Meaning and measurement. European Journal of Political Research, 36(2), 175-205.

• Tresch, A., & Fischer, M. (2015). In search of political influence: Outside lobbying behaviour and media coverage of social movements, interest groups and political parties in six western European countries. International Political Science Review, 36(4), 355-372.

• van den Berg, C. F., Braun, C., & Steen, T. (2014). Consensus politics as administrative practice: The Europeanization of external advice seeking? In H. Vollaard, J. Beyers, & P. Dumont (Eds.), European Integration and Consensus Politics in the Low Countries (pp. 138-157). London: Routledge.

• Van Den Bulck, J. (1992). Pillars and politics: Neo-corporatism and policy networks in Belgium. West European Politics, 15(2), 35-55.

• van der Meer, F., van den Berg, C., van Dijck, C., Dijkstra, G., & Steen, T. (2019). Consensus democracy and bureaucracy in the low countries. Politics of the Low Countries, 1(1), 27-43.

• Van Gyes, G., Terlinden, L., & Vandekerckhove, S. (2017). Belgian social dialogue since the crisis: The ups and downs of an organized system of industrial relations. In I. Guardiancich & O. Molina (Eds.), Talking through the Crisis. Social Dialogue and Industrial Relations Trends in Selected EU Countries. Geneva: International Labour Office.

• Van Haute, E., & Wauters, B. (2019). Do characteristics of consociational democracies still apply to Belgian parties? Politics of the Low Countries, 1(1), 6-26.

• Verschuere, B., & De Corte, J. (2014). The impact of public resource dependence on the autonomy of NPOs in their strategic decision making. Nonprofit and Voluntary Sector Quarterly, 43(2), 293-313.

• Willems, E. (2020). Politicized policy access. The effect of politicization on interest group access to advisory councils. Public Administration. DOI: 10.1111/padm.12651.

• Wonka, A., Baumgartner, F. R., Mahoney, C., & Berkhout, J. (2010). Measuring the size and scope of the EU interest group population. European Union Politics, 11(3), 463-476.

• Wonka, A., De Bruycker, I., De Brièvre, D., Braun, C., & Beyers, J. (2018). Patterns of conflict and mobilization: Mapping interest group activity in EU legislative policymaking. Politics and Governance, 6(3), 136-146.

Appendix – Interest Representation in Belgium

A Bottom-Up Census of Belgian Interest Groups

To map the population of Belgian interest organisations in a bottom-up fashion, we had different sources at our disposal. The most comprehensive database is the Kruispuntbank voor Ondernemingen (Crosspoint Bank for Enterprises – KBO) which is maintained by the Federal Public Service Economy, SMEs, Self-Employed and Energy (FPS Economy, Federale Overheidsdienst Economie, K.M.O., Middenstand en Energie). It registers more than two million corporate entities and, important for our purposes, all established VZWs (non-profits) or foundations. The register uses the second revision of the NACE classification (which is called NACEBEL in Belgium, and uses ISIC classification codes), requiring organisations to indicate in which industrial or other activities they primarily engage in (multiple options can be selected).⁸x NACE is the abbreviation of the French Nomenclature statistique des activités économiques dans la Communauté européenne. This European industry classification system consists of a 6 digit code and is systematically used in most national statistical data-systems (see ec.europa.eu/competition/mergers/cases/index/nace_all). To account for interest organisations, a separate category was created known as S94, which refers to organisations that represent the interests and views of specific constituencies.⁹x The full definition of S94 reads as follows: ‘This division includes activities of organisations representing interests of special groups or promoting ideas to the general public. These organisations usually have a constituency of members, but their activities may involve and benefit non-members as well. The primary breakdown of this division is determined by the purpose that these organisations serve, namely interests of employers, self-employed individuals and the scientific community (group 94.1), interests of employees (group 94.2) or promotion of religious, political, cultural, educational or recreational ideas and activities (group 94.9).’ This initial list contained 19,191 organisations.
The S94 category is further specified under sub-headers including business, employers and professional membership associations, development NGOs, as well as religious orders (such as abbeys, dioceses and other mostly local religious institutions). The latter organisations were deleted from the list (3,358 in total). Furthermore, the list also contained a high number of double entries (3,797 were deleted). Furthermore, we also deleted organisations based on tag-word searches (e.g. ‘EURO’ and ‘Youth House’) (in total 827 entities deleted). The remainder of the list included 11,209 entities.
One feature of the KBO is that it contains data on the connectedness of organisations. Indeed many Belgian organisations are part of a conglomerate structure, which is represented in the KBO as dyadic ties between organisations (Figure A1). The KBO makes a distinction between the ‘maatschappelijke zetel’ (main office/secretariat or main organisation) and the ‘vestigingseenheid’ (sub-entity/branch/ chapter or sub-organisation). The ‘maatschappelijke zetel’ is uniquely identified by the enterprise number. In principle, each organisation (with its different sub-entities) has one enterprise number, also known as the VAT number (value added tax number; used for tax purposes). Each separate ‘vestigingseenheid’ of an organisation has a unique identifier called the entity number. In principle, all entity numbers are linked to a registered main office (which always has an enterprise number). This feature was used to drastically reduce the number of organisations that needed to be checked manually. We did so by separating the so-called ‘main organisations’ (5,024) from the ‘sub-organisations’ (6,185). Yet, as some sub-organisations are active at the national or subnational level, they also needed to be processed; but we managed to do so in a semi-automated way.
This means that we can map whether and how organisations have established distinct branches, which often have their own legal personality and autonomy. For example, according to the KBO, the Ordre des Pharmaciens – Orde van Apothekers (Association of Belgium Pharmacists) has a secretariat which is located in Brussels, in addition to ten branches, one for each province in Belgium. In what follows, we refer to the secretariat as main organisation and to the branches or establishments as sub-organisations. Keep in mind that these sub-organisations are not the same as the members of an organisation. Also the main organisation is not necessarily more important than the sub-organisation. Although this is usually the case, sometimes sub-organisations are highly autonomous entities. The result is that some interest groups have a rather complicated branch structure, which might be typical for Belgium, yet similar structures can probably be observed in various other countries that are characterised by federalism, multilevel governance and neo-corporatism. Yet, one of the conceptual messages we draw from this is that we cannot equate an ‘interest group’ with one organisation, but that in many cases an interest group consists of a complicated patchwork of multiple interlinked organisations.
To process these organisations, we developed an (online) data-processing interface. First, a list of 5,024 organisations was uploaded. For these organisations, coders first determined whether the organisation is indeed ‘in target’ (given the criteria outlined above). Many main organisations – about 80% – did not qualify. This means that for about 80% of the organisations, the matching sub-organisations need not be checked. For the organisations that are in target, the tool compiled a list of matching sub-organisations, drawing not from the limited list of 6,185 sub-organisations, but from the full KBO database. Furthermore, as some organisations can have multiple main organisations (this is sometimes done for administrative reasons), from the uploaded list, the tool also compiled a list of main organisations based on a word similarity match (hereby reducing the list of organisations we had to manually check even further). The coders then arranged these entities in a hierarchical structure (based on whether the organisational entity is active on the local, subnational, national or international level).
Take, for instance, a certain interest organisation that is a conglomerate of three main organisations, to which ten sub-organisations are tied to each. This organisation would take up 33 entries in the registry. In essence, when not in target, the coder only has to process one of these main organisations to reduce the list with 33 entries. When in target, the coder can process these entities more efficiently, compared to processing them one by one, at random. So although we started from a ‘flat’ list of entities, the tool, in an efficient manner, enabled us to capture the hierarchical nature of organisations. All organisations active at the national and subnational level were included in the sample.
The resulting sample (n=1,461) was further triangulated, by comparing it to two other bottom-up sources: SectorLink and Filantropie.be. First, Filantropie.be is a voluntary register with mostly non-business interests, or social profit organisations, and encompasses organisations that are active at the national, subnational and local level (N=2,904 on 15 December 2014). It is an online platform that was developed through cooperation between the Koning Boudewijnstichting and the National Bank of Belgium. The database (www.filantropie.be) provides very detailed and precise information on each organisation (such as its mission, contact information, details about the structure of the organisation, annual accounts and yearly reports) and can be linked with the KBO. Still, it does not allow to distinguish local organisations from those that are organised at a regional or nation-wide level. Another drawback involves that, given the register’s voluntary nature, it is hard to assess which (and how many) organisations are missing, and to find out if there are any systematic or ad hoc biases in the current online database. Compared to the register of the KBO, however, one important benefit is that very small and informal organisations can also be registered (whereas the KBO requires a minimum amount of staff and/or financial resources). All in all, if one seeks to analyse the non-profit (or more specifically social sector) field in Belgium, filantropie.be probably offers the best and most comprehensive starting point.
Second, a similar and complementary initiative is SectorLink (www.bsae.be). It provides an overview of Belgian professional associations, industry groups and business federations. It includes organisations that are recognised as professional associations by the ‘Hoge Raad van de Middenstand’, as well as the member organisations of the main industry peak associations. These lists of organisations were also uploaded in the tool, and matched to the existing list by using a customised automated text-comparison function.
The new organisations were processed according to the same procedure, which resulted in 230 new organisations. The end result was a list of 1,678 organisations (for more detailed information, see www.cigsurvey.eu). After this sampling procedure, the online tool prompted the coder to collect contact data and some additional data from the organisation’s website. To establish the type of organisation, the coder first determined whether the organisation mentioned whether or not it had formal members. If this was the case, the coder coded the type and kind of members (from this can be derived whether the organisation is a business, professional, leisure, identity or labour union). The organisations without formal members were given a separate code. Additionally, for all organisations it was also established whether being politically active was one of their goals. Furthermore, for public interest organisations, it was also gauged what kind of interest they represent (cause or identity). Instead of instructing the coders to categorise organisations in a preset classification, coding separate variables allowed to make several typologies.

Code structure in the KBO

The Online Survey Procedure

The online survey of Belgian interest groups focused on topics such as advocacy strategies, organisational development and management, relations with members and stakeholders, and the challenges that organisations face. For each interest organisation we searched contact data for two (high-level) representatives of each organisation (for instance, the director, chair, president or secretary general). Contacting these high-level interest group representatives consisted of four consecutive steps, namely an invitation, an email reminder, a telephone reminder and one last email reminder. After these four steps, we repeated the procedure for those organisations that did not respond and for which we had a second contact person (so 4×2=8 steps). In total, this survey was sent to the high-level representatives of 1,678 nation-wide and subnational organised interests we identified. The survey was conducted over a time span of 117 days (between January and May 2016) and delivered a response rate of 42% (n=727 organisations that responded to more than 50% of questions).

The Mapping of Advisory Councils and Their Members

Sampling Advisory Councils

Advisory councils should be established at the national, Flemish or Walloon/Francophone level of government and have to deal with the consultancy of non-governmental stakeholders within the framework of policy formulation and evaluation, and implementation. Yet, these councils are not easily detectable and not one data source is available in Belgium. Therefore, the identification procedure of the population of advisory councils in Belgium rests on three data sources in order to come to the most complete picture possible: a website search of all the ministries at the national, Flemish and Walloon/Francophone level; a consultation of Politiek Zakboekje/Mémento Politique 2016; and several parliamentary questions. Furthermore, we identified also a substantial amount of councils while coding (often being sub-councils). At this stage, no distinction was made regarding the tasks nor the type of members of the identified advisory councils. The complete identification procedure was done in June-September 2016. In total, 1,136 advisory councils were identified. Table 1A gives an overview of identified councils by data source and level of government.
This general mapping of advisory councils resulted in a high number of councils that was not relevant for the purpose of this study. To filter these advisory councils out, four sampling criteria were used and Table 2A gives an overview of the number of councils excluded on the basis of each criterion. The criteria are put in hierarchical order, which means criterion one is the most determinant criterion. To assess the four criteria, three data sources were used: the own website of the advisory council or the dedicated webpages of a ministry or agency to that advisory council, Politiek Zakboekje/Mémento Politique 2016, and Het Staatsblad/Moniteur Belge. Het Staatsblad/Moniteur Belge is a particularly useful data source as in the founding law of each advisory council the functions and tasks are precisely listed. This delivers us with the most complete picture as to check the following criteria. First, sufficient information needs to be available in order to assess the subsequent criteria. Second, the advisory body should concern consultancy within the framework of general policy formulation and policy evaluation, or the guidance and monitoring of policy implementation. The policy advice concerning implementation can also be directed at institutions or professionals responsible for policy implementation and execution. All advisory bodies concerned with only the advice for the purpose of the adoption of individual administrative acts are excluded (see Table 2A).
Third, the advisory body has to be permanently established/active. All advisory bodies that are temporal or ad hoc in character will be excluded. In addition, all advisory councils that were found to be abolished during previous legislatures were excluded from the sample. Finally, at least one member in the advisory council has to be a non-governmental actor (e.g. interest group, expert organisation) or be a government representative not dependent on the governmental level on which the advisory council is established. When this is not the case, the advisory body was excluded from the sample.

Table 1A Number of advisory councils by source of identification

Data source of identification	Number of advisory councils	Level of government	Number of advisory councils
Website	463	National	711
Politiek Zakboekje/Mémento Politique	170	Flemish	174
Parliamentary questions	389	Walloon/Francophone	202
Identified while coding	114	Brussels	30
		German-speaking	15
		Local level	4
Total	1,136	Total	1,136

Table 2A Overview of the selection criteria and excluded councils

	Exclusion criterion	Number of excluded advisory councils
1.	No sufficient information availability	26
2.	Administrative acts
	Accreditation	73
	Arbitration, appeal and disciplinary bodies	88
	Examination, appointment/selection and promotion procedures (including internships)	84
	Approval, control and evaluation of projects	13
	Retributions in individual cases	31
	Licences and permits	7
	Management	48
	Social dialogue: employee–employer relationship and negotiations on labour conditions within specific ministries or government agencies	41
	Subsidies	40
	Other	17
	TOTAL ADMINISTRATIVE ACTS	441
3.	Temporal or dead	23
4.	Only government members dependent on the level on which the advisory council is established	26
		Total
	Removed from sample	520
	Remaining in sample	616

Mapping Members of Advisory Councils

Finally, a detailed coding of the 616 advisory councils in the sample was conducted. This involved evidence on the jurisdictional level of political activity (e.g. federal or subnational), some basic legal information (e.g. year of foundation, type of founding law, amendments), the policy domain and evidence on the day-to-day functioning of the council (e.g. staff, meetings, annual budget, gender quota, tasks). At the same time, we mapped and coded all the members of these 616 advisory councils. These members were classified into five categories: interest organisations (1), expert organisations (2), government representatives (3), political party representatives (4) and members of other advisory councils (5). Effective, alternate (i.e. having voting rights), as well as members that have a consultative voice and observers (i.e. having no voting rights) were mapped according to the number of seats they hold. In the end, 2,372 unique organisational members were mapped (thus excluding individual experts).
Coders were instructed to consult three data sources: the own website of the advisory council or the dedicated webpages of a ministry or government agency to that advisory council, Politiek Zakboekje/Mémento Politique 2016 and Het Staatsblad/Moniteur Belge. In Het Staatsblad/Moniteur Belge, every time the composition of an advisory council is changed, the list of members is published here.

Media Attention for Belgian Interest Groups

The Selection of Policy Issues

The starting point for our selection of policy issues is the Benchmark Survey for ‘de Stemtest’, conducted by TNS.¹⁰x Kantar TNS Belgium is a market research company (see tnsglobal.com/office/tns-Belgium). This is an online voter survey of 2,081 eligible Belgian voters conducted in the run-up to the subnational, national and European elections held on 25 May 2014. A separate survey was conducted in Flanders (n=1,053) and Wallonia (n=1,028), which resulted in an average response rate of 17%. Respondents were sampled on the basis of gender, age, education level, language and region. For generalisation purposes, oversampling was done for specific types of voters who are usually underrepresented in public opinion surveys (Lesschaeve, 2018).¹¹x Further discussion of the sample of voters can be found in the online appendix of Lesschaeve, van Erkel, and Meulewaeter (2018). Respondents were approached in two waves to avoid survey fatigue (each wave lasted on average 15 minutes). The statements on the national level were the same in both parts of the country, but each survey also contained custom-made statements for each subnational level. Consequently, while both surveys have some statements on subnational policy issues that are very similar, they also contain a substantial number of statements that have no counterpart in the other region. Voters could either agree or disagree with a policy statement (Lesschaeve, 2018). The statements were developed through consultation with political journalists from De Standaard, La Dernière Heure, La Libre Belgique (newspapers), VRT and RTBF (television) to guarantee resonance among voters and political parties and were refined through the use of focus groups of potential respondents. Moreover, the number of statements per policy domain also reflects the budgetary weight of the policy domain in each of the (sub)national governments’ budgets.
Three criteria were used to evaluate the quality and suitability of each of the 163 survey questions for gauging congruence between public opinion and interest groups. First, the question had to measure citizens’ preferences favouring or opposing a specific policy topic. Hence, the survey question had to be unidimensional and could not be hypothetical or conditional in nature (Flöthe & Rasmussen, 2018; Gilens, 2012). Thus, two Flemish issue statements and one Walloon statement addressing multiple priorities at once or that set conditions were excluded. In this way, all questions polling (broader) political attitudes were also excluded.
Second, a question had to concern an issue that could (possibly) be acted upon by either the national, Flemish or Walloon/Francophone governments, since our measure of positional congruence is applied to interest groups mobilising at the national, Flemish or Walloon/Francophone level (Flöthe & Rasmussen, 2018; Gilens, 2012). Based on this requirement six issue statements labelled as belonging to the national level were excluded because they included European or subnational matters on which the national government could not act independently. In addition, one issue statement labelled as belonging to the Flemish level was excluded because it was a local matter on which the subnational governments could not act alone. Finally, we excluded all survey questions dealing with administrative regulations, the institutional structure of the polity, election rules and party financing, the implementation and ratification of international treaties, the internal functioning and management of various governmental organisations and (the accumulation of) mandates, budget appropriations, etc. These cases were excluded from the sample because they do not address substantive regulations but rather relate to administrative organisation or budgetary allocations. This last step resulted in the exclusion of 42 issues. At this point, 110 policy issues remained in the selection, of which 37 were federal issues, 34 were Flemish issues and 39 were Walloon/Francophone issues. A complete list of policy issues can be consulted below.

Table 3A List of policy issues

ID	Statement	Jurisdiction	Policy domain
2	De pensioenleeftijd mag niet stijgen	National	Social affairs
3	Leefloners moeten verplicht kunnen worden gemeenschapswerk te verrichten	National	Social affairs
5	Wie nog nooit heeft gewerkt, mag geen werkloosheidsuitkering krijgen	National	Labour
6	De lonen moeten bevroren worden als ze sneller stijgen dan in de buurlanden	National	Economic affairs
7	In tijden van crisis mogen lonen niet automatisch worden aangepast aan prijsstijgingen	National	Economic affairs
8	Als er bij de NMBS wordt gestaakt, moet er een minimumdienst zijn	National	Labour
9	De verplichte sluitingsdag voor winkels moet worden afgeschaft	National	Economic affairs
10	Het moet makkelijker worden om werknemers te ontslaan	National	Labour
11	Er mag geen alcohol in drankautomaten zitten	National	Health care
12	Alle veroordeelden moeten hun straf volledig uitzitten	National	Justice
13	De GAS-boetes moeten worden afgeschaft	National	Justice
14	Grote vermogens moeten meer worden belast	National	Economic affairs
15	Klanten van prostituees moeten worden beboet	National	Justice
16	Het stakingsrecht mag niet worden ingeperkt	National	Labour
17	Er moet een rijbewijs met punten komen	National	Justice
18	Draagmoederschap voor homokoppels moet worden toegestaan	National	Rights
19	De federale overheid moet haar aandelen in Belgacom verkopen	National	Economic affairs
21	Er moeten kerncentrales open blijven	National	Energy
22	Een asielzoeker die hier als minderjarige is binnengekomen mag niet meer worden teruggestuurd	National	Migration
23	Jongeren die naar Syrië vertrekken om deel te nemen aan de strijd moeten hun recht op uitkeringen verliezen	National	Rights
27	Alle kernwapens die op Belgisch grondgebied opgeslagen zijn, moeten worden verwijderd	National	Foreign affairs & defence
28	Het Belgische leger moet minder deelnemen aan buitenlandse interventies	National	Foreign affairs & defence
29	Mensen moeten meer belastingen (BTW) betalen op wat ze kopen dan op wat ze verdienen	National	Economic affairs
30	Bedrijfswagens moeten zwaarder worden belast	National	Economic affairs
31	Het Belgische leger moet investeren in een opvolger van het F-16 gevechtsvliegtuig	National	Foreign affairs & defence
34	Een gezin moet voor ieder kind evenveel kinderbijslag krijgen	Flemish	Social affairs
35	Voor hoge inkomens moet de kinderbijslag naar omlaag, voor lage inkomens naar omhoog	Flemish	Social affairs
36	De regels om een privé-crèche uit te baten moeten worden versoepeld	Flemish	Social affairs
37	De regels voor de uitvoer van wapens en militaire onderdelen moeten strenger worden	Flemish	Foreign affairs & defence
38	Er moeten meer technische vakken gegeven worden in het secundair onderwijs	Flemish	Education
39	Pas na de eerste twee jaren van het secundair onderwijs zouden leerlingen een studierichting moeten kiezen	Flemish	Education
41	In plaats van een verkeersbelasting, moeten autobestuurders betalen volgens het aantal kilometers dat ze rijden	Flemish	Transportation
42	De aanleg van de Oosterweelverbinding moet worden stopgezet	Flemish	Transportation
43	Ook werkloze ouderen boven de 60 jaar moeten verplicht begeleid worden naar een nieuwe job	Flemish	Labour
44	Wie meer verdient, moet meer betalen voor de zorgverzekering wie minder verdient, moet minder betalen	Flemish	Health care
45	Leerkrachten in moeilijke scholen moeten een financiële bonus krijgen	Flemish	Education
46	Ook bus- en tramlijnen met weinig passagiers moeten blijven	Flemish	Transportation
47	65-plussers moeten gratis kunnen blijven reizen met bus of tram	Flemish	Social affairs
49	Nederlands kennen mag geen voorwaarde zijn om in aanmerking te komen voor een sociale woning	Flemish	Social affairs
50	De Antwerpse haven mag niet verder uitbreiden	Flemish	Transportation
51	L’organisation de certains cours dans une autre langue en humanités (immersion) doit être rendue plus facile	Francophone	Education
52	Un examen d’entrée doit être mis en place a l’université, dans chaque Faculté	Francophone	Education
54	Les écoles à public défavorisé doivent recevoir davantage de moyens	Francophone	Education
55	Le décret inscriptions (qui organise l’accès des élèves aux écoles) doit être supprime	Francophone	Education
57	Chaque enfant dans une famille doit recevoir le même montant d’allocations familiales	Walloon	Social affairs
58	Les allocations familiales doivent diminuer pour les hauts revenus et augmenter pour les bas revenus	Walloon	Social affairs
61	Les TEC doivent être privatises	Walloon	Transportation
63	Le port du foulard doit être interdit pour les élèves dans l’enseignement officiel	Francophone	Rights
65	Au lieu de la taxe de roulage, les automobilistes devraient payer en fonction du nombre de kilomètres parcourus	Walloon	Transportation
66	La vignette autoroutière doit être mise en place en Wallonie	Walloon	Transportation
71	Un quota de 10% de logements sociaux doit être obligatoire dans toutes les communes	Walloon	Social affairs
72	Les nouveaux immigres doivent suivre obligatoirement un parcours d’intégration	Walloon	Migration
76	La production d’énergie via des panneaux solaires ne doit plus être subsidiée	Walloon	Energy
77	Les allocations de rentrée scolaire doivent être augmentées pour les revenus les plus faibles	Walloon	Social affairs
82	Zodra sociale huurders voldoende verdienen, moeten ze hun sociale woning afstaan	Flemish	Social affairs
83	Vooral een lening voor de renovatie van een oude woning moet fiscaal voordelig zijn	Flemish	Energy
84	Wie een auto ouder dan 10 jaar vervangt door een zuiniger automodel, moet een premie krijgen	Flemish	Environment
85	Alle nieuwkomers moeten deelnemen aan een inburgeringsexamen	Flemish	Migration
86	De overheid moet pensioensparen fiscaal meer aanmoedigen	National	Economic affairs
87	Het leefloon moet stijgen	National	Social affairs
88	Het gebruik van cannabis moet volledig worden verboden	National	Justice
89	Mensen met een hoog inkomen moeten minder geld van de ziekteverzekering terugkrijgen	National	Social affairs
91	Werklozen moeten hun uitkering na een tijd verliezen	National	Social affairs
92	Werkgevers moeten worden verplicht om een bepaald aandeel mensen van vreemde origine in dienst te hebben	National	Labour
93	De treinstations die door weinig reizigers gebruikt worden, moeten ook open blijven	National	Transportation
94	Het rookverbod in de horeca moet worden versoepeld	National	Health care
95	De minimumleeftijd voor GAS-boetes moet hoger liggen dan de huidige leeftijd van 14 jaar	National	Justice
96	Een ouder die thuisblijft met de kinderen moet een inkomen krijgen	National	Social affairs
97	Mensen met een ongezonde levensstijl moeten minder geld van de ziekteverzekering terugkrijgen	National	Health care
98	Asielzoekers die te lang op een beslissing moeten wachten, moeten automatisch een verblijfsvergunning krijgen	National	Migration
99	Er moet een hoofddoekenverbod komen voor leerkrachten in het gemeenschapsonderwijs	Flemish	Education
100	Ouders van kinderen die spijbelen moeten tijdelijk hun kinderbijslag verliezen	Flemish	Social affairs
101	Scholen moeten halalmaaltijden aanbieden aan hun moslimleerlingen	Flemish	Migration
102	Bij smogalarm moeten de beperkingen op het gebruik van de wagen worden verstrengd	Flemish	Environment
103	Straten die opnieuw aangelegd worden, moeten een fietspad hebben	Flemish	Transportation
104	Tijdens spitsuren moet er op de autosnelwegen een rijstrook voorbehouden worden voor carpooling	Flemish	Transportation
106	Scholen moeten kinderen verplichten om ook op de speelplaats Nederlands te praten	Flemish	Education
107	Middelbare scholen mogen geen dure schoolreizen meer organiseren	Flemish	Education
108	Er moeten meer mensen van vreemde afkomst te zien zijn op de openbare omroep	Flemish	Culture
109	Wonen in de stad moet fiscaal aangemoedigd worden	Flemish	Environment
111	Les allocations familiales doivent être liées à la fréquentation scolaire	Walloon	Social affairs
112	Le montant maximal demande par les crèches privées doit être plafonne	Walloon	Social affairs
114	A partir du moment où les locataires sociaux gagnent suffisamment, ils doivent quitter leur logement social	Walloon	Social affairs
115	Seuls les prêts pour la rénovation de vieilles habitations doivent être fiscalement avantageux	Walloon	Economic affairs
116	Les propriétaires de plusieurs biens immobiliers doivent être davantage taxes	Walloon	Economic affairs
117	Il faut imposer des normes d’isolation pour les vieilles habitations	Walloon	Energy
118	En cas d’alerte à la pollution de l’air, les limitations d’utilisation des voitures doivent être renforcées	Walloon	Environment
119	Pendant les heures de pointe, une bande de circulation doit être réservée au covoiturage	Walloon	Transportation
123	Les écoles doivent offrir des repas halal à leurs élèves musulmans	Walloon	Migration
126	La Wallonie doit créer une ville nouvelle pour absorber la population croissante	Walloon	Environment
127	Les règles liées à l’exportation d’armes doivent être assouplies	Walloon	Foreign affairs & defence
128	Les voyages scolaires couteux pour les parents doivent être interdits	Walloon	Education
130	Les parents d’enfants qui brossent les cours doivent temporairement perdre leurs allocations familiales	Walloon	Social affairs

Content Analysis of Newspaper Articles

To identify relevant articles, we applied a computer-automated Boolean keyword search in GoPress with keywords closely related to the 110 policy issues. GoPress is the online press database and press monitoring service for all Belgian newspapers and magazine publishers.¹²x See academic.gopress.be. Keywords were carefully selected based on the name of the policy issue in the online voter survey and extensive desk research including legislative initiatives introduced in parliament on the policy issue. Keywords per policy issue range from one to more than seven, depending on when saturation was reached (i.e. no new articles were found). This saturation point was inductively determined by manually checking the number of (new) relevant articles that could be found by entering a new keyword in the GoPress search tool. We performed searches in both Dutch and French for federal policy issues, and limited ourselves to searches in either Dutch or French for policy issues specifically related to Flemish or Walloon/Francophone competences. In order to validate our French keywords, as the researchers involved in the project are not native speakers and do not follow Francophone news outlets on a daily basis, we contacted an external Walloon academic familiar with these news outlets.
For the selection of news media outlets, we opted to include only the most read newspapers in both regions of Belgium and selected publications that varied in terms of their left-right political orientation (see Binderkrantz et al., 2017). In Flanders, the news media outlets selected were De Standaard (centre-right, 715,100 daily readers) and De Morgen (centre-left, 448,500 daily readers). In Wallonia, the media outlets were Le Soir (centre-left, 639,400 daily readers) and La Libre Belgique (centre-right, 339,700 daily readers).¹³x www.cim.be/nl/pers/bereik-resultaten. The period from which we sampled spanned from one year before the elections to the end of 2017 (1 June 2013 to 31 December 2017, a 4.5-year period). A pilot study performed on 12 policy issues (four issues for each of the three government levels) including the most read mainstream newspapers and two additional high-quality newspapers indicated that the inclusion or exclusion of these additional outlets does not substantially alter the results.¹⁴x De Standaard, De Morgen, La Libre Belgique and Le Soir are the most read high quality newspapers, while the Het Laatste Nieuws, L’Avenir and La Dernière Heure are the most read mainstream newspapers. The media salience of issues across the different news outlets is highly correlated, with an average correlation of 0.72. This suggests that the four news media outlets in our study are representative of the broader news media arena in Belgium implying that our aggregate measure of media salience (the sum of all relevant and more broadly themed articles across media outlets) is reliable and valid.
The advantage of working with a computer-automated script is that it enabled us to collect a high volume of articles without manual intervention. The disadvantage of working with this script is that non-relevant articles in which our keywords appeared (e.g. foreign news, fait divers) were retained. Therefore, in the subsequent coding we manually checked the relevance of each article. Coders could indicate that the article was relevant, had no relevance at all to the policy issue as questioned in the online voter survey or that the article discusses the broader policy theme but not the specific policy issue (e.g. retirement policy is discussed, but not the specific issue of raising the retirement age). Non-relevant articles were excluded from further analyses. This coding was performed by one of the authors, a research assistant and two student assistants. Intercoder reliability checks performed on 817 double-coded articles verifying article relevance resulted in an 81.5% agreement (expected agreement of 52.3%), a kappa of 0.61 and Krippendorff’s alpha of 0.61. Discrepancies were corrected by the authors.

Identifying Interest Organisations

Once articles were mapped, we automatically identified interest organisations with news coverage about one or multiple of the 110 policy issues based on a curated dictionary containing all organisation names and abbreviations. The list of organisations we used is the result of a bottom-up and top-down sampling strategy that was designed to capture the population of Belgian interest groups mobilised at the subnational or the federal level. To facilitate the computer-automated identification of interest groups mentioned in the news media, additional coding was done so that different variations of the name or acronym of an interest group could be stored in a curated dictionary. In total, up to five Dutch and five French full names could be coded for each interest organisation in the dictionary, and four Dutch and four French acronyms could be coded. Data sources used to create the curated dictionary include the website of the interest group as well as the Crossroads Bank for Enterprises. Occasionally, we also manually checked how the interest group was usually mentioned in by the media by performing a quick search in GoPress (e.g. ACV, but also ‘Christelijke vakbond’). Furthermore, in the computer-automated identification Python-script we accounted for the possibility that an interest group name or acronym may be directly followed by a punctuation mark or preceded by an apostrophe.
Many interests groups are identified multiple times within and across different issues. To keep the coding of all media claims by interest groups expressing policy positions feasible, we used a multi-stage stratified sampling approach. First, for interest groups that appeared in up to three articles on a specific issue we sampled all the articles linked to that interest group. Second, if an organisation was mentioned more than three times in relation to an issue, a maximum of three of these articles from both the Flemish or Francophone media were randomly included in the sample. Third, to account for the limitations of the computer-automated identification, coders could also manually add organisations that made relevant claims. In total, we had a sample of 2,740 newspaper articles in which a particular interest group was identified.
Finally, each of these 2,740 articles was manually coded. Specifically, we coded for whether or not the identified interest group made a relevant claim on the specific policy issue at hand. A claim was defined as a quotation or paraphrase that can be connected to a specific interest organisation (De Bruycker 2017; Koopmans & Statham 1999. ). In total, 239 unique interest organisations made 986 claims on 83 issues (leaving 27 issues with no relevant interest group identifications and/or claims).

Data Visualisations

Founding dates by group type


Reference

De Bruycker, I. (2017). Politicization and the public interest: When do the elites in Brussels address public interests in EU policy debates? European Union Politics 18(4), 603-619.

Koopmans, R. & Statham, P. (1999). Political claims analysis: Integrating protest event and political discourse approaches. Mobilization: an international quarterly, 4(1), 203-221.

Lesschaeve, C., van Erkel, P. F., & Meulewaeter, C. (2018). Thinking alike: Two pathways to leadership-candidate opinion congruence. Journal of Elections, Public Opinion and Parties, 28(4), 488-515.

Noten

• * Acknowledgements: This research was funded by grants from the European Research Council (ERC-2013-CoG 616702-iBias) and the Research Foundation Flanders (FWO-V grant G032814N). We are grateful to the anonymous reviewers for their helpful suggestions, they allowed us to significantly improve the manuscript. Earlier versions of this manuscript also benefited from the constructive feedback of Bert Fraussen, and without his contributions to the research on Belgian interest groups, this project would not have seen the light of day. Replication materials are available in the dataverse-account of Jan Beyers at https://dataverse.harvard.edu/; the survey-datasets, the advisory council data, and the media data is available via www.cigsurvey.eu and www.ibias.eu.

• 1 NACE is the abbreviation of the French Nomenclature statistique des activités économiques dans la Communauté européenne. This European industry classification system consists of a six-digit code and is systematically used in most national statistical data systems (see http://goo.gl/8NLquM). The full definition of S94 reads as follows: This division includes activities of organisations representing interests of special groups or promoting ideas to the general public. These organisations usually have a constituency of members, but their activities may involve and benefit non-members as well. The primary breakdown of this division is determined by the purpose that these organisations serve, namely interests of employers, self-employed individuals and the scientific community (group 94.1), interests of employees (group 94.2) or promotion of religious, political, cultural, educational or recreational ideas and activities (group 94.9).

• 2 SectorLink (currently www.bsae.be) provides an overview of Belgian professional associations, industry groups and business federations. It includes organisations recognized as professional associations by the ‘Hoge Raad van de Middenstand’, as well as the member organisations of the main peak business associations. Filantropie.be (currently www.goededoelen.be) is a voluntary register with mostly non-profit organisations and encompasses organisations active at the national, subnational and local level (n=2,904 on 15 December 2014). It is an online platform developed through a cooperation between the Koning Boudewijnstichting and the National Bank of Belgium.

• 3 Also, the Brussels Capital Region and the German-speaking community have their own interest group community. Due to the strongly locally based nature of the latter and the considerable overlap with the Flemish and Francophone/Walloon interest group communities, we decided not to include these smaller communities in our bottom-up mapping.

• 4 There are two important aspects to be aware of when considering Figure 1. First, the evidence only concerns founding dates of groups that currently exist, which are all survivors. It tells us little about the composition of the groups’ system in previous eras and the dynamics associated with organisational mortality and survival in the past. Second, the strong decline in recent founding rates (4) should be dealt with cautiously. Although a possible explanation for this is the financial crisis starting in 2008 and government austerity suppressing organisational establishment and survival (Heylen et al., 2018), an entry lag in public directories for several years must be taken into account (Bevan et al., 2013; Fraussen & Halpin, 2016).

• 5 GoPress is the online press database and monitoring service for all Belgian newspapers and magazine publishers (www.gopress.academic.be). In Flanders, the news media outlets selected were De Standaard (715,100 daily readers) and De Morgen (448,500 daily readers). In Wallonia, the media outlets were Le Soir (639,400 daily readers) and La Libre Belgique (339,700 daily readers). For more information, see https://www.cim.be/nl/pers/bereik-resultaten.

• 6 Keywords were carefully selected based on the name of the policy issue in the online voter survey and extensive desk research including legislative initiatives on the policy issue. The saturation point for identifying key words was inductively determined by checking the number of (new) relevant articles that could be found by entering a new keyword in the GoPress search tool.

• 7 This list of 2,340 organisations resulted from a combination of identified organised interests through the bottom-up mapping and the mapping of advisory councils’ members. Additional coding was done so that different variations on the name and acronym of an interest group could be stored in the curated dictionary.

• 8 NACE is the abbreviation of the French Nomenclature statistique des activités économiques dans la Communauté européenne. This European industry classification system consists of a 6 digit code and is systematically used in most national statistical data-systems (see ec.europa.eu/competition/mergers/cases/index/nace_all).

• 9 The full definition of S94 reads as follows: ‘This division includes activities of organisations representing interests of special groups or promoting ideas to the general public. These organisations usually have a constituency of members, but their activities may involve and benefit non-members as well. The primary breakdown of this division is determined by the purpose that these organisations serve, namely interests of employers, self-employed individuals and the scientific community (group 94.1), interests of employees (group 94.2) or promotion of religious, political, cultural, educational or recreational ideas and activities (group 94.9).’

• 10 Kantar TNS Belgium is a market research company (see tnsglobal.com/office/tns-Belgium).

• 11 Further discussion of the sample of voters can be found in the online appendix of Lesschaeve, van Erkel, and Meulewaeter (2018).

• 12 See academic.gopress.be.

• 13 www.cim.be/nl/pers/bereik-resultaten.

• 14 De Standaard, De Morgen, La Libre Belgique and Le Soir are the most read high quality newspapers, while the Het Laatste Nieuws, L’Avenir and La Dernière Heure are the most read mainstream newspapers.