Content uploaded by Maxime Vandenberghe
Author content
All content in this area was uploaded by Maxime Vandenberghe on Sep 01, 2023
Content may be subject to copyright.
Power-sharing and
the paradox of
federalism
Decentralization and ethno-territorial
conflict in Belgian cabinets
(1979-2018)
Maxime Vandenberghe
Ghent University – FWO Vlaanderen
Examination board
prof. dr. Ronan Van Rossem (Chair) Ghent University
prof. dr. Wilfried Swenden University of Edinburgh
prof. dr. Min Reuchamps Université catholique de Louvain
prof. dr. Patricia Popelier Antwerp University
prof. dr. Bruno De Wever Ghent University
prof. dr. Carl Devos Ghent University
Supervisors
prof. dr. Nicolas Bouteca (Supervisor) Ghent University
prof. dr. Bram Wauters (Co-supervisor) Ghent University
2
3
Power-sharing and
the paradox of federalism
Decentralization and ethno-territorial conflict
in Belgian cabinets (1979-2018)
4
This work was supported by the Flemish Research Foundation (Fonds
Wetenschappelijk Onderzoek, FWO) under project number 3F009819.
Power-sharing and the paradox of federalism: Decentralization and ethno-territorial conflict in
Belgian cabinets (1979-2018)
PhD Thesis, Ghent University, 2023
Public defence: August 30, 2023 (Het Pand, Ghent)
© Maxime Vandenberghe
Printing: Reproduct
All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means
electronic or mechanical, including photocopying, recording or any transformation storage,
without written permission from the author.
5
Acknowledgements (Dutch)
Alle gebreken van dit doctoraat zijn alleen aan mij toe te schrijven. De rest niet. Enorm
veel mensen hebben eraan bijgedragen, elk op hun manier en dikwijls zonder het te weten. Ze
maakten het resultaat een pak beter en de afgelopen jaren een pak mooier. Dikke merci
daarvoor.
Eerst en vooral bedankt aan Nicolas, mijn stilaan eeuwige promotor. Dank voor de zowat zeven
jaar samenwerking, van bachelorproef tot doctoraat, de immense ruimte die je me gaf om mijn
ding te doen en de toegankelijke en sympathieke manier waarop je het promotorschap invulde.
Het zal er wellicht niet van komen, maar je nuchtere aanpak en West-Vlaams
relativeringsvermogen zouden moeten worden gedoceerd in de doctoraatsopleiding (liefst in
het Kortrijks, uiteraard). Ook alvast merci voor al wat nog volgt. Blij dat je de komende jaren
nog niet van me af bent.
Al wat ik (niet) schreef werd sterk beïnvloed door het klein leger aan collega’s en vrienden
binnen en buiten onze vakgroep. Dat ik kon beginnen schrijven heb ik onder andere te danken
aan Hilde Coffé en de warmhartige collega’s van het CDO, die me de kans gaven
onderzoekservaring op te doen terwijl de doctoraatsaanvraag hangende was. Bedankt aan de
oudgedienden die me daarna met open armen ontvingen in de (te) nostalgische gebouwen van
de Universiteitstraat en aan de vele nieuwkomers die GASPAR sindsdien versterkten. Dank
voor alles wat jullie met me deelden, inhoudelijk en persoonlijk. In het bijzonder bedankt aan
Tania en Wendy om zoveel meer te zijn dan ‘gewoon’ het secretariaat van de vakgroep, aan
Lorenzo voor de vele manieren waarop je mij en mijn onderzoek hebt gesterkt en aan Elizabeth
voor de zwarte humor (en muziek), de escapades in Lauwe city en de vakkundige opfleuring
van vergaderingen die anderen onterecht serieus zouden durven nemen. Benjamin en Elise,
merci voor jullie welgekomen sarcasme en jeugdig enthousiasme (ondanks de leeftijd), het vele
nalezen en meedenken, de betere avonden op de Gentse feesten, de roadtrip naar Mariakerke,
de volle tafels in Gentbrugge en de Foley’s en alle memorabele en vergeten momenten ertussen.
Veel van wat hier staat heb ik te danken aan mijn co-promotor, de leden van de
begeleidingscommissie en de juryleden: professoren Bram Wauters, Wilfried Swenden, Bruno
De Wever, Patricia Popelier, Carl Devos en Min Reuchamps. Jullie twijfelden geruststellend
mee, vergaderden zonder vergadersfeer en deden me niet uitvoeren maar nadenken. Van alle
antwoorden die mensen me (vaak spontaan) gaven, waren jullie lastige vragen de beste. Van de
talloze zaken waarop ik dikwijls heb gevloekt, waren ze het nuttigst. Bedankt daarvoor.
De vaak frustrerende zoektocht naar degelijke bronnen werd een pak aangenamer gemaakt door
6
de vriendelijke behulpzaamheid van Kathleen Bertrem (archief VRT), Fien Dejonckheere en
Hendrik Defoort (Universiteitsbibliotheek Gent). Tijdens de warme zomerweken in de schaduw
van de Reyers- en Boekentoren leerde ik vooral dat ik dringend andere bronnen moest zoeken.
De schat aan stoffige kranten en krakende radio-uitzendingen was zalig om in te ploeteren, maar
bood geen voldoende sterke basis voor mijn dataverzameling (mijn belangrijkste conclusie na
een jaar onderzoek; gelukkig zijn doctoraatsstudenten geen zelfstandigen). De indrukwekkende
politieke jaarboeken van de hand van Mark Deweerdt waren in dat opzicht een zegen. Ik ben
hem enorm dankbaar voor de gedetailleerde achtergrondinformatie die hij me erover verschafte
en voor zijn deugddoende interesse in het onderzoek. Het grondige, nauwkeurige en
enthousiaste werk van Manon Hoedt was cruciaal om het codeboek te verfijnen en de
betrouwbaarheid van de codering te testen.
Mijn onderzoek zal het debat over België niet in een andere plooi leggen, maar het staat er ook
niet langer volledig buiten. Dankzij verschillende mensen maakt het er ondertussen -minstens
een beetje- deel van uit: Arthur Parzysz (Le Soir), Jeroen Van Horenbeek (De Morgen),
Lorenzo Terrière (Het Laatste Nieuws), Benedikte Coussement (Radio 1), Ellen Debackere
(Bruzz), de journalisten van Belga, de organisatie van de Quetelet colleges (UGent) en de Low
Countries Studies (UGent) en de verschillende politici, partijen en medewerkers die me
doorheen de jaren hebben uitgenodigd voor duiding en afstemming. Soms kortstondig bij een
stuk pizza, soms uitvoerig in droge vergaderzalen. Ze pikten het doctoraat op en loodsten het
tot buiten de muren van de universiteit. Als het werk weldra stof vergaart en er niets van de
conclusies blijft hangen, zal het ondanks hun inzet zijn. Daar ben ik hen ongelooflijk dankbaar
voor.
Niemand is zo schuldig aan het feit dat ik erin slaag om maar te blijven plakken aan de
universiteit als mijn beperkte kring familie en vrienden. De lompe eerlijkheid en oprechte
desinteresse van mijn broers was altijd welgekomen, maar niets deed meer deugd dan de stille
trots van mijn vader, de zeer luide trots van mijn grootouders, de onevenaarbare steun van mijn
mama (die zonder overdrijven een dankwoord op zich verdient) en de compagnie van Pauline,
die het laatste anderhalf jaar van het doctoraat zoveel aangenamer heeft gemaakt. Merci voor
alles, gasten. Van alle belastingbetalers die me (ongewild) financierden, zijn jullie mijn
favoriete. Tot slot bedankt aan de twee compagnons op vier poten die niets van mijn doctoraat
snappen, maar er meer aan hebben bijgedragen dan de meeste mensen die het wel begrijpen:
Bruno, die na mijn eerste jaar doctoreren stopte met kwispelen, en Colin, de veel te sympathieke
labrador die gelukkig nog kwispelt voor twee.
Allemaal enorm bedankt.
7
8
9
Summary
Does decentralization fuel or pacify tensions? Autonomy provisions are amongst the
most popular instruments to address ethno-territorial tensions in divided states. Yet, both
politically and academically, their effectiveness is actively debated. Many consider
decentralization to be ineffective at best and counterproductive at worst. This view is echoed in
the so-called paradox of federalism: rather than mitigating conflict, it is thought to exacerbate
tensions and undermine the functioning, stability, and even survival of polities across the globe.
While decades of studies have fed the debate at hand, several empirical blind spots persist.
Notably, most studies have focussed on the way decentralization affects the so-called sub-state
challenge: does decentralization mitigate the regionalist threat from the periphery? Meanwhile,
little is known about its effects on the power-sharing challenge: does decentralization burden
or facilitate inter-segmental power-sharing? This is striking. The inter-segmental cooperation
between politicians is key to the stability of divided states. Its failure can nourish regionalist
dynamics and facilitating such cooperation is an important ambition behind many
decentralization reforms. Some indicators of power-sharing tensions exist, but these are rather
rough and typically focus on the birth and death of power-sharing cabinets (formation duration,
cabinet duration, etc.). What lacks is a fine-grained view of their day-to-day functioning, the
tensions that arise, and the way this is impacted by decentralization. Advancing upon the
important work already done, this work aims to provide exactly that.
Throughout this dissertation, I present an unprecedented longitudinal analysis of four decades
of segmental power-sharing against the background of five decentralization reforms in a least-
likely case for pacification: Belgium (1979-2018). Due to its peculiar combination of
institutional and historical features, the Belgian consociational federation is a theoretical
powder keg. Accordingly, if decentralization fuels tensions, this should be highly apparent in
Belgium. If not, this provides a serious blow to the paradox thesis. Throughout the work, two
questions stand central: How did the ethno-territorial conflict in Belgian power-sharing
cabinets evolve (RQ1)? And how did decentralization impact this conflict (RQ2)? To
answer these questions, I established two major datasets with over 1,000 cabinet conflicts in
twenty-one executives (Chapter 6). The different analyses of conflict frequencies, features,
intensities, and outcomes provide a unique account of the mundane struggle of everyday elite
cooperation in a notoriously divided state. The results disprove the paradox thesis on all
levels.
The introductory discussion of ethno-territorial conflicts’ morphology and prevalence
(Chapter 7) demonstrates that these clashes are a major burden to segmental power-sharing.
They account for 17% to 29% of all studied conflicts and on average, a new ethno-territorial
10
clash occurred every two to three-and-a-half months. Highlighting their heterogeneity, the
results show that tensions on ethno-territorial issues are far more common than conflicts
in which the Flemings (or Francophones) act ‘as one’. Hence, it does not appear to be
segmental power-sharing as such that triggers most tensions (the fact that both sides must
govern together), but rather the many issues and debates that are inherent to governing divided
polities (ranging from local applications of language laws to major discussions on state reform).
Neither segment is a political monolith, many intra-segmental conflicts are noted, and parties
are rarely engaged in segmental trench warfare. Furthermore, the very nature of ethno-territorial
conflict evolved (Chapter 7). Throughout the decades, the poltergeist got a friendlier face. It
became less pernicious. Ethno-territorial conflicts became less intense (with some nuances), are
less commonly fought along segmental lines, and are less commonly full-fledged conflicts
(clashes in which segmental blocks clash on ethno-territorial issues).
Meanwhile, ethno-territorial conflict did not become increasingly prevalent. Comparisons
of consecutive cabinets and thorough examinations of the absolute and relative frequencies of
conflicts and conflict sub-types (e.g. intense conflicts, full-fledged conflicts) indicate that
decentralization did not come with growing levels of ethno-territorial conflict (Chapter 8). None
of the 21 indicators validates the paradox thesis. The picture is marked by strong fluctuations
and the only long-term tendency that can be discerned is that of a tentative decline of ethno-
territorial conflict frequencies. Confirming this tendency, the 2011 agreement on the sixth state
reform kicked off an unprecedented period of lasting stability.
Next, and using a variety of multivariate Poisson and logistic regression analyses, I demonstrate
that this general pacification statistically relates to the five decentralization reforms under
study (Chapter 9). Decentralization not only relates to (i) decreasing frequencies of ethno-
territorial conflicts. It is also associated with lower odds that (ii) conflicts are ethno-territorial
in nature, (iii) that ethno-territorial issues are at stake when tensions arise, and (iv) that the
Flemish and/or Francophone coalition parties engage in conflict by acting ‘as one’ segmental
front. Of course, decentralization is not a panacea. Controlling for other factors of relevance,
the analyses show that the electoral saliency of decentralization, the proximity of federal
elections, and the participation of regionalist parties in federal cabinets positively relate to some
forms of ethno-territorial conflict too. However, the results are very nuanced. Ethno-territorial
conflicts are not just derivates of electoral incentives and concerns.
Having established the link between decentralization and conflict declines, I explore whether it
burdened the process of conflict resolution at the central policy-level. While there are fewer
conflicts, decentralization might have made it harder to settle the ones that still arise. As such,
pacification might have come at the expense of the functionality of the federal policy-level
(Chapter 10). Discerning different indicators of functionality, my examination of the outcomes
11
of conflicts provides a grim yet nuanced view of politicians’ conflict-solving abilities. It
regularly resembles a three-legged race. Tying the Flemish and Francophone segments
together prevents the minority from being regularly trampled, but when partners start to fight,
it is often hard to get things done and overcome immobilism. Ethno-territorial conflict
resolution is hard, but not harder than the settlement of other clashes and not increasingly harder
over time. If anything, consociational power-sharing appears to be quite consistently -if not
comfortably- numb. As such, decentralization did not make it harder to solve ethno-
territorial tensions and the pacification tendencies did not come at the expense of the
general functionality of the central government.
In sum, the relationship between sub-state autonomy and federal stability is compelling.
There are fewer ethno-territorial clashes than in the past, the ones that still arise have a friendlier
face, the pacification trends are statistically related to decentralization, and (partly) hollowing
out the central level did not burden its functionality or the conflict resolution process in the long
run. This clear rejection of the paradox thesis in a highly favourable context is quite surprising.
In making sense of these results, the work concludes by providing four tentative explanations
(Chapter 11). In doing so, I provide some counterweight to conventional views on the
destructive role of bipolar federalism, the idea that politicians in split party systems have little
incentives to be prudent, and the idea that decentralization emptied the Belgian federal level to
such a degree that the cost of non-agreement declined to problematic levels. Highlighting the
fact that this work is only one small piece of a large puzzle, I finish by highlighting some of
its limitations and presenting some avenues for future research.
12
13
Summary (Dutch)
Zorgt decentralisatie voor pacificatie of escalatie? Het introduceren en versterken
van deelstaatautonomie is één van de meest populaire recepten om communautaire spanningen
in verdeelde staten aan te pakken. De effectiviteit van dergelijke hervormingen is voorwerp van
hevig politiek en academisch debat. Volgens velen is decentralisatie ineffectief of zelfs
contraproductief. Deze visie weerklinkt in de zogenaamde paradox van federalisme: de these
dat decentralisatie conflicten niet verzacht maar aanwakkert, en zo het functioneren van staten,
hun stabiliteit en zelfs hun voortbestaan ondermijnt. Ondanks de talloze studies die het debat
doorheen de decennia wisten te voeden, blijven belangrijke empirische blinde vlekken bestaan.
De meeste onderzoeken richtten zich op de impact van decentralisatie op de zogenaamde
deelstaatuitdaging: slaagt decentralisatie erin de regionalistische druk vanuit de periferie te
verminderen? Veel minder aandacht werd geschonken aan haar effecten op een minstens even
uitdagende opdracht: machtsdeling. Bemoeilijkt of faciliteert decentralisatie de gezamenlijke
machtsuitoefening tussen politici van diverse segmenten? Dit hiaat is opvallend. De
samenwerking tussen politici van diverse segmenten is cruciaal voor de stabiliteit van verdeelde
staten. Het falen ervan kan regionalisme voeden en het faciliteren van dergelijke samenwerking
is een van de belangrijkste doelstellingen van veel decentralisatiehervormingen. Empirische
indicatoren van de spanningen die dergelijke machtsdeling kunnen opwekken zijn schaars,
eerder oppervlakkig en vooral gericht op de geboorte en dood van regeringen waarin diverse
segmenten samenwerken (formatieduur, regeringsduur, enz.). Wat ontbreekt is een fijnmazig
beeld van het dagdagelijks functioneren van dergelijke regeringen, de spanningen die daarbij
ontstaan en de manier waarop decentralisatie deze dynamieken beïnvloedt. Voortbouwend op
de belangrijke studies die reeds werden uitgevoerd, bestaat de ambitie van dit werk erin
dergelijk beeld te bieden.
Het doctoraat omvat een unieke longitudinale analyse van vier decennia van machtsdeling tegen
de achtergrond van vijf staatshervormingen in een context die bijzonder resistent is voor
pacificatie: België (1979-2018). Door haar eigenaardige combinatie van institutionele en
historische kenmerken is de Belgische federatie een theoretisch kruitvat. Als decentralisatie
spanningen aanwakkert, zou dit in België bijzonder duidelijk tot uiting moeten komen. Vanuit
die optiek zou de afwezigheid van escalatie-dynamieken sterk tegenbewijs vormen voor de
paradox van het federalisme. Twee vragen staan in dit werk centraal: Hoe evolueerde het
communautair conflict in Belgische federale regeringen (OV1)? Welke invloed had
decentralisatie op dit conflict (OV2)? Om deze vragen op afdoende wijze te beantwoorden
werden twee datasets opgesteld van ruim 1000 conflicten in eenentwintig regeringen
(Hoofdstuk 6). De analyses van conflictfrequenties, -kenmerken, -intensiteiten en -uitkomsten
bieden een unieke blik op de dagdagelijkse samenwerking tussen politici in een sterk verdeelde
14
staat. De resultaten gaan over de hele lijn in tegen de paradox-these.
De inleidende bespreking van de morfologie en prevalentie van communautaire conflicten
(Hoofdstuk 7) laat zien dat deze spanningen zwaar doorwegen op de federale machtsdeling.
Zo’n 17% tot 29% van alle bestudeerde conflicten is communautair en gemiddeld werd om de
twee tot drieënhalve maand een nieuwe communautaire botsing vastgesteld. De resultaten
geven duidelijk blijk van de heterogeniteit van dergelijke conflicten. Spanningen rond
communautaire thema’s komen veel vaker voor dan conflicten waarin de Vlamingen (of
Franstaligen) als ‘één blok’ optreden. Het lijkt dus niet zozeer de machtsdeling tussen
segmenten op zich te zijn die de meeste spanningen veroorzaakt (het feit dat beide zijden samen
moeten regeren), maar eerder de vele kwesties en debatten die eigen zijn aan het besturen van
verdeelde staten (gaande van lokale toepassingen van taalwetten tot fundamentele discussies
over staatshervorming). Geen enkel segment is een politieke monoliet, er worden veel
conflicten opgemerkt binnen segmenten en partijen zijn zelden betrokken bij het soort
communautaire loopgravenoorlogen waarbij een of beide zijden front vormen. Bovendien
evolueerde de aard van het communautaire conflict zelf (Hoofdstuk 7). De kwelduivel kreeg
een vriendelijker gelaat. Communautaire conflicten werden minder intens, werden minder
vaak uitgevochten tussen Vlaamse en Franstalige blokken en komen minder vaak voor in hun
meest vergevorderde vorm (waarbij dergelijke blokken botsen over communautaire kwesties).
Uiteraard kunnen hierbij enkele nuances worden gemaakt, maar het algemeen beeld is eerder
duidelijk.
Ondertussen is het niet zo dat communautaire conflicten steeds vaker opdoken. Vergelijkingen
van opeenvolgende kabinetten en grondige analyses van de absolute en relatieve frequenties
van conflicten en subtypes (bv. intense conflicten) tonen aan dat decentralisatie niet gepaard
ging met de toename van communautaire conflicten (Hoofdstuk 8). Geen enkele indicator
biedt bevestiging voor de paradox-stelling. Het beeld is er een van sterke fluctuaties doorheen
de tijd en de enige trend die op lange termijn kan worden vastgesteld is die van een voorzichtige
afname van communautaire conflictfrequenties. Het Vlinderakkoord van 2011 over de zesde
staatshervorming bevestigde deze tendens en luidde een ongekende periode van duurzame
stabiliteit in.
In Hoofdstuk 9 laat ik met behulp van verschillende multivariate regressieanalyses zien dat deze
algemene pacificatie effectief samenhangt met de vijf staatshervormingen die tussen 1979
en 2018 plaatsvonden. Decentralisatie is gelinkt aan (i) afnemende communautaire
conflictfrequenties en lagere kansen dat (ii) conflicten communautair van aard zijn, (iii) over
communautaire thema’s gaan en dat (iv) Vlaamse en/of Franstalige coalitiepartijen als één front
tegenover elkaar staan. Uiteraard is decentralisatie geen wondermiddel. De analyses
controleren op andere relevante factoren en tonen aan dat de nabijheid van federale
15
verkiezingen, de mate waarin decentralisatie electoraal wordt benadrukt en de federale
regeringsdeelname van regionalistische partijen positief samenhangt met bepaalde vormen van
communautaire conflict. De resultaten zijn echter zeer genuanceerd en tonen duidelijk aan dat
communautaire conflicten niet louter het product zijn van electorale zorgen en incentives.
Nadat ik het verband tussen decentralisatie en pacificatie heb vastgesteld, onderzoek ik of deze
dynamiek de conflictresolutie op centraal beleidsniveau bemoeilijkt heeft (Hoofdstuk 10).
Decentralisatie bracht misschien wel het aantal conflicten naar beneden, maar kan het
ondertussen lastiger hebben gemaakt om de conflicten die nog opduiken op te lossen. Op die
manier zou de vastgestelde pacificatie ten koste zijn gegaan van de functionaliteit van het
federale beleidsniveau. De analyse van conflictuitkomsten, waarin verschillende indicatoren
van functionaliteit worden onderscheiden, levert een grimmig maar genuanceerd beeld op van
het conflictoplossend vermogen van politici. In bepaalde mate doen de resultaten denken aan
een driepotige race. Door het Vlaamse en Franstalige segment aan elkaar vast te binden kan
men vermijden dat de minderheid regelmatig onder de voet wordt gelopen, maar zodra
spanningen tussen de partners opduiken, is het vaak erg moeilijk om tot tastbare oplossingen te
komen en het immobilisme te overwinnen. Het oplossen van communautaire conflicten is
moeilijk, maar niet moeilijker dan het oplossen van andere conflicten en bovendien (en
belangrijker voor deze studie) niet moeilijker doorheen de tijd. Voor zover de onderzochte
regeringen verlamd lijken te zijn door besluiteloosheid en dysfunctionaliteit, is die intertie
eerder consistent. Decentralisatie maakte het dus niet moeilijker om communautaire
spanningen op te lossen en de pacificatietendensen gingen niet ten koste van de algemene
functionaliteit van de centrale overheid.
Samengevat legt deze studie een duidelijke relatie bloot tussen deelstaatautonomie en
federale stabiliteit. Er zijn minder communautaire botsingen dan in het verleden, de
spanningen die nog bestaan hebben een vriendelijker aanzicht, de pacificatietrends zijn
statistisch gerelateerd aan decentralisatie en het (gedeeltelijk) uithollen van het centrale niveau
tastte de conflictoplossende functionaliteit van het federale niveau niet aan. Gezien de
Belgische context erg vatbaar is voor conflict, is deze duidelijke ontkrachting van de
paradoxthese verrassend. In de conclusie van dit werk worden vier verklaringen voor deze
opvallende resultaten besproken (Hoofdstuk 11). Daarbij ga ik in tegen conventionele
opvattingen over de destructieve aard die bipolair federalisme zou kenmerken, het idee dat
politici in geregionaliseerde partijsystemen weinig prikkels zouden ervaren om zich
constructief op te stellen en de these dat decentralisatie het Belgische federale niveau dermate
uitkleedde dat de kosten van onenigheid problematisch klein werden. Het werk wordt
afgesloten door te benadrukken dat het slechts een klein stukje van een grote puzzel vormt,
diverse beperkingen heeft en veel ruimte laat voor toekomstige onderzoekspistes.
16
17
List of tables
Table 1.
Gopress keyword search
Table 2.
Conflict indicators and intensity levels
Table 3.
Ethno-territorial issues
Table 4.
Intercoder reliability test (N = 93)
Table 5.
Dataset comparison: yearbooks vs. news archives (1995–2006)
Table 6.
Ratio between ethno-territorial (ETC) and non-ethno-territorial conflict
(non-ETC) over time: fictional examples
Table 7.
Descriptive statistics of dataset 1 (monthly ethno-territorial conflict
frequency) and dataset 2 (conflict characteristics)
Table 8.
Conflict outcome types and functionality dimensions
Table 9.
Frequency table: issues of cabinet conflicts (news archive data: N= 1013,
1995-2018; yearbook data: N= 328, 1979-2006)
Table 10.
Frequency table: conflict sides of cabinet conflicts (news archive data:
N= 1013, 1995-2018; yearbook data: N= 328, 1979-2006)
Table 11.
Frequency table: intensity of ethno-territorial cabinet conflicts (news
archive data: N= 169, 1995-2018; yearbook data: N= 96, 1979-2006)
Table 12.
Descriptive statistics (dataset 1: monthly ethno-territorial conflict
frequency; dataset 2: conflict characteristics)
Table 13.
Poisson regressions with monthly number of ethno-territorial conflicts
(ETCs) as dependent variable (incidence rate ratios and Std. Errors)
Table 14.
Logistic regressions (Odds ratios and CI)
Table 15.
Outcomes of ethno-territorial conflicts by type and functionality
dimension: percentage of all ethno-territorial conflict outcomes (N= 96)
18
19
List of figures
Figure 1.
Typology of ethno-territorial conflicts (ETCs) by sides and issues
Figure 2.
Visualisation of possible cabinet conflicts
Figure 3.
General election scores of regionalist parties in Belgium (% of seats in the
Chamber of Representatives; 1949-2019)
Figure 4.
Post-electoral cabinet formation duration (days) in Belgium (1946-2019)
Figure 5.
Overview of hypotheses
Figure 6.
Length (number of pages) of the coded chapters of the Res Publica yearbooks
(1979-2006)
Figure 7.
Absolute frequency of coded cabinet conflicts in Belgium (1979-2018)
(Yearbook data: N= 328; News archive data: N= 1013)
Figure 8.
Yearly frequency of cabinet conflicts on linguistic struggles as a percentage of
all ethno-territorial cabinet conflicts (yearbook data: N= 96; news archive data:
N= 169)
Figure 9.
Yearly frequency of cabinet conflicts on state reform as a percentage of all
cabinet conflicts on ethno-territorial issues (yearbook data: N= 80; news archive
data: N= 159)
Figure 10.
Yearly frequency of cabinet conflicts on distributive issues as a percentage of
all cabinet conflicts on ethno-territorial issues (yearbook data: N= 80; news
archive data: N= 159)
Figure 11.
Yearly frequency of cabinet conflicts along segmental lines as a percentage of
all ethno-territorial cabinet conflicts (yearbook data: N= 96; news archive data:
N= 169)
Figure 12.
Typology of ethno-territorial conflicts (ETCs): prevalence in Belgian cabinets
as a percentage of all ethno-territorial conflicts (news archive dataset (1995-
2018; N= 169) and yearbook dataset (1979-2006; N= 96) reported in that order)
Figure 13.
Yearly frequency of full-fledged ethno-territorial conflicts as a percentage of
20
all ethno-territorial cabinet conflicts (yearbook data: N= 96; news archive data:
N= 169)
Figure 14.
Yearly frequency of intense ethno-territorial conflicts (levels 1 and 2) as a
percentage of all ethno-territorial cabinet conflicts (yearbook data: N= 96; news
archive data: N= 169)
Figure 15.
Ethno-territorial cabinet conflict: yearly absolute frequency (1979-2018)
Figure 16.
Ethno-territorial cabinet conflict: yearly relative frequency (% of all cabinet
conflicts, 1979-2018)
Figure 17.
Ethno-territorial cabinet conflict frequency: yearly equivalent by cabinet (1979-
2018)
Figure 18.
Intense ethno-territorial cabinet conflicts (0 excluded): yearly absolute
frequency (1979-2018)
Figure 19.
Intense ethno-territorial cabinet conflicts (0 excluded): relative frequency (%
of all cabinet conflicts, 1979-2018)
Figure 20.
Intense ethno-territorial cabinet conflict frequency: yearly equivalent by
cabinet (1979-2018)
Figure 21.
Ethno-territorial cabinet conflicts by type: yearly percentage of all cabinet
conflicts (yearbook data: 1979–2006; news archive data: 2007-2018)
Figure 22.
Yearly ethno-territorial cabinet conflict frequency, decentralization degree and
saliency, regionalist party strength, regionalist cabinet participation, and federal
elections (Belgium, 1979-2006)
Figure 23.
Outcomes of ethno-territorial cabinet conflicts (yearbook data; N= 96)
Figure 24.
Outcomes of ethno-territorial cabinet conflicts in Belgium by year: total
functionality level (0-500) and percentage of outcomes meeting different
functionality dimensions (1979-2006; N= 96)
Figure 25.
Outcomes of ethno-territorial (N= 96) vs. other cabinet conflicts (N= 229) in
Belgium by state reform period: total functionality level (0-500) and percentage
of outcomes meeting different functionality dimensions (1979-2006)
21
22
23
Table of contents
ACKNOWLEDGEMENTS (DUTCH) ...................................................................................... 5
SUMMARY ............................................................................................................................... 9
SUMMARY (DUTCH) ............................................................................................................ 13
LIST OF TABLES ................................................................................................................... 17
LIST OF FIGURES .................................................................................................................. 19
TABLE OF CONTENTS ......................................................................................................... 23
CHAPTER 1. INTRODUCTION ........................................................................................ 29
1. POWER-SHARING AND THE PARADOX OF FEDERALISM .................................................... 29
2. RESEARCH FOCUS ........................................................................................................... 34
2.1. The Belgian case: a canary in the coal mine ......................................................... 34
2.2. Cabinet conflicts .................................................................................................... 35
2.3. Period under study (1979-2018) ............................................................................ 36
3. OUTLINE OF THE STUDY ................................................................................................. 37
CHAPTER 2. CONCEPTS AND DEFINITIONS ............................................................. 39
1. DIVIDED STATES ............................................................................................................. 39
1.1. Divided states: definition and application ............................................................. 40
1.2. Terminology ........................................................................................................... 47
2. ETHNO-TERRITORIAL CONFLICT ..................................................................................... 48
3. CABINET CONFLICT ........................................................................................................ 51
4. DECENTRALIZATION ....................................................................................................... 55
CHAPTER 3. THE BELGIAN POWER-SHARING CHALLENGE: A HISTORICAL
BACKGROUND ..................................................................................................................... 57
1. FRANCOPHONE RULE IN A UNITARY STATE ..................................................................... 57
2. FLEMISH EMPOWERMENT AND THE FRANCOPHONE FEAR OF MINORIZATION .................. 59
2.1. Shifting balance of power ...................................................................................... 59
2.2. Post-war tensions and traumas .............................................................................. 61
2.3. Economic transition ............................................................................................... 63
2.4. Linguistic tensions ................................................................................................. 64
2.5. Power-sharing crises within parties and the split of the party system .................. 67
3. POWER-SHARING IN A CONSOCIATIONAL FEDERATION ................................................... 68
3.1. Imposing power-sharing ........................................................................................ 69
3.2. Facilitating power-sharing: segmental autonomy ................................................. 70
24
CHAPTER 4. THE PARADOX OF BELGIAN FEDERALISM: A BRIEF OVERVIEW
OF EXISTING INDICATORS ............................................................................................. 81
1. DECENTRALIZATION AND THE SUB-STATE CHALLENGE .................................................. 81
1.1. Regionalist parties’ strength and positions: a paradox after all? ......................... 82
1.2. Mainstream parties’ positions: a cautious reversal of the paradox ...................... 83
1.3. Public identification, support for independence, and autonomy: the paradox of
indifference ....................................................................................................................... 84
2. DECENTRALIZATION AND THE POWER-SHARING CHALLENGE ......................................... 87
2.1. The birth of power-sharing executives ................................................................... 87
2.2. The death of power-sharing executives .................................................................. 88
2.3. The life of power-sharing executives ..................................................................... 89
CHAPTER 5. DECENTRALIZATION, CONFLICT PREVALENCE, AND
CONFLICT ENDINGS: A THEORETICAL FRAMEWORK ........................................ 95
1. DECENTRALIZATION AND CONFLICT (OUTCOMES) .......................................................... 95
1.1. Pacification ............................................................................................................ 95
1.2. Escalation ............................................................................................................ 102
2. CONTEXTUAL FACTORS ................................................................................................ 107
2.1. Constants: the Belgian case ................................................................................. 108
2.2. Varying factors: issue saliency, regionalist and electoral pressures .................. 120
3. OVERVIEW ................................................................................................................... 125
CHAPTER 6. DATA AND METHOD .............................................................................. 129
1. MAPPING CABINET CONFLICTS AND CONFLICT FEATURES ............................................. 129
1.1. Sources and sample selection .............................................................................. 130
1.2. Cabinet conflicts: operationalization .................................................................. 137
1.3. Conflict features: operationalization ................................................................... 140
1.4. Comparing the yearbook and news archive data ................................................ 151
1.5. Data availability ..................................................................................................... 154
2. METHOD AND ANALYSES .............................................................................................. 154
2.1. Conflict frequency evolutions .............................................................................. 154
2.2. Explanatory analyses ........................................................................................... 156
2.3. Conflict outcomes ................................................................................................. 161
CHAPTER 7. MORPHOLOGY AND PREVALENCE OF ETHNO-TERRITORIAL
CONFLICTS ......................................................................................................................... 165
1. PREVALENCE: HOW COMMON ARE ETHNO-TERRITORIAL CONFLICTS? .......................... 165
2. THE (CHANGING) MORPHOLOGY OF ETHNO-TERRITORIAL CONFLICTS .......................... 166
25
2.1. What ethno-territorial issues are ethno-territorial conflicts about? ................... 166
2.2. Sides: which actors stand against each other in ethno-territorial conflicts? ...... 173
2.3. Typology: how frequent is each type of ethno-territorial conflict? ..................... 177
2.4. Intensity: how intense are ethno-territorial conflicts? ........................................ 178
3. INTERMEDIATE CONCLUSIONS ...................................................................................... 180
CHAPTER 8. EVOLUTION: ETHNO-TERRITORIAL CONFLICT FREQUENCIES
DECLINED OVER TIME ................................................................................................... 183
1. FREQUENCY OF ALL ETHNO-TERRITORIAL CONFLICTS .................................................. 183
2. FREQUENCY OF INTENSE ETHNO-TERRITORIAL CONFLICTS ............................................ 188
3. FREQUENCY OF ETHNO-TERRITORIAL CONFLICTS BY SUB-TYPES .................................... 191
4. INTERMEDIATE CONCLUSIONS ...................................................................................... 193
CHAPTER 9. EXPLANATORY ANALYSES: DECENTRALIZATION IS RELATED
TO ETHNO-TERRITORIAL PACIFICATION .............................................................. 195
1. VARIABLES (DESCRIPTIVES) ......................................................................................... 195
2. POISSON REGRESSION: THE FREQUENCY OF ETHNO-TERRITORIAL CONFLICTS ............... 198
3. LOGISTIC REGRESSION: THE ETHNO-TERRITORIAL NATURE OF CONFLICTS ................... 199
4. INTERMEDIATE CONCLUSIONS ...................................................................................... 203
CHAPTER 10. CONFLICT OUTCOMES: DECENTRALIZATION DID NOT
BURDEN CONFLICT RESOLUTION ............................................................................. 207
1. AGGREGATED RESULTS: HOW ARE ETHNO-TERRITORIAL CONFLICTS SOLVED? ............. 207
2. EVOLUTION OF (ETHNO-TERRITORIAL) CONFLICT OUTCOMES AGAINST THE BACKGROUND
OF DECENTRALIZATION ........................................................................................................ 211
3. INTERMEDIATE CONCLUSIONS ...................................................................................... 215
CHAPTER 11. CONCLUSION: NO POWER-SHARING PARADOX ........................ 219
1. MAIN RESULTS ............................................................................................................. 220
1.1. Ethno-territorial conflict: a multifaceted burden ................................................ 220
1.2. Pacification over time .......................................................................................... 222
1.3. Decentralization relates to pacification ............................................................... 222
1.4. Decentralization did not burden conflict resolution ............................................ 223
1.5. The paradox of pacification ................................................................................. 225
1.6. The paradox of centralism ................................................................................... 228
2. LIMITATIONS AND FUTURE RESEARCH .......................................................................... 229
REFERENCES ..................................................................................................................... 235
APPENDICES ...................................................................................................................... 277
26
27
Part I.
Introduction
28
29
Chapter 1.
Introduction
1
1. Power-sharing and the paradox of federalism
‘The unitary state, with its structure and method
as currently regulated by the laws, has been overtaken by the events.’
On the 18th of February 1970, Prime Minister Gaston Eyskens shared these historical
words with parliament – and as such, with the Belgian population at large. The Flemish-
Francophone cooperation on which Belgian politics relies proved too hard. Fueled by an
accumulation of distrust and tensions, governing Belgium was considered to be impossible
without major reforms. The search for pacification led to a fundamental institutional
transformation. A peculiar ‘consociational federation’ was established (Deschouwer, 2002,
2005, 2006). On the one hand, a series of consociational mechanisms effectively forced both
segments to cooperate and govern together at the federal level. Majoritarianism is no longer an
option. On the other hand, and crucially, consecutive decentralization reforms were used to
facilitate this forced power-sharing. At least in certain domains, the fighting dogs were
separated. Newly established sub-state governments allowed the Flemish and Francophone
segments to ‘go their own way’ and adopt their own policies – thus lifting the very need to
cooperate in the first place. Driven by this logic (amongst others), the autonomy of the sub-
states was boosted throughout six major waves of decentralization, the last of which was agreed
upon in 2011. Gradually, Belgium became one of the most decentralized countries in the world
(Fabre, 2009; L. Hooghe et al., 2016; Shair-Rosenfield et al., 2021; Watts, 2013).
But did it work? More than fifty years after their initiation, the track record of these
decentralization reforms is still heavily contested. While pacification was one of their
primary goals, these reforms have often been blamed for deepening divisions, feeding
intransigence, and fuelling tensions (e.g. Caluwaerts & Reuchamps, 2015; De Winter &
1
Some of the ideas presented here were previously discussed (usually more briefly) in Vandenberghe (2022b,
2022c) and are expected to be covered in two articles that are under review at the time of publication
(Vandenberghe, 2022d, 2023).
30
Baudewyns, 2009). Tellingly, and for the first time in history, a growing set of scholars and
politicians has even started to promote reforms to re-strengthen the Belgian state, e.g. through
the re-centralization of certain policy domains, the creation of a federal electoral district, or the
introduction of a hierarchy of norms (Dodeigne, Gramme, Reuchamps, & Sinardet, 2016;
Dodeigne, Niessen, Reuchamps, & Sinardet, 2022; Dodeigne, Reuchamps, & Sinardet, 2015;
Pascolo, Vermassen, Reuchamps, & Caluwaerts, 2021; Sinardet, Dodeigne, & Reuchamps,
2013). As a result, the debate on the future of Belgium is now wider than ever.
2
Meanwhile, the
never-ending search for stability continues. For example, aiming to transform Belgium into a
‘country of cooperation and respect’, the current cabinet (De Croo) has promised to prepare a
seventh state reform (De Croo & Magnette, 2020, pp. 7-8). For the first time, this preparation
also included a participatory trajectory, allowing citizens to express their views on the future of
Belgium (De Clercq, 2023). In the meantime, and more traditionally, federal departments were
asked to provide lists of institutional bottlenecks while different parliaments established special
committees and used expert hearings to evaluate Belgium’s functioning and future.
3
This quite fundamental debate is not unique to Belgium. Facing similar (or worse)
tensions, countries across the globe have used the decentralization recipe too. And across
the globe, its effectiveness is contested. In fact, the pacifying impact of decentralization is
subject to an extensive and ongoing academic debate (Amoretti & Bermeo, 2004; L. Anderson,
2013; Bermeo, 2002; Brancati, 2006, 2009; Erk & Anderson, 2010b; Horowitz, 2000, 2007;
Keil & Anderson, 2018; Swenden, 2013b; Watts, 2015; Wolff, 2013). On the one hand,
decentralization is praised and promoted by many as a promising tool for ethno-territorial
conflict management (e.g. L. Anderson, 2016; Bermeo, 2002; Keil & Anderson, 2018; Lijphart,
1981c; Stepan, 1999). On the other hand, an extensive chorus of scholars has highlighted the
2
What used to be a valence issue (pleas for state reform were almost always pleas for decentralization) is
now a highly polarized position issue (Stokes, 1963). Valence issues are issues on which only one position
is politically considered to be possible (acceptable), while position issues are issues on which opposing
political positions are taken.
3
Federal parliament: Verslagen namens de parlementaire commissie belast met de Evaluatie van de
Staatshervormingen sinds 1970 uitgebracht door de heren Loones (K.) en Calvo (K.), mevrouw Pas (K.), de
heren Eerdekens (S.) en Van Rompuy (S.) en mevrouw Durenne (S.), Parl.St. Senaat 2021-22, nr. 7-280/2-7
en Parl.St. Kamer 2021-22, nr. 55-2602/2-7. // Flemish parliament: Verslag van de hoorzittingen namens de
Werkgroep Institutionele Zaken uitgebracht door Karl Vanlouwe, Peter Van Rompuy en Caroline Gennez
over de staatsstructuur en het toekomstig staatsmodel, Parl.St. Vl.Parl. 2021-22, nr. 1257/1.
31
potential counter-productiveness of such reforms. In the wording of Dawn Brancati (2006), it
might fuel the fire rather than dampen the flames of ethno-territorial conflict and secessionism.
This view is commonly known as the paradox of federalism (Erk & Anderson, 2010b): rather
than being a solid tool for pacification, providing autonomy is considered to be a recipe for
growing tensions and secessionism.
4
While the debate on the conflict-fuelling or stabilizing impact of decentralization is alive
and kicking, its empirical basis is not in proportion to its importance. To date, several major
blind spots remain. Notably, there is a strong imbalance in the indicators that have been studied.
Of course, in studying the effects of decentralization, “it is impossible to establish a firm set of
criteria by which to judge success and failure” (Burgess, 2006, p. 269). But to date, most
attention has gone to only one set of indicators, namely those grasping the sub-state challenge
(as I call it): the regionalist threat from the periphery. Popular indicators include the public
support for secessionism, the electoral strength of regionalist parties, and the intensity of sub-
state identities (e.g. Brancati, 2006, 2008, 2009; Cagiao y Conde & Gagnon, 2021; Guibernau,
2006; Kymlicka, 2001a). Scholar’s focus on sub-state tensions and pressures is also reflected
in numerous studies on the impact of decentralization on the survival chances of states or the
prevalence of violent ethnic conflicts and rebellions (e.g. Bakke & Wibbels, 2006; Bermeo,
2002; Brancati, 2006, 2009; Hale, 2004; McGarry & O'Leary, 2009; Niessen, Reuchamps,
Stjepanović, & Habra, 2020; Rothchild & Hartzell, 1999; Walsh, 2018).
While much attention has gone to the effects of self-rule provisions on self-rule demands, far
fewer studies have addressed their impact on the power-sharing challenge: the challenge of
different ethno-territorial segments governing together. In this respect, existing works are
scarce and often rather theoretical in nature. Insofar as empirical analyses exist, they typically
provide a rather superficial picture. Scholars often build on general historical overviews,
4
Note that the concept of federalism does not only refer to such autonomy provisions but rather to a particular
combination of autonomy and cooperation: “self rule plus shared rule” (Elazar, 1987, p. 12). However, most
works on the effects and paradox of federalism are primarily preoccupied with the effects of the former (self-
rule provisions; decentralization). Indeed, this is reflected in the very title of one of the most authoritative
works on the paradox thesis: The Paradox of Federalism: Does self-rule accommodate or exacerbate ethnic
divisions? (Erk & Anderson, 2010b) As Brancati (2006, p. 654) notes, a growing chorus of scholars is
replacing the term federalism with the term decentralization for various reasons, e.g. because they wish to
include countries in their analyses that have sub-state governments with decision-making powers but do not
(formally) describe themselves as federal.
32
assessing the paradox thesis by referring to the most eye-catching crises and major political
events. In this respect, situations like the 2010-2011 cabinet formation crisis in Belgium or the
2017 suspension of the Northern Ireland Assembly are often-noted (for an overview, see e.g.
Abts, Dmitriy, & Marc, 2012; M. Hooghe, 2012; Murtagh, 2021). Sometimes, more detailed
historical reconstructions are used to induce general lessons on the art of power-sharing (e.g.
McEvoy, 2015). Some quantitative indicators of power-sharing tensions exist too. However,
these mostly grasp how power-sharing executives start or end. For instance, many reflections
on Belgium refer to data on the frequency of early cabinet terminations and the duration of
cabinets and cabinet formations (Caluwaerts & Reuchamps, 2015; De Winter & Dumont, 2014,
2021; Deschouwer, 1994; Dumont & De Winter, 1999; Dumont, De Winter, & Dandoy, 2001;
Louwerse & Van Aelst, 2013). Accordingly, the study of segmental power-sharing seems to be
burdened by the blind spot that long characterized the literature on coalition politics: a strong
preoccupation with the birth and death phases and a lack of attention to “what occurs between
the beginning and the end of coalitions, on what we may call coalition governance.” (Müller,
Bergman, & Strøm, 2008, p. 9) This is problematic because in assessing the power-sharing
challenge, we should avoid being blind to the central phase in the life cycle of coalition politics
(Strøm, Müller, & Bergman, 2008).
5
Addressing this imbalance, recent analyses have shed
some light on the dynamics of coalition governance in power-sharing regimes. For instance,
Bahtić-Kunrath (2011) studied the adaption rates of draft laws (2006-2010) to conclude that
power-sharing elites pushed Bosnia’s consociational system to its extremes, while Wilford
(2010, pp. 149-151) described several examples of major reforms that were (not) agreed upon
by the power-sharing partners in Northern Ireland (e.g. the inability to agree on certain
educational reforms or a Bill of Rights).
Explorations like these are surely of relevance. However, to understand the day-to-day
cooperation between segmental leaders and the way it is impacted by decentralization, more is
needed. What lacks is a systematically reconstructed overview of tensions based on solid
quantitative indicators. Accordingly, our knowledge about the ethno-territorial clashes that
burden power-sharing is still rather limited. Many questions remain. How common are
these clashes? Is power-sharing an increasingly conflictual practice? Or is ethno-territorial
conflict in decline? Is joint decision-making a non-stop fight between irreconcilable segmental
‘blocks’ (e.g. Flemings vs. Francophones) or are other lines of alignment more common (e.g.
5
As elaborately discussed in the work of Strøm et al. (2008), this life cycle consists of four phases: formation,
governance, termination and elections.
33
liberals vs. social democrats)? How frequent are clashes within segments, and what are ethno-
territorial disputes usually about? Which issues are most conflict-prone and how are conflicts
commonly settled? In consensus, as the consociational logic prescribes, or in less
accommodative ways? Is conflict resolution increasingly difficult? How did tensions evolve
and, most importantly, what is the role of decentralization in explaining these dynamics? Did it
facilitate or burden inter-segmental power-sharing? In short, did it work? In the absence of solid
and fine-grained data, these questions are largely left subject to speculation and political
instrumentalization; cherry picking and unsubstantiated claims. As former Prime Minister Mark
Eyskens (a Flemish Christian democrat) often stated when talking about Belgium: ethno-
territorial tensions did not yield bloodshed but bullshit (Van de Voorde, 2003, p. 336).
Advancing upon the important work already done, this doctoral thesis aims to help us overcome
the phase of speculation (or at least add to this effort) and provide more evidence-based answers
to the questions raised above. Studying four decades of power-sharing tensions in twenty-one
Belgian cabinets (Appendix 1), I provide a systematic overview of over 1,000 coalition
conflicts, the topics and (ethno-territorial) nature of these tensions, their intensity, the actors
involved, the way they are handled and solved, and how this evolved throughout time and
against the background of decentralization. Using a combination of descriptive and explanatory
analyses (Poisson and logistic regression), I provide a generic assessment of the paradox thesis
that answers the two central research questions of this work: How did the ethno-territorial
conflict in Belgian power-sharing cabinets evolve (RQ1) and how was this affected by
decentralization? (RQ2)
Why is it important to answer these questions? For at least three reasons, shedding light on the
impact of decentralization on power-sharing conflicts is well worth the effort. (1) To begin
with, understanding these dynamics helps us make sense of the stability of divided states.
Inter-segmental cooperation is widely considered to be key to the functioning and survival of
these countries. As discussed throughout this work, this point is stressed most famously by
consociational theory (Lijphart, 1968a, 1969a, 1985b, 2002), which presents day-to-day elite
cooperation in power-sharing arrangements as the most pertinent recipe for pacification. (2)
Additionally, understanding power-sharing tensions helps us make sense of an important
trigger behind autonomist dynamics. Power-sharing tensions can nourish the sub-state
challenge because they are often presented as evidence of the inefficiency of the state, the
inability to govern a multinational polity together, and the way central policy gridlocks ‘hold
up’ regions wishing to advance (see also Chapter 2 and 5). Recent discourse analyses have
convincingly shown that regionalist parties across Western Europe strongly and deliberately
emphasize such elements of victimization and marginalization in an attempt to gain support for
their party and its ethno-territorial agenda (Abts, Dalle Mulle, & Laermans, 2019; Dalle Mulle,
2016, 2018; Dalle Mulle & Serrano, 2019). They are not alone in doing so. Many moderate
34
politicians have also raised autonomy demands by referring to power-sharing conflicts and in
Belgium, anti-consociational sentiments and the desire to bypass the very difficult art of
cooperation is a key driver behind the decentralization process (Popelier, 2021). (3)
Furthermore, and evidently related, studying the power-sharing challenge allows for a more
solid evaluation of decentralization reforms. The management of this power-sharing
challenge; the defusal of ethno-territorial clashes between cooperating politicians is one of the
key ambitions behind these reforms (in Belgium and other polities alike). Accordingly, to truly
evaluate whether they delivered on these ambitions, we must gain sight into the prevalence,
evolution, and driving factors of power-sharing conflicts. Understanding these tensions helps
us make sense of the successfulness of decentralization reforms.
2. Research focus
2.1. The Belgian case: a canary in the coal mine
In studying the paradox of federalism, this work focuses on four decades of cabinet
conflicts in Belgium (1979-2018). In evaluating the paradox thesis, the Belgian case is
particularly relevant. For reasons discussed elaboratively throughout this work, Belgium is
considered to be particularly prone to the paradox of federalism. Its peculiar institutional
architecture is a unique combination of features that are traditionally linked to ethno-territorial
tensions. It combines a bipolar, centrifugal, consociational, homogeneous federal system with
a list-PR system, and a split party system (De Winter, Swyngedouw, & Dumont, 2009;
Deschouwer, 2002, 2005, 2006; Elazar, 1987; Horowitz, 2000, 2002; Shair-Rosenfield et al.,
2021; Stepan, 1999; Swenden, 2006; Watts, 2008). Belgium is a theoretical powder keg. From
a research point of view, this conflict proneness has multiple assets. It not only increases
the chances of mining enough ethno-territorial conflicts (observations) for statistical analyses,
but it also characterizes Belgium throughout the whole period under investigation. The
institutional factors listed above are thus constants in the explanatory analyses, which adds to
their robustness. Furthermore, because of this cocktail of risk factors, Belgium can safely be
considered a least-likely case for pacification (Eckstein, 1975), or at least a very-unlikely case
for pacification. This susceptibility to the paradox of federalism increases the room for
(cautious) generalization beyond its borders. Of course, prudence is advised. But it would be
quite damaging to the paradox thesis if the results from Belgium fail to support it (Levy, 2008,
p. 12). If decentralization brought pacification in the flammable context of Belgium, it is likely
to be a stabilizing force in other (less conflict-prone) contexts too. In different words, and thanks
to its particular vulnerability to the toxic gas of ethno-territorial tensions, the small state of
Belgium can act like a canary in the coal mine. As with the latter, its relevance is far greater
35
than its size would suggest.
2.2. Cabinet conflicts
This work assesses the impact of decentralization on ethno-territorial power-sharing
tensions by studying cabinet conflicts. This focus reflects the key role of executives in power-
sharing regimes. Indeed, although this power-sharing can take many shapes (Lijphart, 1979,
p. 500), it typically refers to “meaningful cross-community executive power sharing” (O’Leary,
2005, pp. 12-13). Such cabinets are typically the beating hearts of everyday segmental power-
sharing. In Belgium, for instance, the federal cabinet is continuously engaged in high-level
decision-making between politicians of the two main segments (Flemings and Francophones),
the Council of Ministers is characterized by linguistic parity (French-Dutch), and when the
Belgian parties split along linguistic lines (between 1968 and 1978; cf. De Winter et al., 2009,
pp. 72-76), the federal coalition became the primary meeting place of segmental elites: “the
government level is where the two party systems join” (Deschouwer, 2002, p. 75). As such, the
Belgian cabinet is also the most important arena of ethno-territorial conflict management. Due
to the split of the party system, it is there that ethno-territorial tensions usually end up (as it is
no longer an option to settle them within parties):
“When before the split a linguistic issue might be solved by reaching a
compromise within parties, almost every linguistic issue that comes to the
surface now ends on the table of the government. It is simply impossible to
avoid these issues.” (Deschouwer, 1994, p. 46)
In focussing on clashes in power-sharing executives, this doctoral thesis also adds to the
emerging study of cabinet conflicts. For decades, and as noted above, the coalition politics
literature was preoccupied with the birth and death phases of coalitions (coalition formations
and terminations). Their actual lifetimes; the everyday functioning of these cabinets was largely
ignored (as noted by e.g. Müller et al., 2008; Müller & Strøm, 2000).
6
It is only recently that
6
In this respect, Müller and Strøm (2000, p. 13) talked about a Hollywood-like bias in their introduction to
the seminal Coalition Governments in Western Europe: “In many ways”, they noted, “the cabinet coalitions
literature resembles the romantic Hollywood films of the 1950s. Much is made of the courtship process and
‘who gets whom’, whereas relatively little light is shed on how such alliances actually work or their prospects
for long-term success.”
36
scholars have started to devote attention to these “more mundane aspects of actually ‘living
together’” (Müller et al., 2008, p. 30) (e.g. T. e. Bergman, Bäck, & Hellström, 2021). As a
result, the study of cabinet conflicts is still underdeveloped. A growing chorus of scholars has
started to investigate these tensions, but both conceptually, theoretically, and empirically, there
is still much room for improvement (Andeweg & Timmermans, 2008; Blondel & Müller-
Rommel, 1993; Marangoni & Vercesi, 2015; Moury & Timmermans, 2013; Nousiainen, 1993;
Timmermans & Moury, 2006). Existing (operational) definitions often fail to discern conflicts
from mere disagreements, coding choices are often unclear or suboptimal, and scholars often
fish with wide-meshed nets, only exposing a small set of the most major clashes
(Vandenberghe, 2022b). Building on and adding to these studies, this work presents a new
definition of cabinet conflicts and detailed coding instructions. Hand-coding 1,743 pages of
political data yearbooks and 9,547 news agency articles allowed me to systematically harvest
data on more than 1,000 cabinet conflicts of varying sizes – ranging from long-forgotten
quarrels between frustrated ministers to the regime-threatening gridlocks that became part of
our collective memory. The result is an unprecedented longitudinal view of more than a
thousand clashes in the twenty-one segmental power-sharing cabinets that governed Belgium
between the beginning of 1979 and the end of 2018.
2.3. Period under study (1979-2018)
The timeframe on which this work sheds light is the 1979-2018 period. This focus is
based on both pragmatical and theoretical considerations. From a pragmatical perspective, no
source allowed me to provide a similarly fine-grained, systematic view of conflicts in the pre-
1979 period. The data sources I use (political yearbooks in combination with digital news
archives) failed to cover earlier years, and explorative coding efforts of alternative sources (e.g.
printed newspapers) provided insufficient results (see Chapter 6).
From a theoretical perspective, however, 1979 is a true milestone. It is the first full year since
the split of the last traditional party in Belgium (De Winter et al., 2009). Hence, the analyses
cover a period in which the abovementioned cocktail of risk factors is constant. This is in line
with my ambition to study the paradox thesis in a very favourable setting. Meanwhile, the 1979-
2018 period covers five out of six decentralization reforms in Belgium, which provides
sufficient variation on the decentralization variable (L. Hooghe et al., 2016; Shair-Rosenfield
et al., 2021). Only the first state reform (1970) is left out of sight. The lack of data on the unitary
period (pre-1970) and this first decentralization package is unfortunate, but not problematic. In
assessing the impact of decentralization, the 1970 reform is arguably less relevant. It established
the predecessors of the now very powerful Communities, but it only provided these sub-states
with extremely limited powers (e.g. L. Hooghe, 2004). De facto, decentralization only really
37
gained track during the reforms that followed (1980, 1988, 1993, 2001, and 2011). These
decentralization reforms established the Flemish, Walloon (1980), and Brussels-Capital Region
(1988), introduced separate sub-state cabinets (1980), allowed for directly elected Flemish and
Walloon sub-state parliaments (1993, first elections in 1995), and provided the sub-states with
ever-increasing fiscal autonomy and substantial packages of competences. If decentralization
fuels tensions and the paradox thesis holds, it should be particularly notable in the post-1979
period.
3. Outline of the study
This work proceeds as follows. The second chapter of this introductory part discusses
the central concepts in this work (Chapter 2). Consequently, readers are provided with a solid
basis and background to dig into the rest of the study. A brief historical overview presents some
of the most pertinent dynamics in the ethno-territorial history of Belgium and sketches the
development of the power-sharing challenge since its creation (Chapter 3). Next, Chapter 4
exposes what we already know about the paradox of Belgian federalism by discussing existing
indicators. Interestingly, different indicators provide a different picture.
Part II forms the theoretical corps of this work. Chapter 5 presents the contrasting hypotheses
on the effect of decentralization (pacification vs. escalation) and the expected effects of two
sets of contextual factors: constants (the peculiarities of the Belgian case and its conflict
proneness) and the varying factors that are included in the explanatory analyses (regionalist
party strength, their cabinet participation, the electoral saliency of decentralization, and the
proximity of federal elections).
In Part III, the empirical corpus of the dissertation, I elaborate on the data, coding efforts, and
analyses used to answer the research questions (Chapter 6). The empirical analyses are
presented in four main components. First, I make sense of the challenge at stake by assessing
the general morphology and prevalence of ethno-territorial conflicts (Chapter 7). I show that,
while these clashes are a major burden to power-sharing dynamics, they became less disruptive
over time. Next, and answering the first research question (RQ1), I analyse the evolution of
conflict frequencies against the background of five decentralization reforms (Chapter 8).
Contradicting the paradox thesis, these analyses show that conflict levels generally decreased.
Third, and assessing whether this overall pacification tendency actually relates to
decentralization, Chapter 9 presents the results of the explanatory regression analyses.
Answering the second research question (RQ2), they clearly demonstrate that this is truly the
case. Pacification relates to decentralization. Finally, to strengthen my evaluation of the paradox
of federalism, conflict outcomes are studied (Chapter 10). It is sometimes hypothesized that
38
conflicts might be less frequent but harder to solve. As such, pacification might have come at
the expense of the functionality of the federal policy level. The results refute this expectation.
In Belgian cabinets, conflict resolution is not increasingly difficult but rather consistently (if
not comfortably) numb.
As such, this study disproves the paradox thesis from different angles. In the conclusion, I
reflect upon these rather surprising findings and their implications for a series of public and
scholarly debates. Stressing the limitations of this work, I also identify some fruitful avenues
for further research (Part IV, Chapter 11).
39
Chapter 2.
Concepts and definitions
7
Throughout this work, Belgium will be referred to as a ‘divided state’ that consists of different
‘segments’ and was transformed by five waves of ‘decentralization’ in an attempt to pacify
‘ethno-territorial’ ‘cabinet conflicts’. These concepts are widely used in both academic and
societal spheres. However, their omnipresence should not withhold us from clarifying their
meaning (on the contrary). Conceptual vagueness is never an asset, but given the contested
nature of some of these terms, terminological clarity is even more important. The aim of this
chapter is to provide exactly that.
1. Divided states
Of all the key concepts used throughout this dissertation, the ‘divided state’ notion is
arguably the most controversial. It is a heavily charged term, burdened by ideological
connotations and prone to political recuperation. After all, the degree to which a state is
considered to be divided is subject to fierce debates between regionalists and many of their
opponents. Nationalists on both sides have a strong tendency to stress the coherence of the
preferred nation(-state) and downplay the unity of the policy level they reject – either the centre
or the sub-state level (Breuilly, 1982; Guibernau i Berdún, 1996; Hobsbawm, 1990; Hroch,
2015). As references to ‘divided states’ often resonate in such discourses and debates, it seems
appropriate to start this section by clarifying what I don’t mean when referring to the
concept. I do not refer to the idea that there is no common societal ground at all, that no societal
divisions can be found within the sub-groups (segments), or that the existing divisions between
them are desirable, undesirable, historically predetermined, or demarcation lines between
‘nations’ (whatever those may be).
8
Nor should the ‘divided state’ concept be understood in
7
The definitions of ethno-territorial conflict and cabinet conflict presented here were previously discussed
in less elaborate terms in a research note in Politics of the Low Countries (Vandenberghe, 2022b) and in an
article in Ethnopolitics (Vandenberghe, 2022c).
8
Defining nations is no easy task and debates on their meaning yielded numerous definitions and
controversies (for an overview, see e.g. Hroch, 2015, Ch. 1). Reflecting on the state-of-the-art, Hroch (2015,
40
ideological terms, i.e. as a reference to the ideological divisions between different regions or
sub-groups such as the depiction of ‘right-wing Flanders vs. left-wing Wallonia’ or ‘progressive
Scotland vs. conservative England’ (Béland & Lecours, 2016; Billiet, Abts, & Swyngedouw,
2015).
9
With these disclaimers in mind, the following paragraphs present the definition of
divided states to which this work adheres and illustrates why the concept applies to the Belgian
case.
1.1. Divided states: definition and application
Of course, all societies or states are ‘divided’ to some extent. For evident reasons, the
idea(l) of a perfectly homogeneous state is a myth. Indeed, in every polity, the cultural and
biological sources of heterogeneity are manifold: race, sex, age, religion, language, cultural
traditions, ethical views, political ideology, etc. When do these inevitable forms of internal
diversity turn into what most scholars call a ‘divided society’ (or a ‘divided state’, as I will call
it for reasons discussed below)? Following Arend Lijphart (1977a, pp. 3-4), a key author within
the literature on these cases, they can be defined by two main characteristics: (1) segmental
cleavages and (2) a segmentation of society.
10
Both elements, as well as their applicability to
the Belgian case, are discussed throughout the following paragraphs.
First, divided states are divided by ‘segmental cleavages’, a term Lijphart borrowed from
Harry Eckstein (1966). In his study on Norway, Eckstein discerns between three types of
political divisions. Segmental cleavages are one of these, defined as the phenomenon in which
p. 3) assertively noted: “A closer examination of a number of randomly selected studies of the concepts of
nation and nationalism reveals that each author interprets these terms quite differently. This is accompanied
by efforts to be original, which have reached a global scale, with regard not only to definitions and
interpretations but also to the creation of new terminology.” These discussions are closely linked, and indeed
inherently intertwined with the ongoing debate on the origins of nations (as is also extensively demonstrated
by e.g. Ichijo & Uzelac, 2005).
9
For a general reflection on the left-right divisions between Flanders and Wallonia, see e.g. Billiet et al.
(2015); Bruno De Wever (2011).
10
A similar description can be found in his introduction to the 1981 work Conflict and Coexistence in
Belgium. The Dynamics of a Culturally Divided Society, in which he refers to divided societies as “societies
that are sharply divided along cultural, religious, or ethnic lines into virtual sub-societies with their own
political parties, interest groups, and media of communication” (Lijphart, 1981b, p. 3)
41
“political divisions follow very closely, and especially concern, lines of objective social
differentiation, especially those particularly salient in a society” (Eckstein, 1966, p. 34).
Examples include differences in race, region, sex, generation, and language. The presence of
deep, sharp or salient cleavages is an important element of most definitions of divided societies
(e.g. Belmont, Mainwaring, & Reynolds, 2002; McCulloch, 2014; Rabushka & Shepsle, 1972;
Reilly, 2001, 2002). In this respect, two remarks must be made.
(i) In the descriptions of some scholars, the divisions at stake are not only highly salient
or deep, but the dominant or primary political divisions. For instance, according to Rabushka
and Shepsle (1972, pp. 20-21), one of the defining characteristics of ethnically divided societies
(a sub-type), is “the practice of politics almost exclusively along ethnic lines”: most political
conflicts are perceived in ethnic terms, and ethnic communities act cohesively on almost all
political issues. Similarly, Donald L. Horowitz describes severely divided societies as societies
“in which ascriptive cleavages are highly salient in politics” and even “more salient than
alternative cleavages such as social class” (Horowitz, 2014, p. 7; emphasis added). In my view,
however, this interpretation is unnecessarily restrictive. It also comes with a range of
puzzles. For instance, how should this saliency be measured or assessed? Does it refer to a
state’s political history as a whole, or should different periods be discerned? The balance
between cleavages can change throughout time, and the relevance of existing cleavages can
decrease as new cleavages arise (e.g. Elchardus & Pelleriaux, 1998; Inglehart, 1977, 1997;
Lipset & Rokkan, 1967). Was Belgium not ‘divided’ before the ethno-territorial cleavage
became predominant throughout the 1960s (Deschouwer, 2002; 2012, p. 39; L. Hooghe, 2004;
Huyse, 1981)? Were many of the divided polities studied so extensively today not ‘divided’
before this period, when the revival of centre-periphery tensions and the related growth of
regionalist movements swept states across the world (Guibernau, 2001)? And, in more narrow
terms, are societies or states no longer ‘divided’ when socio-economic cleavages rise to the top
of the agenda for some years, e.g. because of an economic crisis? Was Belgium no longer a
divided country between October 2014 and December 2018, when the Michel I cabinet agreed
to avoid ethno-territorial debates and focus on socio-economic reform in word and deed (Benoît
Rihoux, Vandeleene, De Winter, Baudewyns, & Deruette, 2016, p. 30)?
11
Rather than trying to
come up with arguments to answer such troublesome questions positively, I refute the idea that
lies at their basis. In my perspective, the saliency of the divisions of interest need not be greater
11
Michel I was a coalition of the Flemish nationalist N-VA, the Flemish liberals (Open VLD), the Flemish
Christian democrats (CD&V), and the Francophone liberals (MR).
42
than that of other divisions for a state to be called a ‘divided state’. It suffices that segmental
cleavages exist and that the divisions at hand translate into a segmentation of society (Lijphart,
1977a, pp. 3-4).
(ii) It is important to note that segmental cleavages do not refer to the mere existence of social
divisions, but to forms of differentiation that result in political competition and conflict,
whereby the conflict “not only pit distinct segments against one another but especially concern
segmental conflicts.” (Eckstein, 1966, p. 34; emphasis in original) Applying this criterion to the
ethno-territorial division and its historical role in Belgium (Witte, 1993b; Witte, Craeybeckx,
& Meynen, 2009; Witte & Van Velthoven, 2010), there is a language-based social division that
resulted in forms of political conflict and competition in which the Flemish and Francophone
segments were repeatedly pitted against one another (often concerning ethno-territorial issues).
The second defining element in Lijphart’s (1977a, p. 4) definition of divided societies is
that political parties, media, interest groups, voluntary associations and schools within
them tend to follow the lines of the segmental cleavages. Put differently, the divisions must
be reflected at the level of societal organizations (see also e.g. Rabushka & Shepsle, 1972, p.
21). As Ian Lustick (1979, p. 325) formulates it, the “ascriptive ties [must] generate an
antagonistic segmentation of society”.
12
Put rather bluntly, the idea is that there is a set of
“virtually separate subsocieties” which can be labelled segments (e.g. Lijphart, 1977a; Lijphart,
1981b; 2008, p. 67).
In Belgium, such a segmentation of society is clearly visible at different levels. At the political
level, the split party system is a first and major case in point (De Winter et al., 2009, pp. 72-76;
Verleden, 2009). In Belgium, parties compete in separate party systems divided along language
lines. All parties are thus non-state-wide parties, that is “parties which participate in elections
in only part of the state in contrast to state-wide parties” (De Winter, Gomez, & Lynch, 2018,
p. 142). Flemish parties choose not to contest elections in Francophone Belgium, and vice
12
Lustick (1979, p. 325) used the term deeply divided societies as a synonym for plural societies, communally
divided societies, etc. According to his definition, societies are deeply divided “if ascriptive ties generate an
antagonistic segmentation of society, based on terminal identities with high political salience, sustained over
a substantial period of time and a wide variety of issues. As a minimum condition, boundaries between rival
groups must be sharp enough so that membership is clear and, with few exceptions, unchangeable.”
43
versa.
13
Accordingly, voters in Francophone Belgium can only vote for Francophone parties,
while voters in Flanders can only vote for Dutch-speaking parties (with minor exceptions).
14
Cross-segmental interaction between these parties appears to be rather limited. Even parties that
belong to the same ideological party family (e.g. the Flemish and Francophone liberal parties)
only have weak organizational ties, which strongly eroded over time (Dandoy & De Decker,
2009; De Winter, 2006; Deschouwer, Pilet, & van Haute, 2017).
15
From a less formal but very
telling perspective, the language divide is even notable on easily accessible social media
platforms with low thresholds for communication. For instance, recent analyses show that
Flemish and Francophone candidates almost exclusively interact within their own segmental
‘bubbles’ on Twitter (Coffé, Cioroianu, & Vandenberghe, 2019). Furthermore, and as explained
more thoroughly in Chapter 5, the political system at the federal level entrenches the political
segmentation between Dutch- and French-speaking politicians: MPs are divided into two
13
German-speaking parties exist too, but these are mostly ‘local’ branches of the Francophone parties
(Dandoy & Joly, 2018, fn. 5). Also, these parties do not play any role in central politics. More generally, the
German-speaking minority is not really involved in the historical ethno-territorial struggle (which is rather a
bipolar struggle between the Flemish and Francophone sides). Accordingly, and perfectly grasping this
dynamic, Asbrock and Van Hiel (2017) talk about this minority as an ‘insider with an outsider’s perspective’.
14
There are several exceptions to this logic. First, in the Brussels-Capital Region (and six surrounding
municipalities with so-called ‘language facilities’), Dutch- and French-speaking parties compete for the votes
of both Dutch- and French-speaking voters. Second, there is one state-wide party: the Marxist PVDA/PTB
(Partij van de Arbeid van België/Parti du Travail de Belgique). Being the only remaining statewide party of
significance, it strongly emphasizes the importance of cross-segmental cooperation. For an overview of its
origins, evolution, and organizational model, see e.g. Delwit (2021a, 2022). Third, some parties sometimes
present lists in other language areas. The most notable example is arguably that of the UF (Union des
Francophones), a collaboration of the major Francophone parties that presents lists in Flemish Brabant to
represent Francophones and defend or strengthen their position (Lanneau, 2015). Mostly in reaction to the
UF’s presence in Flanders, the radical right and separatist VB (Vlaams Blok/Vlaams Belang) chooses to
present candidates in electoral districts across Wallonia (e.g. Thijs, 2019).
15
For a detailed account of the relation between the Flemish (Open VLD, Open Vlaamse Liberalen en
Democraten) and Francophone (MR, Mouvement Réformateur) parties, see Biard (2022). Note that, when it
comes to cross-segmental ties, the Flemish and Francophone ecologist parties are an important exception:
“alors que les deux partis écologistes sont nés indépendamment l’un de l’autre, de chaque côté de la frontière
linguistique, ils ont très tôt développé des relations privilégiées, bien plus significatives que les partis
traditionnels.” (Dandoy & De Decker, 2009, p. 26).
44
language groups in the parliament, linguistic parity characterizes the federal Council of
Ministers, and both sides obtained far-reaching veto-rights by the establishment of special
majority requirements and an alarm bell procedure that can put the legislative process on hold
(Deschouwer, 2012; Popelier, 2021). Some authors have dubbed the resulting situation as one
of a “permanent diplomatic conference” between the representatives of both sides
(Deschouwer, 2013a, p. 217). Such “international politics” metaphors are not new. Similar
parallels have long been used to describe consociational democracies like Belgium, where
interactions between different segments are mostly limited to the negotiations between their
leaders (just like the contacts between states on the international scene) (Lijphart, 1968c).
16
A segmental divide also cuts across the media landscape(s) in Belgium. Separate Dutch and
French radio stations have existed ever since the emergence of the medium in the early 1920s,
and the Belgian public broadcaster (eventually) aired distinct Dutch and French television
programs before splitting along linguistic lines altogether (Saeys, 2007). The resulting Flemish
and Francophone broadcasters obtained a monopoly in their respective language community
and the programs they showed were specifically aimed at either the Dutch-speaking or French-
speaking public (Van Gorp & Sinardet, 2017, p. 238). Commercial radio and television stations
are monolingual too, just like the newspapers. In terms of media consumption, citizens typically
stick with the media of their own region (e.g. Dewachter, 1996, pp. 134-135; Thijssen, Sinardet,
& Dandoy, 2015). For instance, the 2014 PARTIREP data indicate that 66.0% of the Flemish
respondents and 79.6% of the Walloon respondents never opt for Francophone or Dutch media,
respectively (or do so less than once a year) (Thijssen et al., 2015).
17
As a result, they are rarely
confronted with politicians or political news from the ‘other side’. After all, ‘Belgian’ sections
of the news coverage on both sides typically focus on intra-segmental events and in information
programs, the number of ‘guests’ from the other segment is remarkably low (Dewachter, 1996,
16
A similar logic can be found in the work of Verba (1965, p. 470). At one point, he discussed “a political
system made up of two closed camps with no overlapping of membership. The only channels of
communication between the two camps would be at the highest level - say when the leaders of the two camps
meet in the governing chambers - and all conflict would have to be resolved at this highest level. Politics
comes to resemble negotiations between rival states; and war or a breakdown of negotiations is always
possible.”
17
For a similar analysis, see e.g. Dewachter (1996, pp. 134-135).
45
pp. 134-135; Sinardet, 2007, 2012).
18
Unsurprisingly then, most citizens are far less informed
on the politics across the language border than on the situation ‘on their side’ (Dewachter, 1996,
p. 136). In sum then, Flemings and Francophones can be described as the kind of ‘communities
of communication’ that the American sociologist Karl W. Deutsch (1953) calls nations: groups
that can communicate more easily, intensively, and in a more complex way within their group
than with members of another language group. I do not wish (nor need) to enter the contentious
debate on ‘nations’ and their meaning, but the fact that this term is used in this respect highlights
the relevance and pertinence of the divided communication networks.
19
On a larger societal level, similar divisions can be observed. In several instances, authors have
talked about a “social duality”, referred to the existence of “two separate societies” (Billiet,
Maddens, & Frognier, 2006, pp. 914-915), or went as far as stating that the other community is
almost a foreign country (“L’autre communauté du pays est quasiment l’étranger"; Dewachter,
1996, p. 136). The language divide not only characterizes the operational and organizational
scope of many civil society organizations (with many organizations limiting their action range
to one of the segments). It also marks professional and individual relations. For instance, the
level of intersegmental marriages is strikingly low (below 10%; Algemene Directie Statistiek -
Statistics Belgium, n.d.-a), there is little overlap in the labour market, with only around 2% of
the people working in Flanders or Wallonia living in the other region (Algemene Directie
Statistiek - Statistics Belgium, n.d.-b), and rather few citizens cross the language border to go
shopping, to visit friends from the other language group or to enjoy their spare-time (Thijssen
18
To give just one example: based on an analysis of news programmes, weekly political debate programmes
(2003-2004) and electoral debate shows during the 2007 federal election campaign, Sinardet (2012) noted
that television broadcasters show a strong tendency to fixate on politicians of their ‘own’ community. The
coverage of Dutch-speaking ministers on the Flemish side (as a percentage of the total speaking time of the
cabinet) was 80.1% on VRT (public broadcaster) and 81.9% on VTM (commercial broadcaster). The
coverage of French-speaking ministers on the Francophone side was 70.7% on RTBF (public broadcaster)
and 81.2% on RTL-TVI (commercial broadcaster) (Sinardet, 2012, p. 184).
19
(Belgian) historians like Lode Wils (2020, p. 28) have explicitly referred to Deutsch when discussing the
(consequences of this) ‘systematic dismantling of the contacts and the facilities for communication between
Dutch and Francophone citizens’.
46
et al., 2015).
20
,
21
,
22
Evidently, in analyzing statistics on personal relations and individual
behaviour, one might question the relevance of a ‘segmental’ frame and point towards the
importance of elements like proximity. But given Belgium’s small scale, these statistics are
certainly not without meaning. At the very least, it seems safe to state that Belgium adheres to
the second criterion for divided states (Lijphart, 1977a, pp. 3-4): the segmentation of society.
The different language groups “live side by side in isolation” (Popelier, 2021, p. 104).
“Modern Belgian society is extremely segmented. The two language groups
have their own society, their own parties and party system, their own political
system, their own communication system, their own education system.
Members of the segments can obtain all they want without leaving their
subculture. This looks very much like classical pillarization, but then on a
territorial basis.” (Deschouwer, 2002, p. 81)
Interestingly, this divide also appears to mark the way citizens ‘imagine’ Belgium. To give
just one telling example, Reuchamps, Kavadias, and Deschouwer (2014) asked 5269 first-year
higher education students in the human and social sciences (from across the country) to ‘draw
Belgium’ on a blank sheet of paper. While no instruction of this kind was given, a majority of
students (55.8%) spontaneously drew a horizontal line across their map, thus reflecting the
language border between North and South. Belgium, it seems, not only pertains to the definition
of divided states. It is also imagined as one.
20
In 2018, for instance, 93.1% of the ‘first spouses’ and 96.2% of the ‘second spouses’ that resided in
Flanders before the marriage, married to a partner that resided in Flanders too. On the Walloon side, the
numbers are 90.37% and 93%, respectively (Algemene Directie Statistiek - Statistics Belgium, n.d.-a).
21
More specifically, in 2018, 1.6% of the people working in Flanders lives in Wallonia, and only 2.3% of
the working people in Wallonia live in Flanders (Algemene Directie Statistiek - Statistics Belgium, n.d.-b).
22
Using the 2014 PARTIREP data, the authors noted that a minority of the Walloon respondents visit Flanders
more than once a year to enjoy their spare time or vacation (48.2%), to visit Dutch friends (19.1%) or to go
shopping (26.1%). Regarding the visits of Flemish respondents to Wallonia, the numbers are respectively
41.2%, 17.3% and 16.8% (Thijssen et al., 2015, p. 137).
47
1.2. Terminology
Before ending this section on the divided state concept, a final note on the terminology
at hand is warranted. The literature is scattered with a wide range of concepts, sometimes
referring to the same phenomenon and sometimes used to distinguish phenomena. One example
of this is the distinction between divided and deeply divided societies.
23
As noted by Guelke
(2012, p. 9), most scholars restrict the use of the latter term to states facing violent conflicts or,
at the very least, the threat of such conflicts: e.g. Northern Ireland, Cyprus, Iraq, South Africa,
Ethiopia, Sri Lanka, Bosnia-Herzegovina, and the collapsed states of Yugoslavia and
Czechoslovakia. Notwithstanding the relevance of this distinction (to which I will adhere
throughout this work, except in citations), both concepts are often used interchangeably, too –
even within single works.
Also, authors have used a wide variety of alternative terms to describe the polities at hand.
A brief look at the work of Lijphart himself suffices to demonstrate this point. In his early
reflections on the writings of Almond (1956) (cf. Chapter 5), he adopted the latter’s terminology
by discussing states with ‘fragmented political cultures’ (Lijphart, 1968b, 1969a). These were
described as states characterized by ‘clear and distinct subcultures’ (Lijphart, 1968c, p. 201);
instances in which “a society is divided by sharp cleavages with no or very few overlapping
memberships and loyalties” (Lijphart, 1969a, p. 208). In his seminal study of Dutch politics, he
used the Dutch concept verzuild (pillarized) as a translation of ‘fragmented’, and described
verzuildheid (the situation of pillarization) as a situation of ‘strong divisions at the basis’
(Lijphart, 1968c, pp. 11, 201). Elsewhere, and particularly in his later work, the author opted
for the concept of ‘plural societies’ (e.g. Lijphart, 1968a, 1977a; Lijphart, 1980, 1981a, 1989),
of which the most satisfactory description is the one we used above to define ‘divided societies’
(Lijphart, 1977a, pp. 3-4).
24
Notwithstanding the manifold alternatives, the latter term – divided societies – is arguably
among the most popular concepts within the literature. It has eventually surfaced in Lijphart’s
studies (e.g. Lijphart, 1981b, 1985a, 2002, 2004, 2008), and can also be found in several other
23
Other terms, such as severely divided societies, are used as well (Horowitz, 1993, 2014).
24
In other instances, Lijphart (1968a) sketched ‘plural societies’ more briefly as “societies with clearly
discernible racial, linguistic, and religious differences”.
48
authoritative works within the field (e.g. Horowitz, 1993, 2000; Lustick, 1979; Nordlinger,
1972; Reilly, 2001). While this concept seems to prevail within the literature, questions can be
raised regarding the use of the term society. Arguably, in states like these, it is not so much
society that is divided, but rather the state itself. These states are not home to a divided
society, but rather to different societies as such. Given the societal segmentation, they not
only lack many of the typical objective grounds that characterize a society, but often lack its
subjective characteristic too (the shared identity or sense of belonging together): “In many of
those cases referred to as divided societies, the groups in conflict reject precisely a status as
members of the same societies.” (McCulloch, 2009, p. 27) Accordingly, and following
McCulloch (2009, p. 28), I find it more accurate to use terms like ‘divided polities’ or ‘divided
places’.
25
Given the specific focus of this work on one state (Belgium), I will stick with the
‘divided state’ notion.
2. Ethno-territorial conflict
To conceptualize ethno-territorial conflicts, we build on the first defining element
of divided states: segmental cleavage(s). As noted above, Eckstein (1966, p. 34) defined these
as situations in which “political divisions follow very closely, and especially concern, lines of
objective social differentiation, especially those particularly salient in a society”. Another
defining element of segmental cleavages, and in fact the most relevant one at this point, is the
fact that the divisions at hand must result in political competition and conflicts, which “not only
pit distinct segments against one another but especially concern segmental conflicts.” (Eckstein,
1966, p. 34) In this respect, Eckstein (1966, pp. 34-35) presents the example of the male-female
divide that features every society and state. Even when this division correlates with political
behaviour or preferences (e.g. when women vote differently than men), it only becomes a
‘segmental cleavage’ if two additional criteria are met: (a) both sides must be involved in
political conflict, and (b) at least in some way, the conflict at hand must involve the sex roles
within society.
26
In other words, conflicts must be segmental (a) in terms of the factions that
25
According to McCulloch (2009, p. 28), a divided polity refers to a unit of governance (a state or a region),
while a divided place refers to a “piece of territory” that can be larger or smaller than a unit of governance.
26
Consecutive studies have indeed confirmed the existence of a so-called “gender gap” in electoral
behaviour, with women being generally more likely to favour left-wing parties than men (at least in recent
decades) (Abendschön & Steinmetz, 2014; Giger, 2009; Inglehart & Norris, 2000; 2003, Ch. 4; Studlar,
49
oppose each other, and (b) in terms of the issues at stake.
Translating this logic to the ethno-territorial divide, the conflicts that turn this division into a
segmental cleavage, must (a) pit ethno-territorial segments against one another (e.g. Flemings
vs. Francophones in Belgium, Catholics/Irish nationalists vs. Protestants/unionists in Northern
Ireland) and (b) especially concern ethno-territorial issues, that is issues concerning the roles
and positions of the ethno-territorial segments (e.g. the issues of sub-state autonomy, language
policy, or the political representation of segments). To a large extent, these issues coincide with
those pertaining to the traditional centre-periphery cleavage (Lipset & Rokkan, 1967).
27
Note that Eckstein considers both criteria to be necessary conditions. In my view, however, (a)
all conflicts along segmental lines are ethno-territorial conflicts (whatever the issue at stake),
just like (b) all conflicts on ethno-territorial issues (whatever the opposing sides). (a) For
instance, when socio-economic or foreign policy debates turn into a tense stand-off between
two segments, the conflict’s ethno-territorial nature is evident. In the Belgian context, the thorny
issue of the tobacco advertisement ban (and related, the financial future of the Francorchamps
race circuit) provides us with a good historical example. This ban was supported by the Flemish
coalition parties but fiercely opposed by the Francophones, as they feared that the racing circuit
of Francorchamps (situated in Wallonia) wouldn’t survive without the sponsorship of the
tobacco sector. The resulting debates triggered major political clashes between the Flemish and
Francophone coalition partners in the late 1990s (Deweerdt, 1998a, 1999). (b) Similarly,
conflicts on ethno-territorial issues like decentralization or sub-state representation are still
ethno-territorial clashes when they are fought between politicians of the same segment. Hence,
I adhere to a broader definition, according to which the abovementioned elements are sufficient
rather than necessary conditions. Accordingly, ethno-territorial conflicts are defined as all
conflicts in which segments are pitted against one another and/or conflicts concerning
ethno-territorial issues. Drawing on these two dimensions, I propose the following typology
of ethno-territorial conflicts (Figure 1).
McAllister, & Hayes, 1998).
27
The centre-periphery cleavage, a product of the emergence of the national state and liberal democracy,
refers to “the conflict between the central nation-building culture and the increasing resistance of the
ethnically, linguistically, or religiously distinct subject populations in the provinces and the peripheries”
(Lipset & Rokkan, 1967, p. 14; emphasis in original).
50
Why is it important to distinguish both elements? In contemporary public and political
debates, there appears to be a strong tendency to depict all kinds of ethno-territorial conflicts as
clashes ‘between segments’, that is ‘between Flemings and Francophones’. Even when this is
not the case. Such dynamics seem to be nourished by the tendency of journalists to extrapolate
characteristics from part of the group (e.g. a statement or position of one politician or party) to
Flanders and Francophone Belgium as a whole and to present both communities as single actors
(homogenization) stuck in a conflictual relationship (conflictualization) (Sinardet, 2008).
28
A
typology like the one proposed here allows for a more nuanced assessment of the troubles at
hand. It exposes whether the ethno-territorial challenge is actually a problem of cooperation
between leaders of different segments or rather a problem of troublesome topics. Put differently,
is segmental power-sharing itself difficult or are the issues that are inherent to governing
divided polities simply very explosive (e.g. the regulation of languages or the optimal set of
competences of sub-states)? As such, discerning types of ethno-territorial conflicts helps us
better understand the nature of the ethno-territorial fire. It opens the door towards a more refined
problem analysis. In its turn, this improved knowledge allows for more targeted responses to
28
For example, in Flemish current affair programs, ‘the terms ‘Di Rupo’, ‘PS’, and ‘Wallonia’ have become
synonyms’, and Francophone broadcasts show a similar tendency to extend the stances of certain parties to
la Flandre and les Flamands as a whole (Sinardet, 2008, p. 66).
Figure 1. Typology of ethno-territorial conflicts (ETCs) by sides and issues
Segmental side(s)
Other issue(s)
factionalized
ETCs
full-fledged
ETCs
Ethno-territorial
issue(s)
non-
ETCs
thematic
ETCs
Non-segmental sides
51
dampen the flames.
As a final note, it should be stressed that ethno-territorial conflicts and political conflicts more
generally can be fought between different actors (citizens, their representatives, other
politicians, etc.) and can take many forms, ranging from short verbal strikes or destabilizing
political gridlocks to far-reaching acts of violence and rebellion. Illustrating its breath, the genus
of ethno-territorial clashes includes both the peaceful and very technical Brussels-Halle-
Vilvoorde debates that plagued Belgian politics for years (Blero, 2015; Devos & Bouteca, 2008;
Mosselmans, 2009) and the troubles in Northern Ireland, which led to the death of about 3700
men, women, and children (English, 2003, pp. 378-381; McKittrick, Kelters, Feeney, Thornton,
& McVea, 1999; McKittrick & McVea, 2012, pp. 374-375). As McGarry and O’Leary (1993a,
p. 2) describe in the introductory chapter of The politics of ethnic conflict regulation:
“Some parts of the world appear to be trapped in deadlock ethnic wars, where
no faction is sufficiently powerful completely to control or crush its
opponent(s) […]. In happier zones ethnic communities have been able to
negotiate agreed changes to their political systems, notably in Belgium, Canada
and Switzerland. Yet other sites of ethnic conflict live in a twilight world
between deadlocked war and permanent negotiation.”
Given the diversity of ethno-territorial conflicts, it is warranted to explicitly delineate the
specific type(s) of conflict under study. The focus of this work is on ethno-territorial conflict in
the political arena. More specifically, and for reasons discussed in Chapter 1, to make sense of
ethno-territorial conflict at the power-sharing level, I assess cabinet conflicts (in Belgium). The
definition of such cabinet conflicts is addressed in the following section.
3. Cabinet conflict
Problems are a cabinet’s gasoline, and the process of problem-solving is what keeps it
rolling. In the words of Nousiainen (1993, p. 263 ; 259), the agenda of cabinets “is based on
problems that have become politicized”, and governmental decision-making can be seen as “a
problem-solving activity on a broad societal scale.” It is only logical that this continuous search
for solutions and agreements often yields tensions. Empirical analyses of such cabinet conflicts
are on the rise (Andeweg & Timmermans, 2008; Blondel & Müller-Rommel, 1993; Marangoni
& Vercesi, 2015; Moury & Timmermans, 2013; Nousiainen, 1993; Timmermans & Moury,
2006). In defining these conflicts, the conceptualizations they provided are a solid starting point.
Cabinet conflict has been defined before in very broad terms as “any situation in which cabinet
members differ in opinion, preference, interest or activity” (Blondel & Müller-Rommel, 1993,
52
p. 316; Appendix II). Casting the conceptual net less widely, Marangoni and Vercesi (2015, p.
21) define conflict in more fine-grained terms as “any quarrel or explicit disagreement between
two or more executive members and/or coalition (individual or collective) party actors.” Two
criteria can be distilled from these definitions. First, there must be a disagreement of some
kind. Obviously, there can be no conflict without disagreement and for evident reasons,
conflicts can only be mapped and studied when they are explicit (i.e. publicly visible). Second,
the dispute must be between coalition partners, that is two or more executive members
and/or coalition (individual or collective) party actors. Concerning these criteria, four important
remarks must be made.
First, disagreement (the first criterion) need not concern the content of the government
policy itself, and it need not imply that actors have incompatible preferences. Next to so-
called policy conflicts, cabinets can also face office conflicts (conflicts on ‘who gets what and
who gets how much’) and procedural conflicts (meta-conflicts about the intra-coalitional
mechanisms to cope with conflicts) (Vercesi, 2016, p. 180).
29
An example of the former is a
clash about a cabinet reshuffle and the (re-)allocation of portfolios. A procedural conflict is e.g.
a conflict about the arena in which state reform disputes should be settled (private cabinet
meetings, parliament, etc.). Using less self-evident terms, Marangoni and Vercesi (2015, p. 25)
made a similar classification of conflict issues into what they call “three macro-categories”:
policy issues, structure of cabinet, and coalition equilibria. So, policy conflicts are only one
type of cabinet conflict. Similarly, disagreement need not imply incompatible preferences. As
Luebbert (1984, p. 241; 1986, pp. 62-63) noted in the context of coalition formations, parties
can have convergent preferences (a focus on the same issues with compatible positions),
tangential preferences (a focus on different issues that is not incompatible), or divergent
preferences (a focus on the same issues with incompatible positions – or a focus on different
issues with mutually exclusive preferences). Divergent preferences are typically considered to
be the riskiest and most lethal for cabinets, but tangential preferences can also give rise to
serious conflicts about the cabinet’s priorities and, related, about the allocation of the scarce
resources at hand (Andeweg & Timmermans, 2008, p. 276). So, conflict is inherently
29
Similarly, Andeweg and Timmermans (2008, p. 270) refer to meta-conflicts as conflicts over (one of) the
choices that structure any process of conflict management. In their view, these processes involve two basic
choices: the choice of the arena for conflict resolution (“who are recognized as participants in the resolution
of the conflict?”) and the choice of procedures – most notably the decision rules used to settle the conflict
(“what are the mechanisms by which the conflict should be resolved?”).
53
characterized by disagreement, but its subject need not be the actual policy of government and
it need not imply that actors have incompatible preferences.
Second, concerning the second criterion, it is important to stress explicitly that cabinet
conflicts can include clashes between collective or individual party actors (of governing
parties). This formulation prevents me from blindly excluding situations in which parties
engage in conflict through individual party actors. However, not all individual actors should
be automatically seen as agents representing their party. Scholars have long warned us against
treating parties as unitary actors (Daalder, 1983; Katz & Mair, 1993; Laver & Schofield, 1990).
Depending on the situation, politicians and even entire party factions can deviate from the
general party line. Real-life parties “are not monolithic structures” (Bolleyer, 2012, p. 316). For
instance, Katz and Mair (1993) have famously discerned three faces of parties (the party on the
ground, in central office, and in public office) and many modern parties have even been shown
to operate like franchise organizations such as McDonald’s or Subway (Carty, 2004). What
matters here is that a clash is only considered to be a cabinet conflict when the actors involved
are either cabinet members or actors that represent their party at large. Scholars can decide
whether this criterion is met based on an assessment of the particular context at hand. In
partitocratic Belgium, for instance, deviant behaviour is less likely due to the extensive power
of the parties and the way they streamline political communication (Baudewyns, Brans,
Reuchamps, Rihoux, & Van Ingelgom, 2022; De Winter & Dumont, 2006). Note that this is a
relevance criterion that is introduced with the aim of grasping those conflicts that are generally
perceived as cabinet conflicts by the public and politicians alike.
To prevent purely intra-party conflicts from distorting the analyses (which focus on
interactions between coalition partners), conflicts between party actors are only cabinet
conflicts when the actors involved belong to different parties (Figure 2). For instance, a clash
Figure 2. Visualisation of possible cabinet conflicts
Note: Conflicts between party actors are only considered to be cabinet conflicts
when the actors belong to different parties
Cabinet members Coalition party actors
Minister Prime Minister Party PPG leader
…
Secretary of state Minister Party chair(wo)man …
54
between the leaders of the parliamentary party groups (PPGs) of two coalition parties is a
cabinet conflict. But when the party chair(wo)man clashes with the leader of her parliamentary
party group (PPG), this is considered to be a purely internal party affair rather than a coalition
conflict.
30
Third, clashes between executive members and/or coalition (individual or collective) party
actors can occur in a wide range of arenas. Of course, tensions can arise during cabinet
meetings, but the conflict population under study also includes e.g. spectacular attacks in
parliamentary sessions (plenary sessions or committees), broadcasted disputes between cabinet
members during televised debates, and frustrated rants in blog posts or newspaper interviews.
It is not the location that turns a conflict into a cabinet conflict, but the actors involved. When
the Prime Minister starts yelling at a clumsy IT staff member during a cabinet meeting, or when
two cleaning ladies start fighting in the meeting room of the Council of Ministers, these are
evidently not cabinet conflicts.
Fourth, the two abovementioned criteria (disagreement between certain actors) are too
inclusive. They fail to distinguish disagreements from disputes. The former is inherent to
(coalition) governance. As Jean-Luc Dehaene (Deputy Prime Minister at the time) once phrased
it: ‘it is normal that coalition partners hold different opinions on certain dossiers, otherwise, we
would all be part of the same party’ (JVD, 1989). But not all differences turn ugly. Defining
conflicts as disagreements automatically puts every political negotiation (however constructive
and respectful) in the conflict basket. To avoid this distortion, we must recognize that conflict
does not refer to disagreement as such but rather to the way it is handled (i.e. to the way
politicians disagree). Accordingly, to discern the manifold quarrels and discussions that are
inherent to coalition governance from actual cabinet conflicts, a third criterion is needed:
disagreement must be antagonistic or hostile in one way or another. Put differently, and
echoing consociational terminology (Lijphart, 1968a, 1969a, 1977a, 2002), the pathway of
accommodative politics must be left. As discussed in Chapter 6, which covers the
operationalization of this criterion, this shift from constructive to confrontational behaviour can
take many shapes. Several indicators are used to grasp it (forms of swearing, blocking,
threatening, etc.).
30
For an extensive theoretical and empirical analysis of the role (tensions) of PPG leaders in Belgium and
their relation with the party chair(wo)men, see de Vet (2019).
55
In summary, then, cabinet conflicts are defined as any explicit and antagonistic
disagreement between two or more cabinet members and/or relevant coalition party
actors.
4. Decentralization
Decentralization is used throughout this work to refer to the “transfer of political
authority from the central government to the regional/substate level” (Keil & Anderson,
2018, p. 94). Such reforms come in many shapes. The minimal definition that is embraced here
refers to what Keil and Anderson call their “central feature”.
31
As such, the term is used as a
synonym for self-rule or autonomy provisions, two common concepts within the field. The
decentralization concept and (different views on) its conceptualization has been thoroughly
discussed in a wide range of works (e.g. Elazar, 1976; L. Hooghe, Marks, & Schakel, 2010;
Keil & Anderson, 2018; Rodden, 2004; Schneider, 2003; Treisman, 2002, 2007). Although
engaging in all conceptual controversies would lead me too far, some points do deserve some
attention.
Decentralization should be conceptually distinguished from “non-centralization”: the
freedom of sub-states from control by other government levels, whereby “a central government
is not able to control or revoke the jurisdiction of the constituent units” (Watts, 2013, p. 31).
Put differently: in non-centralized political systems, power is diffused in such a way that it
cannot legitimately be revoked or re-centralized (Elazar, 1976, 1987). This feature is
traditionally used to define a federal system (and distinguish it from unitary and confederal
systems). The former refers to “a political organization in which the activities of government
are divided between regional governments and a central government in such a way that each
kind of government has some activities on which it makes final decisions.” (Riker, 1975, p.
31
The authors opt for a wider definition themselves. Including additional criteria, they “define
decentralization as a distinct arrangement that offers autonomy to a specific group in a designated territory
in order to legislate on those affairs that affect them directly without providing them with full legislative or
territorial independence.” (Keil & Anderson, 2018, p. 89) Elsewhere, they “refer to decentralization as
arrangements which involve the transfer of political power (administrative, fiscal and/or legislative
competences) to a minority group (or multiple groups) residing within a particularly defined geographical
area.” (Keil & Anderson, 2018, p. 91)
56
101).
32
Some scholars use decentralization to refer to such federal systems (e.g. Brancati, 2006).
Following Daniel J. Elazar (1976, p. 13), I concur with this approach: non-centralization is not
the same as decentralization. However, I do not go as far as Elazar, who argues that both
phenomena are mutually exclusive. Adhering to the idea of a minimal definition, I do not
consider the absence of non-centralization as a defining feature of decentralization. As the
Belgian case shows, authority transfers from the centre to the sub-states (the most basic
description of decentralization) can be done in a non-centralized system too.
Many authors adhere to more detailed definitions, adding additional criteria such as the idea
that autonomy should be provided to minority groups in designated territories (e.g. Keil &
Anderson, 2018). However, such conceptual appendices seem to be of little theoretical value.
Of course, these elements characterize many situations of decentralization, but they are not
defining features. Their inclusion in definitions appears to be a reflection of the cases authors
have in mind (e.g. Catalonia, Scotland, Québec) rather than a result of conceptual prudence.
32
While numerous alternative definitions have been proposed, the descriptions of Riker (1964, 1975) capture
the defining feature of federations (the key element that differentiates them from unitary and confederal
polities): the shared sovereignty between different policy levels. In the words of Ronald Watts (2013, p. 20),
“the fundamental distinction between federal and unitary systems lies in the location of ultimate political
authority or sovereignty. In unitary political systems, the ultimate authority, constitutionally or in practice,
lies with the central government. There may be administrative, legislative, or financial decentralization to
constituent units, but that occurs at the discretion or will of the central government, which may, if it so
determines, overrule constituent units on any matter.” Federations, which he considers to be “a particular
species of federal system”, are those polities “in which neither the federal nor the constituent units are
constitutionally subordinate to the other (i.e., each has sovereign powers derived from the constitution that is
not unilaterally amendable rather than power derived from another level of government).” (Watts, 2008;
2013, p. 22) In a similar vein, see e.g. Swenden (2013b, pp. 63-64), Kymlicka (2001a), and the ‘federal
principle’ described by Wheare (1963).
57
Chapter 3.
The Belgian power-sharing challenge: A historical
background
The history of Belgium is defined by the power-play between its Flemish and Francophone
segments. The resulting tensions and the reforms made in search of lasting stability
fundamentally changed the very nature of the country. This chapter presents a brief historical
overview of the development of the power-sharing challenge in Belgium. As such, it provides
readers with a general canvas; a frame of reference that adds some context to the analyses and
results presented below. I sketch this history by discerning three phases. As democratization
reforms strengthened the political power of the Flemish segment, (1) the era of Francophone
rule in a unitary state gave place to (2) a period marked by the Francophone fear of being
overruled by the Flemish majority. As the resulting tensions challenged the very survival of the
Belgian state, the search for stability intensified. (3) The result came in the form of a twofold
decision. On the one hand, power-sharing was institutionalized. On the other hand, a
decentralization process was kicked off to facilitate this forced cooperation. The formerly
unitary kingdom of Belgium was turned into a consociational federation.
1. Francophone rule in a unitary state
The power-play between the Flemish and Francophone segments started with an
overt imbalance in favour of the latter. After its birth, the Belgian state was long ruled by
Francophone elites. Having seceded successfully from the United Kingdom of the Netherlands
in 1830, Belgium was established as a ‘centralized and unitary state in which political power
resided to a small francophone elite.’ (Witte & Van Velthoven, 2010, p. 33) Under the 1831
constitution, all citizens were free to use the language they preferred (language freedom), but
French was the exclusive language in central politics.
33
It was the only language used for all
official matters and procedures, including all communication and interactions with citizens (e.g.
in courts and laws). Francophones, including the French-speaking elites in Flanders, ruled the
33
Art. 23 of the 1831 Belgian Constitution. Currently art. 30: “The use of languages spoken in Belgium is
free; only the law can rule on this matter, and only for acts of the public authorities and for judicial affairs.”
58
country. At the time, the choice for a monolingual regime was self-evident, amongst others (and
ironically) because it was considered to be crucial for the unity of the fragile Belgian state. For
the revolutionaries, French was the ‘obvious’ option. It already was the lingua franca of the
ruling classes, it had a very strong cultural reputation as a superior language (especially vis-à-
vis Dutch; Witte, 2009a, pp. 56-57), and it highlighted the break from the preceding ‘Dutch’
rule.
34
As such, the centrality of French had an important nation-building role and became an
aspect of national pride (Bruno De Wever & Vrints, 2008, p. 330).
However, this Francophone façade was hiding a far more diverse reality. In fact, French
was the language of a minority of citizens (about 40% of all Belgians), concentrated in the
South. Ever since its existence, Belgium has been home to a Dutch-speaking majority,
concentrated in the North (Flanders).
35
And as Van den Wijngaert (2011, p. 19) put it nicely,
the unitary Belgian state could only cover up the heterogeneous character of its population for
a short time. In Flanders, language served as a prominent marker of social class, with Dutch
being the ‘low-status language’ and French being the ‘high-status language’ (Witte & Van
Velthoven, 2010). In contrast to the ‘superior’ French language used by the political, cultural,
economic, and academic elites, the Dutch dialects were considered to be the inferior alternatives
used by the poor, rural, and uneducated citizens: “La langue de la population en est un signe
d’infériorité.” (Deschouwer, 1999-2000) So, Dutch-speaking citizens were a numerical
34
French became popular with the elite in Flanders during the 18th-century Austrian rule, and its position as
a ruling class language was further strengthened during the Napoleonic era (Witte, 2009a, p. 56).
35
There is no official post-1947 data on language use in Belgium (as no subsequent linguistic censuses were
held). However, it is beyond doubt that the Dutch-speaking segment still forms a majority in Belgium.
According to recent statistics on the number of inhabitants in each of the language areas (as measured on
January 1, 2020), the Dutch language area is home to 6.623.505 inhabitants, the French language area has
3.563.903 inhabitants, while the German language area (77.845) and the bilingual language area ‘Brussels
Capital’ (1.211.026) are significantly smaller in terms of population (FOD Binnenlandse Zaken, 2020). Given
the territorial monolingualism of the Dutch, French and German language areas (see below), the number of
residents within them provides a solid indicator of the most frequently used languages in contemporary
Belgium. The exceptional ‘facility communities’, in which a significant share of the population uses the other
language (see below), are too small to blur this picture. In the bilingual Brussels Capital language area, about
85% of the inhabitants indicate that their knowledge of French is ‘good/excellent’, while this is only 15% for
Dutch (Janssens, 2018). In discussing these results, it is important to note that there is a growing presence of
other languages such as English, Arabic, Spanish, etc. Especially in the Brussels Capital area, these dynamics
increasingly challenge the binary Dutch-French perspective (see e.g. Janssens, 2019).
59
majority but a ‘sociological minority’ (Witte & Van Velthoven, 2010, p. 15). To climb the
social ladder, citizens needed to know French. Poor, Dutch-speaking citizens were ruled by
Francophone politicians they could not choose (as voting was a privilege of the rich), subject
to Francophone laws they could not read, and faced Francophone courts they could not
understand. In the terminology of the Czech historian Miroslav Hroch, Flanders was a ‘small
nation’ within the dominant or ‘large’ French-dominated Belgian nation (Bruno De Wever,
Verdoodt, & Vrints, 2019, pp. 7-8).
In response to this context of linguistic inequality, the Flemish Movement emerged.
Initially, it was not anti-Belgian or separatist at all. On the contrary, its struggle for the survival
of Dutch within Belgium was initially presented in a Belgian-nationalist, patriotic frame (Bruno
De Wever & Vrints, 2008, p. 331). Indeed, being part of the dynamic wave of Belgian patriotism
that emerged in the aftermath of the Belgian revolution, the young Flemish Movement was
effectively a Belgian Movement (e.g. Wils, 2020, p. 10). Its first political act took place as soon
as 1840 (ten years after the independence struggle), when a petition was issued to demand some
kind of bilingualism in Flanders (Lernout, 1998, p. 2464). Sixteen years later, Belgian
politicians seemed to recognize the Flemish agitation by establishing a ‘grievances
commission’, but the lack of significant results only fuelled frustrations (Wils, 1998). However,
under continuing Flemish pressure, to use the words of Hooghe (1993, p. 49), the language
policy in Belgium would shift from laissez-faire to language planning. Throughout the
1870s-1880s, a first series of laws were passed to allow for Dutch to be used (next to French)
in criminal law (1873), administration (1878), and education (1883) in Flanders. These laws
were very narrow in scope, insufficiently applied, and all in all marginal in effect (Witte, 1993a,
p. 206). But their very existence was indicative of a key dynamic: the growing power of the
Flemish segment.
2. Flemish empowerment and the Francophone fear of
minorization
2.1. Shifting balance of power
The Francophone rule that characterized the Belgian state would be increasingly
challenged by the political empowerment of the Flemish majority in general and the growth
and radicalization of the Flemish Movement in specific. This dynamic was strongly related to
the democratization process in Belgium, and most notably, to the expansion of voting rights in
1848, 1893 (universal plural vote system for men), and 1918 (universal single vote system for
men). Thanks to these reforms, thousands of non-French-speaking citizens obtained the right to
60
vote. Flemings became more numerous and impactful within political parties, and these parties
could no longer afford to neglect the concerns of Dutch-speaking constituents. The result was
a growing number of Dutch oaths and Dutch interventions in parliament (Witte & Van
Velthoven, 2010, pp. 76-77). As such, and step by step, the extension of voting rights allowed
the Flemish demographic majority to become a political majority (Blommaert, 2011, p. 245).
Throughout the decades, and as the following paragraphs describe, the slowly shifting balance
of power manifested itself in many ways.
To begin with, additional and more impactful language laws were introduced in the late 19th
century and after the First World War. These laws were truly seismic. While the ‘Equality
Law’ of 1898 recognized Dutch as an official language (next to French), a series of language
laws adopted in the 1930s embraced the territoriality principle and introduced monolingualism:
each segment was to be governed in its own language (e.g. Dutch in Flanders) (Witte & Van
Velthoven, 2010, pp. 116-122). The same principle was applied to the educational and judiciary
system. Brussels obtained an officially bilingual status.
Meanwhile, the empowerment of the Flemish segment was increasingly visible on the
political scene. Examples include the unexpected election of the imprisoned WWI collaborator
and Flemish nationalist August Borms in 1928 and the electoral victory of the Flemish
nationalists in 1929 (Wils, 2009, p. 230). But most notably, many Francophones found proof
of the ‘predominance of Flanders’ in the uninterrupted rule of Catholic governments since 1884
(the Catholic party was particularly dominant in Flanders). As stated by Lode Wils (2020, pp.
12-13), ‘the uninterrupted Catholic majority in parliament, which lasted until 1919, cemented
the image of polarization between a ‘Catholic and conservative Flanders’ and a ‘secular
progressive Wallonia’’. These conservative cabinets, which ruled the country for decades, were
accused of favouring Flanders in terms of infrastructure and government spending (Witte &
Van Velthoven, 2010, p. 89). Hoping for change, progressive forces had strongly anticipated
the 1912 elections. But once again, the anticlerical government they hoped for failed to
materialize. The Catholic majority was victoriously reconfirmed. In Wallonia, the electoral
result was interpreted in a North-South frame (Kesteloot, 1998c, p. 3637). The ‘painful
awakening’ of 1912 was given an ethno-territorial dimension (Van Velthoven, 2019, pp. 55-
56). “Catholic Flanders kept the conservatives in government, against the electoral wishes of
leftist Wallonia.” (Witte, 2009b, p. 134)
It was this 1912 defeat that made the majority within the Walloon Movement embrace self-
rule, which was considered to be an effective (if not the only) way to escape the
conservative policies of the Flemish majority (Kesteloot, 1998c, p. 3637). These dynamics
nicely illustrate how power-sharing tensions can nourish the sub-state challenge (as argued in
Chapter 1). In Flanders too, the threat of regionalism was growing. In the context of the
61
First World War, frustrations about the Francophone dominance in the Belgian army and the
disrespect vis-à-vis Flemish soldiers made part of the Flemish Movement radicalize towards an
anti-Belgian and independence-seeking position (Bruno De Wever et al., 2019, p. 26).
36
A
clandestine Flemish nationalist organization was established in the trenches (Frontbeweging,
Front Movement) and taking things to the next level, a small segment of the Flemish Movement
(the so-called activists) even opted for collaboration with the occupying German forces. After
the war and building on these radical segments, the Flemish nationalist Front party was
established (Frontpartij). The program of this party, which was officially labelled Vlaamsche
Front (Flemish Front), was explicitly anti-Belgian (Bruno De Wever, 1994, p. 33; Bruno De
Wever & Seberechts, 1998, pp. 3334-3335). For the first time in history, as Bruno De Wever
and Vrints (2008, pp. 337-338) note, a radical wing of the movement developed a Gellnerian
program: that of a Flemish state for a Flemish people.
37
2.2. Post-war tensions and traumas
Tensions continued to grow in the interbellum and in the aftermath of the Second World
War, when Belgium was subject to a series of ‘political and ideological shockwaves, which
opposed the two communities against each other’ (Van den Wijngaert, 2011, p. 20). It is in this
period that the debate about the power-sharing challenge truly came to the forefront. One
source of frustration was the post-war repression. Due to the collaboration by part of the
Flemish Movement, another troublesome layer was added to the Francophones image of
Flanders. Already being associated with the peasantry, Flanders was now identified with treason
too (Witte & Van Velthoven, 2010, p. 139). In Flanders, many considered the sanctioning of
those who (allegedly) collaborated with the German forces to be an unjust, disproportionate,
and anti-Flemish operation (for a more balanced view, see Huyse & Dhondt, 1991).
38
Agitations
36
It can be noted that the idea of Flemish self-rule was already present in some segments of the Movement
on the eve of the First World War (Bruno De Wever & Seberechts, 1998, p. 3334).
37
Gellner (2008, p. 1) famously defined nationalism as “a political principle which holds that the political
and the national unit should be congruent.”
38
A significant share of the Flemish nationalist movement had opted for collaboration. As such, the war
weakened Flemish nationalism both organizationally and ideologically, as it was burdened by an odium of
Nazism while many leaders and grassroots supporters died, were imprisoned, or lost their political rights
(Bruno De Wever & Vrints, 2008, p. 359). At least within part of the Flemish Movement, the repression on
62
were further enhanced by a related issue that came to dominate the political agenda in the post-
war years: the so-called ‘Royal Question’; a key event that highlighted the limits of
majoritarianism in Belgium. The question at stake was simple: should King Leopold III return
to Belgium (the king was transported to Austria by the Germans in 1944)? Leopold was strongly
criticized for a wide range of elements that associated him with the collaboration (Velaers &
Van Goethem, 2001, p. 240; Witte, 2009e, p. 240): his pre-war policy of neutrality, his conflict
with the Belgian cabinet in 1940, his personal preference for a strong executive power and his
authoritarian views more generally, his marriage to Lilian Baels amidst the war, the close
contacts between his entourage and collaborating organizations, his visit to Hitler in
Berchtesgaden (November 1940), the deportation of the royal family before the liberation of
Belgium, the fact that he congratulated Hitler, never publicly condemned the forced labour
deportations or the persecution of Jews, and (most importantly) his refusal to back the
resistance, the Allied forces, and the associated Belgian cabinet. Arguing amongst others that
the king’s approach safeguarded Belgian civilians from even more bloodshed, the return of the
king was backed by (amongst others) the Christian democrats (CVP/PSC), which managed to
organize a referendum on the issue in March 1950. This resulted in an overall majority in favour
of the king’s return (57.6%), but the outcome echoed Belgium’s divided nature: a majority of
Flemish voters (72%) were in favour of the return, while this was only 42% in Wallonia (and
48% in Brussels). When the CVP/PSC obtained an absolute majority in the 1950 elections and
tried to push through the unconditional return of the king, an extraordinary climate of resistance
arose – most notably in Wallonia. Belgium was effectively on the brink of civil war (Velaers &
Van Goethem, 2001).
“The Walloon industrial belt went on strike on July 6 and on July 26 the
transport sector joined the protest action. There were plans to bring the whole
steel industry to a standstill and a march on Brussels was being prepared for
early August. There were calls for the creation of a Walloon republic and when
state police shot and killed four people in Grace Berleur, the smell of revolution
the streets and in the courts nourished a deep sense of resentment vis-à-vis the Belgian state, anger towards
the Belgian elites, and aversion towards any kind of ‘collaboration’ with or participation in the Belgian
regime (Vandeweyer, Huyse, Hoflack, & Ruys, 1998, pp. 2601-2602). In their seminal analysis of the post-
war repression, Huyse and Dhondt (1991, p. 268) emphasise the distinction between Flanders at large and
Flemish nationalism: There is no proof that the repression had an intentional anti-Flemish nature. It did,
however, deliberately aimed to eliminate Flemish nationalists for a significant period of time (for a similar
conclusion, see Kossmann, 1986, p. 154).
63
was in the air.” (Witte, 2009e, p. 242)
In response to these tensions, Leopold III eventually abdicated in favour of his son Boudewijn.
The monarchy was saved, but the damage was significant. The Royal Question had not only
deepened linguistic divisions within the Christian democratic party (Delwit, 2011b, p. 15). It
also reinforced the sense of antagonism between a dominantly catholic Flanders and a generally
socialist Wallonia more generally. Resentment on both sides was fueled. In conservative
Flemish cycles, frustrations about the Francophone violation of the majority principle ran high.
Raw violence in the streets, they argued, prevailed over the results of democratic elections and
a people’s referendum. In Wallonia, the ‘brutal’ attempt to ‘impose’ the king by simple majority
was seen as additional proof for the idea that the Francophone minority was being pushed
around by the increasingly dominant and assertive Flemish majority.
2.3. Economic transition
Another element that added to this perception and the call for state reform was the
economic “reversal of fortunes” between Flanders and Wallonia (L. Hooghe, 2004, pp. 56-
58). The North was traditionally the poor and agrarian part of Belgium, while the highly
industrialized South used to be Belgium’s economic powerhouse. But from the 1950s onwards,
and against the background of de-industrialization dynamics, this changed dramatically. The
economic centre of gravity eventually shifted towards the North (Quévit, 1978) and during the
1960s, Flanders surpassed Wallonia in per capita GDP (Buyst, 2011).
39
This economic
transition went along with demographic evolutions that added to the Francophone fear of
minorization: the Flemish population increased at a faster pace than the Walloon population (L.
Hooghe, 1993, pp. 51-52). As described by Deneckere, De Paepe, De Wever, and Vanthemsche
(2014, p. 294), these dynamics also impacted the self-images on both sides of the language
border. While the Flemish movement used to be driven by feelings of deprivation, it was now
marked by confidence, self-praise and criticism against the Francophone South. Meanwhile, the
latter’s explicit feelings of superiority turned into a self-image of an underdog.
39
From the 1950s onwards, Wallonia’s (largely industrial) economy faced a sharp decline as the coal mining
and steel industries took heavy blows. Flanders, on the other hand, attracted investments and benefited
strongly from the changing context (the breakthrough of oil, the emergence of a European market, the
growing importance of consumer durables, etc.) (Buyst, 2011).
64
The resulting frustrations boosted the Walloon Movement throughout the 1960s
(Kesteloot, 1998a, 1998c; Van Dam, 1998). In these years, it even acquired –at least in part–
the characteristics of a mass movement (Kesteloot, 1993, p. 42). In this respect, the historically
disruptive strike against the Eenheidswet (1960-1961) was crucial. This ‘strike of the century’
was not only one of the most significant outbursts of the class conflict in Belgian history. It also
fed inter-segmental divisions and reaffirmed the image of Flanders imposing its conservative
policies on Wallonia – thus preventing the latter from realizing its true potential (Witte, 2009d,
p. 368). The failed opposition against the law, which was promoted by the right-wing
government led by a Flemish and Christian democratic Prime Minister (G. Eyskens, CVP/PSC),
was far more intensive, exhaustive, and long-lasting in Wallonia than in Flanders. In the words
of André Renard, the influential leader of the Walloon socialist union and the future president
of the federalist-socialist party Mouvement Populaire Wallon: “On veut punir les Wallons parce
qu’ils sont socialiste” (CRISP, 1961). In Wallonia, pre-revolutionary tensions began to rise: the
strike spread to all sectors, socialist mayors refused to execute the orders of the central
government, barricades isolated several places in the industrial belt (the Borinage), brutal street
fights left many wounded, and the sabotage of bridges, electricity and rail infrastructure
increased (Meynen, 2009, pp. 278-279). Meanwhile, opposition in Flanders was more moderate
and less widespread, which strongly nourished Flemish-Francophone animosities in leftist
movements and unions too (Delwit, 1998; Van Velthoven, 2019). It was in this context that the
Walloon labour movement linked the socio-economic and regionalist issues (Pilet, 2005, p.
271; van Haute & Pilet, 2006, p. 302).
Experiences like these, and the growing power (display) of the Flemish segment more
generally, triggered a double reaction in Francophone Belgium. Both were aimed at avoiding
their minorization and pacifying tensions at the federal level. On the one hand, there were
growing calls for the establishment of a Walloon sub-state with substantial economic autonomy.
Self-rule was seen as an elegant way to escape Flemish conservatism, implement socialist
policies, and recover economically. On the other hand, as the Flemish demographic majority
began to display its power, devices to block this majority were urgently demanded
(Deschouwer, 2012, p. 39). The call for consociational vetoes was on the rise.
2.4. Linguistic tensions
The image of Flemish assertiveness was further strengthened by two major
demonstrations in the early 1960s: the so-called ‘Marches on Brussels’. They were a response
to the loss of Flemish territory due to the spread of French, which gradually turned Flemish
towns bordering Wallonia and Brussels into Francophone ones. Such dynamics were facilitated
by the 1932 law on the language of administration, which stipulated that the linguistic status of
65
municipalities would depend on the results of the language census (Witte & Van Velthoven,
2010, p. 119). Residents were surveyed and if 30% of the respondents used the other language,
external bilingualism would be implemented. The language of administration would even shift
altogether if the linguistic majority shifted. As a result, “Flemish nationalists now perceived a
Francophone threat on their boundaries. After each language census some Flemish territory was
lost, especially around Brussels” (L. Hooghe, 1993, pp. 49-50).
40
The linguistic censuses on
which these shifts were based, were also highly contested. There were many complaints of false
statements, intimidations, pre-filled forms, etc. (De Metsenaere, 1998).
41
Meanwhile, the
existing language laws were hardly practised or cleverly bypassed in the officially bilingual
Brussels agglomeration (Craeybeckx, 2009, p. 176), which quickly became a Francophone
stronghold. In 1961 and 1962, these Flemish grieves turned into protest. In two Marsen op
Brussel (Marches on Brussels), respectively 63.400 and 41.000 demonstrators demanded
(amongst others) the ending of the language censuses, the abolition of the 1932 provisions
shifting language provisions, the limitation of the Brussels agglomeration to the 19
municipalities it covered (and genuine bilingualism within it), and in general: a fixed language
border (Haagdorens & Wouters, 1998).
42
This border would eventually be established in 1962-
1963 (Witte & Van Velthoven, 2010, p. 151).
43
Since then, Belgium is divided into four
language areas: a Dutch, a French, a German and a bilingual Brussels Capital language area.
The linguistic segments were now officially anchored.
In several ways, the compromise on the language border sowed the seeds for future
tensions, many of which will resurface throughout this work. To begin with, several
municipalities were transferred from Flanders to Wallonia (Komen/Comines and
40
As expressed more bitterly by Witte (1993a, p. 217), the language law of 1932 turned the language censuses
into “referendums in the French speakers’ favour and a weapon in the armoury of linguistic struggle.”
41
The Flemish discontent with these censuses was clearly expressed in 1960, when the controversial census
was boycotted by over 500 Flemish municipalities (Witte & Van Velthoven, 2010, p. 150).
42
These are the numbers reported by the gendarmerie (which are lower than the estimates of the organization
behind the Marches) (Haagdorens & Wouters, 1998).
43
The decision to exclude the linguistic chapter from the censuses was already taken in 1961 (Witte & Van
Velthoven, 2010, p. 150). The first state reform (1970) constitutionally anchored the language border
(Deweerdt, 1998b).
66
Moeskroen/Mouscron) or vice versa (six municipalities of the Voeren/Fourons region).
Tensions in these towns would regularly run high, often compromising the stability of the
federal cabinet. Additionally, the compromise included the establishment of so-called ‘facility’
municipalities. After all, in certain towns, there was a significant presence of residents of
another language group (e.g. Francophones in towns within the Dutch language area). Some of
these towns were given the facility status. There, public services including official
communication and primary education are not only available in one language. They can be
provided to residents in the other language –if they ask so. The interpretation and application
of these facilities, particularly with regards to the six facility communities bordering Brussels,
would continue to burden inter-segmental relations and haunt federal executives throughout the
years (e.g. Koppen, Distelmans, & Janssens, 2002; Witte & Van Velthoven, 1998, 2010). In
reaction to these 1962-1963 language laws, and as an illustration of the resentment it triggered,
a new political party entered the scene: the Front démocratique des francophones (FDF).
44
Presenting itself as the defender of the Francophones in Brussels and its periphery, its main
objective was twofold: the abolition of the newly adopted language laws, and the organization
of people’s referenda in the municipalities surrounding Brussels on the expansion of the
Brussels Capital area (Delwit, 2011b, p. 18).
The fixed language border largely put a stop to the growing presence of French in Flanders. By
the late 1960s, only a few Francophone bastions were still in place. The most prominent of these
was the Catholic University of Leuven (Flanders), which became the scene of ethno-
territorial controversies (Gevers & Vos, 1998, pp. 2915-2916; Witte, 2009c, pp. 375-376; Witte
& Van Velthoven, 1998, pp. 3028-3029). Resentment about this bilingual institution in the heart
of Flanders (an aberration by then) erupted when the bishops announced the expansion plans of
the university. The establishment of new Francophone departments in the triangle between
Brussels, Leuven, and Wavre was seen as a major threat to the Dutch nature of a large part of
Flanders. The doom scenario of an advancing Francophone elite resurfaced. In the revolutionary
spirit of May 1968, Flemish students took to the streets with slogans that traumatize
Francophones until this day: ‘Leuven Flemish!’, ‘Walloons go home!’, and ‘Walloons out!’
45
44
The party, established in 1964 as the Front Démocratique des Bruxellois, originated in the grievances and
initiatives of Francophone political and academic circles (the ULB and the UCL) (Pilet, 2005, p. 267).
45
In Dutch: Leuven Vlaams, Walen go home, Walen buiten. In Flanders, this struggle is commonly referred
to as Leuven Vlaams (Leuven Flemish). In Francophone Belgium, and highlighting the divergent perceptions
67
Soon, tensions spilt over to the political sphere. The ethno-territorial gridlock kept the federal
executive hostage and Prime Minister Vanden Boeynants eventually had to accept the early
resignation of his cabinet. Eventually, the university was split along linguistic lines: the Flemish
Katholieke Universiteit Leuven (KULeuven) remained in Leuven, while the Francophone
Université Catholique de Louvain (UCL) received funding to establish a new campus in
Wallonia (Witte, 2009d, pp. 375-376). The tensions were pacified, but the damage was
significant.
2.5. Power-sharing crises within parties and the split of the party system
These events were not only highly traumatic for the Francophones but also triggered
two important changes in the party landscape. First, the establishment of a new regionalist
party. Embittered by these events, several Francophone Christian democrats decided to join the
regionalist Parti Wallon (PW), which consequently turned into the Rassemblement Wallon
(RW) (Pilet, 2005, p. 271; van Haute & Pilet, 2006, p. 302).
46
Second, the Leuven issue was
the direct cause of the first cracks in the Belgian party system. It resulted in the first break-up
of a state-wide party along linguistic lines: the Christian democratic party (CVP/PSC). The
split of the largest party in Belgium perfectly reflected the political state of mind at the
time. And in a sense, the state of this party reflected the condition of Belgium at large. The
1968 break-up of the CVP/PSC reconfirmed the image of diverging segments, fed the
increasingly contentious debates on the possibilities and means of Flemish-Francophone
power-sharing, and added to the growing idea that segmental autonomy can ease tensions. The
recipe applied to the party would later be applied to the country at large: pacification
through separation.
The divorce was part of a larger process of party disintegration. As a matter of fact, the gradual
split of the Belgian party system occurred in several waves (De Winter et al., 2009, pp. 72-
76). First, the regionalist parties that were established in response to the growing post-war
tensions paved the way: The Volksunie (VU) in Flanders, the Rassemblement Wallon (RW) in
of history, it is known as Walen Buiten (Walloons out).
46
In the built-up towards the 1965 general elections, different left-wing Walloon regionalist lists emerged
and embraced l’action commune wallonne. After their modest success (two candidates were elected), the
different lists united under the banner of the Parti Wallon (June 1965) (Pilet, 2005, p. 271; van Haute & Pilet,
2006, p. 302).
68
Wallonia, and the Front Démocratique des francophones (FDF) in the Brussels Capital area
(see also Deschouwer, 2009; van Haute & Pilet, 2006). Second, the three traditional parties in
Belgium were facing increasingly strong segmental pressures (the Christian democrats, liberals,
and social democrats). None of these parties, in the words of Deschouwer (2002, p. 73), could
survive these growing tensions. In 1968, 1972, and 1978 respectively, they split into separate
Flemish and Francophone parties, whose organizational and personal ties eroded over time
(Dandoy & De Decker, 2009; De Winter, 2006; Deschouwer et al., 2017). Third, when new
parties joined the party system in the late 1970s, they confined themselves to one community
too. This applied both to the radical right and separatist Vlaams Blok (VB) and to the left-wing
ecologist parties (Ecolo in Francophone Belgium, Agalev in Flanders). The result is a very
peculiar phenomenon: both sides have their own, distinct multiparty system, parties only
compete within their own segment, and voters have no option but to vote for parties on ‘their
side’ (with minor exceptions, cf. footnote 14 in this work).
3. Power-sharing in a consociational federation
As the Flemish underdog was gaining economic, political, and demographic power,
there was a growing perception of a Flemish-dominated Belgian state (L’Etat Belgo-
Flamand) (De Winter & Baudewyns, 2009, p. 285). The combination with the range of
traumatizing experiences discussed above led to two far-reaching conclusions. First, segmental
majoritarianism is to be avoided. There is a need for structural power-sharing. Second, given
the divergent preferences on both sides, this power-sharing can only work if the segments don’t
need to agree on everything. Allowing them to choose their own policies on certain issues might
avoid recurrent tensions and pacify ongoing ones. As such, decentralization was used
(amongst others) to facilitate the forced cooperation that power-sharing prescribed. With the
first state reform (1970), these conclusions were put into practice. Belgium was turned into a
‘consociational federation’ (Deschouwer, 2002, 2005, 2006). Reflecting the Francophone
aversion to Flemish majoritarianism, this system strongly renounces the Westminster model of
majority rule: “the opposite of consociational democracy is majoritarian democracy”, just like
“the opposite of federalism is unitary government” (Lijphart, 1979, p. 500). Consociational
democracy can be defined by four principles: grand coalition, segmental autonomy, mutual
veto, and proportionality (e.g. Lijphart, 1977b; Lijphart, 1979, 1985a). In more general terms,
it combines power-sharing with segmental autonomy, but scholars have mostly focussed on the
former element (as noted by e.g. L. Hooghe, 2004, p. 32). The following paragraphs provide a
more detailed discussion of the way both elements (power-sharing and autonomy) were applied
in Belgium and how they developed throughout the years.
69
3.1. Imposing power-sharing
The 1970 reform forced both segments to cooperate. The power-sharing requirement
was institutionalized with a range of devices. First, it introduced linguistic parity in the
Council of Ministers, which decides by consensus. With the possible exception of the Prime
Minister, this council must be composed of an equal number of Dutch- and French-speaking
members.
47
Second, two language groups were established in parliament: reflecting the
predominant bipolarity of Belgian politics, all federal MPs are formally part of either the Dutch
or French language group.
48
Third, certain contentious issues became the subject of ‘special
majority laws’. These issues, which are specified in the constitution, can only be passed or
changed if (a) a majority of each language group is present in each chamber, (b) a majority of
votes in each language group (in each chamber) is a yes-vote, and (c) an overall majority of
two-thirds is reached. Fourth, the ‘alarm bell procedure’ was introduced to further avoid
majoritarianism (on all issues): when three-quarters of either language group argues that a bill
‘could seriously jeopardize the relations between the communities’, the legislative procedure
can be blocked. The issue is then transferred to the Council of Ministers (characterized by its
linguistic parity), which has thirty days to propose its advice or amendment(s) to the chamber
in which the alarm bell was rung. These consociational crown jewels turned both language
groups into very strong veto players. Imposing the Flemish majority, or threatening to do so,
was no longer an evident option. As such, the 1970 reform anchored the “institutional obligation
47
Only ministers are formally part of the Council of Ministers. The constitutionally required language parity
only relates to them, and does not concern the secretaries of state. The latter are not officially part of the
Council of Ministers, but in practice, they often join in, most notably when the issues at stake have to do with
their area of competence. Note that, to meet the requirement of the parity rule, it suffices that at least one
Dutch-speaking and at least one Francophone party joins the cabinet. It does not preclude the establishment
of a government that fails to represent a majority of Dutch-speaking and/or Francophone voters.
48
A Dutch and French language group was established in both the Senate and the Chamber of
Representatives. In the Senate, the respective strength of the language groups is fixed (35 Dutch-speaking
Senators, 24 French-speaking senators). The sole Senator for the German-speaking community does not
belong to a language group. In the Chamber of Representatives, the balance between the language groups is
not fixed in advance. MPs elected in the German language area are included in the French language group.
MPs elected in the constituency of Brussels-Capital are assigned to one of the language groups based on the
language in which they swear the oath (first). Before the sixth state reform, the latter rule applied to all MPs
elected in the larger, now abolished, Brussel-Halle-Vilvoorde constituency.
70
to opt for consensus” (Deschouwer, 2006, p. 901). Or more accurately, it left the elites with
two options: governing together or not governing at all (Deschouwer, 2013a, p. 214). Of
course, given the existing tensions, it was clear that this inter-segmental power-sharing was
particularly difficult on certain issues. In these cases, decentralization provided an elegant way
to bypass the very need for cooperation.
3.2. Facilitating power-sharing: segmental autonomy
Segmental autonomy was used (amongst others) to facilitate the often difficult
practice of segmental power-sharing. It allowed politicians to evacuate contentious issues
from the central policy level towards the sub-states, thus lifting the very need for cooperation
in the first place. The first state reform (1970) established three communities (initially called
‘language communities’) and prescribed the establishment of three regions (which effectively
came into existence in the 1980s). Initially, the power of these communities was very limited
in scope. Indeed, they had no directly elected sub-state councils, no separate executives (the
decrees were implemented by the national government), and very few competences. This
illustrates a dynamic that would resurface throughout the decades, as “Each state reform in
Belgium is a contradictory combination of strong autonomist and strong unitarian tendencies.”
(L. Hooghe, 1993, p. 55) As the years went by, the sub-states in Belgium would be gradually
strengthened, resulting in autonomy levels that are particularly high in comparative perspective
(see also Chapter 5) (Fabre, 2009; L. Hooghe et al., 2016; Shair-Rosenfield et al., 2021; Watts,
2013). Five additional state reforms are responsible for this. They were agreed upon in 1980,
1988-1989, 1993, 2001, and 2011. Throughout the remainder of this chapter, I provide a brief
summary of the main developments and discussions that marked these decades of institutional
engineering.
From its very start, the process of state reform was burdened by major discord. For the
Flemish politicians, the optimal system was that of a bipolar federation based on two large
Communities (a Francophone and a Flemish one). In contrast, most Francophone leaders
preferred a tripartite federation based on three Regions (Flanders, Wallonia, and Brussels).
These opposing visions are a product of several elements. First, they are strongly linked to the
contrasting nature of the historical struggle of Flanders and Wallonia. The Flemish linguistic-
cultural struggle and the Walloon socio-economic struggle. As stated by L. Hooghe (1993, p.
46), Flemish nationalism ‘imagined its community’ predominantly along cultural-linguistic
71
lines, while Walloon nationalism imagined it along socio-economic lines.
49
The establishment
of three Communities echoed the Flemish demand for cultural autonomy, while the creation of
three Regions (a Flemish, Walloon and Brussels-Capital Region) complied with the Walloon
aspirations for socio-economic self-rule (e.g. Swenden & Jans, 2009, pp. 18-19). Second, these
opposing preferences on the fundaments of the Belgian federation are strongly related to the
divergent visions on the status of Brussels. The dominant Flemish demand was that of a bipolar
Belgium with a distinct, peculiar statute for Brussels (the shared capital). In contrast, the
Francophones opted for a tripartite Belgium with Brussels as a full-fledged sub-state next to
Flanders and Wallonia (une région à part entière) (Van Velthoven, 2005, p. 62; 2011, p. 335).
The latter was as strongly advocated by the Francophones as it was opposed by the Flemings
(de Metsenaere et al., 1998, p. 650). After all, they consider(ed) Brussels to be an integral part
of Flanders, as is famously echoed in the cry that ‘Flanders does not let go of Brussels’
(Vlaanderen laat Brussel niet los).
50
The idea of a tripartite federation (Flanders, Wallonia,
Brussels) was also feared because of the double threat it poses:
“In an autonomous Brussels region the Flemish minority would be cut off from
Flanders, which might put pressure on them to assimilate. On the national level
Flanders could be pushed into a permanent minority position by a Francophone
Brussels region and a Francophone Walloon region.” (L. Hooghe, 1993, p. 54)
In an attempt to reconcile these different logics, Belgium was transformed into a complex
asymmetric federation, which is home to two types of sub-states: three Communities (a
Flemish, French and German-speaking Community) and three Regions (a Flemish, Walloon
and Brussels-Capital Region). While the latter are territorially confined, the Communities are
non-territorial in nature; the Flemish and French Communities overlap in Brussels, where
49
This expression refers to the work of Benedict B. Anderson (1983), who famously described nations as
‘imagined communities’.
50
This famous statement goes back to 1954, when Lode Craeybeckx (the major of Antwerp) proclaimed:
‘Antwerp does not let go of Brussels’ (Antwerpen laat Brussel niet los) (Ruys & Garré, 1998, p. 817). The
Flemish claim on Brussels, or in softer terms, its attachment to the Flemish-Brussels interaction, was further
emphasized by e.g. the Flemish choice to make Brussels its capital as well as the location of its government,
parliament and administration (Witte & Van Velthoven, 2010, p. 181).
72
residents of both Communities are located (Swenden & Jans, 2009, pp. 18-19).
51
The first state
reform only established the Communities. Due to major disagreement, most notably on the
position of Brussels, the establishment of the three Regions was postponed. Article
107quater of the Constitution prescribed their establishment, but the political negotiations that
specified their competences, organization, and status would continue until the 1980s. Both the
tripartite Leburton cabinet (1973-1974) and the Catholic-Liberal-RW Tindemans I government
(1974-1977) failed to deliver, and the attempt to reach an encompassing state reform (the
Egmont Pact) resulted in a traumatic miscarriage that lead to the split of the last traditional party
(the social democrat party) (Witte, 2009d, pp. 377-378). After two additional failures (under
Martens I and Martens II, 1979-1980), it was only in 1980 that a second state reform was
agreed upon. This reform not only strengthened the Communities, amongst others with separate
governments and the so-called person-related matters (Popelier & Lemmens, 2015, pp. 26-27),
it also established the Flemish and Walloon Regions. But in lack of consensus, the most
problematic part of the triplet (Brussels) remained in the womb. A compromise was only
reached with the third state reform (1988-1989).
This third state reform was a victory for the federalists (Witte, 2009d, p. 382). To begin with,
a range of additional powers was transferred to the sub-states, including education policy,
economic policy (including the ‘national sectors’: coal, steel, etc.), and new competences
regarding scientific research, public works and transportation (Senelle & van de Velde, 1998,
p. 2835; Swenden, Brans, & De Winter, 2009, pp. 4-5). The financing of the sub-sates was
described in the notorious ‘special financing law’ (Bijzondere financieringswet - BFW) (for an
analysis, see e.g. Decoster & Sas, 2011; Decoster & Sas, 2012). But most importantly, the third
state reform settled several major sources of frustration. First, there were ongoing linguistic
struggles in the facility communities (most notably in Voeren and the communities bordering
Brussels). Arguably, the most explosive debates were about the appointment of French-
speaking José Happart as mayor of Voeren, a town that was transferred to Flanders with the
establishment of the language border. The refusal of Happart to speak Dutch triggered major
crises, which were only settled with the so-called ‘pacification law’ of 1988. With this major
agreement, tensions were “solved in a spirit of compromise where no one felt unduly hurt”
51
As noted by Reuchamps (2017, p. 4), this led to a particular paradox: “the Flemings prefer linguistic ties,
given by the Community, but need the Regions to ensure clear boundaries and obtain more autonomy,
whereas the Francophones prefer a regional vision in order to recognise Brussels as a fully-fledged Region,
but need the French Community in order to link Brussels and Wallonia.”
73
(Witte, 2009d, p. 382).
52
Second, the regionalization of the ‘national economic sectors’ (most
notably the steel sector) made an end to years of disputes about the federal financing that was
needed to compensate for their decline. This economic transition mostly impacted the heavily
industrialized Wallonia. The Flemish parties started refusing to pay for these terminal sectors,
using the rather clear slogan ‘no more Flemish money for Walloon steel’ (“geen Vlaamse frank
meer voor het Waalse staal”). The resulting debates were a major burden on the stability of
consecutive power-sharing executives and culminated in the resignation of the cabinet of Prime
Minister Mark Eyskens, who announced it in his famous poetic style: ‘It is September, 21. The
first day of fall. The leaves are falling. The cabinet fell too.’ (M. Eyskens, 2014)
Decentralization was considered to be a solid way to evacuate this explosive issue from the
federal agenda. Third, a compromise was reached on the establishment of a Brussels sub-
state. The Flemings accepted the existence of a Brussels Region as long as its status deviated
from that of the Flemish and Walloon Region (Popelier & Lemmens, 2015, pp. 27-28).
However, the subtle elements that differentiate the Brussels-Capital Region from its Walloon
and Flemish counterparts are largely symbolical in nature; they are the forgotten souvenirs of a
compromise that is only remembered by a handful of scholars.
53
As such, the Francophones
52
The most important elements of this compromise, which was no part of the state reform s.s. but strongly
linked to it, can be summarized as follows (Witte, 2009d, pp. 382-383; Witte & Van Velthoven, 2010, pp.
172-173). On the one hand, the linguistic frontier remained intact, the Flemish authorities were now in control
of all Flemish municipalities, and they were given the possibility to dismiss mayors and heads of social
welfare councils that fail to meet the language requirements. On the other hand, the system of language
facilities was now constitutionally anchored (and thus made permanent) and Francophone council members
and aldermen in these municipalities were now sacrosanct. This is the result of one of the most typical
examples of the (often surreal) ethno-territorial compromises that characterize Belgian history. It was decided
that aldermen in these communities were to be elected directly too (just like the council members).
Accordingly, the very fact that they were elected was proof of an ‘irrefutable assumption’ that they knew the
language of the respective region (Witte, 2009d, p. 383; Witte & Van Velthoven, 2010, p. 173).
53
Several subtle elements expose the distinctiveness and the somewhat inferior status of the Brussels-Capital
Region. To begin with, it is not just labelled a Brussels Region, but the Brussels-Capital Region (with
‘Brussels’ as an adjective, and ‘Capital Region’ as a single, distinct notion). As noted by De Bruycker (1989,
p. 56, fn. 177): “nul doute que celle-ci a été soigneusement choisie ce qui permet aux négociateurs flamands
d’insister sur le caractère spécifique de cette région par rapport aux deux autres.” Also, it did not have any
constitutive autonomy until the sixth state reform (for an analysis, see e.g. Peiffer & Sautois, 2014).
Furthermore, the Brussels-Capital Region issues ‘ordinances’ rather than ‘decrees’. “While they also have
force of law”, Popelier (2019, p. 28) notes, “they are treated in a somewhat inferior way.” In this respect,
74
scored a major victory with the 1988 compromise (Witte, 2009d, p. 383; Witte & Van
Velthoven, 2010, p. 181): the Brussels-Capital Region was de facto established as a third region,
with its own elected council and executive, and just about the same competences as the other
Regions. Being established as a ‘nearly full-fledged third region’, the Brussels-Capital Region
also included strong guarantees for the Flemish minority in Brussels (Van Velthoven, 2005, p.
62; 2011, p. 335).
54
As noted by Deschouwer (2013a, p. 214), “this very pragmatic solution for
Brussels has made nobody very happy”. Throughout the years, debates on the status and borders
of Brussels would sporadically resurface.
While all primary components of Belgian federalism were now established, the reform process
continued. Lengthy and often difficult discussions on a new state reform (Witte, 2009d, pp.
385-386) resulted in one of the most far-reaching institutional transitions of all. With the fourth
state reform of 1993, Belgium finally come out of the closet and officially labelled itself ‘a
federal state, composed of the communities and the regions’ (Art. 1 of the Constitution). Apart
from this rather symbolical gesture, the Saint Michael’s and Saint Quinten’s agreements (1992)
were ground-breaking in more palpable ways too. They established direct elections for the
Regional councils (parliaments) and abolished the double mandate (members of sub-state
parliament were no longer members of national parliament), reformed both the Senate and the
Chamber of Representatives, split the bilingual province of Brabant into Flemish Brabant and
Loumaye (1989, p. 11) talks about a ‘tempered equivalence’: “L'ordonnance de la Région de Bruxelles-
capitale a une force juridique équipollente à la loi et au décret tempérée par un contrôle juridictionnel
restreint. On pourrait parler d'une "équipollence tempérée". Ordonnances can be reviewed against the special
majority law on the Brussels institutions and against the Constitution, in so far as they do not fall under the
jurisdiction of the Constitutional Court, against the provisions in the special majority law on the Brussels
institutions. Also, certain ordinances (not all of them) are prone to a form of administrative supervision: to
protect the role of Brussels as an international and capital city, they can be suspended by the federal
government, and annulled by the Chamber of Representatives (with a majority in both language groups). In
practice, however, this system has never been applied (Popelier, 2019, p. 35). Generally speaking, it is safe
to state that the distinct status of ordonnances as opposed to decrees and laws is of little practical relevance,
and indeed largely symbolical in nature (see e.g. De Bruycker, 1989, p. 53; Sautois, 2014, pp. 151-152).
54
The council of the Brussels-Capital Region was divided into two language groups, Dutch-French parity
was introduced in its executive, and an alarm bell procedure was introduced (de Metsenaere et al., 1998, p.
651). It is important to note that these elements clearly mirror the 1970 mechanisms (introduced to protect
the Francophone group at the central level). As such, these protection tools for the Flemish minority in
Brussels should not only be perceived as compensation for the 1988 Flemish concessions, but rather as a
crucial, constitutive part of the Belgian communal equilibrium at large.
75
Walloon Brabant, and transferred significant competences to the sub-states, including the power
to sign international agreements on issues residing within the scope of their respective
competences (in foro interno, in foro externo) (Senelle & van de Velde, 1998, pp. 2836-2839;
Swenden et al., 2009, pp. 3-6; Witte, 2009d, pp. 386-387).
The scope of this reform, which was initially announced by the reformers as being the final
round (L. Hooghe, 2004, p. 77), sharply contrasted with that of its successor (the fifth state
reform). One explanation for this is that the preferences between North and South diverged
(which burdens negotiations ever since). Specifically, while there was very little
Francophone appetite for consecutive steps, Flemish parties continued to raise
increasingly far-reaching autonomy demands. For example, a commission on state reform
was soon created in the newly established Flemish parliament (1995-1999), the head of the
Flemish government (Luc Van den Brande, a Christian democrat) called for a confederal
transition, and the first directly elected Flemish Parliament adopted the notorious ‘five
resolutions’ of 1999, in which it pleaded for a far-reaching strengthening of Flemish autonomy
within a ‘2+2’ federation (a bipolar Flemish-Francophone model, with a special statute for
Brussels and the German-speaking Community) (Ruys, 1998, p. 592; Van Velthoven, 2011, p.
349; Witte, 2009d, p. 388). These calls eventually resulted in the fifth state reform (2000-2001).
Apart from reforming the Brussels-Capital Region, it also strengthened the Region’s fiscal
autonomy and further extended the list of sub-state competences (including municipal and
province laws, developmental aid, science policy, agriculture and fishery, etc.) (Witte, 2009d,
p. 389).
55
What followed was a troublesome decade that culminated in the 2011 agreement on the
sixth state reform. The difficult road towards this compromise was marked by a range of
55
Concerning the Brussels-Capital Region, several means were introduced to strengthen the position of the
Flemish segment in Brussels (most notably a guaranteed number of 17 out of 89 seats in the Regional
Parliament, and an incentive to appoint a Flemish alderman or president of the local council for social
welfare), and a very peculiar range of mechanisms was adopted to avoid the radical right Vlaams Blok from
obtaining an actual veto position and trigger a complete gridlock of the Region (Jacobs & Swyngedouw,
2003). These mechanisms ended up weakening the positions of the Dutch-speaking segment in general, as
they effectively gave up the double majority rule. Referring to the 2001 Lombard agreement on Brussels,
Jacobs and Swyngedouw (2003, p. 136) use the term lombarding to describe the steps taken by a range of
parties to “institutionally ostracize” one party which they consider to be undesirable (in this case the radical
right Vlaams Blok).
76
noteworthy events. To begin with, the regionalist party scene changed significantly. As a direct
result of the 2001 state reform, major tensions within the Flemish nationalist Volksunie (VU)
party eventually led to its implosion (e.g. Delwit & van Haute, 2002; Govaert, 2002; van Haute,
2005; van Haute & Pilet, 2006). Two successors immediately emerged: the progressive Spirit
party, which opted for a cartel with the Flemish Social Democrats (SP.a) and the Nieuw-
Vlaamse Alliantie (N-VA), a more hardline Flemish nationalist party that joined forces with
the Flemish Christian democrats (CD&V) between 2004 and 2008 (e.g. van Haute, 2011). After
three years of joint conservative opposition against the progressive Verhofstadt II government
(liberals and social democrats), the cartel of CD&V and N-VA won the 2007 elections. It
marked the beginning of a long-lasting period of ethno-territorial negotiations. Against the
background of the N-VA’s continuous uprise, tensions and gridlocks on ethno-territorial issues
with great symbolical value resulted in two record-breaking government negotiations (194 days
in 2007 and 541 days in 2010-2011) (for an analysis, see M. Hooghe, 2012).
One of the most contentious issues at stake was that of the Brussels-Halle-Vilvoorde electoral
district (BHV), which not only encompassed the 19 municipalities of the bilingual Brussels-
Capital language area but also spanned 35 municipalities of the Dutch-speaking language area.
Most Flemish parties saw it as a vehicle for the spread of French in the Flemish towns bordering
Brussels, as the Francophones in these towns could vote for famous Francophone candidates in
the capital (at least for federal and EU elections) (Goossens & Cannoot, 2015, p. 35).
56
As the
Flemish demand for the split of BHV was met with consistent Francophone resistance, the issue
turned into the ‘Voeren of the 21st century’ (Devos & Bouteca, 2008). The strategy of certain
Flemish parties to opt for a majoritarian logic (first in discourse, later in action) put the
consociational Belgian framework to the test (Sinardet, 2010). The result was traumatizing
gridlock.
Apart from settling the issue of BHV (the electoral district was split, except for the six facility
communities surrounding Brussels), the 2011 Butterfly Agreement transformed the Belgian
56
While BHV (although not in that name) is as old as the Belgian state itself, it only became the subject of
political conflict in the 1960s and, more particularly, due to the 1962-1963 language agreement (Sinardet,
2010, p. 356). However, until the early 2000s, the topic was hardly salient (outside of the Flemish
Movement). It was only when the Constitutional Court ruled that the 2002 electoral reform (which introduced
provincial constituencies, with BHV as the sole exception) was unconstitutional, that ‘the formerly un-sexy
issue turned into a mobilization theme’; ‘the hottest issue’ of Belgian politics (Devos & Bouteca, 2008, pp.
1, 5).
77
federation in several ways (for an overview, see e.g. Goossens, 2017; Goossens & Cannoot,
2015; Sautois & Uyttendaele, 2013).
57
Amongst others, the fiscal autonomy of the Regions was
significantly boosted (Decoster & Sas, 2013, pp. 5-6), an unprecedented reform degraded the
Senate into an indirectly elected and non-permanent body of little legislative and political
relevance, a principal agreement was struck to introduce coinciding (EU, federal, and sub-state)
elections in the future, and a range of powers were transferred from the federal to the sub-state
level – including elements in the field of justice and social security (family allowances). The
latter marks a historical evolution: while Francophone and left-wing parties had always opposed
any split of social security issues, this holy principle was now breached (Béland & Lecours,
2018; Bouteca & Devos, 2014, p. 293).
After the sixth state reform, the ethno-territorial rollercoaster was temporarily halted.
With the traumatizing debates on BHV and the sixth state reform in mind, there was little
political appetite to kick off a new round of negotiations. Even the Flemish nationalist N-VA,
by now the largest party in Flanders and Belgium, agreed to put its ethno-territorial agenda ‘on
hold’ when entering the federal Michel I government (2014-2018) (Benoît Rihoux et al., 2016,
p. 30).
58
Recently, however, and against the background of troublesome power-sharing
experiences, calls for a seventh state reform are getting louder. The search for lasting stability
and a more functional power-sharing regime has triggered a range of initiatives. As noted in
Chapter 1, the current De Croo cabinet has promised to prepare a seventh state reform that
would turn Belgium into a ‘country of cooperation and respect’ (De Croo & Magnette, 2020,
pp. 7-8), an online ‘Dialogue platform’ was established to gain input from ordinary citizens (De
Clercq, 2023), federal departments are studying the institutional architecture of the country, and
57
The compromise on the BHV constituency can be summarized as follows (Goossens, 2017, p. 323;
Goossens & Cannoot, 2015, pp. 35-36). The former electoral district of BHV was split into two new
constituencies: the province of Flemish-Brabant (consisting of the former communal constituencies Halle-
Vilvoorde and Leuven) and a Brussels-Capital constituency. However, French-speaking citizens in the six
facility communities surrounding Brussels obtained so-called ‘special modalities’: they can choose to vote
for candidates of the Flemish-Brabant constituency or (Francophone) candidates of the Brussels-Capital
constituency. Related to this compromise, agreements were reached on the BHV judicial district and the
settlement of disputes regarding the application of the language laws in the six facility municipalities
surrounding Brussels (for an overview, see e.g. Goossens & Cannoot, 2015, pp. 35-38).
58
Michel I was a coalition of the Flemish nationalist N-VA, the Flemish liberals (Open VLD), the Flemish
Christian democrats (CD&V), and the Francophone liberals (MR).
78
different parliaments have invited scholars and stakeholders to evaluate the Belgian federal
system.
59
The debate on power-sharing in Belgium is alive and kicking and the need for change
is widely acknowledged.
Far less consensus exists on the preferred nature of the reforms. Some parties adhere to the
idea that decentralization can pacify tensions. This position is most prominently stressed by the
Flemish nationalist demands for a so-called confederal model (N-VA) or even Flemish
independence (VB) (Pascolo et al., 2021). Others blame the past state reforms for going too far
and undermining the very incentives for cooperative behaviour. Tellingly, calls to re-strengthen
the Belgian state are on the rise, be it through the re-centralization of certain policy domains,
the introduction of a federal electoral district, or the adoption of a hierarchy of norms (Dodeigne
et al., 2016; Dodeigne et al., 2022; Dodeigne et al., 2015; Pascolo et al., 2021; Sinardet et al.,
2013). Meanwhile, scholars put the decentralization process and its repercussions under
study (e.g. Caluwaerts & Reuchamps, 2015; De Winter & Baudewyns, 2009; De Winter &
Dumont, 2014; M. Hooghe, 2012; Reuchamps, 2013). Scholarly evaluations are often quite
pessimistic, depicting Belgium as e.g. a “federal system without federal loyalty” (M. Hooghe,
2012) or a “disintegrative model” (Popelier, 2021, p. 110). A quick look at the titles used in
studies on Belgian politics suffices to sketch this general state of mind: Is Belgian
consociational federalism digging its own grave? (Caluwaerts & Reuchamps, 2015), Still
consociational? (Caluwaerts & Reuchamps, 2020), Un vecteur d’instabilité? (Sinardet, 2011),
Falling apart together? (Deschouwer, 2002), Will it stay or will it go? (Swenden & Jans, 2009),
are we heading towards the breakdown of a nation-state in the heart of Europe? (De Winter &
Baudewyns, 2009), or shifting from exceptional complexity to regime breakdown? (De Winter
& Dumont, 2021)
Clearly, the debate on decentralization and its impact on segmental power-sharing in Belgium
continues. The aim of this work is to add to our understanding of these dynamics. In doing so,
it is worthwhile to first describe what we already know. Accordingly, the following chapter
59
Federal parliament: Verslagen namens de parlementaire commissie belast met de Evaluatie van de
Staatshervormingen sinds 1970 uitgebracht door de heren Loones (K.) en Calvo (K.), mevrouw Pas (K.), de
heren Eerdekens (S.) en Van Rompuy (S.) en mevrouw Durenne (S.), Parl.St. Senaat 2021-22, nr. 7-280/2-7
en Parl.St. Kamer 2021-22, nr. 55-2602/2-7. // Flemish parliament: Verslag van de hoorzittingen namens de
Werkgroep Institutionele Zaken uitgebracht door Karl Vanlouwe, Peter Van Rompuy en Caroline Gennez
over de staatsstructuur en het toekomstig staatsmodel, Parl.St. Vl.Parl. 2021-22, nr. 1257/1.
79
provides a brief summary of some of the most prominent indicators of the paradox of federalism
(in Belgium).
80
81
Chapter 4.
The paradox of Belgian federalism: a brief overview of
existing indicators
In Belgium as in many countries, the repercussions of decentralization have been subject to an
extensive and expanding range of studies. Focussing on the most prominent indicators, this
chapter provides a brief overview of the resulting picture. Most of what we know concerns the
sub-state challenge. In this respect, existing data and analyses yield mixed conclusions. Support
for the paradox thesis varies depending on the indicator at hand. Our knowledge of the power-
sharing challenge is less developed. Data on the birth and death of power-sharing executives
exist, but our understanding of their actual lifetime is still very limited.
1. Decentralization and the sub-state challenge
Thanks to a wide range of theoretical and empirical works, we already have a solid
view of the nature, pertinence, and evolution of the sub-state challenge in post-war
Belgium. For instance, several exhaustive investigations provide a solid view of the regionalist
parties, their emergence, strength and political impact (e.g. Buelens & Van Dyck, 1998; De
Winter, 1998b; Delwit & van Haute, 2021; Govaert, 2002; Pauwels, 2011; Pilet, 2005; van
Haute, 2005, 2011; van Haute & Pilet, 2006; Wauters, 2005). Meanwhile, different datasets
provide a solid indication of the positions of parties on the decentralization issue (e.g. the
Chapel Hill Expert Surveys; Bakker et al., 2015; Polk et al., 2017), and of the saliency of this
topic within their electoral manifestos (the MARPOR data; Volkens et al., 2019). Moving
beyond parties, scholars have recently tapped into the sub-state identities of MPs and their
preferences on the future of the Belgian federation (e.g. Dodeigne et al., 2016; Dodeigne et al.,
2022; Dodeigne et al., 2015; Sinardet et al., 2013), while an even more elaborate range of
studies has exposed the presence of sub-state identities amongst citizens, the level and evolution
of public support for autonomy or independence, citizen arguments towards (de)centralization,
and the (limited) relevance of the state reform issue in explaining voting behaviour (e.g.
De Winter, 2007; Deprez & Vos, 1998, 1999; Deschouwer, De Winter, Reuchamps, Sinardet,
& Dodeigne, 2015; Deschouwer & Sinardet, 2010; Maddens, Beerten, & Billiet, 1994;
Reuchamps, Boerjan, Niessen, & Randour, 2021; Swyngedouw, Abts, Baute, Galle, &
Meuleman, 2015; Swyngedouw, Abts, & Galle, 2014; Swyngedouw & Rink, 2008).
Throughout the following paragraphs, I provide a brief overview of some of the most
82
prominent indicators. This summary not only shows that we already have a quite solid view
of the sub-state challenge and the way it evolved against the background of decentralization. It
also illustrates that there is no such thing as “the” paradox. The resulting picture is rather blurry,
with different indicators yielding different results. Support for the paradox thesis varies
depending on the indicator at hand. Adding to the nuance, clear-cut conclusions are further
complicated by changing trends over time.
1.1. Regionalist parties’ strength and positions: a paradox after all?
For years, decentralization was considered to be very successful in undermining
regionalist parties in Belgium. As their electoral strength eroded against the background of
the decentralization reforms they asked for, these parties were long considered to be “victims
of their own success” (De Winter, 1998b; Deschouwer, 2009; Pilet, 2005; van Haute, 2005; van
Haute & Pilet, 2006).
60
In the words of Newman (1995), they were winning the policy wars but
losing the electoral battles. As their raison d’être was largely taken away, some of these parties
simply faded away (the Rassemblement Wallon, RW). Others attempted to avoid irrelevance by
raising demands. Most notably, the Volksunie (VU) shifted from federalism to confederalism
(late 1970s/early 1980s) and even embraced separatism (late 1990s) (van Haute & Pilet, 2006,
p. 306). As such, pacification (electoral decline) and escalation (regionalist radicalization) went
hand in hand.
In recent decades, and adding to this ideological radicalization, the (Flemish) regionalists
experienced an electoral renaissance. Indeed, as Figure 3 shows, their decline has been altered
by the skyrocketing scores of the N-VA and the (re-)emergence of the VB. In 2010, and for the
first time since 1974, these parties passed the 20%-ceiling. Less than a decade years later, they
stand historically strong (30% of all seats). The “rebirth of Flemish nationalism” (Rochtus,
2012) is even more prominent within the Flemish parliament, where the two Flemish nationalist
parties more than doubled their number of seats since the first direct election (26/124 seats in
1995) and came close to obtaining a majority at the most recent election (58/124 seats in 2019).
Meanwhile, the regionalist FDF scored “extremely well” in Brussels, where it aims to defend
the interests of Francophone residents (Delcorps, Dujardin, & Maingain, 2014; van Haute,
2016). The idea that regionalist parties were (electoral) victims of their (policy) success made
room for reflections on their renewed success (Baudewyns, Dandoy, & Reuchamps, 2015; M.
60
This decline did not hit the separatist VB (Vlaams Blok/Vlaams Belang).
83
Hooghe & Stiers, 2022; Rochtus, 2012; van Haute, 2016, 2021). These dynamics strongly fed
the support for the paradox thesis in public, political, and scholarly reflections alike (e.g.
Caluwaerts & Reuchamps, 2015; Huysseune, 2017; Reuchamps, 2013).
Figure 3. General election scores of regionalist parties in Belgium
(% of seats in the Chamber of Representatives; 1949-2019)
Note 1: Seat shares are preferred over vote shares. This avoids distortion due to the 1995 reform
(when the number of seats declined from 212 to 150).
Note 2: Flemish parties: Vlaamse Concentratie (VC; 1949), Volksunie (VU; 1954-1999), Nieuw-
Vlaamse Alliantie (N-VA; 2003-2019), Vlaams Blok/Vlaams Belang (VB; 1978-2019). Francophone
parties: Rassemblement Wallon (RW; 1968-1985), Front démocratique des francophones/
Fédéralistes Démocrates Francophones (FDF; 1965-2015); Démocrate Fédéraliste Indépendant
(Défi; 2015-2019).
Sources: Parlgov (Döring & Manow, 2020) and Federal Public Services Home Affairs (2019).
Exceptions: (1) Blaise, de Coorebyter, and Faniel (2007, p. 25) for the for the 2007 result for the N-
VA (which contested elections in cartel with CD&V). (2) FDF/Défi strength after it became part of
the MR (Mouvement Réformateur): based on Delcorps et al. (2014).
1.2. Mainstream parties’ positions: a cautious reversal of the paradox
Throughout the decades, parties in the North and South fundamentally shifted
their positions by (incrementally) raising autonomy demands. For long, the radicalization
of the traditional parties was seen as additional evidence for the paradox thesis (e.g. De Winter
& Baudewyns, 2009; Delwit, 2011b; Deschouwer, 2012; L. Hooghe, 2004). Some mainstream
parties in Flanders even embrace(d) confederalism. While their plea is one for additional sub-
0
5
10
15
20
25
30
35
VC/VU/N-VA FDF/Défi
VB RW
Total regionalist parties
Decentralization deal
84
state autonomy within Belgium (not one for an actual confederal an association of sovereign
states established by means of an international treaty; e.g. Romainville, 2015), the very usage
of this terminology is itself indicative of the political atmosphere. Although party positions on
decentralization can be explained by many factors (Toubeau & Wagner, 2015), the pressure of
party competition is generally considered to be one of their most prominent driving forces. The
radicalization of the Belgian mainstream parties is clearly related to the presence of regionalist
challengers. Attempting to defuse the (electoral) threat posed by the latter, mainstream parties
embraced part of their programme. In this respect, Alonso (2012, p. 37) talks about “pro-
periphery convergence”. Within the PSO theory of party competition (Position-Salience-
Ownership) of Meguid (2005, 2008), this tactic is known as the “accommodative strategy”.
61
Such dynamics of party competition were long blamed for feeding ‘ethnic outbidding’
dynamics: an inflation of ethno-territorial claims, radicalism, and intransigence (Rabushka &
Shepsle, 1972) (see also Chapter 5).
However, and contradicting the paradox thesis, this seemingly unstoppable inflation of
demands is now being countered by a tendency of ideological pacification. The case against
so-called ‘dysfunctional decentralization’ is gaining popularity, and a growing set of parties
and politicians now favours a re-strengthening of the Belgian state, e.g. through the re-
centralization of certain policy domains (Dodeigne et al., 2016; Dodeigne et al., 2022; Dodeigne
et al., 2015; Pascolo et al., 2021; Sinardet et al., 2013).
1.3. Public identification, support for independence, and autonomy: the
paradox of indifference
Interestingly, decentralization did not come with eroding levels of public
identification with the Belgian polity or growing levels of support for sub-state independence
(for a solid overview, see e.g. Caluwaerts & Reuchamps, 2015; Dalle Mulle, 2018, Fig. 9.1;
Deschouwer et al., 2015; Reuchamps, 2017; Swyngedouw et al., 2015). Arguably, the best-
known work on identities in Belgium is provided by the Institute for Social and Political
61
According to the PSO theory (Position-Salience-Ownership), mainstream parties can react to challenging
niche parties in three ways: (1) ignore the niche party in an attempt to stifle the debate (dismissive strategy),
(2) adopt a position similar to that of the niche party (accommodative strategy), or (3) adopt a contrasting
position on the issue (adversarial strategy) (Meguid, 2005, 2008). The latter resembles the “anti-periphery
polarization” strategy described by Alonso (2012, p. 37).
85
Opinion Research/Pole Interuniversitaire Opinion publique et Politique (ISPO/PIOP) (for a
background, see ISPO, 2022). The percentage of voters identifying itself primarily with
Belgium evolved as follows: 52% (1995 elections), 56% (1999 elections), 54% (2003
elections), 52% (2007 elections), 45% (2010 elections), and 56% (2014 elections). For primary
identification with Flanders, this is respectively 25%, 27%, 30%, 34%, 29%, and 28%.
Exposing the relevance of dual identities, the bulk of voters (79.7% in 2014) feel both Flemish
and Belgian (to varying degrees; 38.7% put them on an equal level).
62
Segmental identities are
not only subordinate to state-wide or dual identities. According to some assessments, identity
might also not be the most important element in explaining the ethno-territorial views of
citizens. A recent thematic analysis of three citizen forums on Belgian federalism (organized in
2017-2018) indicates that citizens make more references to efficiency arguments than to
identity considerations when legitimizing their view on (de)centralization (Reuchamps et al.,
2021).
Meanwhile, support for independence has been stably marginal for decades, fluctuating
between 5 to 10% (depending on the surveys/data) (Swyngedouw et al., 2015).
63
At the 2003
general elections, that is after five major state reforms, independence was only supported by
2.5% of Flemish voters. Against the background of tense ethno-territorial negotiations on a
sixth state reform, independence support temporarily grew towards 9.6% (2007) and 11.9%
(2010), but once an agreement on this sixth wave of decentralization was found, it almost halved
(polling at 6.4% in 2014). If anything, such dynamics are supportive of the thesis that autonomy
provisions can boost the kind of state loyalty that decreases the incentives for secession (see
also Chapter 5) (cf. Hirschman, 1970).
64
Clearly, consecutive waves of far-reaching autonomy
provisions did not result in a separatist tsunami. According to Lecours (2021), it is exactly the
dynamic nature of the decentralization process; the fact that it came in waves (rather than being
62
For an overview and comparison of different measures of territorial identities in Belgium, see e.g.
De Winter (2007). For a brief summary of data on earlier decades, see e.g. Caluwaerts and Reuchamps (2015,
pp. 283-284); Reuchamps (2017).
63
Of course, a more considerable part of voters is in support of more sub-state competences (around 50%
before the sixth state reform and 36.4% in its aftermath) (Swyngedouw et al., 2015).
64
According to Hirschman (1970), there are two types of reactions to unsatisfactory situations in firms,
organizations, and countries: exit (leaving) and voice (speaking up and trying to improve the situation).
Feelings of loyalty, it is thought, can impact people’s response and make them prefer voice over exit.
86
implemented at once) that accounts for the low levels of independentism. Interestingly, the
marginality of separatist sentiments is rather odd, given the recent electoral victories of the
regionalist parties in Flanders. However, this electoral rise was not driven by separatist
sentiments but by other issues (e.g. the party’s socio-economic profile). In 2014, less than 11%
of the N-VA voters supported Flemish independence and even in the tense 2010 elections, only
one in five voters considered it to be one of the three most vote-determining issues
(Swyngedouw et al., 2015). Accordingly, the true paradox is not that decentralization boosted
separatist tendencies, but what Huysseune (2017) calls the “Flemish paradox”: the
“hegemony of pro-independence parties in a region largely indifferent towards independence”.
The low levels of support for independence sharply contrast with similar cases like
Catalonia or Scotland, where support rates fluctuated between 20%-35% for decades (Dalle
Mulle, 2018; Figure 9.1). Regarding the core challenge at stake, Belgium differs from countries
like Spain, the United Kingdom, and Canada. The main threat to the stability and survival of
these states is primarily a regionalist one. The struggle they face is one between the central state
and a self-confident periphery that aims for further autonomy or even separation (most notably
the Basque Country and Catalonia in Spain, Scotland and Wales in the UK, and Quebec in
Canada).
65
This is not the case in Belgium, where the most prominent threat to the state is the
tension at the central level itself. Especially in recent decades, it is arguably not the challenge
from below (the periphery) that lies at the heart of the debate, but rather the challenge at the top
(the centre), where both segments struggle to cooperate and where power-sharing often takes
the form of deadlock and conflict. In other words, debates on Belgium’s chances of survival do
not (yet) centre around the direct threat of regionalism (will Flanders go its own way?), but
rather concern the problem of ‘governability’ (can we still do it together?).
66
It is this power-
65
In Northern Ireland, both debates prevail and are inherently connected. On the one hand, there is the sub-
state challenge of Irish nationalists seeking an exit from the United Kingdom and unification with the
Republic of Ireland. On the other hand, political struggles in contemporary Northern Ireland also take the
form of a power-sharing challenge; the challenge of power-sharing and cooperation between its two
constitutive segments (i.e. Catholics/republicans and Protestants/unionists). For a background, see e.g.
(Ferriter, 2019); McKittrick and McVea (2012).
66
Of course, the latter can feed the former and the debate might shift in the future. Indeed, recent polls
indicate that the Flemish nationalists might obtain an absolute majority of seats in the 2024 Flemish elections
(IPSOS, 2023). However, and especially given its linkage to the power-sharing challenge; this would not
make the latter less prominent (but rather add to it).
87
sharing challenge; this challenge of conflict and cooperation between Flemish and Francophone
politicians that seems to be the nub of the problem. But strikingly, and as the following section
described, our understanding of this challenge is still very limited.
2. Decentralization and the power-sharing challenge
The importance of the power-sharing challenge sharply contrasts with the limited
attention it has received (for Belgium, see e.g. Caluwaerts & Reuchamps, 2015; De Winter &
Baudewyns, 2009; Deschouwer, 2002, 2006; L. Hooghe, 1993, 2004; Huyse, 1981). Until this
day, profound and systematic analyses of this conflict’s prevalence and evolution in different
polities are yet to be made. The following paragraphs provide a general summary of some of
the existing indicators, which mainly shed light on the birthing and dying phases of power-
sharing executives (rather than their day-to-day life).
2.1. The birth of power-sharing executives
Of all indicators used to prove that Belgium is prone to paradox-like dynamics, the most
popular one is arguably the duration of cabinet formations. Since the early 2000s, the birthing
process of power-sharing cabinets has been especially painful. In line with the paradox thesis,
the duration of cabinet negotiations clearly increased against the background of decentralization
(Figure 4). Belgium made international headlines due to the traumatizing formation crises
following the elections of 2007 (194 days), 2010 (541 days), and 2019 (306 days).
67
Ethno-
territorial tensions have a large share in these dynamics. When ethno-territorial issues are at
stake, negotiation rounds take longer and more formation attempts are needed (De Winter &
Dumont, 2014, p. 245). These tense negotiations translate into remarkably long periods of
caretaker cabinets. Between 2007 and 2020, Belgium was governed by a caretaker cabinet for
over four years (1485 days) (Dandoy & Terrière, 2021, p. 123). From a comparative
perspective, this situation is highly exceptional. Belgian cabinet negotiations are characterized
by the largest number of failed attempts of all West-European countries (De Winter & Dumont,
67
For a reflection on these crises, which were all burdened by ethno-territorial debates, see e.g. M. Hooghe
(2012); Benoit Rihoux, Vandeleene, De Winter, Baudewyns, and Deruette (2020, 2021); Swenden (2013a).
88
2021, p. 115) and while the duration of negotiations is also rising elsewhere, no other country
is burdened by an increase as spectacular as the one noted in Belgium (Louwerse & Van Aelst,
2013, p. 14). Unsurprisingly, references to the increasingly difficult coalition formations are
particularly salient in scholarly writings on the (detrimental) effects of decentralization.
Arguably, they even lie at the basis of most recent works on the topic.
2.2. The death of power-sharing executives
A very different picture arises when we consider data on the duration of cabinets and
the frequency of early cabinet terminations due to ethno-territorial conflicts. Indeed, the
pessimistic conclusions on the troublesome birth of coalitions stand in sharp contrast with the
observation of increasing life expectancies and the decreasing number of ethno-territorial
fatalities.
Initially, the emergence of the ethno-territorial debate to the forefront of Belgian politics, which
accelerated throughout the 1960s, was followed by a troublesome era of instability and
extremely low cabinet duration levels. Between June 1968 and December 1981, Belgium was
governed by 13 cabinets in 13 years. The most notorious era of cabinet turnover was the 1977-
1981 period, which is commonly known as a period of “malgoverno” in Belgian political
history (Heylen & Van Hecke, 2008). Although the regularity of early cabinet terminations
continued, this tense era of short-lived cabinets was followed by a period of somewhat greater
0
100
200
300
400
500
600
1946
1949
1950
1954
1958
1961
1965
1968
1971
1974
1977
1978
1981
1985
1987
1991
1995
1999
2003
2007
2010
2014
2019
Days
Election year
Figure 4. Post-electoral cabinet formation duration (days)
in Belgium (1946-2019)
Decentralization deal
Source: De Winter and Dumont (2014), own data for 2019
89
stability (at least in terms of cabinet duration), with seven cabinets governing Belgium in the
almost eighteen years between December 1981 and July 1999 (Dumont & De Winter, 1999, pp.
48-49).
68
Since then, and showing that this modest stability continued, eight coalitions were in
charge until the resignation of Michel I in December 2018 (the end of this work’s focus). In
terms of the life expectancy of power-sharing executives, no paradox-like dynamic is
found.
Meanwhile, the number of cabinet terminations due to ethno-territorial tensions clearly
declined. All elections after 1965 were early elections until 1995 (Deschouwer, 2013a, p. 213).
Many of these were the direct result of ethno-territorial disputes, which were long “extremely
effective coalition killers” (Deschouwer, 1994, p. 46). In fact, most of the early cabinet
terminations in the 1946-1999 period were either the result of ethno-territorial tensions or at the
very least linked to such debates (Dumont et al., 2001, p. 21). In recent decades, such clashes
were far less fatal. Ethno-territorial conflicts only killed one of the eight coalitions established
between the elections of 1999 and 2019 (namely Leterme II due to the Brussels-Halle-
Vilvoorde issue; cf. Blero, 2015; Devos & Bouteca, 2008). Hence, and refuting the paradox
thesis too, the coalition-killing impact of ethno-territorial debates was greater in the past.
2.3. The life of power-sharing executives
Summarizing the previous sections, while it is more difficult to establish power-sharing
executives (increasing formation duration), their life expectancy increased (increasing cabinet
duration, fewer ethno-territorial fatalities). Put differently, it is harder to mount the horse but
easier to stay in the saddle. Of course, good riders are not only recognized by their ability to
mount horses or avoid falling. The most important question is how smoothly they ride. And in
this respect, our knowledge is rather limited.
Lacking firm quantitative indicators, many authors base their reflections on general, broad
historical overviews. Typically, summaries like this refer to the most memorable conflicts,
deadlocks, or protests (such as the ones included in the very valuable overview of ‘major
68
Some scholars argue that the increased cabinet stability in the following period was not a result of ethno-
linguistic pacification but rather a reflection of parties’ fear of long-lasting cabinet formation negotiations,
their economic repercussions, and the possibility of not being invited to such future negotiations (Dumont &
De Winter, 1999, pp. 48-49).
90
instances of disruptive protest concerning territorial conflict’ provided by L. Hooghe, 2004, pp.
63-64). Chapter 3 introduced readers to this history and such major events in Belgium. An in-
depth historical assessment of the questions under study here would surely be of great
relevance. But sporadic references to significant historical events in constrained documents like
book chapters or papers are no such assessment. They do not provide a sufficiently strong and
reliable basis for empirical claims on the conflictual nature of segmental power-sharing and its
evolution. What is needed is a systematic mapping of conflicts, their features, and their
(potential) relation with decentralization.
Several works have already focused on cabinet conflicts and conflict endings (e.g. Luebbert,
1984; Luebbert, 1986; Moury & Timmermans, 2013; Nousiainen, 1993; Timmermans, 2006).
However, the resulting empirical picture is still quite limited. For example, Timmermans and
Moury (2006, p. 396), expose 18 cabinet conflicts in the three Belgian coalitions between 1992
and 2003 and conclude that “the recent institutional arrangements of federalization, with their
transfer of competencies on a number of economic and cultural matters to the regions, seems to
pay off in national coalition politics.”
69
Similarly, but covering the 1961-1987 period,
Nousiainen (1993, pp. 270-271) exposed 37 conflicts in Belgian cabinets. A large minority of
38% of these were about the “language and nationality problem”, as he calls it.
To truly draw conclusions on the impact of state reform in a given polity, longer periods of
time (covering more decentralization waves) need to be studied. Also, existing studies expose
rather limited numbers of clashes, which indicates that only the most salient crises were
covered. This leaves us blind to the small-scale tensions that truly grasp the everyday power-
sharing spirit of politicians. The small-N problem also prevents us from running the kind of
statistical analyses that help us grasp the role of decentralization (if any) while controlling for
other factors. Furthermore, and as elaborated upon in Chapters 2 and 6, there is still much room
for improvement when it comes to the conceptualization and operationalization of cabinet
conflicts (Vandenberghe, 2022b, pp. 116-117). In short, to truly assess whether the paradox of
federalism manifested itself, more extensive and fine-grained analyses are needed. This work
aims to provide exactly that. Before addressing these analyses and the data they build on (Part
III, the empirical part), I present the effects that are expected according to existing literature.
69
The analysis covers three cabinets: Dehaene I, Dehaene II, and Verhofstadt I. Timmermans and Moury
(2006, p. 397) found 2 clashes about linguistic issues, while 2 other conflicts revolved around institutional
debates (a category that does not appear to cover the debates on state reform covered in this work).
91
As the discussion presented in the following chapter shows, the debate on the repercussions of
decentralization is far from settled.
92
93
Part II.
Theoretical framework
94
95
Chapter 5.
Decentralization, conflict prevalence, and conflict endings:
A theoretical framework
70
Scholars have long debated the effect (if any) of decentralization on ethno-territorial conflict
and conflict endings. This chapter provides a discussion of the theories at hand and the
contrasting conclusions that characterize the literature. First, the hypotheses on the
repercussions of decentralization are addressed. Second, I tap into the relevance of contextual
factors. Specifically, I explain why certain (static) features make the Belgian federation
particularly conflict-prone and I take into account the role of issue saliency, regionalist and
electoral pressures. Four variables allow me to do so: (1) regionalist party strength, (2)
regionalist party cabinet participation, (3) the saliency of the decentralization issue, and (4)
the proximity of federal elections. This chapter ends with an elaboration on their expected role
and a visualization of the hypotheses under study.
1. Decentralization and conflict (outcomes)
1.1. Pacification
Decentralization is often embraced by policymakers and scholars alike for its presumed
ability to de-escalate tensions and abate secessionist dynamics (e.g. L. Anderson, 2016;
Bermeo, 2002; Keil & Anderson, 2018; Lijphart, 1981c; Stepan, 1999). To begin with, it can
satisfy the autonomy demands of significant segments of the population and regionalist
movements (many of which do not per se embrace secessionism).
71
This is one reason why
70
The arguments presented here were previously described (often less elaboratively) in a publication in
Ethnopolitics (Vandenberghe, 2022c) and are expected to be covered in two articles that are under review at
the time of publication (Vandenberghe, 2022d, 2023).
71
Support for independence is not even shared by all members of the regionalist party family (insofar as they
can be called a family; cf. ). Many regionalist parties are not pro-independence but rather in favour of sub-
state autonomy or cultural protection within the existing state (e.g. De Winter & Türsan, 1998; Massetti &
96
decentralization is often argued to hamper regionalist parties’ electoral strength; the so-
called ‘accommodation thesis’ (Massetti & Schakel, 2016, 2017; see also Rudolph &
Thompson, 1985). For example, and as discussed in Chapter 4, regionalist parties in Belgium
have long been considered “victims of their own success” (De Winter, 1998b; Deschouwer,
2009; Pilet, 2005; van Haute, 2005; van Haute & Pilet, 2006), because the consecutive
decentralization reforms eroded their raison d’être. In Wallonia, regionalist parties appear to
have perished for good.
72
Electoral declines like these should come with less tension at the
federal level, because bluntly stated, regionalist parties are territorial agitators. As explained
more elaboratively below, they push territorial issues to the top of the political agenda, they
stress ethnic divisions, and they can trigger dynamics of ethnic outbidding (Horowitz, 2000;
Meguid, 2005; Müller-Rommel, 1998; Rabushka & Shepsle, 1972).
Meanwhile and related, decentralization allows ethno-territorial groups to protect their
culture and interests, and avoids their marginalization, exploitation, or elimination by the
central government. For instance, Bermeo (2002) uses the “Minorities at Risk” project data
(MAR) (Gurr, 1993) to demonstrate that the levels of political discrimination, economic
discrimination, and minority grievances concerning political and economic rights are lower in
federal regimes than in unitary regimes.
73
Sub-state autonomy also provides an important check
on the central government. As such, it takes away or at least lessens minority segments’ fear
of being overruled or even weeded out. Doing so can mitigate the so-called security dilemma
that can nourish ethnic conflicts and encourage the tendency of segmental or sectarian groups
to turn violent (Lake & Rothchild, 1996; Posen, 1993; W. Rose, 2000). Adding to this, as Keil
and Anderson (2018, p. 94) phrase it, self-rule also offers sub-state groups “legitimate channels
to articulate their grievances”. It allows their concerns and frustrations to be voiced through
formal procedures and institutions (e.g. sub-state parliaments), thus avoiding attempts to be
heard by causing disruption (e.g. protest or rebellion). The very establishment of such
Schakel, 2016; O. Mazzoleni & Mueller, 2016).
72
As discussed in Chapter 4, this dynamic stopped in recent times. Scholars are now talking about the
renewed success of the regionalist parties (van Haute, 2016).
73
Although this relationship is sometimes challenged (e.g. Hale, 2000), the victimization, exploitation, and
marginalization of regions are typically considered to be a key factor behind autonomist tendencies (Rudolph
& Thompson, 1985) and recent discourse analyses have shown that regionalist parties in Western Europe
strongly and deliberately stress the victimization and marginalization of their region in an attempt to gain
support (Abts et al., 2019; Dalle Mulle, 2016, 2018; Dalle Mulle & Serrano, 2019).
97
institutions also serves an important symbolical purpose. In many states, formally
recognizing the existence and relevance of sub-state groups and their identities is a big deal. Its
significance, both emotionally and legally, should not be underestimated.
74
“By restructuring the state to accommodate this identity, the state explicitly
recognises it as legitimate and moves from viewing it as something to be
repressed, to framing it as officially accepted, a reality which informs the
character of the state.” (Walsh, 2018, p. 7)
Such dynamics are in line with the widely debated politics of recognition (Gutmann, 1994;
Taylor & Gutmann, 1992) that are often discussed in the context of migration-based
multiculturalism too. In his recent discussion of multinational federalism, Alain-G. Gagnon
(2021, p. 12) even talks about its “potential to institutionalize a politics of recognition”. Some
scholars argue that such accommodative policies can nourish feelings of loyalty towards the
state, which further decreases the incentives for exit (secession) (cf. Hirschman, 1970).
75
Meanwhile, the case for decentralization is often made with negative arguments: What is the
alternative? Scholars often point towards the apparent lack of other ways to accommodate
diversity and pacify tensions without breaching the framework of liberal democracy (Kymlicka,
2001a; McGarry & O’Leary, 1994). Also, the refusal to provide substantive autonomy to sub-
state groups may make them pursue secession (or any other goal, we might add) “in less happy
ways” (Tierney, 2009, p. 251). For these and other reasons, autonomy provisions have often
been praised for their ability to abate armed rebellion or other violent forms of ethno-territorial
74
To give just one example, once autonomy is provided to peoples, revoking it might violate certain
international treaties. For instance, according to the (then) UN Independent Expert on the promotion of a
democratic and equitable international order (Prof. dr. Alfred de Zayas), the decision of the Spanish
authorities to suspend the autonomy of Catalonia following the 2017 referendum on independence was at
odds with the existing international framework: “Denying a people the right to express themselves on the
issue of self-determination, denying the legality of a referendum, using force to prevent the holding of a
referendum, and cancelling the limited autonomy of a people by way of punishment constitutes a violation of
Article 1 of the ICCPR and of the International Covenant on Economic, Social and Cultural Rights.” (United
Nations, 2017; emphasis added).
75
Hirschman (1970) famously argued that there are two types of reactions to unsatisfactory situations in
one’s firm, organization, or country: exit (leaving without attempting to remedy the flaws) and voice
(speaking up and trying to improve the situation). The idea is that loyalty can impact people’s response,
causing them to opt for voice rather than exit.
98
conflict. Indeed, as demonstrated by examples across the globe (Ghai, 2000, pp. 496-497) and
confirmed by certain large-N empirical analyses of deadly conflicts of varying sizes (e.g.
Bermeo, 2002), the perspective or adoption of self-rule provisions often brought de-
escalation.
76
These arguments mostly concern the impact of decentralization on the sub-state challenge; the
way autonomy provisions can temper regionalist demands and the pressure from the periphery
more generally. But in several ways, decentralization can also be expected to pacify power-
sharing tensions. To begin with, reducing the regionalist threat can have a tempering effect
in itself, amongst others because it might decrease the saliency of ethno-territorial debates and
establish the kind of trust that can facilitate inter-segmental cooperation. But providing
autonomy can also facilitate power-sharing in other ways.
First, and most basically, decentralization diminishes the number of issues the federal
government must address, thus leaving federal politicians with less matter to fight about. As
Swenden and Jans (2009, p. 888) put it when reflecting on Belgium: “the emptying of
distributive prerogatives at the federal level, […] left the linguistic communities with little to
disagree about at the federal level.”
Second and related, decentralization can lift the need to share power when it becomes too
difficult. It ‘can remove multiethnic tensions from the politics of the center’ (Swenden, 2006,
p. 288) and “prevents ethnic groups from fighting each other over what they perceive as unfair
treatment by another group” (Brancati, 2006, p. 656). In practice, and as the historical summary
provided in Chapter 3 shows, competences are often transferred from the federal policy level
to the sub-states exactly because policymakers consider them to be too prone to ethno-territorial
tensions. When autonomy is provided, “the obligation to find one single and common policy is
taken away and therefore the chances of gridlock are substantially reduced.” (Deschouwer &
76
Examples range from the Israeli-Palestinian conflict and the Bougainville rebellion in Papua New Guinea
to the violent struggle of the Tamil Tigers in northeast Sri Lanka (Ghai, 2000, pp. 496-497). Providing a
large-N analysis, Nancy Bermeo (2002) famously assessed the data in the Minorities at Risk project (MAR)
(Gurr, 1993) and concluded that the mean armed-rebellion score for unitary democracies is about twice the
score for federal democracies. As Bermeo (2002, p. 100) states, this result only arises when India (an outlier)
is excluded from the analysis: “The divergent Indian case deserves a great deal more attention, but otherwise
the differences are dramatic.”
99
Temmerman, 2012, p. 501)
77
When cooperation bogs down into conflict, decentralization can
bring pacification.
As Dalle Mulle (2018) vividly describes in his study of regionalist parties in Catalonia,
Flanders, Northern Italy, and Scotland, this line of reasoning is a major part of these parties’
discourse (see also Abts et al., 2019; Dalle Mulle, 2016; Dalle Mulle & Serrano, 2019). Using
an instrumental rather than a traditional argument for decentralization (or even
secession), they often point out the (perceived) indecisiveness and inefficiency of the central
state, which is argued to be the result of cultural differences and opposing policy views. As
noted in Chapter 2, it is well-known that regionalists movements often link the centre-periphery
divide to the ideological left-right divide, e.g. by talking about ‘right-wing Flanders vs. left-
wing Wallonia’ or ‘progressive Scotland vs. conservative England’ (Béland & Lecours, 2016;
Billiet et al., 2015; Massetti & Schakel, 2015).
78
These divisions are typically invoked as an
argument for sub-state autonomy, as it would be less difficult to reach consensus and meet the
demands and priorities of citizens at the more homogeneous sub-state level. However,
regionalists are not alone in stressing this rationale. The idea that decentralization can foster
pacification by separating the fighting dogs also prevails in other spheres. Many observers
and scholars have expressed similar views and the idea that providing segmental autonomy is
key to the political stability of divided states also lies at the heart of consociational theory (e.g.
Lijphart, 1968a, 1969a, 1977a, 2002, 2004). Relatedly, economists have proposed the well-
known trade-off between the scale (size) of a state and the heterogeneity cost (the
difficulties resulting from divergent preferences) (Alesina & Spolaore, 1997).
79
This very trade-
off has been stressed by regionalists, including Bart De Wever, the chairman of the Flemish
77
The comment of Deschouwer and Temmerman (2012, p. 501) referred to the autonomy provisions that
characterize consociational democracy, but the same logic evidently applies to decentralization in non-
consociational contexts.
78
Several studies have indicated that the political values and preferences of sub-groups in Belgium show a
larger overlap than often presumed (for Belgium, e.g. Billiet, 2011; Billiet et al., 2006; Walgrave, Lefevere,
et al., 2019). For a general reflection on the left-right divisions between Flanders and Wallonia, see e.g. Billiet
et al. (2015); Bruno De Wever (2011).
79
In a similar vein, scholars have argued that the very process of nation-building in the first half of the 19th
century can be seen as a result of the trade-off between the benefits of market sizes and the costs of population
heterogeneity (e.g. Alesina, Spolaore, & Wacziarg, 2000).
100
nationalist party N-VA (Dalle Mulle, 2016, 2018). In a 2007 TV interview, to give just one
example, he stated:
‘In my opinion, you must always balance the scale and the cost of heterogeneity. […]
Belgium is a unique case. It is a very small country. Doing things together comes
with little economies of scale. […] But the cost of heterogeneity, that is the price you
must pay to bring different visions together, is very very high. Because we are two
democracies. And to me, that is the death sentence of this country in the 21st century.’
("De keien van de Wetstraat," 2007)
Regionalists are not alone in embracing the baseline behind this reasoning. Many mainstream
parties embrace it too. Discussing contemporary debates in Belgium, some scholars have even
argued that the “recurring incongruities form the major argument used by the Flemings, who
claim more competences for the regions, supposed to be more internally homogeneous in their
priorities and goals.” (Walgrave, Zicha, Hardy, Joly, & Van Assche, 2014, pp. 145-146;
emphasis added). They continue:
“Flemish parties state that it has become increasingly difficult to strike a
national compromise between Flemish speakers and Francophones because the
two groups care about different issues; furthermore, even when they care about
the same issues, they have different opinions on how to solve them.”
Third, decentralization increases the visibility of intra-segmental divisions, that is the
divisions between parties and politicians belonging to the same segment (e.g. disputes amongst
Flemings). In this respect, decentralization allows for a shift in the nature of debates. Federal
discussions about the optimal distribution of competences and power between policy levels
(Who should decide?) can be turned into debates about the actual content of policy at the sub-
state level (What should be decided?). In other words, decentralization can activate or
accentuate lower-level cleavages and “reduce the energy expended at the center in inter-ethnic
cleavages” (Horowitz, 2000, pp. 618-619). For evident reasons, it is harder to scapegoat other
segments for sub-state tensions than for disputes at the power-sharing level. For example, it is
easy to blame the Flemish parties for federal gridlocks, but it is hard to blame them
(convincingly) for a policy conflict within the Walloon cabinet. Meanwhile, such debates and
disputes at the sub-state level undermine the idea of segments being harmonious or
homogeneous entities. This dynamic is particularly persistent in homogeneous federations like
Belgium, where sub-state borders tend to overlap with segmental borders (as discussed below)
101
(Swenden, 2006).
80
In these cases, governing a sub-state (e.g. Flanders) implies governing a
segment (e.g. the Flemings), so sub-state politics is fought between politicians of the same
segment (e.g. Flemish politicians), and sub-state tensions are tensions within segments.
Decentralization increases the prominence of sub-state politics, and in doing so, it also boosts
the saliency of intra-segmental divisions. After all, media attention strongly follows power (e.g.
Habermas, 1991; Wolfsfeld, 2011). The more powerful a policy level becomes, the greater its
news value (and thus, the more it should be covered).
81
All these dynamics “foster intra-group
competition, at the expense of an exclusive focus on inter-group competition” (Horowitz, 2002,
p. 25). This might provide a valuable counterweight to forms of ethno-territorial determinism
and confrontational behaviour.
Fourth, and more unconventionally, I suspect that decentralization can diminish the number
of clashes at the federal level by allowing sub-states to act as lightning rods. It provides
politicians with an alternative route to fight federal coalition partners. The latter need not be
attacked directly (e.g. by another minister) but can be fought convincingly in an indirect way
by letting sub-state actors charge (e.g. sub-state MPs or cabinet members, whose prominence
increases as decentralization goes on). In partitocratic Belgium (Baudewyns et al., 2022; De
Winter & Dumont, 2006), party leaders have enough control to rigorously organize and
coordinate such charges. Evidently, such proxy warfare is far less disruptive for the federal
coalition. Note that this logic applies to conflicts of all kinds, ranging from small sneers to great
crises. In Belgium, for instance, the far-reaching and typically mediatized ‘conflict of interest’
procedure allows sub-states to block federal legislative procedures (for 120 days in theory)
when the initiative harms its interests.
82
By using such procedures, parties can canalize grave
80
The overlap at hand is not perfect (e.g. the Walloon region includes the German-Speaking Community).
81
To date, however, Belgian media cover federal politics more extensively than sub-state politics. On
Flemish TV news shows, to give just one example, Flemish politicians get far less speaking time than
politicians that are active at the federal level (De Smedt, Hooghe, & Walgrave, 2011; Walgrave & De Swert,
2005).
82
In practice, creative political mathematics allows such conflicts to block legislative procedures for
significantly longer periods (excluding certain days or periods from the calculation can prolong the procedure
for months). For instance, during the crisis on BHV (Brussels-Halle-Vilvoorde), the ‘conflict of interest’
initiated by the Walloon parliament lasted for 285 days, which triggered major criticism in Flemish nationalist
spheres (e.g. Vuye & Wouters, 2017, p. 60).
102
conflicts that would otherwise endanger the very survival of the federal cabinet. Similarly, less
severe attacks on coalition parties are often launched via sub-state politicians. Such dynamics
appear to have been overlooked in previous studies, but they might reduce the number of clashes
at the power-sharing level itself significantly.
Based on these arguments, it is expected that decentralization reforms will result in less ethno-
territorial conflict. Meanwhile, in terms of conflict resolution, pacifying dynamics like this
should have rendered the process of intersegmental conflict accommodation less difficult in the
long run.
H1: Decentralization pacified ethno-territorial tensions
H2: Decentralization facilitated the resolution of ethno-territorial conflicts
Despite the many reasons to expect pacification, the recent literature on Belgium appears to be
particularly paradox-minded (e.g. Caluwaerts & Reuchamps, 2015; De Winter & Baudewyns,
2009). The following section discusses the main arguments on which this scepticism is based.
1.2. Escalation
According to the paradox of federalism, providing (additional) sub-state autonomy is
counterproductive: it exacerbates the very divisions and conflicts it was meant to temper (Erk
& Anderson, 2010b). There are several reasons to expect this logic to apply to power-sharing
tensions too.
To begin with, and in sharp contrast to the logic discussed earlier in this Chapter,
decentralization is sometimes argued to strengthen autonomy demands. After all, the very
existence of segmental autonomy not only stresses ethno-territorial divisions but also explicitly
acknowledges and institutionalizes them. Many authors argue that entrenching or cementing
differences “can lock in place the very identities and factors which minorities claim to make
them different”, “erode any attempts to foster a sense of unity between minority communities
and the host state.” (Keil & Anderson, 2018, p. 99), and feed the kind of banal nationalism that
has been famously discussed by Michael Billig (1995, 2017). As such, every divided state is
prone to the following puzzle:
“the more it recognises and affirms the demand for self-government”, the more
it succeeds “in accommodating national minorities, the more it will strengthen
the sense that these minorities are separate peoples with inherent rights of self-
government, whose participation in the larger country is conditional and
103
revocable.” (Kymlicka, 1998, p. 136)
Decentralization has also been argued to provide regionalist parties with a more fruitful
context for electoral success (Brancati, 2006, 2008, 2009).
83
It “whets the appetite of would-
be secessionists” (Keil & Anderson, 2018, p. 98) and gives these actors valuable resources
(political, administrative, fiscal, and economic resources) that can be used to actively challenge
the centre and boost sub-state nationalism even more (for an early discussion of this criticism,
see Nordlinger, 1972).
84
Accordingly, decentralization can strengthen regional identities,
increase regionalist party support, fuel autonomy demands or separatism, and/or turn secession
into a more viable alternative (e.g. Brancati, 2006, 2008, 2009; Guibernau, 2006; Hale, 2000;
Kymlicka, 2001b, Ch. 5). Power motives add to these dynamics. After all, the search for
pacification is not the only rationale behind decentralization reforms.
85
According to some
scholars, these reforms are rational actions of politicians and parties that try to maximize their
own electoral gains (Meguid, 2009; O'Neill, 2003; Sorens, 2009). As Alonso (2012, p. 8) puts
it rather bluntly: “State parties will not opt for devolution unless they see net electoral gains
resulting from such decision. Devolution is intended to save the threatened state party, not the
state.” From this perspective, it is only logical that actors with a strong electoral position in a
certain sub-state raise additional autonomy demands. For them, more sub-state autonomy
implies more power. For example, studies on Belgium have referred to the dominant position
of the Christian democrats in Flanders and the Social-democrats in Wallonia to explain both
parties’ key role in the Belgian process of state reform (e.g. Deschouwer, 2006; Huyse, 1981,
83
Recent analyses indicate that this effect is limited to sub-state elections (Massetti & Schakel, 2017).
Although the authors report an overall null effect of decentralization on regionalist party strength in national
elections, they find that this is actually the net result of an empowering effect on new regionalist parties in
combination with an accommodating effect on old regionalist parties.
84
Based on these concerns, Nordlinger (1972) famously excludes federalism from his list of desirable
conflict-regulating practices. For a more elaborate discussion of the way political entities can actively (agent
of political socialization) or passively (structure of social interaction) foster feelings of identification, see e.g.
Martínez‐Herrera (2002).
85
Most notably, decentralization is often promoted for its (presumed) economic and democratic benefits. For
these and other reasons, it has received strong support from numerous international agencies, including the
World Bank and the United Nations (Treisman, 2007, pp. 2-3).
104
p. 124).
86
If decentralization leads to growing autonomy demands and autonomist pressures, this should
fuel conflict at the federal power-sharing level. There are at least three reasons to expect this.
First, state reform negotiations are particularly conflict-prone. This is unsurprising, given
their nature. Debates like these go to the very heart of sovereignty debates and questions of
political loyalty. As such, they are not only burdened with technical complexity and high legal
thresholds (e.g. because they often require constitutional reform) but also prone to sharp
ideological red lines and deep-rooted emotions. Hence, debates on territorial reform are political
minefields. Calls for further reforms force politicians to cross these fields, thus increasing the
risk of explosions. Second, and paralysing negotiations even more, the growing autonomy
demands of some parties might increase the fear of secessionism in other parties. In itself,
decentralization already makes the perspective of a break-up more realistic: secession “is more
likely where the cost of disentanglement is too small, in other words where the centre has been
left with few competencies of significance.” (Swenden, 2013b, p. 69) To those adhering to the
survival of the state, insights like these sound very alarming – especially against the background
of growing autonomist demands. Such dynamics might seriously erode the kind of trust and
indulgence on which power-sharing essentially relies. It is not hard to imagine them coming
with growing tensions. Third, the increasing autonomist pressure can result in an ethnic
outbidding dynamic: an inflation of claims, intransigence, and increasingly radical party
stances (e.g. Rabushka & Shepsle, 1972). In the fight for votes, parties might struggle to present
themselves as the purest and least compromising spokesperson of their own segment. Such
dynamics come at the cost of federal stability.
Caluwaerts and Reuchamps (2015, p. 279) present several additional reasons why
decentralization deals can “accommodate political conflicts that are currently on the agenda”
but “renders the process of intersegmental conflict accommodation increasingly more
difficult in the long run.”. First, decentralization makes it harder to buy off peace. Package
86
As described by Deschouwer (2006, p. 901): “Moving towards more regional autonomy […] created two
entities in which the two major political forces of the country – the Christian Democrats and the Socialists –
became even stronger. It is a feature that has certainly facilitated state reform and that therefore also explains
why these two political families were the driving forces behind decentralisation. […] Taking away powers
from the centre therefore meant a transfer of those powers to a level where both major political forces would
still be able to control them.”
105
deals, in which all actors get some sweets, are harder to reach because decentralization leaves
the central level with fewer and fewer competences and resources (sweets) to buy off peace.
This logic applies to all decentralization reforms, but it is particularly evident in countries
whose sub-states heavily rely on federal grants as a source of income; i.e. where sub-states
enjoy little fiscal autonomy (for a comparison of countries, see Blöchliger & King, 2007;
Blöchliger & Rabesona, 2009; Watts, 2008). In these countries, the federal level pays a
particularly (and literally) high price for strengthening the sub-states. As Caluwaerts and
Reuchamps (2015, p. 288) phrase it, the resulting budgetary burden decreases the financial
manoeuvrability of the federal government. This makes it less evident to seal deals by
increasing the transfers from the centre towards the sub-states. Similarly, there are also fewer
competences left to include in such agreements, and the remaining competences are typically
the ones whose potential decentralization is more contested or controversial (e.g. social security
in Belgium; Béland & Lecours, 2018; Dandoy & Baudewyns, 2005).
87
After all, and for evident
reasons, the low-hanging fruit is typically picked at an early stage of the decentralization
process. The most undisputed policy domains and issues are easily agreed upon, but as
decentralization continues, negotiations will cover more controversial grounds and politicians
will be faced with the bitter residue of previous discussions (issues that failed to be settled).
‘Hollowing the centre in return for peace’ (L. Hooghe, 2004) is thus increasingly more difficult.
This not only implies that ethno-territorial clashes will become increasingly hard to solve. It
also means that unresolved issues will continue to burden power-sharing dynamics and flare up
again. Kicking the can means facing it more than once.
Additionally, Caluwaerts and Reuchamps (2015) argue that decentralization lowers the cost
of non-agreement. To understand this argument, it is important to know that the default option
faced by politicians (what happens when no agreement is found) is a crucial determinant of
negotiation dynamics. Before decentralization, the fear of a ‘generalized policy paralysis’ is
thought to have provided the pressure needed to push negotiating elites towards
accommodation.
88
87
For example, for the first time in Belgian history and after years of contestation, the 6th state reform (2011)
decentralized some social security competences (the decentralization of family allowances) (Béland &
Lecours, 2018; Goossens & Cannoot, 2015).
88
A generalized policy paralysis refers to the kind of failure of joint decision-making that “entails a broad
and generalised blockage of the wider decision-making process.” (Jans, 2001, p. 44)
106
“A general policy deadlock implied that the ethnonational representatives had
to forsake to all other policy programs they might have planned and envisage
elections. Non-agreement rendered the political resolution of other societal
problems impossible and brought the political system in a state of paralysis.
Most ethnonational actors preferred to tune down their preferences and to
accept a compromise rather than to confront the consequences of a general
policy paralysis. The consequences of non-agreement rendered the
establishment of an agreement valuable in itself, irrespective of its actual
content.” (Jans, 2001, p. 43)
Decentralization is thought to have shifted this default option. As it decreased the relevance of
the central policy level, the failure to reach federal agreements might no longer result in such
generalized policy paralysis (see also L. Hooghe, 2004). Because many competences and
resources were stripped from the central level, central deadlocks are simply thought to be less
disruptive or problematic. After all, even when the federal level is struck by political deadlocks,
the country can still function thanks to the increasingly strong sub-states (Caluwaerts &
Reuchamps, 2015, p. 290). The policy carousel never stops. An example of this is the major
2010 formation crisis in Belgium, when it took politicians 541 days to reach an agreement:
“Belgium has no less than six governments and […] five of those —the regional and community
governments—continued working as they had been newly composed after the elections of 2009
and their term ran until 2014.” (Devos & Sinardet, 2012, p. 168) This shift towards a far less
disruptive default option took away an important incentive for politicians to reach consensus
when tensions arise and to avoid confrontation in the first place. And as noted by e.g. Horowitz
(2002, p. 24), incentives “are the key to accommodation in difficult conditions”.
To the arguments presented by Caluwaerts and Reuchamps (2015) to expect paradox-like
dynamics, we might add that decentralization increases the room for segmental
interpretations of certain debates. Providing voice to sub-state interests and geographically
concentrated issues has been argued to make inequality issues more susceptible to being framed
and politicized in geographical terms, as tense debates on intergovernmental transfers, different
taxation rates, or national spending illustrate (Bakke & Wibbels, 2006, pp. 13-14). And finally,
autonomy provisions face the same problem that was pointed out by Sisk (1996, p. 2) in the
context of secession: it “does not solve problems of multiethnic coexistence, it only rearranges
107
the configuration of minorities and majorities.” Providing minority groups with their own sub-
state creates new minorities or places them under the authority of another segment (e.g.
Flemings in Brussels, Francophones in an increasingly autonomous Flemish sub-state). As
such, and rather than fully dampening the fire, it typically shifts rather than eliminates
frustrations, tensions, and the subjects of potential discrimination (similarly, see Horowitz,
2003). For all these reasons, the expectation is that decentralization will result in more ethno-
territorial conflict while also making the process of ethno-territorial conflict resolution harder.
H3: Decentralization fuelled ethno-territorial tensions
H4: Decentralization burdened the resolution of ethno-territorial conflicts
2. Contextual factors
It is widely established that decentralization is only one piece of an extensive puzzle
and certainly not a panacea (e.g. Bermeo, 2002; Keil & Anderson, 2018; Kymlicka, 2001b,
Ch. 5; Seymour & Gagnon, 2012). Its effect also depends upon the context in which it is
implemented. In this respect, scholars have identified several institutional and historical
features that increase the risk of conflict and instability: e.g. bipolar/multipolar nature of
polities, the degree of sub-state autonomy, the centrifugal/centripetal nature of federalization,
and the presence or absence of vetos (e.g. Horowitz, 2000; Swenden, 2006, 2013b; Watts, 2008,
2015). The focus on a single case (Belgium during a certain timeframe) implies that several of
these key factors are kept constant. Put differently, this work provides an assessment of the
role of decentralization against a stable institutional background. In a first section, I discuss
these constants and why they make the Belgian context particularly paradox-prone. Several
features of Belgian federalism are addressed. The discussion of these elements strongly builds
on standard concepts and reflections within the study of comparative federalism (e.g. Horowitz,
2000; Swenden, 2006, 2013b; Watts, 2008, 2015). Next, and given its importance, particular
attention is devoted to the consociational democracy that characterizes the Belgian system.
Furthermore, this work also assesses the role of several factors that do vary over time.
Echoing the idea that parties react to such incentives, the aim is to take into account issue
saliency as well as regionalist and electoral pressures. Four variables allow me to do so.
Specifically, in making sense of ethno-territorial conflict dynamics, I tap into the effect of (1)
the strength of regionalist parties and (2) their cabinet participation, (3) the saliency of the
decentralization issue, and the (4) proximity of federal elections. The presumed role of these
variables is discussed in a second section.
108
2.1. Constants: the Belgian case
2.1.1.
Belgian federalism
Due to its very nature, Belgian federalism is particularly conflict-prone from a
comparative perspective. To begin with, it is a centrifugal federation (e.g. Brans, De Winter,
& Swenden, 2009; Swenden, 2006; Swenden & Jans, 2009). Put differently, it is the result of a
process of disintegration (decentralization); a textbook example of the so-called holding
together federalism that stands oppose coming together federalism (Stepan, 1999). The latter
refers to federations that are the result of centripetal dynamics, i.e. a positive choice of certain
polities to join forces and cooperate. In Belgium, the choice for federalism was a negative one
(an attempt to separate the fighting dogs and guarantee the survival of the state). For evident
reasons, centrifugal federations are more conflict-prone than their centripetal counterparts.
Centrifugal dynamics typically come with multiple rounds of risky negotiations (or at least calls
for such negotiations). And as discussed above, the lingering threat of regionalism in general
and separatism specifically obviously burdens the atmosphere in power-sharing executives.
Second, the Belgian federation is one of the few countries that manage to survive despite having
a so-called core ethnic region (Flanders), that is “a single ethnic federal region that enjoys
dramatic superiority in population” (Hale, 2004, p. 166).
89
Analyzing the fate of ethno-federal
states across the globe since World War II, Hale finds that all cases that collapsed possessed
core ethnic regions, while no federation without these kinds of sub-states has faced such a
dramatic ending. He presents three main rationales that help explain why core ethnic regions
increase the chance of state collapse. To begin with, and building on theories of revolution
(Brinton, 1965; Tilly, 1975), they make states very susceptible to the rise of dual-power
situations: situations of multiple sovereignty whereby a challenging authority competes with
the central state for control over its territory. Additionally, the presence of core ethnic regions
can be (perceived as) a threat to minority sub-states. Their strong bargaining position can make
other sub-states fear that the central government will be more responsive to the demands of the
core ethnic region than to those of other sub-states, implying that the latter will “permanently
get the short end of the federal stick” in negotiations on federal resource allocations and in
89
According to Hale (2004, p. 169), core ethnic regions exist when a region possesses half of the federation’s
population or at least 20% more inhabitants than the second-largest region. Representing roughly 60% of the
total Belgian population, Flanders clearly meets this criterion.
109
policy-making in general (Hale, 2004, p. 175). This can undermine the credibility and
effectiveness of central efforts to build trust, which are key to the capacity of states to avoid
secession or conflict. Finally, Hale argues that core ethnic regions come with a destabilizing
effect that is often ascribed to federalism as such (e.g. Kymlicka, 2001b, Ch. 5), namely that it
facilitates the collective imagining of secession by a segment. It makes the idea more
conceivable that this group can govern itself in a future separate state. As Kymlicka (2001b, p.
115) states, the idea is that it provides segments “with the experience of self-government, so
that they will feel more confident of their ability to go it alone, and with an already recognized
territory over which they are assumed to have some prima facie historical claim.” In a similar
vein, Keil and Anderson (2018, p. 98) state that territories that gain decentralized structures
“become mini-statelets, organized with political and bureaucratic resources, which whets the
appetite of would-be secessionists.” The same concern is raised by Hale when it comes to the
presence of core ethnic regions. Evidently, all these dynamics not only increase the chance of
state collapse but also strongly undermine the power-sharing ambience. Decentralization might
add to this dynamic, as it can strengthen the core ethnic region’s position even more.
90
Third, scholars usually agree that the division of segments is an effective way to increase
stability (Horowitz, 2000). It highlights intra-segmental divisions and competition, thus
providing some counterweight to tendencies of homogenization and slowing down the
development of sub-national as ‘imagined communities’ (B. Anderson, 1983). From this
perspective, the homogeneous federal system installed in Belgium is problematic. In
homogeneous federations, sub-state borders tend to coincide with segmental borders (Swenden,
2006). Of course, most federations face the problem of inequality issues being politicized and
framed in geographical terms, as tense debates on transfers or national spending illustrate
(Bakke & Wibbels, 2006). And cabinet members in all kinds of power-sharing coalitions might
be tempted to evaluate proposals based on which segment they will benefit most (or least).
91
But when the geographical borders of sub-states and the segmental borders overlap, such
90
Note that decentralization reforms do not per se strengthen core ethnic regions in relative terms, i.e. in
comparison to the position of other sub-states (as such a reform can also strengthen the position of the latter).
91
To give just one example, studying social policy-making in power-sharing coalitions in Northern Ireland
(2007-2011), Gray and Birrell (2012, p. 23) observe that “Disputes between ministers also often reflect
communal interests rather than any socio-economic interests; thus welfare delivery and social policy
innovations are often analysed in terms of which community is likely to benefit most or be disadvantaged.”
110
tensions are highly prone to segmental interpretations, e.g. to be portrayed as conflicts between
segments (‘the Flemings vs. the Francophones’ in Belgium). According to Horowitz (2000, p.
619), the only way that homogeneous federations can reduce conflict at the central level is by
stressing and playing out divisions within the segments (sub-ethnic or lower-level ethnic
divisions). But efforts to do so are particularly hard, due to the persevering bipolar nature of the
Belgian system.
Indeed, despite the presence of four language areas and two types of sub-states, Belgium is
typically considered to be a dyadic, bicommunal, or bipolar federation (e.g. Brans et al.,
2009; Duchacek, 1988; Niessen et al., 2020; Peters, 2006; Popelier, 2021; Popelier & Lemmens,
2015; Watts, 2008). This is not only part of the Belgian hardware, as it is institutionally
anchored in e.g. the parity of the federal Council of Ministers and the division of MPs in the
Chamber of Representatives into two language groups (Deschouwer, 2012; Popelier, 2021, pp.
100-103). It is also part of the software of Belgian politics. The dominant dynamic of ethno-
territorial debates throughout the contemporary history of Belgium is effectively a dyadic one;
a struggle between Flemish and Francophone actors (both in the way clashes are fought and
perceived). Bipolar arrangements are usually very negatively evaluated – both in theory and in
reality. Surely, as cases like Czechoslovakia (segregation in 1993), Serbia and Montenegro
(split in 2006), and Pakistan (secession of East Pakistan in 1971) illustrate, “the experience of
bipolar or dyadic federal systems is not encouraging” (Watts, 2008, p. 184). The main problem
with bipolarity is that it (presumably) evokes a permanent stand-off between the same groups.
In contrast to multi-unit federations, dyadic federations “do not give rise to floating coalitions,
but focus conflict along a single axis.” (McGarry & O'Leary, 2009, p. 20) Put differently, it
leaves little room for ‘tension-reducing coalitions’ (Elazar, 1987, p. 244): the kind of
alliances that shift across negotiations, which avoids animosities between two sides from
reaching problematic levels (opponents in one clash are allies in other disputes, which alleviates
tensions). In this context, consociational mechanisms such as the parity rules and mutual vetoes
that can be found in Belgium (see below) can be a recipe for “cumulatively sharpening
frustrations” (Watts, 2008, p. 184). Furthermore, in a context of bipolarity, it is also harder to
play out intra-segmental divisions at the federal level: “if there are only a few units,
homogeneous states […] are likely to exacerbate rather than alleviate ethnic conflict”
(Horowitz, 2000, p. 619). This reflection was written with the Nigerian context in mind, but as
Swenden (2006, p. 256) explicitly notes, it also provides a clear warning for the bipolar Belgian
federation. Some Belgian politicians are well aware of the danger of bipolar dynamics. For
instance, former Prime Minister Herman Van Rompuy (Flemish Christian democrats; CD&V)
has explicitly argued that we should ‘not to underestimate bipolar federalism. Few countries
survive such a dynamic: Czechoslovakia was split. Cyprus too. And in Ireland, the battle
111
between Catholics and Protestants claimed thousands of deaths’. (Hoflack, 2021, p. 215)
92
In
recent times, this logic has often been evoked as an argument to transform Belgium into a
federation with four or even ten sub-states (based on the provinces and Brussels) (e.g. De
Grauwe & Van Parijs, 2021).
93
Fifth, scholars often accentuate the importance of electoral systems in making or breaking the
incentives for accommodative elite behaviour. As summarized by Horowitz (2002, p. 23),
“electoral rewards […] can provide the motivation ethnic leaders otherwise lack”. But in
Belgium, he argues, such incentives are absent. Belgium combines a list-PR electoral system
with political parties that are divided along segmental lines. State-wide parties are often
considered to be stabilizing factors for divided countries, as they can try to win support from
all segments, balance their concerns, strengthen feelings of loyalty vis-à-vis the federation at
large (e.g. Linz & Stepan, 1992), and nurture the kind of common nationality that Kenneth C
Wheare (1963) considers as crucial to make federal states work (namely a shared nationality
that exists alongside segmental nationalities, rather than replacing them). In Belgium, there are
no state-wide parties to fulfil this role (except for the Marxist party). As discussed elaboratively
in Chapter 3, the traditional Belgian parties (liberals, Christian democrats, social democrats)
were torn apart between 1968 and 1978. Ever since, all parties are split along linguistic lines.
There are separate Flemish and Francophone liberal parties, social-democratic parties, etc. (cf.
De Winter et al., 2009, pp. 72-76; Verleden, 2009). These parties are thought to have very few
incentives to reach out to the other side, and many reasons to opt for confrontation. They are
considered to only represent and address their own segment, they fear being called traitors by
competitors on their side when seeking compromise, and they cannot be electorally sanctioned
or held accountable by the other side because they choose to only present lists in districts within
their own segment (e.g. Caluwaerts & Reuchamps, 2015; Deschouwer, 2002, 2006, 2013b;
92
Note that most killings took place in Northern Ireland (most notably Belfast), while the number of deaths
in the Republic of Ireland was far smaller (English, 2003; McKittrick et al., 1999; McKittrick & McVea,
2012).
93
Pacification is not the only rationale behind proposals like these. Some scholars have presented a similar
plea for a four-unit Belgian federation based on the concern that the current situation in which competences
are scattered across three communities and three regions leaves little room for coherent policy-making (e.g.
Van Parijs, 2018).
112
Reuchamps, 2013; Swenden, 2006, 2013a).
94
Accordingly, Horowitz (2002, p. 22) argues, the
combination of a list-PR electoral system and parties divided along segmental lines “does
nothing to foster compromise on ethnic issues” in the pre-electoral arena (cf. also Horowitz,
1991, pp. 167-176). Rather, and in contrast to systems with state-wide parties or e.g. alternative
vote systems (e.g. Horowitz, 2004), the combination of Belgium’s electoral and party system
is typically considered to be a recipe for conflict (if not disaster).
95
This is even the case in a
context like Belgium, where (at least in recent decades) ethno-territorial issues are not really
vote-determining issues (e.g. Swyngedouw et al., 2015, pp. 11-14). While few voters are
preoccupied with these debates, there are at least two reasons why compromise aversion can
still arise: (i) because politicians think otherwise and/or (ii) because they want to avoid losing
credibility. After all, making concessions and being perceived as a party that fails to stand its
ground can also deter voters that don’t care about ethno-territorial issues at all. Voters are quite
good at evaluating whether or not governing parties fulfil their election pledges (Duval & Pétry,
2020; Thomson, 2011; Thomson & Brandenburg, 2019). They also tend to reward parties that
deliver (cf. retrospective pledge voting). Abandoning election promises once in office fuels
distrust in political parties (R. Rose & Wessels, 2019) and government parties that fulfil a higher
share of election pledges are more likely to prevent electoral losses (e.g. Matthieß, 2020).
Finally, another feature of the Belgian federal system that makes it particularly conflict-prone
is its high degree of decentralization. To be sure, although several major policy domains still
pertain to the federal level (Reybrouck & Sottiaux, 2019), the degree of autonomy of the
Belgian sub-states is particularly high in comparative perspective (certainly in terms of
competences) (Fabre, 2009; L. Hooghe et al., 2016; Shair-Rosenfield et al., 2021; Watts,
2013).
96
For instance, a comparison of 42 democracies between 1950 and 2006 based on the
94
Several exceptions can be noted, see fn. 14 in this work.
95
The alternative vote arrangement is thought to reward political moderation because the high threshold for
election (an absolute majority of votes cast) should give political parties incentives to aim for the marginal
votes of voters of other segments (the kind of electorate that won’t support ethnic hardliners that opposes
their interests) (Horowitz, 2004). For further reflections on the potential of electoral system reform to nurture
stability in divided states, see e.g. Reilly (1997, 2001, 2002); Reynolds, Reilly, and Council (2000); Wolff
(2005).
96
Because residual competences also pertain to the federal government (Belgian Constitution, art. 35),
several new challenges such as artificial intelligence, drones, telecommunication, and medical/ethical issues
113
authoritative Regional Authority Index (RAI) indicates that Belgium is amongst the countries
with the largest RAI score in general, the greatest self-rule and shared-rule scores specifically,
and the broadest regional policy portfolio scope (Marks, Hooghe, & Schakel, 2008). Similar
results arise when comparing the more recent RAI scores (until 2018) of Belgium to those of
94 other countries across the world (L. Hooghe et al., 2016; Shair-Rosenfield et al., 2021). If
the stripping of the central policy level effectively increases the risk of conflict, as discussed
above, this dynamic should thus be particularly visible in Belgium. Note that the large degree
of decentralization is not really a constant in this work (which covers the 1979-2018 period). It
only characterizes the state in recent decades. Nevertheless, it serves as a noteworthy addition
to the many troublesome features discussed above and it evidently adds to the idea that paradox
dynamics should be particularly salient in Belgium.
2.1.2.
Consociational democracy
Another key feature of the Belgian system that characterizes it throughout the period
under study is its consociational nature (Deschouwer, 2002, 2004, 2006, 2012; Huyse, 1970;
Lijphart, 1981b). Consociational democracy is described and advocated most thoroughly by
Lijphart (e.g. 1968b, 1969a, 1977a, 1977b, 2002), who once labelled Belgium “the most
thorough example” of such a system (Lijphart, 1981b, p. 1). As discussed in Chapter 3,
consociational democracy strongly renounces the Westminster model of majority rule: “the
opposite of consociational democracy is majoritarian democracy, just like the opposite of
federalism is unitary government” (Lijphart, 1979, p. 500).
97
It can be defined by four
principles: grand coalition, segmental autonomy, mutual veto, and proportionality (e.g.
Lijphart, 1977b; Lijphart, 1979, 1985a). In more general terms, the consociational system
combines power-sharing with segmental autonomy, but scholars have mostly focussed on the
former element (as noted by e.g. L. Hooghe, 2004, p. 32). This focus on consociational power-
sharing also echoes in the seminal description of consociational democracy as “government by
elite cartel designed to turn a democracy with a fragmented political culture into a stable
democracy.” (Lijphart, 1969a, p. 216) Although this power-sharing can take many shapes
(Lijphart, 1979, p. 500), it typically refers to “meaningful cross-community executive power
have added some new competences to the federal government’s agenda (e.g. Stefan Sottiaux in Justaert,
2019).
97
After all, and as Belgian history illustrates (Chapter 3), majoritarianism is considered to be an agitating
factor in divided states (Lijphart, 2008).
114
sharing” (O’Leary, 2005, pp. 12-13).
In Belgium, such consociational democracy “has come in waves, pacifying cleavage after
cleavage” (Deschouwer, 2002, p. 68). Having successfully pacified the labour-capital and
church-state cleavages, consociational recipes were used again to address the growing tensions
between the ethno-territorial segments. In 1970, the first state reform established a particular
consociational system. It introduced segmental autonomy (three Regions and three
Communities), linguistic parity in the Council of Ministers, two language groups in parliament,
special majority laws, and an alarm bell procedure (Deschouwer, 2012; Popelier, 2021). At the
federal policy level, this arrangement left Flemish and Francophone politicians with two
options: governing together or not governing at all (Deschouwer, 2013a, p. 214). As it combines
federal with consociational systems, Belgium has often been dubbed a ‘consociational
federation’ (Deschouwer, 2002, 2005, 2006).
98
Consociational democracy is one of the most prominent examples of the many recipes for
conflict management in divided states (Cordell & Wolff, 2016; McCulloch, 2009; McGarry &
O’Leary, 1994, 1993b; Reynolds, 2002; Smooha & Hanf, 1992; Wolff & Yakinthou, 2012).
The key to this recipe is a specific kind of elite behaviour. In fact, despite its emphasis on the
importance of institutions, “the essential characteristic of consociational democracy is not so
much any particular institutional arrangement as the deliberate joint effort by the elites to
stabilize the system.” (Lijphart, 1969a, p. 213) As in football, the rules of the game set some
boundaries, but it is the attitude of players that determines how tough games really are.
Many tackles are prohibited, but this does not keep them from happening. In a similar vein,
consociational theory stresses that (party) elite behaviour is absolutely imperative for the
98
Federalism and consociationalism share a wide range of features, and essentially, they are both non-
majoritarian forms of government that combine self-rule and shared rule (Elazar, 1987). Nevertheless, there
are important differences (for an elaborate account, see Bakvis, 1987; Lijphart, 1979, 1980, 1985a). Not all
consociational democracies are federal and not all federations are consociational (Swenden, 2006, p. 7):
“Consociational regimes can be unitary (at some point in their history, the Netherlands or South Africa were
consociational); federal regimes may not be consociational (for instance, the USA federation).” In
distinguishing both phenomena, most authors mainly point out the fact that consociational systems allocate
autonomy to groups who live dispersed throughout a territory while the entities are (typically but not per se
exclusively) territorially based in federations (groups are geographically concentrated) (Bakvis, 1987;
Kincaid, 2011; Lijphart, 1979, 1985a; Swenden, 2006). As summarized by Kincaid (2011, p. viii), both “are
non-majoritarian; however, they structure and operate non-majoritarianism differently.”
115
stability of divided states.
99
This pivotal element of elite cooperation goes back to the very origins of consociational
theory. In his reflection on the classification of political systems presented by Almond (1956),
Lijphart (1968c) first coined the term ‘consociational democracy’ to describe the seemingly
odd cases that deviated from the idea that major societal divisions come with major political
instability. Several countries, namely Switzerland, Austria, Belgium, and the Netherlands were
identified as countries that contradict this logic. In making sense of these “deviant cases of
fragmented but stable democracies” (Lijphart, 1969a, p. 211), the behaviour of the political
elites within these countries was considered to be crucial. It was “the missing link” between a
plural society and political stability (Bogaards, 1998, p. 475). In very rough terms, segmental
elites can essentially act in two ways. They can either behave in a competitive way, which
aggravates tensions, or they can opt for cooperative behaviour, i.e. “make deliberate efforts to
counteract the immobilizing and unstabilizing effects of cultural fragmentation.” (Lijphart,
1969a, pp. 211-212; emphasis removed) The latter kind of accommodative elite cooperation
has been presented as the primary distinguishing feature of consociational democracies
(Lijphart, 1977a, p. 1). It implies that elites “accept to play the democratic game in a different
way”, which includes “the willingness not to be the final winner and thus […] the acceptance
that the sharing of power and therefore the finding of compromises on important issues is
necessary.” (Deschouwer & Temmerman, 2012, p. 501) Consociational theory holds that such
elite attitudes can have the same pacifying effect as cross-cutting cleavages (Lijphart, 1968c).
If such cross-cutting cleavages are absent, the political awareness that their absence comes with
great risks can turn into a “self-denying prophecy” (Lijphart, 1968c, pp. 188-189). Being
concerned about these risks, leaders might then be urged to search for ways to avoid instability
or violence and act more accommodatively (see also e.g. Dahl, 1966, p. 358). Such dynamics
might be facilitated by traumatic experiences or emerging forms of violence. Examples of such
situations that might have been a leverage for accommodative elite behaviour in Belgium
include the mass demonstrations and street riots in favour of the Catholic University of Leuven
becoming monolingual (Leuven Vlaams/Walen Buiten), the often fierce riots and physical
encounters in Voeren in the 1980s and 1990s, the brutal 1970 killing of FDF-politician Jacques
Georgin in the streets of Brussels, and the emergence of private militias such as the Flemish
99
On the role of political parties and party elites in consociational democracy, see e.g. Luther and
Deschouwer (2004).
116
VMO (Vlaamse Militanten Orde) (Bart De Wever, 1998; Delcorps, 2012; Deweerdt, 1998b, p.
2962; Gevers & Vos, 1998, pp. 2915-2916; Kesteloot, 1998b; Witte, 2009c, pp. 375-376; Witte
& Van Velthoven, 1998, pp. 3028-3029).
100
Additionally, troublesome dynamics of violence
abroad can add to the self-denying prophecy, as do very basic, non-violent domestic gridlocks.
The general trauma after the long-sought agreement on a sixth Belgian state reform (2011)
provides a good example of the latter.
The attractiveness of the consociational model, also as a prescriptive tool, lies in the possibility
to achieve stability through the action of political elites, in spite of dissension at the mass level
(Bakvis, 1987). However, scholars remain strongly divided on the ability of consociational
systems to pacify tensions. Indeed, critics have attacked consociational theory from several
angles and raised a wide range of conceptual, theoretical, empirical, and ideological concerns
(Andeweg, 2000; Barry, 1975a, 1975b; Bogaards, 2000, 2006; Lustick, 1979; Tsebelis, 1990;
Van Schendelen, 1985).
101
Arguably, one of the most pertinent critiques is that consociational
theory fails to adequately take into account the incentives that guide the behaviour of politicians
(Horowitz, 1985, 1991, 2002). In this respect, many scholars are “skeptical that cultural group
leaders generally prefer self-restraining accommodationism to outbidding aggressiveness.”
(Lustick, 1997, p. 113) As explained throughout the following paragraphs, there are several
reasons to expect the exact opposite, as politicians face strong incentives to avoid
accommodation and cultivate intransigence. Accordingly, the functionality of the power-
sharing it prescribes is often feared to be very weak (McCulloch, 2021, pp. 2-3). This
functionality refers to “how consociational institutions work, whether they work as intended,
and whether they facilitate elite cooperation.” (McCulloch & Vandeginste, 2019, p. 1178)
Similarly, McEvoy and Aboultaif (2022, p. 242) describe it as “the utility of the particular
political system to secure the goals of elite cooperation and joint decision-making. A functional
power-sharing system is one that proves useful for its intended purpose, ostensibly to promote
100
FDF-politician Jacques Georgin (Front démocratique des francophones) was beaten to death with sawn-
off table legs by nine members of the Flemish nationalist VMO militia (Vlaamse Militanten Organisatie,
later Vlaamse Militanten Orde). The attack occurred in the linguistically contested territory of Brussels, while
Georgin was pasting party posters in the Houba de Strooperlaan/avenue Houba de Strooper. The VMO was
an extreme-right Flemish nationalist organization that dissolved in 1983 after being convicted in court as a
private militia (Bart De Wever, 1998). For an overview of instances of ethno-territorial violence in Belgium,
see Delcorps (2012).
101
For consociational rebuttals, see e.g. Lijphart (1985b, 2000, 2002); O’Leary (2005).
117
elite cooperation, help the state transition to democracy, and secure peace.” As this description
shows, these intended purposes differ, ranging from the maintenance of peace to the providence
of good governance (see also McCulloch, 2021, p. 5). The focus of this work is on the latter. In
this respect, the fear is that power-sharing resembles a three-legged race: obliging
segmental elites to cooperate prevents the minority from being regularly trampled, but also
makes it hard to move forward and ‘get things done’. In this respect, consociational power-
sharing has been blamed for only leaving room for lowest common denominator decisions
(McLaughlin, 2005) and for resulting in a brinkmanship-style politics (Roeder & Rothchild,
2005, p. 9). Nagle (2020, p. 137) even talks about ‘zombie power-sharing’. In his view,
consociations are typically “neither dead or alive, but walking dead” as they “tend to develop
ossified properties rendering them resistant to practically any reform”. In sum, the forced
cooperation between politicians that are strongly divided along ideological, ethnic, and/or other
lines might well result in forms of immobilism, legislative deadlocks, and forms of
intransigence that can even challenge the survival of power-sharing cabinets or systems as a
whole (see also Barry, 1975a, 1975b; Horowitz, 1985, 2002; Horowitz, 2014; Tsebelis, 1990).
Why might we expect dynamics like these to occur? This expectation rests on two legs. (i)
On the one hand, consociationalism provides politicians with the tools to provoke and maintain
immobilism. (ii) On the other hand, and as already partly addressed elsewhere in this work,
political elites face strong incentives to be intransigent and confrontational.
(i) To begin with, the consociational toolbox allows politicians to create immobilism
whenever they want. It provides politicians with an anti-compromise arsenal of veto powers
and procedures that allows them to block any demand and paralyze any negotiations.
Mechanisms such as the special majority laws or the alarm bell procedure in Belgium “empower
these elites with the means to back their demands with political brinkmanship”, which
“transforms the bargaining room into a new battlefield.” (Roeder & Rothchild, 2005, p. 9) As
the consociational guarantees make immobilism both likely and hard to overcome, calls for
institutional reform will grow (Horowitz, 2014, p. 7). But as Horowitz argues, putting these
calls into practice is extremely tough exactly because of these arrangements and the high
threshold for agreement they impose. Such dynamics certainly seem to be at play in Belgium,
where Flemish parties often demand reforms due to the (perceived) difficulty to strike
compromises with the Francophones at the federal policy level (Walgrave et al., 2014, pp. 145-
146). If anything, the resulting stalemates appear to feed frustrations and mutual distrust.
(ii) Apart from providing the tools needed to trigger immobilism, consociational democracy is
also burdened by the fact that politicians have many incentives to avoid accommodation and
cultivate intransigence. For one, and as noted by Horowitz (2002, pp. 21, 29), there is usually
not one set of politicians that can be identified as the ‘leaders of the segment’, but rather
118
different groups contesting for segmental leadership. Accordingly, political leaders always risk
facing criticism by ‘counter-elites’ who might depict agreements as a ‘sell-out’ by ‘traitors’.
The threat of such counter-elites is present in most consociational democracies, especially as
cabinets in such systems are usually not all-encompassing ‘grand coalitions’. They only include
some parties of each segment. Meanwhile, others compete for the support of the same group
from the banks of the opposition, often by besieging their competitors in government. Such
centrifugal competition for group allegiance is thought to strongly burden the search for
compromises across group lines (Horowitz, 2002, p. 21). Illustrating just how omnipresent the
pressure of competitors can be, even the very choice to join certain coalitions in the first place
can trigger fierce backlashes. A clear and rather extreme example of this is the Belgian Michel
I-cabinet (2014-2018). This centre-right coalition of four parties included the Flemish liberals,
Flemish Christian democrats, and the Flemish nationalist N-VA party, but only encompassed
one Francophone party (the liberal MR). The latter faced strong opposition in its segment, as it
was blamed for betraying the Francophone interests by joining a coalition with Flemish
nationalists that was only backed by 20 Francophone MPs in the Chamber of Representatives
(out of 63 Francophone seats in total). Perfectly grasping the hostile atmosphere, the cabinet
was even nicknamed the ‘kamikaze coalition’ (Benoit Rihoux, Vandeleene, De Winter,
Baudewyns, & Deruette, 2015, p. 38). As Charles Michel (MR), the Prime Minister of this
coalition and later president of the European Council, recalls:
‘From one day to the next, I became a political pariah because the MR was the
only Francophone party joining the coalition. (…) I was convicted a priori, not
because the policies were good or bad, but because I was the only one who
dared to join a coalition with the N-VA. (…) I was mentally capable of coming
to terms with the criticism on the Walloon side, but perhaps I should have been
more sensitive to it. The caricatures were so exaggerated.’ (Hoflack, 2021, pp.
100-101)
While the fear of being called out and losing votes can paralyze negotiations, it might be
considered more rewarding to cultivate radicalism and immobilism. The abovementioned
ethnic outbidding dynamics, i.e. spirals of radicalization and intransigence (Rabushka &
Shepsle, 1972), are the logical result. As Deschouwer and Temmerman (2012, p. 502)
summarize: “If subgroup elites are afraid of electoral sanctions, consociational democracy will
not work.” Rather, the “obligation to share power is then more likely to result in gridlock, in a
situation where subgroup elites are using their veto against any cross-border agreement.” Again,
voters need not actually be preoccupied with ethno-territorial issues or reward confrontational
behaviour for these dynamics to occur. As noted in the context of Belgium’s split party and
electoral system (see point 2.1.1. in this Chapter), it suffices that political leaders believe that
119
such attitudes are beneficial (or that they will lose credibility when they compromise).
Meanwhile, and adding to this dynamic, the consociational rules also increase the threshold
for agreements, which might decrease the pressure to settle conflicts or avoid them in the first
place. After all, politicians know that a clash only really becomes decisive when one of the
conflict sides reaches the qualified majority. As long as neither side has the ability to impose
its will, disputes can be easily bolstered and tensions can be safely cultivated. This is why, from
a game-theoretic perspective, measures like qualified majorities have been argued to reduce the
consequences of disagreement. Accordingly and paradoxically, “the adoption of measures that
reduce the consequences of disagreement” is argued to “increase the frequency of
disagreement.” (Tsebelis, 1990, p. 22)
Taking these considerations into account, it should not surprise that segmental leaders are not
always more dovish than the people they rule and represent (Horowitz, 2002, pp. 19, 21).
Indeed, while consociational theory states that elite cooperation can provide a pacifying
counterbalance to the presence of societal divisions (Lijphart, 1968a, 1969a, 1977a),
consecutive studies have shown that the highly educated (elites) are not per se less ethnocentric
(Horowitz, 1991, pp. 140-141, fn. 44-50). In Belgium, MPs have been found to hold more
radical views on the state reform issue than their voters – at least in recent times (Reuchamps,
Sinardet, Dodeigne, & Caluwaerts, 2017). It is widely established that segmental leaders can
even trigger conflicts themselves, and critics have argued that consociational theory fails to
explain or acknowledge such elite-initiated conflicts (Tsebelis, 1990). One clear and notorious
example of this is the debate about the electoral district called BHV (Brussels-Halle-Vilvoorde).
While few citizens actually understood the debate on the split of this district, it triggered a range
of fierce political clashes that challenged the very survival of the Belgian state (Blero, 2015;
Devos & Bouteca, 2008).
As a final note, it seems useful to stress that these concerns about elite behaviour in
consociational regimes are particularly pertinent in the context of ethno-territorial
conflicts. After all, such clashes are especially resistant to accommodation. “If it is true that
inter-group conflict involves a conflict for control and ownership of the state, for group
superiority, and for group success […] then compromise will be difficult to achieve.”
(Horowitz, 2002, p. 23) Even if these preconditions are toned down, the conclusion still holds:
“The divisive issues are not easy to compromise.” In view of all those elements, it should not
come as a surprise that Belgium is widely considered to be particularly vulnerable to joint-
decision traps (Benz, 2011; Scharpf, 1988, 2006) and that the (counter)productiveness of the
Belgian consociational system is subject to growing debates and that its very chances of survival
are now openly questioned (e.g. Caluwaerts & Reuchamps, 2015; Deschouwer, 2002, 2005,
120
2006).
102
2.2. Varying factors: issue saliency, regionalist and electoral pressures
Ethno-territorial conflict is not only expected to depend on decentralization and the
institutional-historical context in which such reforms are implemented (a particularly conflict-
prone context in Belgium). Of course, “institutional design makes a difference” (Belmont
et al., 2002, p. 3), but its role nor potential should be overstated. Institutional engineering is
always tricky, but particularly so when it is admired as a holy grail. As rightly stated by Wilfried
Swenden (2013b, p. 72): “There are limits to what institutional design can do”. As the literature
overview clearly indicates, the behaviour of politicians is also thought to depend upon the
electoral and regionalist party pressures they face, as well as the saliency of the decentralization
issue. Taking these elements into account allows us to obtain a more robust understanding of
the conflict dynamics at hand.
Firstly and secondly, both the strength of regionalist parties and their federal cabinet
participation are expected to fuel ethno-territorial tensions in power-sharing cabinets. As
Breuning and Ishiyama (1998, p. 5) note in their analysis of rhetorical strategies of regionalist
parties in Flanders: “Although political parties may reflect the environment from which they
arose, they also nurture and affect that environment.” In this respect, regionalist parties are not
only “children of the centre-periphery divide” (van Haute & Pilet, 2006, p. 303) but also
nurturers of this divide. They were the ones who truly introduced territorial politics as a new
dimension of party competition (Meguid, 2005, 2008; Meyer & Miller, 2015; Wagner, 2012).
Emerging as so-called niche parties, they “politicize[d] sets of issues which were previously
outside the dimensions of party competition” (Meguid, 2005, p. 347).
103
And ever since, it has
102
As Benz (2011, p. 208) put it: “If any structure of policy-making is prone to end in the joint-decision trap,
it is the setting of constitutional negotiation in Belgium.”
103
Discussions on the niche party concept resulted in a plethora of definitions (for overviews, see e.g.
Bischof, 2017; Meyer & Miller, 2015). These parties have been described before as e.g. parties that de-
emphasize economic concerns and focus on a limited number of non-economic issues (Wagner, 2012), parties
that emphasize policy areas neglected by others (Meyer & Miller, 2015) or parties characterized by non-
centrist (“niche”, e.g. ecologist) or extremist ideologies (Adams, Clark, Ezrow, & Glasgow, 2006; Ezrow,
2008). According to the impactful work of Meguid (2005, pp. 347-348; emphasis in original), they can be
defined by three main features: “First, niche parties reject the traditional class-based orientation of politics.
121
been part of their raison d’être to articulate frustrations at the existing order (Müller-Rommel,
1998), often by presenting their segment as a victim of the state in which it is ‘trapped’ and
forced to share power with opposing segments(Abts et al., 2019; Dalle Mulle, 2016, 2018; Dalle
Mulle & Serrano, 2019).
104
As such, they make an instrumental case for autonomy by linking
the identity issue to the interests of people (autonomy as a means to serve the interests of the
segment, not as an end in itself). More generally, regionalist parties might be considered the
anti-thesis of one of the success factors for consociational regimes: a strong allegiance to the
system and its stability on part of the political leaders (Lijphart, 1968c; 1969a, p. 216). Their
presence can also widen the segmental divisions that trigger ethno-territorial conflict in the first
place:
“By appealing to electorates in ethnic terms, by making ethnic demands on
government, and by bolstering the influence of ethnically chauvinist elements
within each group, parties that begin by merely mirroring ethnic divisions help
to deepen and extend them. Hence the oft-heard remark in such states that the
politicians have created ethnic conflict.” (Horowitz, 2000, p. 291)
As opposition parties, they can go all out. And when joining federal cabinets, they are
under great pressure to ‘deliver’ and must contain the loss of credibility. After all, federal
cabinet participation is seen as “a fundamental betrayal of core autonomist principles” (Elias &
Instead of prioritizing economic demands, these parties politicize sets of issues which were previously outside
the dimensions of party competition. […] Second, the issues raised by the niche parties are not only novel,
but they often do not coincide with existing lines of political division. […] Third, niche parties further
differentiate themselves by limiting their issue appeals. They eschew the comprehensive policy platforms
common to their mainstream party peers, instead adopting positions only on a restricted set of issues. […]
[They] rely on the salience and attractiveness of their one policy stance for voter support.” A wide range of
authors, including Meguid, classify regionalist parties within the niche party category (Adams et al., 2006;
Meguid, 2005, 2022; Meyer & Miller, 2015). However, several authors have challenged the usefulness of
this categorization and/or argued that many regionalist parties’ niche status has eroded over time (e.g. Elias
& Tronconi, 2011; Hepburn, 2013). This ‘mainstreaming’ is often linked to (lengthy) periods of government
participation, but such processes are not homogeneous and should not be taken for granted (Bottel,
Mazzoleni, & Pallaver, 2016).
104
Scholars traditionally consider the actual or perceived exploitation and marginalization of certain regions
to be a key factor behind autonomist tendencies and ethno-territorial conflict (Gurr, 1970; Rudolph &
Thompson, 1985). In the words of Will (Kymlicka, 2001a, p. 92): “As long as national minorities feel that
their interests cannot be accommodated within existing States, they will contemplate secession.”
122
Tronconi, 2011, p. 362) in the eyes of many regionalist voters, party members, and associates.
The need to compensate for such pressures is great. In this respect, it is no coincidence that
governing regionalist parties often control territorial, cultural and institutional portfolios and
generally control a disproportionately high share of the Minister posts (Terrière, 2021; Terrière
& Bouteca, 2021). These portfolio allocations not only stress regionalist parties’ preoccupation
with their core business, but also further increase the pressure to deliver, thus boosting the
chances of ethno-territorial conflict. More generally, within the field of coalition politics, the
inclusion of “extreme” or “anti-system” parties has long been considered a factor that
increases instability and decreases cabinet duration (Laver & Schofield, 1990). Relatedly, and
as regionalists typically have very outspoken/radical programmes (especially on ethno-
territorial issues), their inclusion in cabinets might also come with larger levels of polarization,
i.e. a greater “degree of ideological differentiation among political parties” (Dalton, 2008, p.
900). Although the relevance of ideology in explaining cabinet stability is sometimes contested
(e.g. Schofield, 1987), polarization is usually associated with instability and conflict. Scholars
have exposed empirical links between party polarization and the likelihood of legislative
proposals facing gridlock (Jones, 2001) and a wide range of studies have referred to ideological
divisiveness when making sense of coalition conflicts and cabinet durations (e.g. Browne,
Gleiber, & Mashoba, 1984; Laver, 1974; the overview in Müller et al., 2008, pp. 21-23;
Warwick, 1979, 1992). The basic idea is that cabinet stability is at least partly determined by
the positions of its constituent parties, whereby stability decreases as the policy distance
between coalition parties increases (Warwick, 1994).
105
More recent empirical analyses
conclude that coalition parties’ preference profiles can in fact help us make sense of the chance
of cabinets facing conflictual terminations (Damgaard, 2008, p. 320) and that the hazard ratio
of cabinet replacements is higher when coalition parties’ preference range is larger (Saalfeld,
2008, p. 351). These studies typically focus on (general) left-right stances. However, given its
particularly contentious nature, it seems warranted to translate these insights to the domain of
ethno-territorial politics.
Meanwhile, competitors often respond to regionalist parties by embracing their agenda,
taking more radical positions, and putting greater salience on ethno-territorial issues themselves
(Deschouwer, 2013b; M. Mazzoleni, 2009). It is widely known that regionalist parties across
105
This line of reasoning fits within the so-called “attributes approach”, which explains cabinet duration and
stability based on the attributes of cabinets – in contrast to e.g. the “critical events approach” (for a discussion,
see e.g. Krauss, 2018; Warwick & Easton, 1992).
123
Western Europe have historically pushed mainstream parties towards radicalization, thereby
paving the road for decentralization reforms (e.g. De Winter & Türsan, 1998). Within the
famous PSO theory of party competition (Position-Salience-Ownership) proposed by Bonnie
M. Meguid (2005, 2008), this reaction is known as the “accommodative strategy”.
106
Similarly,
Alonso (2012, p. 37) talks about “pro-periphery convergence”. Such responses can lie at the
basis of the so-called ‘ethnic outbidding’ dynamics discussed above: an inflation of increasingly
far-reaching claims and radical intransigence (Rabushka & Shepsle, 1972).
H5: Regionalist party strength is positively related to ethno-territorial conflict
H6: Regionalist party’s federal cabinet participation is positively related to ethno-
territorial conflict
Thirdly, and for at least two reasons, the saliency of the decentralization issue is also expected
to fuel ethno-territorial conflict. More specifically, tensions in power-sharing cabinets are
hypothesized to be higher when parties devoted more attention to decentralization in the last
electoral campaign period (as noted in Chapter 6, electoral manifesto data is used to grasp this).
A first reason to expect this is that issues that are more salient in the run-up towards election
day will typically be high(er) on the political agenda in the legislative period that follows.
For instance, previous studies have indeed found that the topics stressed in preceding electoral
platforms of parties strongly resonate in parliamentary debates (e.g. Louwerse, 2012;
Vliegenthart, Walgrave, & Zicha, 2013) and that government parties’ electoral manifestos are
often congruent with the cabinet agenda (e.g. Borghetto & Belchior, 2020; Borghetto &
Carammia, 2014). Evidently, these effects are not always confirmed unconditionally. Cabinet
agendas are products of many factors, including (changes in) public concerns and hard-to-grasp
elements such as unexpected events (e.g. Froio, Bevan, & Jennings, 2017). But insofar as
electoral agendas translate to cabinet agendas, state reform issues should be higher on the
cabinet’s agenda when decentralization was a salient issue in electoral manifestos. And when
it is, the chance of ethno-territorial collisions increases.
106
According to the PSO theory (Position-Salience-Ownership), mainstream parties can react to challenging
niche parties in three ways: (1) ignore the niche party in an attempt to stifle the debate (dismissive strategy),
(2) adopt a position similar to that of the niche party (accommodative strategy), or (3) adopt a contrasting
position on the issue (adversarial strategy) (Meguid, 2005, 2008). The latter resembles the “anti-periphery
polarization” strategy described by Alonso (2012, p. 37).
124
A second (and evidently related) reason to expect that the electoral saliency of decentralization
issues positively relates to ethno-territorial conflict, is that the pressure to deliver on topics
increases when parties emphasized these topics before the elections. This is especially so when
politicians adhere to the idea of promissory representation, which corresponds with the
traditional delegate model of representation, mandate theory (Pitkin, 1967, p. 146) and the
responsible party model (cf. Schattschneider's work; for a reflection, see White, 1992).
Promissory representation focuses on “the normative duty to keep promises made in the
authorizing election” (Mansbridge, 2003, p. 516). Scholars have often found and stressed that
politicians effectively act on their election promises (Artés, 2013; Bara, 2005; Naurin, 2011;
Thomson, 2001; Thomson et al., 2017). For instance, assessing the fulfilment of over 20,000
pledges made in 57 electoral campaigns in 12 countries, Thomson et al. (2017) conclude by
explicitly challenging the image of parties being promise breakers.
107
As noted above, voters
are generally quite good at evaluating whether cabinet parties fulfil their election pledges
(Duval & Pétry, 2020; Thomson, 2011; Thomson & Brandenburg, 2019) and they also seem to
reward such behaviour (cf. retrospective pledge voting). Distrust in parties is fuelled when they
abandon election promises after taking power (R. Rose & Wessels, 2019) and governing parties
that fulfil a higher share of election pledges are more likely to prevent electoral losses (e.g.
Matthieß, 2020). Insofar as they are at play, dynamics like these might nurture politicians’ urge
or tendency to deliver. And this urge is thought to increase the risk of ethno-territorial collision
and decreases the room for compromise, because it pushes ethno-territorial debates to the
political forefront while also nurturing the kind of intransigence that can complicate, paralyse,
and escalate negotiations.
H7: The saliency of decentralization is positively related to ethno-territorial conflict
Fourthly, direct electoral pressures should also burden accommodative power-sharing between
segmental leaders. More specifically, I take into account the proximity of federal elections,
which has received surprisingly little attention in previous studies on (ethno-territorial) conflict.
The ethnic outbidding dynamics mentioned above (Rabushka & Shepsle, 1972) are generally
thought to be most notable in the run-up towards elections, when party competition truly
107
Electoral pledges refer to “commitments in parties’ programs to carry out certain policies or achieve
certain outcomes.” (Thomson et al., 2017, p. 528) Evidently, the degree of pledge-keeping strongly differs
between contexts. For a recent overview of the latest comparative research, see Naurin, Thomson, Cotta, and
Russo (2020).
125
flourishes. This is certainly the conventional wisdom in Belgium, where conflictual behaviour
is presumably strengthened by the fact that parties choose to compete only within their own
language group (as discussed above) (Caluwaerts & Reuchamps, 2015; Deschouwer, 2002,
2006, 2013b; Reuchamps, 2013; Swenden, 2006, 2013a).
108
As explained above, the
combination with the Belgian list-PR system is argued to provide parties with very few pre-
electoral incentives for accommodation (Horowitz, 1991, pp. 167-176; 2002, p. 23). Rather, in
the build-up towards the election, this separated competition within each segment is thought to
create a major cleavage between both sides, which must be bridged after election day
(Deschouwer, 2002, p. 79). If parties fear being sanctioned for accommodative behaviour, this
fear and the resulting confrontation-seeking attitudes should become increasingly noticeable as
judgment day approaches.
H8: Election proximity is positively related to ethno-territorial conflict
3. Overview
In summary, this chapter discussed the two contrasting hypotheses that can be found
within the literature (decentralisation as pacification vs. the paradox thesis) and reflected upon
the role of several contextual factors. Most notably, it explains why the paradox thesis is
108
Several exceptions can be noted (see fn. 14 in this work).
Figure 5. Overview of hypotheses
Decentralization Ethno-territorial conflict
(frequency and/or endings)
H1-2
(-)
H3-4
(+)
Regionalist parties
(strength & cabinet participation)
Decentralization saliency
Election proximity
H5-6
(+)
H8
(+)
H7
(+)
126
particularly likely to hold in the context of Belgium and why both the saliency of
decentralization and electoral and regionalist party pressures should fuel tensions. The
hypotheses discussed throughout this chapter are visualized in Figure 5.
127
Part III.
Empirical section
128
129
Chapter 6.
Data and method
109
This chapter presents the measurements and data of cabinet conflicts and their features. The
approach described here can easily be applied to various sources, periods, policy levels, and
countries. A descriptive overview of the data provides readers with a general picture. Finally,
I discuss the methods and analyses used to study whether decentralization pacified or fuelled
ethno-territorial tensions.
1. Mapping cabinet conflicts and conflict features
As described in Chapter 1, empirical analyses of cabinet conflicts are scarce. This
should not come as a surprise, as the mapping of cabinet conflicts and conflict features comes
with many challenges. In facing these challenges, existing works provide a solid and welcome
point of departure. However, they also left room for improvement. Indeed, and as discussed
throughout this chapter, many of them build on (coding) choices that are either unclear (not
mentioned) or suboptimal. Building on and adding to previous works, this doctoral thesis uses
a detailed codebook, provides an intercoder reliability test, operationalizes conflict intensities
directly, and includes several variables to keep track of cases of doubt. The resulting datasets
might add to the field in several ways. First and importantly, the underlying methodology is
easily applicable to other sources, periods, policy levels, and countries. Second, they open many
scholarly doors by allowing us to study and (re-)evaluate a wide range of questions and theories
– including those on cabinet conflict. What is the role of cabinet composition, coalition
agreements, external factors, etc. (for factors, cf. T. Bergman, Gerber, Kastner, & Nyblade,
2008)? Does ideological heterogeneity matter? (e.g. Müller et al., 2008, pp. 21-23; Warwick,
1979, 1992) Are leftist cabinets more conflict-prone than right-wing executives? Does the
profile of the Prime Minister matter? (e.g. Timmermans, 2006) Or that of the party leaders?
What about the role of approaching elections or the presence and usage of social media?
109
The discussion presented here was previously described (often less elaboratively) in publications in
Politics of the Low Countries (Vandenberghe, 2022b) and Ethnopolitics (Vandenberghe, 2022c), and are
expected to be covered in two articles that are under review at the time of publication (Vandenberghe, 2022d,
2023).
130
(Zeitzoff, 2017; Zhuravskaya, Petrova, & Enikolopov, 2020) Who is more inclined to engage
in conflicts? And which parties are mostly involved? Why are some periods characterised by
intense conflicts while others are not? These are but some of the many questions that can be
addressed (or re-assessed) using the data presented here. In their turn, these answers might
nourish public and political debates alike and provide them with a more factual basis. Perhaps,
this is the data’s greatest contribution.
Throughout all coding efforts, four golden rules were respected (both in identifying
conflicts and in coding conflict features). (1) First, context-blind rule fetishism was avoided.
Both coders (myself and the second coder for the inter-coder reliability test) always kept the
specific context in consideration and opted for the code that best grasped the situation at hand
– even if this implied exceptional or deviant coding choices. For validity’s sake, case-specific
deviations were always accepted (as long as they are reported in the logbook). (2) Second, code
choices always needed to apply to the ‘core’ of the conflict (even when they don’t apply to all
aspects or phases of a conflict). For instance, when a fierce clash between the Flemish
regionalists (N-VA) and Christian democrats (CD&V) suddenly triggers a side remark of the
MR (Francophone liberals), which publicly regrets the childlike behaviour of its coalition
partners, the conflict was still to be coded as a clash between N-VA and CD&V (intra-Flemish).
(3) Third, deciding is key. No combinations of codes could be used and code “T” (doubts) was
maximally avoided (without impairing the validity of the coding effort). (4) Fourth and finally,
prudence was preferred over trigger happiness. When doubting between two codes, coders were
instructed to opt for the least heavy one: no segmental sides rather than segmental sides, one
conflict rather than two conflicts, no imposition (outcome) when there is doubt, etc. Because of
this reluctance rule, the resulting picture might be an underestimation of the harshness of real-
life politics. But this is preferred over the alternative: the risk of systematic exaggeration.
1.1. Sources and sample selection
Most existing empirical studies on cabinet conflicts have used (amongst others)
quantitative content analyses of printed newspapers or keyword searches in digital newspaper
archives to track disputes (e.g. Blondel & Müller-Rommel, 1993; Marangoni & Vercesi, 2015;
Moury & Timmermans, 2013; Timmermans & Moury, 2006). My own exploration of such
sources thought me some noteworthy lessons. Importantly, unless a very large team of coders
is used, it is infeasible to read and code all articles on all pages of every day’s newspaper of
every single day. Longitudinal analyses of printed newspapers only seem to be feasible when
coders focus on samples: certain journals, certain issues (e.g. the newspaper edition of every
other day in the week), certain pages, and perhaps even certain articles (based on their titles).
My own attempts to find Belgian cabinet conflicts this way only exposed the most significant,
131
eye-catching conflicts but failed to provide a consistent view of the smaller tensions that truly
grasp the everyday power-sharing atmosphere (grumpy sneers, short-lived criticism, etc.). The
sample approach only exposed some of these small tensions, while many clashes of equal
magnitude remained hidden. Only an infeasible reading effort could overcome this.
Numerous other sources proved to be prone to the same feasibility problem. Others were
manageably short but failed to report most small clashes, or were considered unfit for reasons
of data availability (e.g. only provided a fragmented view of some years, interrupted by
problematic gaps). For example, I explored the Political Data Yearbooks of the European
Journal of Political Research (EJPR), high-standard magazines (e.g. Knack), parliamentary
records (e.g. plenary debates), and the radio and TV archives of the Flemish public broadcaster
(e.g. radio news, Het Jour naal, Villa Politica, De Zevende Dag, Terzake). Facing these
difficulties and limitations, I opted for alternative sources.
Unfortunately, no single source covered the entire post-1979 (let alone post-war) period.
Hence, two sources were used. The bulk of coding is based on the political yearbooks of the
Res Publica journal (1979–2006; 1762 pages). These yearbooks only go as far as 2006.
110
The
most recent period (1995–2018) is coded using keyword searches in the Gopress digital news
archive. The following sections discuss both sources (yearbooks and news archives). The
extensive overlap between both datasets (12 years) allows for a nuanced comparison and
interpretation, which is presented later in this chapter.
1.1.1.
Political yearbooks (1979-2006)
The bulk of the coding is based on the political yearbooks of Res Publica. This is the
predecessor of the journal Politics of the Low Countries, an academic journal that specializes
in Belgian and Dutch politics and that is the official journal of the political science associations
of Flanders (VPW), Francophone Belgium (ABSP), and Luxembourg (LuxPol) in cooperation
with the Dutch Political Science Association (NKWP) ("Politics of the Low Countries," n.d.).
In the chapters labelled ‘Overview of the Belgian Political Events’ (Overzicht van het Belgische
Politieke Gebeuren), these 28 yearbooks provide systematically available expert summaries of
110
There is also an overview of the political year 1978 (Deweerdt, 1979). However, this year was not fully
included in the dataset/analyses because I want to keep constant the risk factors that make Belgium
particularly paradox-prone. The split of the traditional parties is a key factor, but this process was only
completed when the social democratic party split at the end of 1978 (De Winter et al., 2009; Van Bosstraeten,
2008; Verleden, 2009).
132
Belgian politics that are comprehensive while maintaining an impressive eye for detail and
lower-level incidents (small sneers, frustrations, etc.). An overview of the yearbook chapters,
their authors, and their length is provided in Appendix 2.
Figure 6. Length (number of pages) of the coded chapters
of the Res Publica yearbooks (1979-2006)
Being part of the Winkler Prins editorial staff since February 1976, the main author of the
yearbook overviews (M. Deweerdt) already had years of experience by the time he started
writing the reports for Res Publica. Throughout the year, he collected thematic maps with
newspaper articles on e.g. foreign policy, defence, institutional debates, etc. The bulk of these
maps consisted of articles from the Flemish newspapers for which he worked as a journalist
(De Standaard and De Financieel-Economische Tijd/De Tijd). These articles were always
complemented with articles from Francophone newspapers, most notably Le Soir and La Libre
Belgique. This time-consuming effort allowed him to construct a solid overview of the political
year. In doing so and ‘from the beginning, the goal was to provide an as comprehensive
overview as possible of what was considered to be relevant and ‘to be conserved for history’’
(Deweerdt, 2021; personal correspondence by mail).
Throughout the years, the ‘overviews of Belgian political events’ in the Res Publica
yearbooks became more lengthy (Figure 6). However, there is no statistically significant
correlation between the length of the coded chapters and the number of ethno-territorial
conflicts I found in them (Pearson’s R= -0.217; p= 0.268). Contrarily, a positive correlation
does exist between this length and the overall number of observed clashes (incl. non-ethno-
territorial ones). This correlation is significant at the .05-level (two-tailed; p= 0.037). This adds
a relevant nuance to the analyses of the relative frequencies of ethno-territorial conflicts (their
frequency as a % of all cabinet conflicts). But it does not make these analyses worthless. The
0
20
40
60
80
100
120
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
133
correlation is only moderate (Pearson’s R= 0.396) and as discussed in the empirical chapters,
analyses of the absolute ethno-territorial conflict frequencies lead to the same conclusion on the
paradox thesis.
1.1.2. Digital news archives (1995-2018)
The conflict data covering the most recent period (1995-2018) is based on a coding
effort of digital news archives (Gopress, recently re-branded as Belga.press). Why are such
archives relevant sources? Similar archives are available in most countries and typically cover
many years if not decades. Hence, the usage of digital news archives is a fruitful avenue for
comparative and longitudinal reflections. It also avoids the feasibility problem that characterises
e.g. printed newspaper analysis (where no prior keyword selection can downsize the sample of
articles). Finally, using such archives has been shown to result in a wide selection of numerous
clashes of varying sizes. One example is the study of Marangoni and Vercesi (2015), who
exposed 851 Italian cabinet conflicts between 1996 and 2011. In using digital news archives as
a source, sample selection is a key issue (with evident repercussions). Here, two choices were
made.
(1) First, I focus on news agency articles. Specifically, I searched the full Dutch selection of
Belga, the largest and most authoritative Belgian news agency. For several reasons, news
agency articles are preferred over journal or magazine articles. First, most incidents are covered
by various journals or magazines, resulting in a large selection of similar articles without added
value. Second, using digital news archives avoids the issue of having to choose a selection of
specific journals or magazines (to keep coding feasible). Third, news agency articles include all
articles and news releases of all days (which is a major asset). Finally, as compared to journals
and magazines, news agencies like Belga also use more straightforward headlines, which
further facilitates data collection.
Of course, reading all articles of the Belga news agency would be infeasible. Hence, a keyword
search was used in an effort to exclude irrelevant articles. In this respect, a second choice needed
to be made. (2) Which keywords should be used? Acting as a sieve, this choice determines
which situations will be exposed and as such, it has a major impact on the resulting cabinet
conflict data. Previous studies typically use keywords referring to cabinet conflicts directly,
such as ‘contrast’, ‘conflict’, ‘disagreement’, ‘struggle’, or ‘against’ (Marangoni & Vercesi,
2015, n. 11). While this approach has an evident advantage (it strongly limits the number of
articles that should be read), it also comes with two major shortcomings. First, it is hard to
establish an exhaustive list of relevant concepts. Second, it leaves us blind to the many conflicts
that are not explicitly labelled as such. Not all articles on conflicts actually mention such terms.
To avoid missing conflicts, a more thorough two-step approach was used. To begin with, a
134
keyword search with Boolean operators was used to find all articles about a certain coalition
(Table 1). To exclude irrelevant articles, this search had to apply to articles’ texts and titles. In
a second phase, I read all headlines of the resulting selection of tens of thousands of articles.
Articles were selected when their headline included any of the conflict indicators discussed
below, any pleas against something (e.g. ‘Proposal X faces criticism’), or any potential rebuttals
against sneers (e.g. ‘Proposal X is not unrealistic, says Prime Minister’). Essentially, the only
excluded articles were the ones whose titles fail to indicate any kind of disagreement and articles
that were certainly not about the coalition partners (e.g. ‘North Korea threatens the US’). In this
phase, and to avoid missing conflicts, being overly inclusive was preferred to being overly
restrictive. The gold was sieved from the resulting selection in a third phase, when the 9,547
resulting articles were fully read to find and code conflicts by hand. As no Dutch Belga titles
were available for the period between 15 May 1997 and 31 December 2000, French archives
were searched. As a result, the final selection includes 1,213 French articles.
Table 1. Gopress keyword search
Boolean operators
Seven distinct references to (Belgian) cabinets (separated by OR), all cabinet members (ministers, secretaries of state,
Prime Minister) (separated by OR), all names of coalition parties (separated by OR) and all names of coalition party
presidents (separated by OR), excluding (NOT) nine irrelevant recurring article titles (separated by OR).
For a cabinet with n parties and m cabinet members (ministers and secretaries of state): (“reference to cabinet
1”OR“reference to cabinet 2”OR(…)“reference to cabinet 7”OR“name party 1”OR“name party 2”OR(…)“name
party n”OR“name PM”OR“name cabinet member 1”OR“name cabinet member 2”OR(…)“name cabinet member
m”)NOT(“recurring article title 1”OR“recurring article title 2”OR(…)“recurring article title 9”)
Example: Leterme I
("meerderheid"OR"regering"OR"federale regering"OR"Belgische regering"OR"federale coalitie"OR"Belgische
coalitie"OR"wetstraat"OR"PS"OR"MR"OR"VLD"OR"Open Vld"OR"CD&V"OR"CDH"OR"FDF"OR
"Leterme"OR"Reynders"OR"Onkelinx"OR"Dewael"OR"Vandeurzen"OR"Milquet"OR"Vervotte"OR"De Gucht"
OR"Magnette"OR"Laruelle"OR"De Crem"OR"Arena"OR"Turtelboom"OR"Van Quickenborne"OR"Michel"OR
"Laloux"OR"Delizee"OR"Delizée"OR"Wathelet"OR"Chastel"OR"Devlies"OR"Schouppe"OR"Fernandez"OR"Cler
fayt"OR"Maingain"OR"Reynders"OR"Di Rupo"OR"Beke"OR"Thyssen"OR"Somers")NOT ("persselectie"OR
"krantentitels"OR"program of the day"OR"BELGA CORRECTION"OR"1LEAD"OR"Bilan de l'actualité"OR"titres
journeaux"OR"Revue sélective de la presse"OR"Actualité internationale pour la semaine")
135
1.1.3 An indirect view
Two remarks should be made concerning the usage of yearbook and news archive data.
First, all coding efforts are interpretations (by the conflict coder) of interpretations of
reality (by the author of the yearbook or the Belga news agency journalist(s)). The latter also
base their writings on others’ interpretations, e.g. the analyses of fellow journalists or the
statements of the politicians that were interviewed. Surely, even the stories that politicians
themselves share with the press (e.g. right after a tense closed meeting) are an interpretation of
reality. As such, there is not just an issue of ‘double’ interpretation, but a genuine reality of
interpretation accumulations. However, this is unavoidable. There is no ‘direct’ source
exposing cabinet clashes. We are not flies on the cabinet’s walls. Going back in time is
impossible and even conflicts that occur now are only visible to us through reports by others.
Although it serves as a relevant nuance, the potential presence of blind spots should not
withhold us from studying what is visible. The question, then, is which reports to use to obtain
a trustworthy summary of political events. In this respect, the basis for the data presented here
is rather solid. On the one hand, I use academic reports that are systematically written over time
by country experts (Res Publica yearbooks). These reports are published in a publicly available,
established journal whose primary audience consists of country experts (which is illustrated by
the fact that the yearbooks are written in Dutch). On the other hand, I use articles from the Belga
news agency, an internationally renowned source of information that is known for its balanced
and non-partisan presentation of events. Of course, neither of these sources is perfect. But the
search for information sources that are free of errors (the holy grail) is a challenge that is
evidently beyond the scope of this work.
Second, both datasets rest on media sources. These are well-fit for mapping and coding
conflicts. It is widely known that the media focuses on conflict in political news (cf. Lengauer,
Esser, & Berganza, 2012) and uses strategy and game-frames (cf. Aalberg, Strömbäck, & de
Vreese, 2012). If anything, this is an asset. Using media sources thus ensures that few conflicts
slip through the net and remain unexposed. Rather, a significant bycatch was thrown overboard
after checking for conflict indicators myself. I am aware of the debates on the potentially
growing media focus on conflict (e.g. Vliegenthart, Boomgaarden, & Boumans, 2011), which
could distort the data. To some extent, this problem is unavoidable. All sources can be prone to
such trends – including statements by politicians themselves (perhaps, their focus on conflict
increases too). As noted, the only view we have is an indirect one. However, and interestingly,
the resulting data provides a somewhat comforting picture. As Figure 7 shows, the number of
observed conflicts does not show a continuous rise. Strong fluctuations are noted, and while the
numbers did grow during the 1990s, this apparent trend of increasing conflict levels (1) might
of course reflect growing tensions (the early 2000s are well-known for their troubled nature)
136
and (2) did not persist.
111
The most recent years covered by the data (2017 and 2018) even show
the lowest conflict levels in more than a decade (except for 2011). Furthermore, the pacification
tendencies are even more clear-cut in the analyses (on ethno-territorial conflicts) presented
below. If there would be a growing media focus on conflicts, this would (only) imply that this
pacification is even sharper in reality. Note that in Figure 7, and as explained more elaboratively
below, axis ranges are adapted to compensate for the quantitative difference between the
datasets.
112
Figure 7. Absolute frequency of coded cabinet conflicts in Belgium (1979-2018)
(Yearbook data: N= 328; News archive data: N= 1013)
Finally, the use of media sources might receive criticism because their segmental framing
might distort the data. For instance, studies have shown that the Belgian media tend to
extrapolate statements and actions from one politician or party to Flanders and Francophone
Belgium as a whole, often presenting both segments as homogeneous players in a conflictual
111
The Verhofstadt I cabinet (greens, social democrats, liberals) was burdened by numerous quarrels and
small-scale tensions between the coalition partners. The general atmosphere of explicit hostility was
euphemistically dubbed the ‘open debate culture’ by Verhofstadt (Mouton & Vanpeteghem, 2003, p. 287).
112
For the 12 overlapping years between both datasets (1995-2006), the news archive data exposes 2.80 times
as many conflicts (N= 439) as the yearbook data (N= 157). Hence, its axis is about 2.80 times as large (max.
85) as that of the yearbook data (max. 30).
0
10
20
30
40
50
60
70
80
0
5
10
15
20
25
30
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Yearbook data News archive data
137
relationship (Sinardet, 2008, 2012).
113
Using such media sources to code whether conflicts are
fought along segmental lines (between the Flemish and Francophone coalition partners) might
thus result in distortion. However, the media’s terminological trigger happiness does not
really damage this work’s data on segmental sides. Usually, the extensive selection of articles
and the detailed conflict descriptions provided a solid sight of the specific parties and
individuals involved in the conflicts. With few exceptions, there was no need to rely on the
sporadically used segmental frames of the Belga news agency. Furthermore, the risk of
distortion is further limited thanks to the golden rule to prefer prudence over trigger happiness.
As noted above, coders were always instructed to opt for the least heavy code when two codes
might be possible (e.g. no segmental sides or segmental sides).
1.2. Cabinet conflicts: operationalization
Recall that cabinet conflicts are defined as any explicit and antagonistic disagreement
between two or more cabinet members and/or relevant coalition party actors (Chapter 2).
Many situations evidently fit this description. But sometimes, it is unclear whether all elements
of the definition apply. As discussed below, I use several variables to keep track of such
cases of doubt. This approach deviates from standard practices (explicitly tracking doubtful
situations is surprisingly uncommon in previous studies). But it is important. Situations are
often unclear and highlighting uncertainties strengthens the transparency and provides a
welcome counterbalance against the (unconscious) tendency to interpret the data as an
undistorted reflection of reality.
To decide whether disagreement is antagonistic, I inductively constructed six conflict
indicators (Table 2). Note that combinations are possible (e.g. threatening with resignation).
Of course, these indicators grasp conflicts of varying significance, ranging from small-scale
113
For example, scholars have noted that in Flemish current affair programs, ‘the terms ‘Di Rupo’, ‘PS’, and
‘Wallonia’ have become synonyms’, and Francophone broadcasts show a similar tendency to extend the
stances of certain parties to la Flandre and les Flamands as a whole (Sinardet, 2008, p. 66). In regionalist
parties’ discourses too, the Francophone social democratic party PS (Parti Socialiste) is often depicted as the
embodiment or the spokesperson of an essentially left-wing Wallonia (Abts et al., 2019; Dalle Mulle, 2018).
Such dynamics reflect the fact that the PS is the main political force in Wallonia – both historically and
recently (Delwit, 2021b). Elio Di Rupo was the president of the PS between 1999 and 2019 (during the 2011-
2014 period, Di Rupo was Prime Minister of Belgium).
138
quarrels to major political crises. Many studies only focus on major clashes. One example is
the set of 44 conflicts in Dutch and Belgian cabinets (1989-2003) discussed by Timmermans
and Moury (2006) (see also Moury & Timmermans, 2013). But existing works often fail to
pinpoint the line that divides major from minor conflicts. When is a conflict ‘major’ enough to
be included? And more fundamentally, should the significance of conflicts really be used as an
additional criterion? Why exclude smaller clashes? Arguably, we should either code according
to the definition and include them or change the definition to exclude them. As there is no
reason to state that minor conflicts are not conflicts and given the importance of exposing small
tensions to grasp the general power-sharing atmosphere, I code every situation that meets the
cabinet conflict definition (whatever its significance). Of course, differentiating major from
minor clashes can be relevant in certain analyses. But doing so is impossible when the latter are
a priori excluded. Rather, such a comparison necessitates an intensity variable. I included such
a variable, whose operationalization is discussed below.
Table 2. Conflict indicators and intensity levels
In identifying conflicts, many situations were deliberately and systematically ignored. This
applies to speculations (e.g. references to conflicts that were ‘in the air’ or ‘bound to occur’),
Indicator
Description
Intensity
level
Verbal
Pejorative criticism, swearing, scapegoating, anger, etc.
0
Block
Manifest blocking (attempts) (restrictive coding): empty
chair tactics, sharp veto against X despite urgency, etc.
1
Impose
Imposing or trying to impose X on partner (restrictive
coding): faits accomplis, sharp ultimatum for X, etc.
1
Threat
Threatening partner (whatever the threat). Any reference
to future retaliations against the behaviour of partner(s) (if
they do X, we will Y).
1 (2 if
survival)
Branding
Situation is labelled a conflict (or crisis, tensions,
escalation, etc.) by coalition partner(s) or third actor(s).
1 (2 if
‘crisis’)
Survival
Resignation of the cabinet, a coalition party or a cabinet
member is in the balance, openly questioned (incl.
threats), or effectively occurring.
2
Other
Residual category (logbook)
case-
specific
Details: codebook (available on demand)
139
vague references to distrust or displeasure (e.g. ‘there was distrust between the coalition
parties’), and the mere presence of acts or procedures that are part of ongoing investigations
and inquiries (e.g. the establishment of an investigation commission in parliament or the
presence of criticism against the government in a report of such a commission). Because of
their exceptional context, which deviates from everyday power-sharing, conflicts that were part
of the ongoing formation negotiations were not included either (e.g. ‘the party was attacked for
trying to establish an executive that isn’t backed by a Flemish majority in parliament’). In the
context of cabinet formation negotiations, only clashes that clearly concerned the caretaker
cabinet were taken into account (e.g. ‘in strong terms, the party was blamed for being
responsible for the fact that the present cabinet fails to be backed by a Flemish majority in
parliament’). In the final stages of the formation negotiations, the cabinets that are about to be
launched are often already attacked by the opposition parties in spe (who target the first known
choices of the cabinet, its composition, etc.). While doing so, some of the latter are still part of
the caretaker government. However, their attacks against the future cabinet are not included as
cabinet conflicts (as the actors are involved in their role as future opposition parties). After all,
the aim of this study is to gain a better understanding of Flemish-Francophone cooperation in
power-sharing executives, not to expose relations and dynamics between cabinet and
opposition.
Finally, distinguishing consecutive conflicts is not always easy. Political clashes are often
accumulations of successive attacks, interrupted by short breathers or escalating as they go. As
lurking compromises often slip out of politicians’ hands, perspectives of stability can quickly
make room for a revival of tensions. As such, the demarcation line between clashes is often
quite blurry. In this work, situations are coded as ‘new’ cabinet conflicts if (1) the cabinet
changes, (2) the core issue changes, if (3) the preceding conflict was clearly settled (the
discontinuation of tensions must be clear), and when (4) the clash disappeared from the radar
for a sufficiently long period. Here, and in line with the prudence rule discussed above, I use a
threshold of 10-14 days: if there is a gap of 10-14 days between both situations in which the
initial conflict was no longer salient (disappeared from the agenda), they are coded as separate
conflicts. This rule was only breached if the situation clearly continued. To expose whether this
was the case, I used all available elements presented in the sources at hand (ranging from
political comments on social media to formal initiatives in parliament). I opted for a range (10-
14 days) rather than a single threshold (e.g. 10 days) to allow for sufficient flexibility and a
case-specific approach. While the 10-14 day threshold is of course arbitrary to some extent (like
all thresholds), it is not groundless. This period ensures that both the coalition partners and the
opposition had the opportunity to raise the issue during the weekly plenary debate in the
Chamber of Representatives (on Thursday), a highly mediatized session that is traditionally a
primetime moment for political tensions. If tensions still exist, they are very likely to be
140
mentioned during this debate or in its aftermath, when parties react to what was said. A silence
of 10 days (or more) indicates that this was not the case.
Surely, the approach presented here is not exhaustive nor free of shortcomings. But it does
provide the most solid, detailed, and comprehensive view of cabinet conflicts in Belgium to
date (Yearbook data: N= 328; News archive data: N= 1013).
114
For example, despite using a
less wide definition, this approach exposed as many conflicts in the eight years from 1980 to
1987 (N= 101) as a manual newspaper analysis did for the twenty-seven years from 1961 to
1987 (N= 101) (Nousiainen, 1993, pp. 269-276).
115
Other analyses exposed far fewer
conflicts in Belgium too (e.g. Timmermans & Moury, 2006).
1.3. Conflict features: operationalization
Throughout the following paragraphs, I present the different variables that grasp the
different conflict features (and variables that were derived from them). An encompassing
frequency table of the variables is provided at the end of this discussion.
1.3.1.
Basic conict info
The dataset keeps track of each conflict’s ID number, starting date (month and year),
cabinet, and information on the yearbook or Belga article(s) in which the conflict was found
(page/date of first article). I also provide a general conflict description, which is a copy of the
parts of the yearbooks or Belga articles that describe the conflicts. If several parts of the
yearbook or multiple Belga articles cover the conflict (throughout time), all relevant coverage
is added to the dataset. As such, especially within the latter, there is sometimes some repetition
in these descriptions. The goal, however, is to provide readers with a solid overview of the
situation at hand. Hence, I occasionally added personal notes on relevant elements sketching
the context at hand. Examples include meaningful references to past events (previous debates
or comments that failed to meet the conflict indicators), the date of events that are mentioned
(if relevant), the function of certain actors, or information on the way the situation was presented
114
Conflicts that might not have been between coalition partners (cases of doubt) are not included in these
numbers, nor in the analyses below. Yearbook data: N = 36; News Archives: N = 77.
115
When excluding cases of doubt on whether the relevance criterion applies (in my dataset), this number
is 99.
141
(e.g. ‘This situation was discussed under the heading ‘Tensions within the cabinet’ in the
yearbook overview’). On average, these conflict descriptions are roughly half a page long (211
words in the yearbook dataset; 355 words in the news archive dataset). Some long-lasting,
regime-threatening crises are described in several pages and in total, almost 850 pages of
conflict descriptions are included in the datasets.
1.3.2.
Conict intensity (0/1/2)
To evaluate the power-sharing challenge, getting a solid view of the prevalence of
(ethno-territorial) conflicts is primordial. However, data on their mere quantity only tells us
so much. As with weaponry, it is not only the rate of fire that counts but also the power of the
blows. Twenty shots with a small-calibre pistol are nothing compared to one blow from a
modern tank gun. Next to mapping the prevalence of (non-)ethno-territorial conflicts, I
therefore want to gain sight of their intensity too. Doing so is not evident, as previous attempts
demonstrate.
Scholars have typically grasped conflict intensity (often dubbed ‘seriousness’) by using
certain proxies (rather than direct indicators). For example, in their seminal chapter on
Conflict Management in Coalition Government, Andeweg and Timmermans (2008, pp. 275-
278) classify conflicts based on the seriousness of the threat they pose to the stability or survival
of the cabinet. They take into consideration three proxies: the preferences of actors, the game-
theoretical nature of clashes, and the players involved. First, and as discussed in Chapter 2,
parties can have convergent preferences (a focus on the same issues with compatible positions),
tangential preferences (a focus on different issues that is not incompatible), or divergent
preferences (a focus on the same issues with incompatible positions – or a focus on different
issues with mutually exclusive preferences) (cf. Luebbert, 1984, p. 241; Luebbert, 1986, pp.
62-63). Convergent preferences are ignored “as these do not involve any serious conflict”, and
conflicts involving tangential preferences (more serious) are discerned from those involving
divergent preferences (most serious) (Andeweg & Timmermans, 2008, p. 276). The rationale
behind this is that situations of tangential preferences can be solved by ‘logrolling’ (except
when the conflict is about the cabinet’s priorities), which is not an option when preferences are
divergent. A second and related proxy they use is the game-theoretical distinction between
constant-sum games and variable-sum games. The latter refer to instances in which the sum of
the payoffs of players can differ depending on the strategies they use. As Andeweg and
Timmermans (2008, p. 276) argue: “Constant-sum games concern conflicts that are most
threatening to cabinet survival. Once such high-risk conflicts have broken out, attempts to
resolve them are likely to resemble crisis decision-making.” Third, the seriousness of conflict
is also thought to depend upon the players involved. Andeweg and Timmermans (2008, p. 277)
consider inter-party conflict to pose a greater threat to governments’ survival than
142
interdepartmental conflict, because a party that loses a conflict can leave the government, while
a losing department cannot. A similar tendency to use the actors involved in a conflict as an
indicator of its seriousness can also be found elsewhere (e.g. Nousiainen, 1993, p. 271). In this
respect, and building on such previous works, Marangoni and Vercesi (2015, p. 23) even talk
about the partyness of conflicts as a proxy of their seriousness.
116
Apart from pragmatical concerns (the workload to map these proxies), this brief overview of
various indicators also shows that this way of grasping conflict intensity quickly turns into a
rather complex puzzle. But most importantly, using proxies rather than direct indicators of
intensity comes with a great risk of avoidable empirical slack – especially when the proxies
are chosen on (largely) theoretical grounds.
117
For instance, intra-party conflicts can be very
‘serious’ too (Marangoni & Vercesi, 2015, p. 23 ; 25) and not all clashes involving divergent
preferences are more disruptive than clashes with tangential preferences. Much depends upon
the attitude and intentions of the actors involved. A highly explosive situation that is handled
with caution can cause less damage than a mismanaged trifle.
For these reasons, a different approach was used to map conflict intensity. Trying to minimize
the room for complexity and distortion, I try to grasp it directly using clear indicators of
its direct manifestations. Specifically, I use the conflict indicators discussed above to discern
three levels of intensity (cf. Table 2). After all, they provide a solid (but of course imperfect)
indicator of the intensity of situations. For example, when coalition parties sneer about the need
for tax cuts (verbal indicator), the situation is arguably less intense than when one party uses an
empty chair strategy to block the legislative process (block indicator). Things get even worse
116
The partyness of conflicts refers to the extent to which partisan actors are involved in the conflict. The
authors position interdepartmental conflicts on one end, party government conflicts in the middle, and inter-
party conflicts on the other end. The latter are considered to be the most dangerous ones because they concern
a struggle between at least two constitutive parts of the coalition (Marangoni & Vercesi, 2015, p. 23).
117
Attempts to minimize this slack would result in an extensive range of complicating exceptions. For
example, intra-party conflicts are typically seen as being less serious, unless the subject at stake is the support
for the cabinet or unless the conflict results in the split of a coalition party –with one component joining the
opposition (Marangoni & Vercesi, 2015, p. 23). Indeed, Marangoni and Vercesi (2015, p. 25) have
themselves described an example of such a serious intra-party conflict, which destabilized the Berlusconi IV
coalition as a whole. Situations like this indicate that the authors had other, a priori indicators in mind when
assessing situations.
143
when a coalition partner threatens to resign if the tax cut fails to be implemented (survival
indicator). Simplicity can be an asset, but it should not come at the price of accuracy. Hence,
and in line with the golden rule discussed above, certain exceptions and additional coding rules
had to be added to account for specific or exceptional circumstances.
1.3.3. Ethno-territorial issues (dummy and detailed variable)
Four variables grasp conflicts’ ethno-territorial nature. As discussed in Chapter 2, and
building on the concept of segmental cleavages presented by Eckstein (1966, p. 34), ethno-
territorial conflicts are defined as conflicts between segmental sides (Flemings vs.
Francophones) and/or on ethno-territorial issues (e.g. language policy, state reform). To identify
ethno-territorial issues, I use the codebook of the Belgian Agendas Project (BAP).
118
This
codebook nicely fits the scope of my research, as it is customised to fit the Belgian context (cf.
Walgrave, Joly, & Sevenans, 2019). The BAP is part of the Comparative Agendas Project
(CAP) (Baumgartner, Breunig, & Grossman, 2019), which collects data on the ‘issue attention’
of different institutions and actors (newspapers, parties, etc.). The project is not preoccupied
with conflicts itself. But the categorisation of issues it presents is useful for my
operationalization of ethno-territorial issues. As shown in Table 3, I combine three codes, two
of which are amended to fit this work: (1) intergovernmental relations (e.g. financial transfers,
decentralisation; code 2001), (2) state reform and constitution (2033), and (3) the promotion
and defence of national culture (e.g. language law; 2311). Conflicts on these topics are coded
as clashes on ethno-territorial issues. Conflicts on topics that are clearly linked to any of these
codes are also included (e.g. clashes on the appointment of a Minister due to his/her legacy of
language law violations).
118
Original data collection: Walgrave, Joly, Hardy, Zicha, Sevenans and Van Assche. Funding: European
Science Foundation (07-ECRP-008), Flemish National Science Foundation (G.0117.11N), Belgian Federal
Science Policy (IUAP P7/46).
144
Table 3. Ethno-territorial issues
Note: Amended extraction of codes from the Belgian Agendas Project (Walgrave, Joly, Van Assche,
& Zicha, 2009). Details and exceptions: see codebook (available on demand).
Ethno-territorial issues are mapped on two levels. First, with a dummy variable (ET_issue:
0/1) that grasps whether the topic meets any of the relevant BAP codes (or clearly relates to
them). Second, with a more detailed variable that distinguishes between (0) non-ethno-
territorial topics and five types of ethno-territorial issues (ET_issue_detail). (1) Issues
concerning the language struggle, such as tensions on language laws, language knowledge of
cabinet members, language balance in the cabinet, language usage in facility communities, the
position of the language border, etc. (2) State reform (all debates excluding the ones on
fiscal/financial issues), e.g. conflicts about the distribution of competences, the need for
decentralization/recentralization, the power and organization of sub-states, intergovernmental
relations, the procedure of state reform (timing, scope, etc.). (3) Fiscal/financial issues such as
clashes about the transfers between Flanders and Wallonia, debates on fiscal autonomy, and
tensions on the special finance law that regulates the federal allocations to the sub-states
(criteria, calculation, etc.). (4) The distribution of assets and liabilities (other than the
fiscal/financial issues discussed in category 3). This fourth type refers to discussions about the
distributions of certain (dis)advantages between segments or sub-states, whereby these trade-
offs are explicitly presented in a segmental frame. Examples include certain tensions about the
inter-segmental distribution of EU funding obtained by Belgium (advantage) or the distribution
of night flights between Brussels and the surrounding Flemish municipalities (disadvantage).
A fifth and sixth category grasp (5) combinations and (6) ethno-territorial issues that do not fit
any of the specific categories (1-4). The latter includes e.g. vague claims about the ‘Flemish
BAP
Summary
Code
2001
Intergovernmental Relations: (between sub-states and between the federal and sub-
state level): administrative relations, finance of various levels of government, financial
transfers, tensions between levels of government, powers and finances,
(de)centralization, implementing a law or directive of another level of government,
money transfers from one level to another, issues of coordination and/or the
distribution of efforts, assets or liabilities (e.g. the distribution of efforts in
implementing a climate agreement).
Code
2033
State Reform and Constitution: all issues of centre-periphery relations, the positions
of the sub-states, the language areas, and the Dutch and French language groups at the
federal level.
Code
2311
Promotion and defence of national culture: defence of the Belgian, Flemish, Brussels,
or Walloon culture and identity, the (Dutch) language and its use.
145
dominance in Belgium’ or the future of Belgium. In line with the golden rules discussed above,
the codebook explicitly stresses that the two latter codes should be avoided when possible.
Note that the detailed issue variable is not based on the BAP codebook. It is an own attempt to
cluster topics based on the most commonly noted themes. This typology was established
inductively by browsing the conflict descriptions once the conflicts were mapped. It is arbitrary
insofar as other typologies are possible, but legitimate insofar as it does what it needs to do:
feed the analyses by providing a more fine-grained picture of ethno-territorial conflicts.
1.3.4.
Segmental sides (0/1/2 and detailed variable)
Segmental sides are operationalized as linguistically homogeneous and encompassing
conflict sides. This means that all parties from at least one segment (encompassing) must stand
aside (act ‘as one’), without any party of the other segment joining their ranks (homogeneous).
This is for instance the case when all Francophone parties attack one Flemish Minister (one
segmental side) or when all Flemish parties clash with all Francophone parties (two segmental
sides). It does not suffice that some Flemish and some Francophone actors stand against each
other, for instance when one of the three Flemish parties attacks one Francophone minister. I
am well aware of the fact that this operationalization sets the bar high. As such, the resulting
picture is arguably an underestimation of the number of conflicts that many would consider as
‘ethno-territorial’. Adhering to the golden rule of prudence, cautiousness was preferred over
ethno-territorial trigger-happiness.
A general variable (SGS2.0) discerns the presence of 0, 1, or 2 segmental sides. It also grasps
when conflicts are described by the news agency or yearbook overview author in segmental
terms that could not be verified based on the information presented in the coded texts (code 6;
e.g. ‘proposal X was fiercely criticised in Flanders’). A second variable (SGS_detail)
provides a more detailed account of the segmental nature of conflict sides. It distinguishes
six types. To begin with, conflicts (0) with segmentally mixed sides. These occur when one of
the conflict sides includes Flemings and Francophones (e.g. a conflict between the Flemish and
Francophone liberals on the one hand and the social democrats on the other). As the cabinet
always includes parties of both segments, this code was also used if the government or its policy
as a whole was attacked. Next, I discern conflicts (1) among Francophones, (2) among
Flemings, (3) between some Francophone and some Flemish actors, (4) between all
Francophone and some Flemish actors, (5) between some Francophone and all Flemish actors,
and (6) between all Francophone and all Flemish actors.
146
1.3.5. Ethno-territorial conflict (dummy)
Based on the variables discussed above, I computed an ethno-territorial conflict
variable. To be considered an ethno-territorial conflict, conflicts must meet the dummy variable
that grasps the presence of ethno-territorial issues and/or be fought along segmental lines.
Concerning the latter, variable “SGS2.0” must be 1 or 2; that is, there must be at least one full
segmental side. To avoid distortion, an exception was made for the Michel I cabinet. Here,
only clashes between two homogeneous and encompassing sides were seen as conflicts along
segmental lines. One segmental side does not suffice. This is because this cabinet included three
Flemish and only one Francophone party (the liberal party MR). Applying the general coding
rule would imply that any clash between MR and some of its coalition partners would
automatically be coded as an ethno-territorial conflict. To avoid this major distortion. a more
restrictive coding rule was needed.
Note that code 6 of the SGS2.0 variable, which grasps segmental terms that could not be verified
based on the information I coded, is not included. Given its ambiguity and the risk of wrongful
extrapolations, such situations are not considered conflicts along segmental lines for the
analyses.
119
This is in line with the general rule of thumb that prudence is preferred over trigger-
happiness.
1.3.6. Conflict outcomes (9 outcomes)
As discussed above, studying the outcomes of conflicts can add substantially to my
evaluation of the paradox thesis. Decentralization has been blamed for burdening the process
of conflict resolution: “granting segmental autonomy […] renders the process of intersegmental
conflict accommodation increasingly more difficult in the long run”; it “undermines the
problem-solving capacity of the other power-sharing mechanisms” (Caluwaerts & Reuchamps,
2015, p. 277). To test whether decentralization has burdened the process of conflict-solving
(H4), I first need to know how conflicts are solved.
To this end, I present a raw typology of conflict outcomes. Specifically, each conflict is
assigned to one of the 9 mutually exclusive outcome categories. In the case of combinations,
the most severe outcome was coded (e.g. when a clash ends with an imposition that triggers a
119
To this end, I computed a new variable for the analyses, in which these clashes (with code 6) are added to
the ‘0’ category (no segmental conflict sides).
147
coalition partner’s resignation, the resignation is coded). The starting point is the distinction
between (1) procedural decisions, (2) compromise, and imposition (Timmermans & Moury,
2006, p. 401). To discern these phenomena, two criteria are used: the presence of consensus on
the decision and the object of the decision. These criteria are discussed consecutively. First, the
question is whether there is consensus on the decision. Decisions are either made with mutual
consent (compromise and procedural decisions) or harshly forced upon coalition partners
(imposition). The latter refers to situations whereby partners impose their will upon others, e.g.
by presenting them with a fait accompli against their will or by threatening them with fierce
retaliation if they fail to give in (e.g. the resignation of the cabinet). The mere lack of explicit
enthusiasm or support on behalf of some coalition parties does not suffice. It’s not about
disliking the taste of a dish, it’s about forced feeding. Additionally, (3) non-segmental and (4)
segmental impositions are discerned. The latter refer to situations in which one segment
imposes its will, e.g. when the Flemish coalition parties use their joint strength to adopt a
proposal against the explicit will of the Francophones. Other impositions are non-segmental,
e.g. when the social democrats use the threat of cabinet resignation to impose their will upon
the liberal parties. Note that such impositions are still possible, even in a consociational system
that’s packed with mechanisms to avoid them. This might seem puzzling, but these mechanisms
are political bazookas. Employing them (to the full) risks triggering even deeper crises, puts the
survival of the cabinet at stake, and would trigger major debates on the future and viability of
Belgium at large. Accordingly, openly frustrated coalition partners often decide that whatever
issue is at stake is not worth the trouble. Rather than throwing all consociational spanners in the
works, they often choose to simply resign to the decision being imposed upon them. One
example is the clash in Martens I (December 1979) concerning the decision NATO had to make
about the manufacturing and stationing of 572 intermediate-range missiles in Western Europe
(c21 in the yearbook data). Determining the Belgian stance vis-à-vis this decision triggered
major tensions. In the context of growing national and international protests against the arms
race, the social democratic parties wanted to postpone the decision to provide room for prior
negotiations with the Warsaw Treaty Organization. Nevertheless, on December 12, Prime
Minister Martens (CVP; Flemish Christian democrats) announced the stance of his cabinet in
the Chamber of representatives: under certain conditions, the cabinet agreed with the
modernization of the European weapon arsenal. This decision was tied to the survival of the
cabinet with a vote of confidence and as such, the social democrats were put with their backs
against the wall. As they wanted to avoid a major crisis, they had no choice but to swallow the
decision that was crammed into their mouth.
The second criterion is the object of the decision. Parties can make a substantive decision on
the nub of the problem (compromises) or a merely procedural decision. The former refers to an
instance in which politicians agree upon a solution for the problem at stake (e.g. the decision to
148
opt for tax cuts in a conflict about their desirability). Whether the solution actually solves the
underlying problem (e.g. low economic growth) does not impact the coding. Assessing this for
each case, insofar as it is possible, would lead us way too far. I want to map how elites settle
tensions, not whether their policies are ‘working’ (whatever that means). Similarly, conflict
outcomes need not be balanced nor satisfy all partners to be compromises. What matters is that
politicians voluntarily agree on a decision about the core problem. This differs from procedural
decisions, which only address how the problem will be handled, e.g. deciding to postpone
decisions, establish a committee to study the matter, or install a work group with politicians or
other stakeholders to evacuate tensions. Such efforts are a well-established practice in Belgium,
which is commonly known as ‘encommisionnering’ (e.g. Bouteca, Devos, Moens, Ossenblok,
& Terrière, 2018, p. 53; Witte, 1990, p. 447). Procedural decisions resemble the negative
decision-making described by Bachrach and Baratz (1963, p. 641). Elsewhere, they are referred
to as implicit compromises (Luebbert, 1984, p. 242; 1986, p. 63) or non-decisions (e.g.
Timmermans & Moury, 2006).
Conflicts can also result in the “ultimate outcome” (Nousiainen, 1993, p. 273): resignations of
(5) a cabinet member, (6) a coalition party, or (7) the cabinet. In this respect, two remarks
should be made. First, at the risk of aggravating legal scholars, cabinet resignations include
both formal and de facto terminations. For example, a clash can result in the Prime Minister
proposing the cabinet’s resignation to the king. The latter can refuse to approve the proposal,
but given the gridlock, it can be decided that the cabinet has to continue with a minimal
programme to bridge the gap until the upcoming elections. In situations like this (e.g. Martens
V), the conflict essentially brought the cabinet to its knees. Second and for the same reason,
resignations of coalition parties need not imply cabinet resignations. Keeping validity in mind,
the aim is to code situations as they were experienced in real-life. For example, the exit of the
Francophone greens (Ecolo) was not perceived to be a clash that led to the resignation of the
Verhofstadt I cabinet (neither by the media, observers, nor by the political actors involved). For
them, this was not a deadly conflict for the cabinet, but for Ecolo. By adhering to the way such
major events are interpreted in reality, I wish to avoid the kind of artificial coding that might
broaden the gap between the findings and their real-life relevance and applicability.
Finally, clashes can also (8) fade away or blow over without any palpable result or decision
whatsoever. Despite its relevance, this option is underexposed (and indeed usually neglected)
149
in existing work. Such clashes can evidently resurface later on, but so can all clashes.
120
To
grasp exceptional and peculiar situations, I also included a (9) residual outcome category. An
example of such a residual situation was the result of one of the many conflicts on Voeren (c3),
where Flemish and Francophone protestors clashed with each other and the gendarmery (see
Chapter 3). In January 1979, several French-speaking demonstrators were bitten by a police
dog. Minister of the Interior R. Boel (Flemish social democrats, BSP) publicly approved of the
police action. Being interviewed by the Francophone public broadcaster (RTBF), he
controversially declared that the dog had acted in self-defence (“a agi dans un état de légitime
défense”). Infuriated by these statements, the Francophone social democrats (PS) responded
with fire and fury. On the order of party president Cools, seven of the Francophone staffers of
Minister Boels resigned (including his Francophone head of cabinet). As this outcome doesn’t
fit any of the specific categories, it was coded under the residual category.
My coding of outcomes comes with several weak spots and nuances. First, as always, there
is a potential gap between the observed and the actual conflict endings. Most notably, clashes
can be settled ‘behind the scenes’, which presumably applies to some of the clashes that blew
over. But by definition, we cannot know which and how many clashes were settled discretely.
Estimating their frequency would be guesswork. Rather than speculating, I choose to focus on
what is observable and verifiable. Here too, the presence of blind spots should not withhold us
from studying what is visible. Second, politicians might continue to negotiate after a conflict
has been settled. For instance, procedural decisions might provide the room needed to continue
discussions and as such, some of them might eventually result in a more encompassing
compromise on the nub of the problem. The dataset, however, reports how elites end conflicts,
not how issues develop after pacification (which would require far more elaborate assessments
based on in-depth reconstructions). Being preoccupied with conflict and conflict resolution
dynamics, my primary concern is to grasp what dampened the flames, not what grows on the
burnt soil in the long term.
120
As noted above, when a clash disappears from the radar long enough (10-14 days, depending on the
situation), the first consecutive clash on the same issue is coded as a distinct conflict. In these situations, the
first clash is considered to ‘blow over’ (except when this would be artificial). This rule only applied to roughly
two dozen of all 328 cases in the yearbook dataset.
150
1.3.7. Intra-party conflicts and cases of doubt (dummies)
Dummy variables grasp whether conflicts are fought between actors of the same party
and identify situations that might not fully fit the definition of cabinet conflicts. First, it is
sometimes unclear whether a clash is actually fought between coalition parties; e.g. “The
proposal of the Prime Minister was called ‘dumb and inhumane’ by several parties”. Second,
in some instances, there is even doubt on whether the conflict is fought between political actors;
e.g. “The proposal of the Prime Minister was called ‘dumb and inhumane’ by many”. The latter
might well be civil society organizations, citizens on social media or other public fora, scholars,
journalists, etc. And third, it is not always clear whether the relevance criterion applies; i.e.
whether all actors that constitute the conflict actually represent their party at large (e.g. “The
PM’s proposal was called ‘dumb and inhumane’ by certain MPs of coalition parties too”). For
all other variables, situations of doubt are identified with code “T” (twijfel, Dutch for doubt).
As noted above, the use of this category is maximally avoided (within the boundaries of
responsible coding).
1.3.8. Logbook and intercoder reliability test
Coding transparency and consistency were enhanced by an extensive logbook covering
troublesome cases and situations of doubt. This logbook is readily available on simple
demand.
121
The resulting dataset on cabinet conflicts was subjected to an intercoder reliability
test, measuring the degree of resemblance (Cohen’s κ) between original coding efforts and
those of an external coder (Table 4). Cohen’s kappa accounts for change agreement (Cohen,
1960) and is the most commonly used statistic to assess nominal agreement between two raters
(Warrens, 2015). Its value is 0 when the observed agreement equals the agreement expected by
chance and 1 when there is perfect agreement. The second coder was a social sciences student
with layman’s knowledge of Belgian politics. She was briefed on the codebook and practised
(a day) before coding a random sample of 93 conflicts: 33 yearbook conflicts (~10% of this
dataset) and 60 news archive conflicts (~5% of this dataset). Following Landis and Koch (1977),
the results expose substantial, almost perfect, or perfect agreement on all but one variable. For
the variables grasping segmental sides on a more aggregated level (SGS2.0), the level of
intercoder agreement is moderate.
121
Notably, the logbook is written in Dutch, just like most conflict descriptions and the codebook.
151
Table 4. Intercoder reliability test (N = 93)(a)
(a) 33 yearbook conflicts (~10% of this dataset), 60 news archive conflicts (~5% of this dataset); (b) According to Warrens
(2015), Landis and Koch (1977) provide the most commonly used benchmark guideline for the interpretation of kappa values
by discerning five levels of agreement: slight (0.00-0.20), fair (0.21- 0.40), moderate (0.41-0.60), substantial (0.61-0.80),
and (almost) perfect agreement (0.81-1.00). (c) Outcome variable: only included in yearbook dataset.
1.4. Comparing the yearbook and news archive data
All variables presented above are included in both datasets, with one exception.
Because of pragmatical limitations, the outcomes of conflicts are only included in the yearbook
dataset (1979-2006). The yearbook overviews provide very solid and comprehensive
summaries of conflicts from beginning to end: how they started, developed, and ended. This is
not the case for the news archives. The daily articles issued by the news agency provide a far
more fragmented view of clashes and their evolution. Coders still need to put these pieces of
the puzzle together themselves. Unravelling the conflict outcomes for the news archive dataset
would necessitate a very labour-intensive additional coding effort in which the conflict’s
development would need to be reconstructed (e.g. with an approach similar to the one used to
construct the yearbooks, that is by using thematic maps with articles on all kinds of issues).
Doing so was considered infeasible within the boundaries of this project.
Table 5 presents a comparison of both datasets (news archive vs. yearbooks) for the 12
overlapping years (1995–2006) (because of parsimony concerns, variables on the dates of
conflicts are not included). It shows that news archive coding captures even more cabinet
conflicts (482) than the already extensive coding of yearbooks (175). This raw quantitative
difference is taken into consideration when presenting results (see below). But in general, as
the analyses presented throughout the following chapters will show, the aggregated results on
most variables are highly comparable (average difference of the visualised variables: 4
Variable
Raw % of
agreement
Cohen’s κ
Sign.
Interpretation of Cohen’s κ
(Landis & Koch, 1977)(b)
Intensity
90.32
0.788
< 0.001
Substantial
Coalition partner (doubt)
100.00
1.000
< 0.001
Perfect
Political actors (doubt)
97.85
0.662
< 0.001
Substantial
Segmental sides (0/1/2/6)
84.95
0.585
< 0.001
Moderate
Segmental sides (detailed)
82.80
0.794
< 0.001
Substantial
Intra-party
100.00
1.000
< 0.001
Perfect
Relevance criterion (doubt)
93.55
0.748
< 0.001
Substantial
Ethno-territorial conflict
82.80
0.616
< 0.001
Substantial
Ethno-terr. issue
91.40
0.762
< 0.001
Substantial
Ethno-terr. issue (type)
92.47
0.816
< 0.001
Almost perfect
Outcome(c)
90.91
0.912
< 0.001
Almost perfect
152
percentage points). Indeed, the values on most variables are very similarly represented,
implying that the conclusions I draw on e.g. the prevalence of ethno-territorial conflicts or the
rather marginal frequency of clashes along segmental lines are backed by both datasets. For
example, the percentage of cabinet conflicts that is ethno-territorial in nature is equally low
(18%), most conflicts do not involve segmental sides (85% vs. 93%), almost no conflicts are
intra-party clashes (2% vs. 1%), the Dehaene I coalition faced far less conflicts than its
successors, etc. The most notable difference between both datasets is their view of the
intensity of clashes (differences up to 21 percentage points). The yearbook dataset provides a
far more ‘fiery’ picture. Presumably, this can be at least partly explained by the fact that the
threshold to be covered by the numerous daily articles issued by the Belga news agency is far
lower than the threshold to be included in the Res Publica yearbooks (which are only issued
once a year). The authors of the latter face a greater need for selectivity. Within this selection
process, capturing situations of historical relevance was the main ambition. As M. Deweerdt,
the main author of the yearbooks, explains: ‘from the beginning, the goal was to provide an as
comprehensive overview as possible of what was considered to be relevant and ‘to be conserved
for history.’’ (Deweerdt, 2021; personal correspondence by mail) This search for long-term
relevance might explain why several smaller tensions were less commonly covered. Where
needed, I stress and reflect upon this difference throughout the analyses.
Table 5. Dataset comparison: yearbooks vs. news archives (1995–2006)
Yearbook data
News archive data
Difference
N
%
N
%
|pp|
Conflicts
175
100.00
482
100.00
0.00
Cabinet
Dehaene I
1
0.57
6
1.24
0.67
Dehaene II
50
28.57
138
28.63
0.06
Verhofstadt I
86
49.14
178
36.93
12.21
Verhofstadt II
38
21.71
160
33.20
11.49
Ethno-territorial
0 No
144
82.29
394
81.74
0.55
1 Yes
31
17.71
88
18.26
0.55
Intensity
Level 0
97
55.43
368
76.35
20.92
Level 1
53
30.29
61
12.66
17.63
Level 2
25
14.29
52
10.79
3.5
T doubt/missing
0
0.00
1
0.21
0.21
Coalition partners
1 yes
157
89.71
439,0
91.08
1.37
153
T doubt/missing
18
10.29
43,0
8.92
1.37
Doubt: political actors
0 no doubt
168
96.00
457
94.81
1.19
1 doubt
7
4.00
25
5.19
1.19
Segmental sides
0 none
148
84.57
446
92.53
7.96
1 one
11
6.29
25
5.19
1.1
2 two
11
6.29
9
1.87
4.42
6 framing
5
2.86
2
0.41
2.45
Segmental sides detailed
0 mixed
61
34.86
110
22.82
12.04
1 intra-FR
12
6.86
56
11.62
4.76
2 intra-FL
22
12.57
84
17.43
4.86
3 partial
41
23.43
151
31.33
7.9
4 full FR
7
4.00
14
2.90
1.1
5 full FL
4
2.29
11
2.28
0.01
6 full FR & FL
11
6.29
9
1.87
4.42
T doubt/missing
17
9.71
47
9.75
0.04
Same party
0 no
157
89.71
454
94.19
4.48
1 yes
3
1.71
5
1.04
0.67
T doubt/missing
15
8.57
23
4.77
3.8
Doubt: relevance
criterion
0 no doubt
159
90.86
393
81,54
9.32
T doubt/missing
16
9.14
89
18,46
9.32
Ethno-territorial issue
0 no
153
87.43
399
82.78
4.65
1 yes
22
12.57
81
16.80
4.23
T doubt/missing
0
0.00
2
0.41
0.41
Ethno-territorial issue
(detailed)
0 not
153
87.43
399
82.78
4.65
1 linguistic
8
4.57
20
4.15
0.42
2 state reform
7
4.00
18
3.73
0.27
3 fiscal/financial
2
1.14
2
0.41
0.73
4 distributive
4
2.29
18
3.73
1.44
5 combination
0
0.00
11
2.28
2.28
6 other
1
0.57
12
2.49
1.92
T doubt/missing
0
0.00
2
0.41
0.41
154
1.5. Data availability
Following an embargo, the datasets will be available on 1 November 2023 in Mendeley
Data (Vandenberghe, 2022a). Meanwhile, data can be made available by the corresponding
author, upon reasonable request.
2. Method and analyses
To answer the research questions, I combine a range of descriptive and explanatory
analyses. (1) First, I present and study the evolution of ethno-territorial conflict frequencies
(Chapter 8). Doing so might falsify the paradox thesis (by showing that no conflict escalation
occurred), but it can never verify it (observed escalations might be due to other factors). (2) To
truly assess whether the growth or decline of conflicts relates to decentralization, explanatory
analyses are needed (Chapter 9). (3) Finally, attention should not only go to the presence or
absence of conflicts but also to the way they are handled or solved once they arise. Several
scholars have argued that decentralization might pacify tensions in the short run while making
it harder to settle the clashes that still arise (e.g. Caluwaerts & Reuchamps, 2015). To assess
this concern, I also study conflict outcomes (Chapter 10). The following paragraphs provide
a general overview of the basics of these analyses.
2.1. Conflict frequency evolutions
To assess how the ethno-territorial conflict in Belgian power-sharing cabinets evolved
(RQ1), conflict frequencies are assessed from three angles: (1) absolute conflict frequencies,
(2) relative frequencies (the number of ethno-territorial conflicts as a percentage of all cabinet
conflicts), and (3) the degree to which a cabinet is burdened by ethno-territorial conflicts. The
latter is measured with so-called yearly equivalents: the number of conflicts in a cabinet
divided by that cabinet’s duration in days, multiplied by 365. This calculation allows for a solid
comparison of cabinets with highly divergent lifespans – ranging from less than three months
(Verhofstadt III) to four years or more (Dehaene II, Verhofstadt I and II, Michel I).
Why do I study both absolute and relative frequencies, rather than only focusing on the
raw number of ethno-territorial clashes? Absolute numbers are evidently relevant, but
analyses of these numbers come with an important downside: they are blind to the more general
political context and, more specifically, to the frequencies of non-ethno-territorial clashes.
Recall that one of the golden coding rules is to always keep the specific context in consideration.
The same applies to the analyses of the resulting data. There is a difference between five ethno-
155
territorial clashes in an explosive year marked by dozens of coalition conflicts and five ethno-
territorial clashes in an otherwise conflict-free year. The pertinence of ethno-territorial clashes
is evidently higher in the latter example. Proportions tell us whether and to what extent ethno-
territorial conflict is a predominant source of distress itself (high percentage of all cabinet
conflicts) or rather a side-product of a larger context of confrontational politics (low
percentage). In other words, it exposes whether there is a problem of segmental politics
specifically or coalition politics more generally. As illustrated in Table 6, this additional
perspective is also relevant when assessing conflict trends. Growing absolute numbers of ethno-
territorial conflict can hide their decreasing predominance (as they represent smaller shares of
all cabinet conflicts; e.g. from year 2 to year 3), decreasing numbers might conceal their
growing predominance, stable numbers can hide both dynamics of growing or diminishing
pertinence, etc.
Table 6. Ratio between ethno-territorial (ETC) and non-ethno-territorial conflict
(non-ETC) over time: fictional examples
Year
Total
cabinet
conflicts
Non-ETC
ETC
% ETC
∆ ETC
vis-à-vis
last year
∆ % ETC
vis-à-vis
last year
Year 1
5
2
3
60
Year 2
7
4
3
42.86
=
-
Year 3
18
12
6
33.33
+
-
Year 4
5
2
3
60
-
+
Year 5
4
1
3
75
=
+
Year 6
12
3
9
75
+
=
Year 7
8
2
6
75
-
=
156
2.2.Explanatory analyses
2.2.1.
Analyses
The explanatory analyses shed light on the issue from different angles. First, ethno-
territorial conflict frequency is assessed by modelling the number of ethno-territorial
conflicts of every month covered by the dataset. Here, each month is a case (N= 336). As this
is a count variable without over-dispersion, Poisson regressions are used.
122
In modelling the
data, we should not be blind to non-ethno-territorial conflicts, which account for at least 70 per
cent of all coalition conflicts in the period under investigation. I account for their presence in
two ways. First, by including the number of non-ethno-territorial conflicts as a control variable.
And second, by presenting a second set of analyses.
For this second set, a second dataset is constructed in which cases are individual conflicts (N=
322). Using this dataset, which includes ethno-territorial and other cabinet conflicts, logistic
regression models estimate the chances of conflicts being ethno-territorial in nature. More
specifically, and taking into account the heterogeneity of these conflicts, I not only assess
whether conflicts are generally ethno-territorial in nature, but also provide specific models to
assess whether ethno-territorial issues are at stake and whether coalition partners fight along
segmental lines (Flemish and/or Francophone blocks). Is decentralization related to decreased
chances that a cabinet conflict is ethno-territorial in nature (rather than non-ethno-territorial)?
Or to the chance that, when a cabinet conflict arises, the Flemish and Francophone segments
stand against each other? Or that ethno-territorial issues are at stake? If the answer to these
questions is no, it would be wrong to call these factors determinants of ethno-territorial conflict
per se. If the answer is yes, this would strengthen the evidence for a negative association
between decentralization and ethno-territorial conflict.
Note that the explanatory analyses are conducted based on the data of one dataset only
(the yearbook data; 1979-2006). Combining datasets to include recent years would necessitate
distinct analyses and result in an overly complicated puzzle (the work already includes two
types of regressions and twenty distinct models). For the explanatory analyses, the news archive
122
Although the dependent variable has many zero values (N= 254), theory does not suggest that excess zeros
might be generated by distinct processes from the count values. Accordingly, no zero-inflated model was
used.
157
dataset (1995-2018) is arguably less relevant. Variation on the decentralization variable is far
more limited in this dataset, as it only covers two decentralization reforms. One of these is the
fifth state reform, which only slightly increased sub-state autonomy (far less than other state
reforms, see below). This is in contrast to the four reforms covered by the yearbook data (1979-
2006), which include the massive decentralization boosts of the second, third, and (to a lesser
degree) fourth state reforms. Hence the choice to focus on the historical relevance of
decentralization and leave the most recent period out of sight.
2.2.2. Variables
The main predictor in these regression models is the degree of decentralization. It is
measured based on the yearly self-rule scores from the Regional Authority Index (RAI) (L.
Hooghe et al., 2016; Shair-Rosenfield et al., 2021). These scores are based on a thorough
evaluation of five dimensions: institutional depth, policy scope, fiscal autonomy, borrowing
autonomy, and representation. The result is a solid and authoritative indicator of sub-state
autonomy that fits the theory under scrutiny here. The RAI is widely used within the field and
includes longitudinal data on ninety-five countries (1950-2018), which provides more room for
future comparative analyses than other datasets (e.g. Dardanelli et al., 2019). The self-rule
scores presented in the existing RAI_country dataset also take the autonomy of the provinces
into account. But these are not sub-states. They represent the second tier of the local government
in Belgium (e.g. Bouvier, 2012; Valcke, Reynaert, & Steyvers, 2011). To avoid distortion, I
recalculated the yearly self-rule scores while excluding the provinces. To this end, I used the
same formula that was applied to construct the RAI (deliberation with one of the scholars
behind the RAI allowed me to avoid errors). I first multiplied the self-rule scores of each sub-
state with the proportion of the countries’ population in this sub-state. Next, I added up these
multiplications.
123
As Appendix 3 shows, excluding the province scores makes a huge
123
The calculation includes the following regions: the Flemish community (regionid in RAI= 402), the
German-speaking community (404), the Brussels-Capital Region (406), its predecessor (the Brussels
Agglomeration, 407), and two peculiar categories (498 and 499). Indeed, to avoid double counting due to the
combination of two types of sub-states in Belgium (regions and communities), the RAI authors made two
adjustments. First, a ‘combined authority’ of the French community and Walloon region was constructed
(498). It presents a score as if these institutions would be merged (as is the case in Flanders, where the Flemish
region and the Flemish community are politically fused). Second, an ‘additional authority’ (499) was built to
make sure that the authority of the Flemish and French communities over Dutch- and French-speaking
residents in the Brussels-Capital Region are solidly taken into account.
158
difference (on average, the self-rule scores are twice as large without this refinement). As such,
it deserves recommendation to break with the very common practice of using general RAI data
to study sub-state autonomy.
Using data provided by Parlgov (Döring & Manow, 2020), regionalist party strength is
grasped by the percentage of seats in the Belgian Chamber of representatives of the RW
(Rassemblement Wallon), the FDF (Front démocratique des francophones), the VB (Vlaams
Blok/Vlaams Belang), the VU (Volksunie), and its successor: the N-VA (Nieuw-Vlaamse
Alliantie).
124
These are all Belgian parties (active at the federal level) that adhere to the
definition of regionalist parties provided by Massetti and Schakel (2016, p. 62).
125
The VB is
sometimes ignored in regionalist party analyses (e.g. Deschouwer, 2009; van Haute & Pilet,
2006), as its electoral strength is largely built on its radical right immigration stances rather
than its ethno-territorial agenda (for a discussion, see e.g. Dandoy, 2014). However, the party
emerged as a separatist breakaway from the VU, presents Flemish independence as a primary
objective, limits its operational scope to Flanders, and presents itself as the protector of the
Flemish identity and interest (the party name literally translates as ‘Flemish interest’). It thus
clearly fits existing definitions of regionalist parties (Massetti & Schakel, 2016, p. 62; Türsan,
1998, pp. 5-6). Furthermore, the VB has often attacked its more moderate regionalist competitor
(the N-VA) for ignoring ethno-territorial debates and downplaying its regionalist agenda.
126
To
accommodate those who dismiss the decision to include the VB, I also present all models
124
FDF strength after it became part of the MR (Mouvement Réformateur): based on Delcorps et al. (2014).
Seat shares are preferred over vote shares. This avoids distortion due to the 1995 reform (when the number
of seats declined from 212 to 150) and accommodates the fact that the cartel period of the FDF makes it hard
to estimate its vote share since 1995.
125
“First, they are self-contained political organizations that contest elections. Second, they are
organizationally present and/or field candidates only in a particular sub-territory (region) of the state. Third,
the territorial limitation of their political/electoral activity is a consequence of their explicit objective of
defending only the identities and interests of ‘their’ region. Fourth, as stated by De Winter (1998: 204),
regionalist parties’ core mission is to achieve/protect/enhance ‘some kind of [territorial] self-government’
for their homeland.” (Massetti & Schakel, 2016, p. 62) Source in citation: (De Winter, 1998a).
126
The VB often stresses that the N-VA failed to use its electoral weight to push through its ethno-territorial
agenda. To give just one example, during a recent presentation of a book on the break-up of Belgium (written
by Gerolf Annemans, a key figure within the party and its former president), the now-president Tom Van
Grieken argued that the VB ‘would never make the same mistake’ (Vlaams Belang, 2023).
159
without the VB. In contrast, the Spirit party, which was part of the Verhofstadt II-cabinet after
siding up with the Flemish social democrats, is not considered a regionalist party. Its ethno-
territorial profile was notably ambiguous and subordinate to its left-wing liberal stances (van
Haute, 2011, p. 211). Also, after a thorough selection process, Spirit did not make Massetti and
Schakel’s (2016, pp. 70-72) list of parties adhering to their definition.
When any of these parties participate in a federal cabinet, that cabinet is considered to have a
regionalist coalition partner. As is standard practice in works, overviews, and datasets of
Belgian cabinets (e.g. CRISP, 2023; De Winter & Dumont, 2014; Döring & Manow, 2020; Pilet
& van Haute, 2008), the FDF is no longer considered to be a distinct coalition party after the
establishment of the liberal MR (Mouvement Réformateur) of which it became a component in
2002 (for a background on the 2002 reform, see Dedecker, 2011; Delcorps et al., 2014). Apart
from the conceptual or theoretical rationales behind this choice, e.g. the idea that only parties
with ministers are to be included (Döring & Manow, 2020), the exclusion of the FDF also
ensures that the analyses only include cases in which there was a deliberate political choice to
include regionalist parties in the cabinet (excluding situations in which FDF politicians were
part of the majority because of their position within the MR). This is not only in line with the
golden rule that conflicts should be coded in a way that best reflects the political reality, it also
fits the theoretical expectations behind regionalist parties’ cabinet participation.
The MARPOR dataset (Volkens et al., 2021a) provides a systematically available measure of
the saliency of the decentralization issue in party manifestoes. MARPOR uses trained coders
to divide manifestos into quasi-sentences and assign them to one of its 56 issue categories. The
per301 category measures the saliency (percentage of quasi sentences) of statements showing
“support for federalism or decentralization of political and/or economic power”, including
claims supporting “the continuation and importance of local and regional customs and symbols”
(Volkens et al., 2021b, p. 13). To grasp the electoral saliency of such claims, I use the average
per301 scores in the manifestos of the Flemish and Francophone traditional parties (social
democrats, Christian democrats, and liberals), green parties, and regionalists (RW, FDF, VB,
VU, and N-VA). I am aware of the criticism that MARPOR has received (from different angles)
(cf. Gemenis, 2013). Many of these critiques target the saliency and issue ownership theories
on which it leans (Budge & Farlie, 1983; Petrocik, 1996). Their main concern is that issue
saliency cannot be considered a (solid) indicator of (let alone substitute for) issue positions.
Here, we are interested in issue saliency itself, so this debate is of little interest to this work.
The data provide the best available estimate of party saliency on territorial politics throughout
the period under study. It is the predominant source of studies on parties’ issue emphasis and
on different occasions, its advocates have thoroughly defended its validity and reliability
(Budge, Klingemann, Volkens, Bara, & Tanenbaum, 2001; Budge & Pennings, 2007; Volkens,
160
Bara, & Budge, 2009). Furthermore, the face validity of the data comforts me in stressing its
usefulness in the context of this doctoral thesis. For example, the traditional parties in Flanders
usually attribute more attention to decentralization than their Francophone counterparts (which
is in line with the greater calls for autonomy in Flanders), the saliency of the per301 category
is almost 2,5 times as large in regionalist party manifestos (average: 6.68 per cent of quasi-
sentences) as in the manifestos of the traditional party families (liberals, Christian democrats,
social democrats; average: 2.87 per cent), and the well-known decline in Francophone calls for
additional autonomy since the 1993 state reform is clearly visible in the CMP data.
127
The proximity of federal elections is measured as the number of months since the last federal
election (starting at 1 for the first month after the election). The result is a measure that increases
as the next federal election gets closer. Note that this measure is prone to some distortion for
situations of early elections, which result in lower numbers (fewer months since the last
election). However, and limiting the distortion, political actors typically only know they will
face early elections in the short run-up towards them. As noted by scholars using a similar
approach (Fernandes, Leston-Bandeira, & Schwemmer, 2018, note 8), this measure of election
proximity thus replicates the information level political elites have when conflicts emerge.
Finally, as discussed below, the frequency analyses also assess the number of new conflicts in
each month. To account for the potential distortion of clashes that are not limited to one month
(e.g. a clash starting in May but continuing in June), these analyses include (5) the number of
ethno-territorial conflicts in the preceding month.
The descriptive statistics of these variables are included in Table 7 (see also Chapter 9), which
reflects the results for the two datasets used here. First, the dataset for the Poisson regressions,
which includes the monthly frequency of ethno-territorial conflicts (N= 336). Second, the
dataset used for the logistic regressions, in which cases are individual conflicts rather than
months (N= 322). For each month (Poisson) or conflict (logistic), the value of each predictor is
measured.
127
Comparison of traditional and regionalist parties: own calculations using Volkens et al. (2021a).
Regionalist parties: VU, N-VA, RW, VB, and FDF. Regionalist average without VB= 7.58. From 1995
onwards, scores of the MR (Mouvement Réformateur, of which the FDF became part) are included in the
traditional parties’ calculation.
161
Table 7. Descriptive statistics of dataset 1 (monthly ethno-territorial conflict
frequency) and dataset 2 (conflict characteristics)
2.3. Conflict outcomes
Finally, I assess whether decentralization has burdened the process of conflict-solving
(H4). To do so, I use a two-step approach. I first show how conflicts are solved, using the
typology of conflict outcomes discussed earlier in this chapter. But these only tell us so much.
So, apart from mapping the raw outcomes of conflicts, I also try to grasp the conflict-solving
functionality. Arguably, functionality does not only come in different degrees, it can also mean
different things to different observers. Some might consider a system to be functional if
politicians manage to settle conflicts in consensus, while others might only be satisfied if
politicians reach substantial decisions. Still others might find the mere survival of power-
sharing executives sufficient. At least to some extent, functionality is in the eye of the beholder.
Accordingly, and providing a more nuanced picture, I measure different dimensions of
functionality and present them all. This calculation is based on the conflict outcome variable
presented above. For each conflict outcome, I track whether (1) the cabinet as a whole survived
the conflict, (2) all cabinet members and parties survived the conflict (no resignations), (3) any
decision was made, whatever the decision, (4) a substantial decision was made, that is a decision
on the actual nub of the problem, and (5) whether the coalition parties ended the conflict in
consensus, that is in mutual consent.
Variable
Mean
SD
Min
Max
N
Number of ETCs in month
0.28
0.52
0
3
336
Decentralization
10.63
|10.53|
2.96
|3.20|
4.41
|4.41|
13.98
|13.98|
336
|322|
Regionalist party strength (% of seats)
12.92
|13.35|
2.26
|2.53|
9.43
|9.43|
16.67
|16.67|
336
|322|
Regionalist party in coalition (dummy)
0.16
|0.17|
0.37
|0.38|
0.00
|0.00|
1.00
|1.00|
336
|322|
Electoral saliency of decentralization
2.42
|2.17|
1.73
|1.59|
0.44
|0.44|
4.93
|4.93|
336
|322|
Months since last election
21.83
|22.39|
13.10
|13.27|
1.00
|1.00|
49.00
|49.00|
336
|322|
Number of ETCs in previous month
0.28
0.52
0
3
335
Note: Results for dataset 2 are reported |between vertical dashes| below those of dataset 1.
162
Table 8. Conflict outcome types and functionality dimensions
Functionality dimension
Conflict outcome
type
Cabinet
survival
Member/
party survival
Decision
Substantive
Consensus
Procedural
X
X
X
X
Compromise
X
X
X
X
X
Imposition
X
X
X
possibly
Resignation
possibly
possibly
possibly
Blow over
X
X
Residual
possibly
possibly
possibly
possibly
possibly
Missing
X
X
unknown
unknown
unknown
Total
/100
/100
/100
/100
/100
The scheme presented in Table 8 is used to track the percentage of clashes that meets each
functionality dimension. Boxes with a dark blue background indicate that all cases of the
respective outcome type meet the functionality dimension. For instance, all compromises meet
all functionality dimensions. When conflict outcomes are unknown (missing), they are only
included in the dimensions that grasp the lack of resignations; the logic being that such
resignations would have certainly been reported in the yearbooks. White boxes show that an
outcome type fails to meet a dimension. All impositions, to give just one example, evidently
fail to meet the consensus dimension. Light blue boxes indicate that the respective outcome
type can meet a functionality dimension. For instance, impositions, resignations, and the
residual outcomes can but need not solve clashes substantively. Determining whether they do
requires a case-specific examination. Doing so exposes that all impositions and half of the
residual outcomes in the dataset settled the respective issues substantively. What about the
resignations in the dataset? In assessing whether they settled a clash substantively, there is a
relevant distinction to keep in mind. Scholars discern between conflicts on policy issues (policy
conflicts; e.g. about the language policy) and conflicts on the structure of the cabinet (office
conflicts; e.g. clashes about cabinet reshuffles or portfolio allocations) (e.g. Marangoni &
Vercesi, 2015; Vercesi, 2016, p. 180). Resignations can only provide a substantive solution to
office conflicts that are about the resignation itself. This only applies to one case: the dispute
about the position of Deputy Prime Minister Guy Coëme, a Francophone social democrat (PS)
who was mentioned in a corruption case (early 1994; c281).
Having checked which outcomes meet which functionality dimensions, I can calculate the
163
percentage of outcomes that meets each dimension (0-100). Combining these percentages
for all five dimensions results in a score of the total functionality level (TFL), which ranges
between 0 and 500. Dividing the score into three equal parts, I can discern clearly low (0-167)
from clearly high scores (333-500). In between, we find the middle range. Of course, this
aggregated score might not appeal to those who consider certain dimensions to be irrelevant.
Yet, its main asset is that it provides a more general picture that opens the room for general
conclusions.
Next to presenting the aggregated results, I also present trends over time (by year and by state
reform period). Did the percentage of clashes that are settled substantively change? Are
consensus decisions in decline? And what about the total level of functionality?
164
165
Chapter 7.
Morphology and prevalence of ethno-territorial conflicts
128
This chapter addresses the general morphology and prevalence of ethno-territorial conflicts.
How common are these clashes, what are they about, who is involved, and how intense are
they? This descriptive overview provides a good stepping-stone for the analyses of the evolution
of ethno-territorial conflict frequencies (Chapter 8) and the explanatory analyses (Chapter 9).
In some respects, the results provide an unexpected counterweight to conventional views on
e.g. bipolar federalism and the (often presumed) absence of ‘tension-reducing coalitions’.
1. Prevalence: how common are ethno-territorial conflicts?
Ethno-territorial conflicts are a serious burden on the political decision-making
process in Belgium. On average, the coding of the yearbooks (1979–2006) exposed a new
ethno-territorial conflict every three to four months (monthly X= 0.28; SD= 0.52). For the news
archives, which cover the most recent period (1995–2018), this number goes up to more than
one new clash every two months (X= 0.59; SD= 0.90). Yearly, the average number of new
ethno-territorial clashes is 3.43 (SD= 2.41; yearbooks) and 7.04 (SD= 4.54; news archives).
This is far lower than the average numbers of non-ethno-territorial clashes (respectively 8.29
and 35.17). Both datasets (news archive and yearbook data reported in that order) indicate that
a significant share of coalition conflicts is ethno-territorial in nature (16.7%–29.3%). Of course,
such clashes come in many sizes and shapes. Throughout the following paragraphs, a more
detailed account of the morphology of ethno-territorial conflicts is provided.
128
Conflicts that might not have been between coalition partners (cases of doubt) are not included in the
analyses. Contrarily, and given the partitocratic nature of the Belgian polity (Baudewyns et al., 2022; De
Winter & Dumont, 2006), cases are included when there is doubt about the applicability of the relevance
criterion. The Belgian partitocracy comes with very low chances of purely individual actions by political
loners. Some of these analyses and results were previously presented (often in less elaborative terms) in an
article in Ethnopolitics (Vandenberghe, 2022c).
166
2. The (changing) morphology of ethno-territorial conflicts
2.1. What ethno-territorial issues are ethno-territorial conflicts about?
Between 15.7% (news archive data, 1995-2018) and 24.4% of all cabinet conflicts
(yearbook data, 1979-2006) certainly concern ethno-territorial issues. But what are these issues
specifically? The overview provided in Table 9 provides an answer to this question. Of all
conflicts on ethno-territorial issues, most concerned the linguistic struggle (32.3% and 25.4%
of all ethno-territorial clashes in the yearbook and news archive data, respectively). Disputes
on state reform issues such as the decentralization of certain competences are frequent too
(28.1% and 21.3%), just like tensions on distributive questions (12.5% and 24.9%). Examples
of the latter are the tensions about the imbalanced spreading of asylum seekers between
Flanders and Wallonia (e.g. conflict c388) or the Flemish consternation about the plans of
Defence Minister De Donnea (a Francophone liberal) to transfer the royal cadet School from
Lier (Flanders) to Laken (Brussels) (conflict c192). Strikingly, at the cabinet level,
fiscal/financial issues such as interregional transfers are hardly explosive. Of all the cabinet
conflicts included in the datasets, only a handful primarily revolved around such debates (5 in
the yearbook dataset, 3 in the news archive dataset).
In the news archive dataset, a significant set of clashes (14.8%) did not really fit any of the
types of ethno-territorial issues and was ascribed to the category ‘other’. Examples include
the frustrated attacks of Francophone coalition parties against Prime Minister Guy Verhofstadt,
who was sporadically blamed for being ‘too Flemish’, for paying more attention to his Flemish
coalition partners, and for ‘not sufficiently being the Prime Minister of all Belgians’ (e.g. c3219
and c3268). At one point, in January 2002, Verhofstadt even threatened with the resignation of
his ‘purple-green’ cabinet (social democrats, greens, and liberals) after such an attack by
Philippe Defeyt, president of the Francophone greens (Ecolo). The latter eventually confirmed
their confidence in the PM, thus defusing the tensions (c3268). Another example is the May
2002 conflict between the Francophone social democrats (PS) and the Minister of Justice (Marc
Verwilghen, Flemish liberals) (c3287). Following the judgment of the juvenile court (on the
Mombaerts case), the latter blamed Francophone judges for being far ‘laxer’ than their Flemish
colleagues. The PS considered such comments to be ‘unacceptable’, ‘dangerous’,
‘inflammatory’, ‘demagogic’, ‘an exploitation of a painful case for political and ethno-
territorial purposes’, and ‘ammunition for the extreme right’. A more recent example of a clash
on an ethno-territorial issue that was ascribed to the ‘other’ category is the criticism against
Theo Francken, Secretary of State for Asylum and Migration for the Flemish nationalist party
(N-VA). Following the 2017 Catalan independence referendum, Francken publicly stated that
167
Catalan minister-president Carles Puigdemont could seek asylum in Belgium. Spanish officials
and the conservative Partido Popular (PP) in particular reacted furiously and Francken’s
comments triggered fierce criticism of the N-VA’s coalition parties too (c1220 in the news
archive data).
129
Table 9. Frequency table: issues of cabinet conflicts (news archive data: N= 1013, 1995-2018;
yearbook data: N= 328, 1979-2006)
All cabinet conflicts
Ethno-territorial cabinet conflicts
News archive
data
Yearbook
data
News archive
data
Yearbook
data
Type of issue
N
%
within
dataset
N
%
within
dataset
N
%
within
dataset
N
%
within
dataset
Non-ethno-territorial
852
84.11
245
74.70
10
5.92
16
16.67
Linguistic struggle
43
4.24
31
9.45
43
25.44
31
32.29
State reform
36
3.55
27
8.23
36
21.30
27
28.13
Fiscal/Financial issues
3
0.30
5
1.52
3
1.78
5
5.21
Distributive issues
42
4.15
12
3.66
42
24.85
12
12.50
Combination
10
0.99
0
0.00
10
5.92
0
0.00
Other
25
2.47
5
1.52
25
14.79
5
5.21
Missing
2
0.20
3
0.91
0
0.00
0
0.00
Total
1013
100
328
100
169
100
96
100
For the three ethno-territorial issues that are most prone to coalition conflicts (linguistic
struggle, state reform, and distributive issues), Figures 8, 9, and 10 expose several interesting
results and trends over time. To begin with, the data show that in most years, ethno-territorial
tensions were not primarily expressions of the traditional language struggle. This is in line
with the analysis that by 1979 (the start of the period studied here), the language struggle that
used to be predominant had already transformed into a broader ethno-territorial struggle (Celis
129
In comparison to the response of other regionalist parties like the Scottish National Party (SNP), the N-
VA’s reaction to the Catalan referendum and the subsequent repression was particularly assertive. For a more
elaborative investigation of these different reactions and the dynamics behind them, see e.g. Sijstermans and
Brown Swan (2022).
168
& Van Daele, 2011). Second and interestingly, no linear trend can be discerned. Linguistic
tensions did not continuously increase or decrease over time. Rather, during the period
under study (1979-2018), the linguistic struggle came in two major waves (Figure 8). A first
wave terrorized segmental power-sharing throughout the late 1970s and the 1980s. The battles
at hand were typically about the status and application of linguistic laws in facility communities
(20 out of 23 clashes on the linguistic struggle between 1979 and the end of 1988). Most
notably, two issues were particularly explosive: the struggles about the Dutch school in
Comines (3 conflicts) and the tensions about the status of the Flemish town of Voeren and the
appointment of José Happart, who refused to speak Dutch, as its mayor (15 clashes). After the
adoption of the 1988 ‘pacification law’, these tensions were defused (see Chapter 3). A second
wave of tensions on linguistic issues arose in the late 1990s/early 2000s, and reached its peak
in 2004/2005, after the 2003 ruling of the constitutional court on the notorious electoral district
BHV (Brussels-Halle-Vilvoorde), which was considered unconstitutional. As the news archive
data show (1995-2018), the ethno-territorial fireworks about BHV would continue to light up
the Belgian political landscape until the 2011 agreement (as part of the sixth state reform). In
its aftermath, the situation de-escalated. The remaining tensions were rather isolated cases.
One example is the threat of the Francophone liberal MR party to withdraw its support for
Minister of Defence Pieter De Crem (CD&V, Flemish Christian Democrats), who was accused
of widening the linguistic imbalance in the highest echelons of the army (October 2012, c918).
Figure 8. Yearly frequency of cabinet conflicts on linguistic struggles as a percentage of all
ethno-territorial cabinet conflicts (yearbook data: N= 96; news archive data: N= 169)
0
10
20
30
40
50
60
70
80
90
100
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Linguistic struggle (yearbook data) Linguistic struggle (news archive data)
169
After the sixth state reform, linguistic tensions declined immediately and disappeared
eventually. This is interesting because there certainly was no lack of linguistic tensions ‘on
the ground’. The ingredients for grand crises were present. To give just one example, linguistic
agitations often escalated in the Flemish facility community of Linkebeek. There, the
appointment of French-speaking Damien Thiéry (MR; Francophone liberals) was fiercely
opposed because he had clearly violated the Flemish interpretation of the language laws (as
described in the 1997 circular letter ‘Peeters’). The conflict started after the 2006 local elections
and continued for another decade. The issue in itself strongly resembled the one of Voeren,
which triggered major cabinet crises in the 1980s. Again, there was widespread agitation about
the appointment of a Francophone refusing to use Dutch (in some situations) as mayor of a
Flemish town. Yet this time, the tensions failed to destabilize the federal cabinets. Why did we
not witness a Voeren 2.0? At least three elements might help explain this remarkable
difference. First, in the aftermath of the tense and regime-threatening cabinet formation of
2010-2011, there was a generalized aversion and avoidance of ethno-territorial debates (see
Chapter 3). The focus was on the implementation of the latest state reform. Ethno-territorial
debates became synonymous with political tensions and chaos. Fear of the dark made most
parties avoid the topic altogether. After the 2014 elections, the coalition partners of the Michel
I cabinet even formally and famously agreed to avoid ethno-territorial debates and focus on
socio-economic reform throughout the legislative period (2014-2019) (Benoît Rihoux et al.,
2016, p. 30).
130
Second, in contrast to the 1980s, the organization of local governments no
longer fell under the authority of the federal government. It was transferred to the sub-states
as part of the fifth state reform (2001), which decentralized the municipal and provincial law
(for a discussion, see Brassinne de La Buissière, 2005). As such, it was not the federal but the
Flemish government who decided about the fate of Thiéry. Nicely illustrating the pacifying
potential of decentralization, the issue never really reached the negotiating table of the federal
cabinet. Third, the public sensitivity to linguistic tensions might have decreased. Younger
generations are well aware of the language border that divides the country (e.g. Reuchamps et
al., 2014), but they are arguably less in touch with these tensions.
131
After all, they themselves
130
Michel I was a coalition of the Flemish nationalist N-VA, the Flemish liberals (Open VLD), the Flemish
Christian democrats (CD&V), and the Francophone liberals (MR).
131
Reuchamps et al. (2014) asked first-year higher education students across Belgium to draw the country.
170
(and even many of their parents) never really experienced the reality of being considered
inferior or socially deprived for using their ‘second class’ language (Dutch) in a centralized
Belgian state in which French was the lingua franca of the elite (Witte, 1993b; Witte & Van
Velthoven, 2010). Only a small segment of Flemings still encounters such situations from time
to time (notably the ones living in Brussels or one of the facility communities). To most
Flemings, the language struggle might be an abstract if not anachronistic debate of the past. As
such, there might also be less room for the political exploitation of the feelings and sensitivities
that fueled political tensions in previous decades.
Figure 9. Yearly frequency of cabinet conflicts on state reform as a percentage of all cabinet conflicts
on ethno-territorial issues (yearbook data: N= 80; news archive data: N= 159)
Interestingly, in relative terms, no paradox can be found when it comes to clashes on state
reform (a second conflict-prone ethno-territorial issue). Rather, such tensions seem to have
become less numerous over time. Hence, conflicts on ethno-territorial issues are less commonly
about state reform. However, this overall tendency is certainly not unambiguous. Rather, their
prevalence appears to follow a certain sequence (Figure 9). Notably, the build-up towards
Although this was not asked nor mentioned, 55.8% of the 4766 participants who actually did so
spontaneously drew a language border.
0
10
20
30
40
50
60
70
80
90
100
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
State reform (yearbook data) State reform (news archive data)
171
state reform agreements is often marked by growing tensions. But once these agreements
are found, they typically decline (sometimes with some delay). This dynamic is noted either
clearly or vaguely for the second (1980), third (1988-89), fifth (2001), and sixth state reform
(2011). In contrast, the route towards the fourth state reform (1993) was particularly smooth.
Of course, numerous factors might account for this observation, but it seems to be no
coincidence that the quiet build-up towards the agreements on this state reform coincided with
the reign of Jean-Luc Dehaene (Prime Minister between 1992-1999), a Flemish Christian
democrat (CVP) who cultivated consociational conflict-solving techniques like secrecy, elite
negotiations, media avoidance, and compromise-willingness (see e.g. Hoflack, 2021, pp. 320-
322). The more recent conflicts about state reform were very isolated incidents, such as the
consternation about the potential existence of secret agreements on state reform in hidden
Atoma notebooks (as stated by N-VA Deputy Prime Minister Jan Jambon during a lecture for
a Flemish nationalist student club, November 2014, c1037), the attack of N-VA president Bart
De Wever on federal Secretary of Foreign Trade Pieter De Crem (CD&V), whose role was
depicted as ‘obsolete’ because ‘foreign trade is a sub-state competence’ (June 2015, c1071), or
the piqued reactions to the N-VA’s announcement that it would re-emphasize its confederal
plans, translate them to legislative initiatives, and actively fuel public debates on the future of
Belgium – despite the ethno-territorial standstill agreed upon within the Michel I cabinet
(January 2016, c1101).
Figure 10. Yearly frequency of cabinet conflicts on distributive issues as a percentage of all cabinet
conflicts on ethno-territorial issues (yearbook data: N= 80; news archive data: N= 159)
0
10
20
30
40
50
60
70
80
90
100
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Distributive issues (yearbook data) Distributive issues (news archive data)
172
A third explosive category, the one of distributive issues, does not show an overall decline.
Here, the most recent data (news archive data, 1995-2018) even shows an upward tendency.
However, this is largely due to the low absolute frequency of ethno-territorial clashes in recent
years. Put differently: insofar as such ETCs still arise in the most recent decades, they are
more often on distributive questions than before. More generally, distributive tensions were
particularly prominent during several periods. First, throughout the 1980s, when the
abovementioned ethno-territorial tensions on the cost distribution for keeping alive the national
economic sectors (most notably the steel sector) triggered various clashes. Additionally, a
couple of clashes during this era were about defence (e.g. the abovementioned conflict c192
about the transfer of the cadet school from Flanders to Wallonia) and about the financing of
education: the Dutch-French distribution key for certain school subsidies (June 1985, c125),
the distribution of austerity efforts between the Dutch-speaking and Francophone education
systems, and the infamous Francophone call for additional funding for their education in
combination with their presumed lack of efficiency and low willingness to cut costs (as
compared to the efforts made in Flanders) (e.g. July 1986, c150; February 1987, c168).
Unsurprisingly in view of these tensions, the third state reform decentralized the educational
policy domain (which is now under the authority of the Communities). Here too, the real-life
potential of decentralization as a pacification tool seems to be nicely illustrated.
Conflicts about the distribution of certain assets or liabilities resurfaced during the mid-1990s
and since then, they regularly burdened central policy-making. There were sporadic tensions
about e.g. segmental imbalances in railway investment plans or schedules (e.g. c3041 and
c3283, July 1996 and March 2002), the ban on tobacco advertising during the late 1990s, which
strongly impacted the Walloon racing sector (e.g. c3091 and c3094, news archive dataset), or
the distribution of climate efforts (e.g. 2004 and 2015, respectively c3371 and c1088). However,
the most common clashes were about two traditionally explosive topics in Belgian politics: the
distribution of night-time flights and the noise regulations for Brussels Airport (numerous
clashes throughout the years) and the federal quota for doctors and dentists. Debates about these
quotas, which limit the number of students that can start med school in Flanders and
Francophone Belgium, together with the Francophone violations of the existing quota, resulted
in recurrent explosions across the timeline. The rather remarkable growth of the relative
frequency of clashes on distributive issues after the sixth state reform (2011), which
typically involved the same topics, is largely a reflection of the very low absolute number of
clashes on ethno-territorial issues in this period.
173
2.2. Sides: which actors stand against each other in ethno-territorial conflicts?
Of all cabinet conflicts, clashes that are fought along segmental lines represent a far
smaller proportion (4.5%–16.2% of all cabinet conflicts; news archive data and yearbook
data reported in that order) than conflicts on ethno-territorial issues. The latter are roughly
three and a half times as frequent in the news archive data (15.7%) and one and a half times as
frequent in the yearbook dataset (24.4%). Furthermore, insofar as clashes along segmental lines
take place, they typically mobilize one segment only, e.g. all Francophone parties vs. one
Flemish party (recall that for Michel I, such clashes are not seen as conflicts along segmental
lines). As Table 10 shows, conflicts in which both segments stand against each other are
particularly exceptional. Such clashes, in which all Flemish parties clash with all Francophone
parties, only account for 1.5%–7.3% of all conflicts in coalitions. So, it seems, the problem is
not so much segmental power-sharing as such (the forced cooperation between two seemingly
incompatible blocks), but rather the many issues and debates that are inherent to governing
divided polities (ranging from local applications of language laws to major discussions on state
reform).
Even in ethno-territorial disputes, Flemish and Francophone actors rarely join forces to
act ‘as one’. While a significant share of these conflicts is fought along segmental lines (29.6
to 55.2%), they typically only mobilize one segment. One example of such a situation is the
case of the weapon deliveries to Uruguay (March 1980, c59). Minister of Foreign Affairs
Simonet (PS, one of the two Francophone coalition parties) unilaterally decided to issue an
export permit, thus painfully and publicly turning a deaf ear to the explicit veto of the two
Flemish coalition parties, who strongly opposed the delivery: the social democratic SP
(Minister of Economic Affairs Claes) and the Christian democratic CVP (Minister of
Developmental Aid Eyskens). The Flemish parties faced the decision in bitterness. On major
debates like Voeren, Komen, or Brussels, ethno-territorial clashes were also often fought with
only one segment fully mobilized. Ethno-territorial clashes in which both segments stand
against each other are far less common. They only represent a small minority of all ethno-
territorial cabinet conflicts (8.9%–25%) and -as noted- an even smaller proportion of all cabinet
conflicts (1.5%–7.3%). Clear examples of such segmental trench warfare are many of the
clashes between 2004 and 2011 on the issue of BHV. Concerning this topic, Flemish parties
often abandoned the consociational consciousness and opted for a more majoritarian logic,
threatening to push through the split of BHV against the will of the Francophones (Sinardet,
2010). Another example of a clash between two segmental fronts ending in overt
majoritarianism goes back to the end of 1979, when the Flemish parties in the Senate joined
forces to approve an amendment to a draft law on the second phase of the second state reform
(draft law 261) that was strongly rejected by the Francophone parties (c22). In response, and
174
creating a theatrical scene of bitterness, the latter ostentatiously left the session.
Clashes within segments are far more prevalent. Between 37.2% and 19.2% of all cabinet
conflicts are fought amongst Flemings (or Francophones). Interestingly, they also clash on
ethno-territorial issues. Roughly stated, ethno-territorial clashes are about as often fought
amongst Flemings (9.5%–5.2%) as amongst Francophones (10.1%–1%). Contrarily and
interestingly, Flemish parties clash far more often on non-ethno-territorial issues than their
Francophone peers. Indeed, when taking all cabinet conflicts into account, the percentage of
disputes amongst Flemings (25.2%–13.4%) is twice as high as that of clashes amongst
Francophones (12.0%–5.8%). Put differently, Flemish politicians are generally more inclined
to clash with one another than their Francophone colleagues, but not on ethno-territorial issues.
Table 10. Frequency table: conflict sides of cabinet conflicts (news archive data: N= 1013, 1995-2018;
yearbook data: N= 328, 1979-2006)
All cabinet conflicts
Ethno-territorial cabinet conflicts
News archive data
Yearbook
data
News archive data
Yearbook
data
Conflict sides
N
%
within
dataset
N
%
within
dataset
N
%
within
dataset
N
%
within
dataset
Mixed sides
243
23.99
120
36.59
28
16.57
16
16.67
Intra-Francophone
122
12.04
19
5.79
17
10.06
1
1.04
Intra-Flemish
255
25.17
44
13.41
16
9.47
5
5.21
Partial segmental sides
292
28.83
75
22.87
58
34.32
19
19.79
Full FR vs. partial FL
42
4.15
15
4.57
21
12.43
15
15.63
Partial FR vs. full FL
14
1.38
14
4.27
14
8.28
14
14.58
Full FR vs. full FL
15
1.48
24
7.32
15
8.88
24
25
Doubt/missing
30
2.96
17
5.18
0
0
2
2.08
Total
1013
100
328
100
169
100
96
100
Note: Michel I (2014–2018): to be conflicts ‘along segmental lines’, two segmental sides must be involved. Hence, not all
clashes with full segmental conflict side(s) are clashes along segmental lines.
Segmentally mixed sides are quite frequently noted too (24% to 36.6% of all clashes), even
when it comes to ethno-territorial tensions (16.6%–16.7%). The fiercely criticized June 2009
statements of Didier Reynders, then president of the Francophone liberals (MR), provide a good
example (c733, news archive dataset). Reynders stated that the Flemings are overrepresented
in the parliament of the Brussels-Capital Region and that there ought to be a debate on this issue
after the elections. This received fierce criticism from both Flemish and Francophone coalition
parties. The Flemish Christian democrats (Deputy Prime Minister S. Vanackere), for instance,
175
recalled that the Francophones have many guarantees and vetoes at the federal level. On a
threatening tone, he declared: ‘Whoever pulls this string risks disentangling the whole Belgian
sweater’; ‘no one can call into question one element with impunity’. Similarly, the president of
the Francophone social democrats (E. Di Rupo) classified the remarks of Reynders as ‘a pre-
electoral provocation’. Stating that he prefers dialogue, he sharply added that ‘the MR wants to
make Francophones believe that their problems will be solved by provoking the Flemings’.
These results go against the quite popular view that the bipolar Belgian system leaves little
room for ‘tension-reducing coalitions’ (Elazar, 1987, p. 244), resulting in a permanent
antagonism between its two main segments. Even in ethno-territorial debates, clashes within
segments and shifting alliances are often noted. Neither the Flemish nor the Francophone
segment is a political monolith.
Figure 11. Yearly frequency of cabinet conflicts along segmental lines as a percentage of all ethno-
territorial cabinet conflicts (yearbook data: N= 96; news archive data: N= 169)
Perhaps, the popularity of the opposite view can be partly explained by the large segments of
cases in which some (not all) of the Flemings clash with some (not all) of the Francophones.
Such situations account for 28.8%–22.9% of all cabinet conflicts and 34.3%–19.8% of ethno-
territorial clashes. One example of such a situation occurred when the Flemish Christian
democrats (CVP) imposed the appointment of T. Van Parys as Minister of Justice despite the
strong resistance of their Francophone sister party (PSC). The latter fiercely opposed the
0
10
20
30
40
50
60
70
80
90
100
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2006
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Conflict along segmental lines (yearbook data) Conflict along segmental lines (news archive data)
176
appointment because of Van Parys’ earlier comments. A while before his appointment, he
triggered controversy and Francophone agitation by pleading to ‘stop waiting for the
counteracting Walloon politicians’ and decentralize the police and justice systems because on
these domains, he argued, ‘there are fewer problems and a greater willingness to solve them in
Flanders’ (c336, April 1998). Another example of a clash with partial segmental sides was the
situation in December 1990, when the continued call for a split of the bilingual province of
Brabant into separate Flemish and Walloon provinces triggered a sharp reaction from the
Francophone Christian democrats (PSC). Fed up with these demands, they made it loud and
clear that to them, the Flemish nationalist coalition partner (VU) should no longer be part of the
Martens VIII cabinet (CVP, PS, SP, PSC, and VU) (c232). When including such cases in the
basket of inter-segmental tensions, a far less pacifistic picture arises: in total, more than one in
three cabinet conflicts are fought between Flemings and Francophones (35.8%–39%). In sum
then, Flemings and Francophones often clash (this happens in about one in three cabinet
conflicts) but in doing so, they rarely act ‘as one’ (i.e. as homogeneous blocks).
How did this evolve over time? Figure 11 shows that ethno-territorial conflicts are less
commonly fought along segmental lines. The highest percentages of ethno-territorial clashes
that were fought along segmental lines are noted in the past. The decline is particularly notable
since 2006. However, the overall picture looks quite blurry. They were very prominent up and
until the first half of the 1980s, less consistently prevailed in the next decade, and regained their
prominence in the mid-1990s (following the first direct elections of the Flemish and Walloon
parliaments in 1995). Here too, a clear dynamic of pacification is noted in the aftermath of the
sixth state reform (unprecedented stability).
Within the population of ethno-territorial conflicts, only two intra-party conflicts were noted
(one in each dataset). The minoritarian (if not marginal) status of intra-party clashes was also
noted in previous empirical studies on cabinet conflicts (Nousiainen, 1993, p. 273). Apart from
the evident ideological ties among party actors, two features of the Belgian political system
can help explain this remarkable lack of friendly fire. First, the split nature of the Belgian
party system (cf. De Winter et al., 2009, pp. 72-76; Verleden, 2009) implies that there is no
need for inter-segmental negotiations within the party. All ethno-territorial debates within
governing parties are discussions among Flemings or among Francophones. Interaction only
turns inter-segmental when the debate between parties starts. Accordingly, it is highly likely
that intra-party clashes on ethno-territorial issues were far more frequent before the split, when
inter-segmental debates were still fought within parties. Several qualitative and historical works
have already shown that the inter-segmental tensions within traditional parties were both
numerous and significant (in terms of intensity) (e.g. Van Velthoven, 2019 on the social
democratic party family). Eventually, they accumulated into their formal split along linguistic
177
lines. The data presented here indicate that since then, the ethno-territorial dynamite within
parties has been defused. Second, Belgium is a textbook example of a partitocracy
(Baudewyns et al., 2022; De Winter & Dumont, 2006). As such, political actors have strong
incentives to contain their discontent or only express it behind the closed doors of the party
organization. Of course, the partictocratic pressure to avoid desertion is not limited to ethno-
territorial debates. It acts as a deterrent against desertion on all kinds of issues. Yet, intra-party
conflict is even more rare on ethno-territorial issues (0.6%–1% of all clashes on such issues)
than on other issues (1.3%–2.8%) such as socio-economic debates or clashes on migration,
international politics, ecology, or defence. Perhaps, this is because clashing with party fellows
on ethno-territorial issues might be perceived as a double form of treason: against one’s party
and against one’s segment. The use of qualitative research on politicians’ perceptions of such
(peer) pressures would be a particularly valuable way to check whether this logic holds.
2.3. Typology: how frequent is each type of ethno-territorial conflict?
In Chapter 2, ethno-territorial conflicts are defined as conflicts (a) along segmental lines
(Flemings vs. Francophones) and/or (b) conflicts on ethno-territorial issues (e.g. language
policy, state reform). Using dimensions (a) and (b), a typology of ethno-territorial conflicts was
proposed. How common is each type? Figure 12 shows that the largest set of ethno-territorial
conflicts is thematic in nature, i.e. clashes about ethno-territorial issues that are not fought
along segmental lines (72.8%–44.8%, bottom right). As the low percentages of factionalized
conflicts show (5.9%–16.7%, top left), segmental trench warfare between the Flemish and
Francophone coalition partners is less common when it comes to non-ethno-territorial issues
(e.g. socio-economic debates).
Figure 12. Typology of ethno-territorial conflicts (ETCs): prevalence in Belgian
cabinets as a percentage of all ethno-territorial conflicts (news archive dataset (1995-
2018; N= 169) and yearbook dataset (1979-2006; N= 96) reported in that order)
Segmental side(s)
Other issue(s)
factionalized
5.92%
16.67%
full-fledged
21.30%
38.54%
Ethno-territorial issue(s)
non-ETCs
thematic
72.78%
44.79%
Non-segmental side(s)
Note: Results of the yearbook dataset are reported below those of the news archive dataset.
178
Meanwhile, between one and two in five ethno-territorial clashes is a full-fledged conflict in
which all Flemish (and/or Francophone) coalition parties join forces in a clash on an ethno-
territorial issue (21.3%–38.5%, top right). In this respect, many (but not all) of the
abovementioned tensions on BHV provide solid examples. Figure 13 shows that, over time,
fewer ethno-territorial conflicts are full-fledged in nature. Such clashes were most commonly
noted throughout the troublesome first half of the 1980s and in the late 1990s/early 2000s.
particularly in the 1980s, it was not uncommon that all ethno-territorial conflicts in a year were
full-fledged versions. In recent times, they went fully extinct. Note that their prevalence
declined since 2005, despite the emergence of the explosive BHV issue. This is because of the
large number of thematic clashes on BHV and the sixth state reform, i.e. clashes that were not
fought with full segment(s) but e.g. between parts of each segment (some Flemish vs. some
Francophone coalition parties) or even within segments. Often, the latter concerns situations in
which parties on the same side of the language border disagree not about the policy stance itself
(the split of BHV) but about the strategy used to make it happen (e.g. the best way to convince
Francophones or the option of unilateral decision-making against the latter’s will).
Figure 13. Yearly frequency of full-fledged ethno-territorial conflicts as a percentage of all ethno-
territorial cabinet conflicts (yearbook data: N= 96; news archive data: N= 169)
2.4. Intensity: how intense are ethno-territorial conflicts?
As noted in Chapter 6, both datasets provide a different picture of the intensity of
clashes. This disparity resurfaces here (Table 11). In the news archive dataset (1995-2018),
two-thirds of all ethno-territorial cabinet conflicts are not intense at all (66.3%). Typically, this
0
10
20
30
40
50
60
70
80
90
100
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Full-fledged (yearbook data) Full-fledged (news archive data)
179
involves short-lived criticism against proposals of coalition parties such as the idea of the
Flemish liberal party (Open VLD) to ‘recentralize’ certain sub-state competences back to the
federal level: ‘Deputy Prime Minister De Croo (Open VLD) pretends to be courageous while
actually being afraid to make a proper choice’; ‘Why doesn’t he go for unitarism all the way?’;
‘It’s tinkering at the margins’, which ‘won’t fundamentally strengthen the efficacy of the
cricked Belgian state structure.’ (P. De Roover, leader of the parliamentary party group of the
Flemish regionalist N-VA) (c1144, July 2016). The news archive dataset is also marked by a
quite logical order: the higher the intensity, the lower the frequency. This is not the case in the
yearbook dataset. Here, clashes with higher levels of intensity (1 and 2) are more frequently
noted (37.5% and 32.3%) than low-intensity clashes (29.2%). Presumably, and as discussed in
Chapter 6, this could be explained by the fact that the threshold to be covered by the numerous
daily articles issued by the Belga news agency (on which the news archive dataset is based) is
lower than the threshold to be included in the Res Publica yearbooks (which are only issued
once a year).
Table 11. Frequency table: intensity of ethno-territorial cabinet conflicts
(news archive data: N= 169, 1995-2018; yearbook data: N= 96, 1979-2006)
News archive data
Yearbook data
Intensity level
N
% within
dataset
N
% within
dataset
0
112
66.27
28
29.17
1
36
21.30
36
37.50
2
20
11.83
31
32.29
Doubt/missing
1
0.59
1
1.04
Total
169
100
96
100
Figure 14 provides a longitudinal overview of the percentages of ethno-territorial clashes that
are intense (levels 1 and 2 are combined, to avoid complexity). In general and especially since
the second half of the 2000s, the intensity of ethno-territorial conflicts decreased. For
decades, the bulk of these clashes was intense. This is no longer the case. Presumably, however,
the combination of datasets partly accounts for this trend.
180
Figure 14. Yearly frequency of intense ethno-territorial conflicts (levels 1 and 2) as a percentage of all
ethno-territorial cabinet conflicts (yearbook data: N= 96; news archive data: N= 169)
3. Intermediate conclusions
Ethno-territorial conflicts are widely considered to be one of the biggest poltergeists
that haunt the Belgian system. Before examining whether the haunting has increased or
decreased throughout the decades (Chapter 8), this chapter exposes how common such conflicts
are and what they usually look like. The results reveal some interesting dynamics.
To begin with, the results demonstrate that the pertinence of ethno-territorial tensions is not an
exaggerated phantasy of regionalist hardliners. Visibly, ethno-territorial politics deserves its
explosive reputation. Of all coalition conflicts in Belgium, 17% to 29% are ethno-territorial in
nature (depending on the dataset) and on average, there is a new dispute every two to three-
and-a-half months.
Next, the analyses indicated that some ethno-territorial issues are more conflict-prone than
others. While the linguistic struggle, debates about state reform, and distributive questions are
particularly explosive, debates that are primarily about fiscal or financial issues such as inter-
regional transfers are exceptional. When it comes to disputes on state reform, an interesting
pattern was noted. Such tensions often accumulated in the build-up towards these reforms and
declined in their aftermath. Generally speaking, however, these tensions have become less
0
10
20
30
40
50
60
70
80
90
100
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Intense ETCs (yearbook data) Intense ETCs (news archive data)
181
common throughout time. Particularly in recent times, ethno-territorial tensions are less often
about linguistic tensions and state reform as such (the usual suspects), and more frequently
about the distribution of assets/liabilities between the different segments.
All in all, the analyses also demonstrate that such tensions on ethno-territorial issues are far
more common than conflicts between segmental sides; i.e. conflicts in which the Flemings
(or Francophones) actually act ‘as one’. Conflicts in which all Flemings and all Francophones
stand against each other are particularly rare, accounting for 1.5% to 7.3% of all cabinet
conflicts (depending on the dataset). Meanwhile, conflict sides often vary throughout time, and
about 60% of all tensions are either fought within segments or characterized by segmentally
mixed sides. Hence, the most pertinent problem does not seem to be segmental power-sharing
as such (the fact that both sides must govern together), but rather the many issues and
discussions that are inherent to governing a divided state (ranging from local applications of
language laws to major discussions on state reform).
Finally, ethno-territorial clashes appear to have become ‘softer’ over time. Their nature has
changed. Compared to the past, ethno-territorial conflicts are less intense, less commonly
fought along segmental lines, and less commonly the kind of full-fledged conflicts in which
segmental blocks clash on ethno-territorial issues. Although this picture comes with several
nuances, it is safe to state that the ethno-territorial poltergeist still haunts Belgian politics,
but now with a friendlier face. While it is far too soon to conclude that decentralization
pacified ethno-territorial tensions (H1), these results do provide some first (cautious) damage
to the idea that it fuelled them (H3).
In several ways, they also contradict the quite common perception that Belgian politics is
the sum of two irreconcilable segments. Arguably, this perception is fueled by the rather large
number of clashes in which some (not all) Flemish coalition partners clash with some (not all)
of the Francophone parties (29%–23% of all conflicts). Taking these into account, almost 40%
of all tensions within coalitions are fought between Flemish and Francophone actors. As
such, it is rather unsurprising that the image of two incompatible blocks is so firmly imprinted
in the minds of many politicians, citizens, and observers. The fact that parties often present
themselves as the voice of their segment evidently adds to this, as does the media’s tendency to
echo such frames (Sinardet, 2008, 2012). But clearly, the results show that there is still some
room for the kind of ‘tension-reducing coalitions’ (Elazar, 1987, p. 244) that are often
thought to be precluded by Belgium’s bipolar nature. The Flemish and Francophone segments
are no political monoliths. They rarely act ‘as one’ and often clash amongst one another. Bipolar
federalism does not imply that all politics is bipolar.
182
183
Chapter 8.
Evolution: ethno-territorial conflict frequencies declined over
time
132
How did the ethno-territorial conflict in Belgian power-sharing cabinets evolve (RQ1)? In
answering this question, I study conflict frequencies from different angles. Absolute and relative
conflict frequencies are assessed against the background of decentralization. I also compare
the conflict proneness of consecutive cabinets. None of the indicators support the paradox
thesis. If anything, most of them expose tendencies of pacification.
1. Frequency of all ethno-territorial conflicts
Did decentralization come with rising or declining levels of ethno-territorial conflict?
Ethno-territorial conflict frequency is presented from different angles. I assess both (1)
absolute and (2) relative frequencies (the number of ETCs as a % of all cabinet conflicts).
These analyses are presented at the yearly level (which is well fit for visualisation). Providing
a more aggregated picture, I also compare (3) conflict frequencies at the cabinet level: are
cabinets increasingly more/less prone to ethno-territorial conflict?
Before presenting the results, an important remark must be made. All figures present the data
of both datasets (yearbooks, archives) within the same graphs. However, it should be kept
in mind that these datasets are not surrogates. As noted in Chapter 6, comparisons of the
overlapping periods indicate that they expose different conflict frequencies (most commonly,
more clashes are exposed by the news archive dataset).
133
To account for this quantitative
132
Conflicts that might not have been between coalition partners (cases of doubt) are not included in the
analyses. Contrarily, and given the partitocratic nature of the Belgian polity (Baudewyns et al., 2022; De
Winter & Dumont, 2006), cases are included when there is doubt about the applicability of the relevance
criterion. The Belgian partitocracy comes with very low chances of purely individual actions by political
loners. Some of these analyses and results were previously presented in Vandenberghe (2022b, 2022c).
133
Deviations between both datasets can partly be attributed to the fact that the archive data expose more
low-intensity conflicts. Note that the trends of intense conflict frequencies (Figure 18) are more in line.
184
difference and to avoid distorted visuals, axis ranges are adapted in the graphs that present
absolute frequencies and yearly equivalents. Different axis ranges are used for each dataset. To
calculate the appropriate ratio between these ranges, I use the comparison of overlapping
periods (1995–2006; cabinets Dehaene II and Verhofstadt I). For instance, in the 1995-2006
period, the news archive data exposes 2.89 times as many ethno-territorial conflicts as the
yearbook data, so its axis is about 2.89 times as large (max. 35) as that of the yearbook data
(max. 12). For transparency’s sake and to satisfy the interested readers, all graphs are also
presented with a single axis (no compensation for the quantitative difference between the
datasets) in the Appendix. References to the specific Appendices are provided throughout the
discussion below.
Figure 15. Ethno-territorial cabinet conflict: yearly absolute frequency (1979-2018)
Figure 15 presents the absolute frequency of ethno-territorial conflicts throughout time.
134
Interestingly, no clear-cut upward trend can be discerned. If anything, there appears to be an
overall yet fuzzy decline in the raw number of conflicts since the late 1970s. However, this
downward trend is blurred by numerous fluctuations. There is an apparent alternation between
conflict and stability, with peaks of eruptions in e.g. 1979, 1986-7 (yearbook data), and 2008–
134
Interested readers can find a graph with a single axis (no compensation for the quantitative difference
between the datasets) in Appendix 6.
0
5
10
15
20
25
30
35
0
2
4
6
8
10
12
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
News archive data
Yearbook data
Yearbook data News archive data
Decentralization deal
185
2010 (archives). Some clear depths can be discerned too, most notably in the early 1990s
(yearbook data) and early 2010s (news archive data). The eruptions of the 1970s and 1980s are
largely driven by traditional clashes on state reform, tensions about weapon exports (important
for the Walloon economy but often opposed by Flemish parties), and fierce battles about
disputed towns at the linguistic border, where linguistic tensions made ethno-territorial
temperatures reach record heights (mostly in the Flemish town of Voeren, less so in Comines).
A description and background of these tensions, which were largely settled with the third state
reform, is provided in Chapter 3.
Meanwhile, the most eruptive years in the news archive dataset (1995-2018) were 2008 and
2010 (16 new clashes in each year), followed by 2009 (15 conflicts) and 2002 (13 clashes). The
major eruptions in 2008, 2009, and 2010 were largely due to traditional clashes on the next state
reform and the notorious Flemish-Francophone tensions on Brussels-Halle-Vilvoorde (BHV).
In 2002, the large number of ethno-territorial battles was the result of an accumulation of rather
random incidents on various topics, including three attacks against the Francophone greens by
the Francophone liberals, who accused them of ‘allying with the Flemish majority to oppose
Walloon projects’, as well as traditional ethno-territorial clashes on state reform, the
decentralization of certain competences such as social security and the national railway
company of Belgium (NMBS/SNCB), and the regulation of the night-time flights and flight
routes of Brussels Airport (Zaventem). The latter is traditionally an explosive topic, which is
essentially about the distribution of noise between (mainly Francophone) residents in Brussels
and the residents in the Flemish towns North of Brussels.
State reform agreements (indicated in the visuals) do not seem to have a clear immediate
impact. In some instances, these deals were directly followed by escalations. In other cases,
conflict levels declined. But usually, these trends were really short-lived and changed as soon
as the following year. Only the 2011 agreement on the sixth state reform seems to be clearly
followed by a lasting period of ethno-territorial pacification and indeed, the longest period of
continued stability in the four decades under study. From an ethno-territorial perspective, both
the Di Rupo (2011-2014) and the Michel I cabinets (2014-2019) were characterized by a
historical lack of tensions. As noted above, their reign was marked by a widespread aversion
and deliberate avoidance of ethno-territorial debates. The focus was on socio-economic
recovery and the implementation of the sixth state reform. Meanwhile, the latter provided the
sub-states with an unprecedented level of autonomy (L. Hooghe et al., 2016; Shair-Rosenfield
et al., 2021; Watts, 2013). The results nicely expose the pacifying results of this cocktail.
In sum, the raw number of ethno-territorial conflicts did certainly not increase as
decentralization progressed (on the contrary) and in the short run, the individual
decentralization deals were not followed by any invariable, long-lasting trend (except for the
186
sixth state reform of 2011).
Figure 16. Ethno-territorial cabinet conflict: yearly relative frequency
(% of all cabinet conflicts, 1979-2018)
The relative prevalence of ethno-territorial conflicts provides a somewhat similar picture
(ethno-territorial conflict as a % of all cabinet conflicts; Figure 16). It shows an overall decline
too. From this perspective, ethno-territorial conflict gradually became more and more of a
side-product of confrontational politics, rather than being the predominant source of
distress itself. This tendency is especially clear since the late 1990s. Furthermore, and
confirming the picture of a more troublesome past, there were only five years in which there
were at least as many ethno-territorial as other conflicts: 1979, 1983, 1985, 1990, and 1994.
Meanwhile, and as with the absolute conflict evolution, the immediate aftermath of
decentralization deals strongly differs between different state reforms. In some cases, such deals
are followed by escalation, in other cases by pacification, and usually, such tendencies are really
short-lived. Again, the data highlight the lasting and historical stability following the
agreement on the sixth state reform (2011). So, in relative terms too, the data undeniably
show that ethno-territorial conflict did not increase as decentralization progressed and that state
reforms did not per se come with a direct aftermath of growing segmental tensions. If anything,
the pertinence of ethno-territorial conflicts declined as sub-state autonomy increased.
The abovementioned pacification tendencies are also apparent when consecutive cabinets
are compared. Figure 17 presents each cabinet’s yearly equivalent of ethno-territorial cabinet
conflicts (number of ethno-territorial cabinet conflicts divided by cabinet duration in days,
multiplied by 365). Here too, axis ranges are adapted to compensate for the quantitative
0
10
20
30
40
50
60
70
80
90
0
10
20
30
40
50
60
70
80
90
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Yearbook data News archive data
Decentralization deal
187
difference between the datasets.
135
The yearbook data expose a general decline of equivalents
throughout the decades. Put differently, Belgian cabinets are increasingly less burdened by
ethno-territorial clashes. Yet again, many fluctuations are observed. The most conflictual
executive was Martens I (1979-1980), a rather short-lived coalition of Christian democrats,
social democrats, and the Francophone regionalist FDF – which was excluded from the
coalition after fierce tensions on the second state reform (January 1980; c29). The relation
between conflict and regionalist parties’ cabinet participation is further explored (more
systematically) in Chapter 9. The second and third most explosive cabinets were led by Prime
Minister Martens too (Martens II and VI), but so were some of the calmest cabinets in the
dataset (Martens IV and VII). In line with previous results, Dehaene I was also particularly
135
For the two cabinets covered by both datasets (Dehaene II and Verhofstadt I), the news archive data
exposes a general yearly equivalent (7.07) that is 3.17 times as high as that exposed by the yearbook data
(2.23). Hence, its axis is about 3.17 times as large (max. 38) as that of the yearbook data (max. 12). Interested
readers can find a graph with a single axis (no compensation for the quantitative difference between the
datasets) in Appendix 7. Raw numbers are presented in Appendices 8 and 9.
0
6
12
18
24
30
36
0
2
4
6
8
10
12
Martens I
Martens II
Martens III
Martens IV
M. Eyskens
Martens V
Martens VI
Martens VII
Martens VIII
Martens IX
Dehaene I
Dehaene II
Verhofstadt I
Verhofstadt II
Verhofstadt III
Leterme I
Van Rompuy
Leterme II
Di Rupo
Michel I
News archive data
Yearbook data
Figure 17. Ethno-territorial cabinet conflict frequency:
yearly equivalent by cabinet (1979-2018)
Yearbook data News archive data
Note: Including the periods in which the coalitions were caretaker cabinets. Vanden Boeynants
II not included in the analysis of yearbook data and Dehaene I not included in the analysis of
news archive data (as these datasets do not cover the full lifespan of these cabinets). Michel I
cabinet: end date equals resignation date, as this marks the end of the coding effort.
188
stable, as were the Di Rupo and Michel cabinets in recent times.
2. Frequency of intense ethno-territorial conflicts
The general frequencies discussed above are of course important, but in assessing
power-sharing dynamics, the intensity of ethno-territorial conflicts is arguably as relevant. In
Figures 18, 19, and 20, the least serious conflicts (those coded with intensity level 0) are
excluded and the only clashes that are visualised are the ones involving impositions, threats,
fierce blocking such as empty chair politics, (speculations on) cabinet resignations, and the
like.
136
Two major observations catch the eye. First, the results are clear: no paradox-
confirming dynamic is found. If anything, the general decline noted above resurfaces here.
This pacification trend is particularly visible in the relative numbers (Figure 19). In an average
136
In Figure 18, axis ranges are adapted to compensate for the quantitative difference between the datasets.
In the overlapping period, the news archive data exposes 1.67 times as many intense ethno-territorial conflicts
(N= 30) as the yearbook data (N= 18), so its axis is about 1.67 times as large (max. 12) as that of the yearbook
data (max. 7). Interested readers can find a graph with a single axis (no compensation for the quantitative
difference between the datasets) in Appendix 10.
0
2
4
6
8
10
12
0
1
2
3
4
5
6
7
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
News archive data
Yearbook data
Yearbook data News archive data
Decentralization deal
Figure 18. Intense ethno-territorial cabinet conflicts (0 excluded):
yearly absolute frequency (1979-2018)
189
year between 1979 and 1999, they accounted for almost one in three conflicts (29%, SD= 19.12;
yearbook data). In some years, they even represented a majority of conflicts between coalition
partners: 1985 (57%), 1990 (67%), and 1994 (75%). Since the turn of the century, the picture
is very different. Both datasets demonstrate that in the last 19 years under study, such intense
ethno-territorial conflicts represent a far more marginal share of all cabinet conflicts (ranging
between 0% and 17%). Figure 20 presents the yearly equivalent of intense ethno-territorial
cabinet conflicts.
137
It painfully demonstrates the way the recent tensions on BHV and the sixth
state reform burdened the Leterme I and Van Rompuy cabinets, thereby breaching the apparent
pacification tendency that was noted (especially since Dehaene I).
Second, there appears to be a cyclical pattern, most notably in the absolute number of intense
ethno-territorial clashes (Figure 18) and the yearly equivalents (Figure 20). Large increases in
137
Here too, axis ranges are adapted to compensate for the quantitative difference between the datasets. For
the two cabinets covered by both datasets (Dehaene II and Verhofstadt I), the news archive data exposes a
general yearly equivalent (2.36) that is 1.27 times as high as that exposed by the yearbook data (1.86). Hence,
its axis is about 1.27 times as large (max. 13) as that of the yearbook data (max. 10). Interested readers can
find a graph with a single axis (no compensation for the quantitative difference between the datasets) in
Appendix 11.
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
News archive data
Yearbook data
Yearbook data News archive data
Decentralization deal
Figure 19. Intense ethno-territorial cabinet conflicts (0 excluded):
relative frequency (% of all cabinet conflicts, 1979-2018)
190
intense conflicts are followed by abrupt declines. It’s heaven and hell. The waves typically
peaked in the periods between the state reforms, while they were at their smallest in the
years in which these reforms are agreed upon. This remarkable dynamic applies to state
reforms three (1988), four (1993), five (2001), and six (2011).
138
These intriguing findings
almost perfectly echo previous reflections on political negotiations in Belgium:
“The end of the Belgian system has been frequently predicted (especially in the
1970s), and generally those were the times when a new compromise was reached.
This fluctuation between conflict and compromise is a fundamental characteristic
of the Belgian consociational democracy, whether the segments are the old pillars
and parties or the language communities.” (Deschouwer, 2006, p. 904)
Elsewhere, Deschouwer (2002, p. 79) described a similar dynamic when assessing cabinet
138
Note that in all of these cases, the decline already started in the previous year (although very marginally
for the fifth and sixth state reform).
0
2
4
6
8
10
12
0
2
4
6
8
10
News archive data
Yearbook data
Figure 5678. Intense ethno-territorial cabinet conflict frequency: yearly
equivalent by cabinet (1979-2018)
Yearbook data News archive data
Note: Including the periods in which the coalitions were caretaker cabinets. Vanden Boeynants
II not included in the analysis of yearbook data and Dehaene I not included in the analysis of
news archive data (as these datasets do not cover the full lives of these cabinets). Michel I
cabinet: end date equals resignation date, as this marks the end of the coding effort.
Figure 20. Intense ethno-territorial cabinet conflict frequency:
yearly equivalent by cabinet (1979-2018)
191
formation negotiations, whereby the Belgian elites manage conflicts
“by using the consociational logic: waiting until there are a lot of problems to be
solved, and then produce an agreement which essentially institutionalizes the non-
agreement by letting each side deal with its own policy. In consociational language
this is known as ‘granting autonomy’, and that is exactly what is done in a federal
state.”
The results presented here show that similar dynamics continue after the birth of the cabinet.
Interestingly, they also indicate that this pattern of consecutive conflict waves coming to shore
stabilized with the agreement on the sixth state reform (Vlinderakkoord, 2011) and the start of
the Di Rupo cabinet. Again, the results highlight that this hard-fought deal was followed by an
unprecedented period of stability.
3. Frequency of ethno-territorial conflicts by sub-types
Unpacking ethno-territorial conflicts and discerning conflict types adds to this picture.
As Appendices 4 and 5 show, the declining numbers of ethno-territorial conflicts are the result
of the decrease in both clashes on ethno-territorial issues and clashes along segmental lines (i.e.
when the Flemish and/or Francophone coalition partners act ‘as a block’).
139
Putting the pieces
of the puzzle together, Figure 21 presents the evolution of the different types of ethno-territorial
conflicts. To avoid chaotic or overly complex visuals, only one dataset is presented fully. The
remaining years are covered by the other dataset (no overlapping periods are shown). In
deciding which dataset to present fully, I choose the visuals that support the paradox thesis
the most. This is in line with the aim of this work to assess this thesis in a very favourable
context: if there is a paradox, it should be apparent here.
Contrarily, however, there is a remarkable decline in the percentage of full-fledged
conflicts, i.e. conflict on ethno-territorial issues and between segmental side(s). In the late
1970s and early 1980s, such clashes sometimes accounted for over 30% of all cabinet conflicts.
Since then, they gradually yet clearly lost relevance and after the latest state reform (2011), they
139
Appendices 4 and 5 present the relative frequency of conflicts on ethno-territorial issues and conflicts on
segmental lines, respectively. For their absolute frequencies and yearly equivalents at the cabinet level, see
Appendices 12 and 13.
192
disappeared altogether. As this type represents the pinnacle of ethno-territorial conflict, these
findings provide a particularly strong case against the paradox thesis. In contrast, thematic
conflicts (issues, not sides) persist throughout time, although in varying proportions. Such
fights, in which ethno-territorial issues give rise to clashes without putting the Flemish and
Francophone segments against one another, appear to be of all times. However, their heydays
are clearly in the past, too. The same applies to factionalized conflicts (sides, not issues), which
are far less frequent in number. In sum, then, the results demonstrate that ethno-territorial
subjects trigger less tensions, that the Flemish and Francophone segments less often fight about
such subjects, and that hostilities between both sides are less common on other policy domains
too (e.g. socio-economic policy, migration). Put differently, pacification is observed on all
indicators.
These conclusions come with many nuances, one being the presence of many fluctuations.
Also, the results for the absolute frequencies and yearly equivalents (reported in Appendices 12
and 13) are somewhat less supportive of the pacification thesis. Yet, they too fail to provide any
evidence for the paradox of decentralization, which implies the kind of escalations (increasing
conflicts levels) that fail to be noted.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Figure 21. Ethno-territorial cabinet conflicts by type: yearly percentage of all
cabinet conflicts (yearbook data: 1979–2006; news archive data: 2007-2018)
Full-fledged Factionalized Thematic Not
193
4. Intermediate conclusions
How did the ethno-territorial conflict in Belgian power-sharing cabinets evolve?
This chapter answered this research question (RQ1) by studying the frequency of these conflicts
throughout four decades and against the background of five decentralization reforms. Apart
from assessing absolute and relative conflict frequencies, I also compared consecutive cabinets
and their conflict proneness. The evolution of different sub-populations was also tracked. Are
intense ethno-territorial conflicts on the rise? What about conflicts on ethno-territorial issues or
disputes along segmental lines? And how did the prevalence of the three types of ethno-
territorial conflicts evolve (full-fledged, thematic, and factionalized clashes)? The resulting
picture sheds light on 21 indicators.
140
The results are clear as a bell. All indicators forcefully contradict the paradox thesis.
Confirming hypothesis 2, decentralization did not come with growing levels of ethno-territorial
conflict. Indicators typically vary over time, with peaks and valleys across the timelines. Also,
state reform deals did not have an invariable impact in the short term. Sometimes, they are
directly followed by pacification. Sometimes, long-lived escalations are noted. Perhaps most
importantly, insofar as any long-term tendency can be discerned, it is that of a tentative decline
of conflict against the background of decentralization. Even in a very unfavourable context
for accommodation (cf. least-likely case; Eckstein, 1975), decentralization certainly did
not fuel the fire of ethno-territorial cabinet conflict. Ethno-territorial conflict frequencies
declined over time and following the 2011 agreement on the sixth state reform, an
unprecedented period of lasting stability was observed. Chapter 7 already demonstrated that the
remaining conflicts are less intense, less commonly fought along segmental lines, and less
commonly full-fledged conflicts. This chapter shows that the poltergeist not only has a
friendlier face. It also makes fewer appearances.
Several additional insights of relevance were revealed too. For instance, the conflict’s relative
prevalence also shows that ethno-territorial conflict gradually became a side-product of
confrontational politics, rather than being a predominant source of distress itself. Until the
mid-1990s, ethno-territorial clashes sometimes accounted for (over) half of the conflicts in
140
The absolute frequency, relative frequency, and yearly cabinet equivalents of (1) ethno-territorial
conflicts, (2) intense ethno-territorial conflicts, (3) clashes on ethno-territorial issues, (4) along segmental
sides, (5) full-fledged conflicts, (6) thematic disputes, and (7) factionalized conflicts.
194
power-sharing executives. They used to be the primary burden of these cabinets. In recent
decades, this is no longer the case. Meanwhile, intense ethno-territorial clashes appeared to
have come in waves, coming and going as the years went by. Their magnitude steadily declined,
and in recent years, the formerly raging sea of ethno-territorial disputes turned into stagnant
water.
While these results solidly debunk the paradox thesis (H3), the question is whether the
observed pacification trends are statistically related to the decentralization process (H1).
Put differently, and using the famous terminology of Brancati (2006), we now know that
decentralization did not fuel the fire, but it is yet to be determined whether it is responsible for
the dampening of the flames. Doing so is the goal of Chapter 9, in which the role of
decentralization is assessed while controlling for a series of other factors.
195
Chapter 9.
Explanatory analyses: decentralization is related to ethno-
territorial pacification
141
How did decentralization impact the ethno-territorial conflict in Belgian power-sharing
cabinets? (RQ2) In this chapter, and using the yearbook data (1979-2006), the relation between
these conflicts and decentralization is assessed from different angles. Specifically, the analyses
show (1) whether and how decentralization is related to the frequency of ethno-territorial
conflict (Poisson regressions) and (2) whether and how it relates to the chance that cabinet
conflicts are ethno-territorial in nature (logistic regressions). In every way, the results strongly
contradict the paradox of federalism. In general, the analyses empirically confirm the negative
association between decentralization and conflict that was visually noted in the descriptive
chapters. Going beyond institutions, the analyses also expose the conflict-fuelling role of issue
saliency, regionalist cabinet participation, and the proximity of elections. No significant results
were found concerning regionalist parties’ strength.
1. Variables (descriptives)
Figure 22 visualizes the yearly number of ethno-territorial conflicts and the main
predictor variables (decentralization degree and saliency, regionalist parties’ strength and
cabinet participation, and the presence of federal elections). The federalization deals of 1980,
1988-9, 1993, and 2001 visibly boosted the degree of decentralization. In terms of autonomy
provisions, the most significant state reforms under study are the second (1980) and third ones
(1988-9), which established the Flemish, Walloon, and Brussels-Capital Regions. The adapted
RAI scores increased from 4.41 to 7.18 (+2.77) with the second state reform and continued to
grow with the third state reform (to 11.19 = +4.01), the fourth state reform (13.08; +1.89), and
141
Conflicts that might not have been between coalition partners (cases of doubt) are not included in the
analyses. Contrarily, and given the partitocratic nature of the Belgian polity (Baudewyns et al., 2022; De
Winter & Dumont, 2006), cases are included when there is doubt about the applicability of the relevance
criterion. The Belgian partitocracy comes with very low chances of purely individual actions by political
loners. These analyses and results are expected to be presented in an article that is under review at the time
of publication (Vandenberghe, 2023).
196
the fifth state reform (13.98; +0.90). The period between these reforms was characterized by
the electoral decay of the regionalist parties. However, due to the growth of the separatist
Vlaams Blok/Vlaams Belang (VB), this decline was altered from the late 1980s/early 1990s
onwards (on the rise of the VB, see Pauwels, 2011; Sijstermans, 2021). Interestingly,
decentralization became far (x5) more salient in party manifestoes since the 1990s. This is
rather surprising, given the pertinence of ethno-territorial clashes in the 1970s and 1980
reported in Chapter 8. However, even then, saliency levels were quite low. Indeed, as Table 12
shows, parties only devote marginal shares of their party manifesto to statements showing
“support for federalism or decentralisation of political and/or economic power” (category
0,00
1,00
2,00
3,00
4,00
5,00
0
2
4
6
8
10
12
14
16
18
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
Decentralization saliency
Other variables
Decentralization degree (a) Regionalist party strength (b)
Ethno-territorial conflict (N) Decentralization saliency (c)
Regionalist party in cabinet
Figure 22. Yearly ethno-territorial cabinet conflict frequency, decentralization degree and saliency,
regionalist party strength, regionalist cabinet participation, and federal elections (Belgium, 1979-2006)
Note: In the case of different regionalist party scores & polarization scores within a given year
(election years and years with different cabinets), the graph presents the scores applying to the
largest part of the year (e.g. election in December 1987: regionalist party strength based on the
previous election – 1985).
(a) Refined self-rule score of RAI – excluding autonomy of the provinces (cf. L. Hooghe et al.,
2016; Shair-Rosenfield et al., 2021). For a discussion of the measurement, see Chapter 6 of this
work; (b) Federal elections; VU (Volksunie), N-VA (Nieuw-Vlaamse Alliantie), FDF (Front
démocratique des francophones), RW (Rassemblement Wallon), and VB (Vlaams Blok/Vlaams
Belang) (% of seats in Belgian Chamber of Representatives); (c) average percentage of quasi-
sentences on decentralization in party manifestos of traditional parties, greens, and regionalists (incl.
VB), using the per301 category of the CMP dataset (Volkens et al., 2021a).
Federal election year
197
per301 in the CMP dataset; Volkens et al., 2021b, p. 13). The analyses presented in this chapter
put the potential relations between these variables and the dependent variables to the test. More
specifically, I provide a view from different angles. First, using Poisson regressions, I assess
the frequency of ethno-territorial conflicts: Is decentralization associated with higher or
lower conflict frequencies? Second, and using logistic regressions, the ethno-territorial
nature of conflicts is studied: Does decentralization increase or decrease the chance of cabinet
conflicts being ethno-territorial in nature?
Table 12. Descriptive statistics (dataset 1: monthly ethno-territorial conflict
frequency; dataset 2: conflict characteristics)
Considering these variables, we might expect multicollinearity issues to be at play – “the
thousand pounds monster in statistical modelling” (Midi, Sarkar, & Rana, 2010, p. 253).
Previous research indicates that strong relations might exist between e.g. regionalist party
strength and decentralization (Brancati, 2006, 2009; De Winter, 1998a; Massetti & Schakel,
2013, 2017) or decentralization saliency (M. Mazzoleni, 2009). Similarly, several authors have
identified empirical links between regionalist cabinet participation and decentralization reforms
or between their presence in executives and their electoral strength (De Winter, 1998a; Elias &
Tronconi, 2011; O. Mazzoleni & Mueller, 2016; Terrière & Bouteca, 2020). However, these
relations do not resurface in the datasets used here. Multicollinearity tests indicate that none of
the models discussed in this chapter is burdened by high correlations between the predictor
variables (VIF <5). The multicollinearity problem only occurs in some of the models in which
Variable
Mean
SD
Min
Max
N
Number of ETCs in month
0.28
0.52
0
3
336
Decentralization(a)
10.63
|10.53|
2.96
|3.20|
4.41
|4.41|
13.98
|13.98|
336
|322|
Regionalist party strength(b)
12.92
|13.35|
2.26
|2.53|
9.43
|9.43|
16.67
|16.67|
336
|322|
Regionalist party in coalition (dummy)
0.16
|0.17|
0.37
|0.38|
0.00
|0.00|
1.00
|1.00|
336
|322|
Electoral saliency of decentralization(c)
2.42
|2.17|
1.73
|1.59|
0.44
|0.44|
4.93
|4.93|
336
|322|
Months since last election
21.83
|22.39|
13.10
|13.27|
1.00
|1.00|
49.00
|49.00|
336
|322|
Number of ETCs in previous month
0.28
0.52
0
3
335
Note: Results for dataset 2 are reported |between vertical dashes| below those of dataset 1.
(a) Refined self-rule score of RAI – excluding autonomy of the provinces (cf. L. Hooghe et al., 2016; Shair-Rosenfield
et al., 2021). For a discussion of the measurement, see Chapter 6 of this work; (b) Federal elections; VU (Volksunie),
N-VA (Nieuw-Vlaamse Alliantie), FDF (Front démocratique des francophones), RW (Rassemblement Wallon), and
VB (Vlaams Blok/Vlaams Belang) (% of seats in Belgian Chamber of Representatives); (c) average percentage of quasi-
sentences on decentralization in party manifestos of traditional parties, greens, and regionalists (incl. VB), using the
per301 category of the CMP dataset (Volkens et al., 2021a).
198
the radical right VB party is not considered to be a regionalist party (Appendices 14 and 15).
2. Poisson regression: the frequency of ethno-territorial conflicts
Table 13 presents the results of the Poisson regressions that model the number of
ethno-territorial conflicts of each month. This allows me to make sense of the frequency
of ethno-territorial conflicts. Here, cases are individual months (N= 335-336). All models fit
the data adequately (residual deviance analyses) and Kolmogorov-Smirnov tests confirm that
the distribution of counts in the dependent variable follows the Poisson distribution. The
bivariate model 1a shows that decentralization is significantly associated with decreases in
the number of monthly clashes. More specifically, the incidence rate ratio (IRR) demonstrates
that, holding other variables constant, a one-unit increase in the decentralization variable is
expected to decrease the number of monthly clashes by a factor of 0.87. Making things more
palpable, such an increase in the decentralization degree roughly corresponds with the impact
of the fifth state reform adopted in the early 2000s (which boosted the decentralization level
from 13.08 to 13.98). This significant relation holds when taking into account the role of
regionalist parties (model 2a) and the other variables under study (model 3a). So even after
controlling for other factors of relevance, the rates of ethno-territorial conflict are lower when
sub-state autonomy levels are higher. In short, this empirically confirms the link between
decentralization and ethno-territorial pacification (H1).
How do the other independent variables relate to ethno-territorial conflict frequency? Models
2a and 3a expose a significant association between the presence of regionalist parties in
cabinets and larger monthly conflict frequencies. According to the rate ratios and holding
the other variables constant, the monthly number of clashes is 1.70 to 1.71 times as high for
cabinets with regionalists as for cabinets without such parties. Put differently, the presence of
regionalist parties in a federal cabinet is associated with an expected increase of at least
70% in the monthly number of ethno-territorial conflicts. Interestingly, the conflict-fueling
potential of regionalist parties is thus weaker when they are seated at the banks of the
opposition. However, the average number of monthly clashes is rather low (0.28), which puts
these rather remarkable statistics in perspective. Finally, there might be some distortion due to
the presence of clashes that cover several months (e.g. a conflict starting in January and
continuing in February). In this respect, however, the results are rather comforting: conflict
frequency is not significantly related to the number of ethno-territorial conflicts in the previous
month. The same non-results are found for the other (control) variables under study. So
contradicting the expectations, the results do not confirm the claims that the electoral
strength of regionalist parties, the electoral saliency of decentralization, or the proximity
of federal elections fuel conflict frequencies. As such, the analyses unexpectedly provide
199
some counterweight to the idea of conflict being a derivate of electoral concerns and incentives.
Highly similar findings arise when the radical right VB party is not considered to be a
regionalist party (and as such, excluded in the calculation of regionalist party strength)
(Appendix 14). The only difference is that in these analyses, I do find an association between
election proximity and conflict frequency (IRR= 1.02).
Table 13. Poisson regressions with monthly number of ethno-territorial
conflicts (ETCs) as dependent variable (incidence rate ratios and Std. Errors)
3. Logistic regression: the ethno-territorial nature of conflicts
Shedding light on the issue from a different angle, a second set of analyses (logistic
regressions) makes sense of the ethno-territorial nature of cabinet conflicts. Does
decentralization decrease or increase the probability that conflicts are ethno-territorial when
they occur? Or fought along segmental lines? Or about ethno-territorial issues? Negative
answers would suggest that decentralization is not per se a determinant of ethno-territorial
Model 1a
Model 2a
Model 3a
Decentralization
0.87 ***
(0.03)
0.89 **
(0.04)
0.86 *
(0.06)
Regionalist party
strength (% of seats)
0.99
(0.05)
0.99
(0.05)
Regionalist party in
coalition (dummy)
1.70 *
(0.25)
1.71 *
(0.27)
Decentralization
saliency
1.06
(0.11)
Election proximity
1.01
(0.01)
Number of non-ETCs
1.02
(0.12)
Number of ETCs in
previous month
1.22
(0.18)
N
336
336
335
AIC
429.45
428.67
430.96
BIC
437.08
443.94
461.47
Pseudo R²
0.06
0.08
0.09
*** p < 0.001; ** p < 0.01; * p < 0.05
Note: intercepts not reported
200
conflict. But now, the results add to the evidence for the link between autonomy and ethno-
territorial stability. Recall that the cases here are not specific months but individual cabinet
conflicts (N= 319-322). Reassuring analyses of the standardized residuals demonstrate that
none of the models contains possible outliers. Table 14 shows the results of logistic regression
models that assess the chance that conflicts are ethno-territorial in nature (models 4a, 5a, and
6a). Echoing the pacification thesis (H1), these odds decrease as decentralization levels
increase. This decrease is significant but not spectacular. Specifically, as the self-rule score
increases with one, the odds that a cabinet conflict is ethno-territorial in nature decreases by a
factor of 0.80 to 0.90 (depending on the model). This indicates that decentralization is in fact a
determinant of ethno-territorial conflicts specifically (rather than being a stabilizing force on
all kinds of cleavages). Contrarily, the participation of regionalist parties in federal
executives strongly increases the odds of clashes within the cabinet being ethno-territorial in
nature. However, this relation is only statistically significant in the analyses that exclude the
VB from the list of regionalist parties (Appendix 15). In summary, these findings are in line
with hypothesis 1 (decentralization is negatively related to conflict). No significant results
were found for any of the other variables.
Of course, ethno-territorial conflicts are no uniform monoliths. As elaborately described in
Chapter 2, they include both conflicts on ethno-territorial issues and conflicts fought between
segmental fronts. Recognizing this heterogeneity and further exploring its relevance, separate
analyses are presented. To begin with, Table 14 also shows the results for the logistic
regressions in which the dependent variable is the presence or absence of (one or more)
segmental sides (models 7a, 8a, and 9a). Do self-rule provisions decrease the odds of segmental
trench warfare? The results do indicate that decentralization comes with lower odds that,
when a cabinet conflict arises, the Flemish and/or Francophone side(s) act as blocks.
However, and notably, this result is only significant when all variables are regressed.
Specifically, model 9a shows that, as the refined self-rule score of the RAI increases with one
unit, the odds of a conflict being fought along segmental lines decrease by a factor of 0.77. This
contradicts the idea that decentralization ‘entrenches’ divisions and incentivizes politicians to
act accordingly: “Group recognition ensures the perpetuation of the differences” and provides
segmental elites “with a vested interest in the continuation of the divided system” (Erk &
Anderson, 2010a, p. 2). Recognizing divisions does not prevent politicians from transcending
them. In Belgium, it even resulted in a context in which such divisions became less politically
shaping.
Contrarily, three variables significantly relate (to varying degrees) to larger odds that
segments stand against each other when conflicts arise. First and most clearly, regionalist
cabinet participation comes with a tripling of these odds (model 9a). When regionalists govern,
201
parties belonging to the same segment are more inclined to join forces against ‘the other side’.
Such segmental phalanxes are also more common when decentralization was more salient in
the build-up towards the preceding elections (in party manifestoes) and when federal elections
are approaching. However, the latter result is far less outspoken (Odds ratio= 1.03). Electoral
proximity matters, but hardly so. As such, and even in a split party system that is thought to
provide little incentives for accommodation (Horowitz, 1991, pp. 167-176; 2002, p. 23), the
competition in the build-up to elections does not really lead to full-blown and increasingly
present ethnic-outbidding dynamics (Rabushka & Shepsle, 1972). For transparency’s sake, it
should be noted that the role of decentralization and election proximity is no longer significant
when the radical right VB party is excluded from the list of regionalist parties (Appendix 15).
Table 14. Logistic regressions (Odds ratios and CI)
DV: Ethno-territorial (0/1)
Model 4a
Model 5a
Model 6a
Decentralization
0.86 ***
(0.79 – 0.92)
0.90 *
(0.82 – 0.98)
0.80 **
(0.69 – 0.92)
Regionalist party strength
(% of seats)
0.90
(0.80 – 1.00)
0.91
(0.81 – 1.02)
Regionalist party in coalition
(dummy)
1.63
(0.85 – 3.12)
1.89
(0.97 – 3.73)
Decentralization saliency
1.28
(0.98 – 1.65)
Election proximity
1.02
(0.99 – 1.04)
N
322
322
322
R² Tjur
0.050
0.072
0.086
DV: Segmental side(s) (0/1)
Model 7a
Model 8a
Model 9a
Decentralization
0.93
(0.85 – 1.02)
0.96
(0.86 – 1.06)
0.77 **
(0.65 – 0.91)
Regionalist party strength
(% of seats)
1.04
(0.91 – 1.19)
1.09
(0.95 – 1.26)
Regionalist party in coalition
(dummy)
2.15
(0.98 – 4.69)
3.03 *
(1.28 – 7.30)
Decentralization saliency
1.60 **
(1.18 – 2.18)
Election proximity
1.03 *
(1.00-1.06)
N
322
322
322
R² Tjur
0.008
0.023
0.061
202
Table 14 also shows the results of the logistic regressions in which the presence or absence of
an ethno-territorial issue is included as the dependent variable (models 10a, 11a, and 12a). Does
decentralization come with lower odds that clashes revolve around such issues? Once again,
the results back the pacification thesis (H1). Decentralization is effectively related to lower
odds that ethno-territorial issues are at stake when coalition partners clash. This finding
continues to be significant when controlling for the electoral and executive role of regionalist
parties (models 11a and 12a), decentralization saliency, and the proximity of elections (model
12a). As the refined self-rule score of the RAI increases with one, these odds decrease by a
factor of 0.79 to 0.83 (depending on the model). As such, and at least to some extent,
decentralization seems to guide ethno-territorial issues away from the cabinet (conflict) agenda.
Interestingly, no significant results are found for the other variables. Hence, the analyses
do not provide any support for the idea that regionalist parties are ethno-territorial agenda-
setters. Their strength or cabinet participation does not appear to push ethno-territorial issues to
the forefront of political debates (or at least, to the part of the agenda that ignites into tensions).
Surprisingly, approaching elections fail to do so too. Also, when decentralization is more salient
in the build-up towards federal elections, this does not translate into greater odds that cabinet
clashes will concern ethno-territorial issues. Similar findings arise when the radical right VB
party is not considered a regionalist party (Appendix 15).
DV: Ethno-terr. Issue (0/1)
Model 10a
Model 11a
Model 12a
Decentralization
0.80 ***
(0.74 – 0.87)
0.83 ***
(0.76 – 0.91)
0.79 **
(0.68 – 0.93)
Regionalist party strength
(% of seats)
0.94
(0.83 – 1.05)
0.94
(0.83 – 1.06)
Regionalist party in coalition
(dummy)
1.63
(0.83 – 3.20)
1.71
(0.85 – 3.43)
Decentralization saliency
1.11
(0.82 – 1.48)
Election proximity
1.00
(0.98 – 1.03)
N
319
319
319
R² Tjur
0.090
0.102
0.104
*** p < 0.001; ** p < 0.01; * p < 0.05
Note: intercepts not reported
203
4. Intermediate conclusions
By the start of this chapter, the paradox thesis (H3) was already debunked. In previous
chapters, I showed that there are fewer ethno-territorial clashes than in the past (Chapter 8) and
that the ones that still arise have a friendlier face (less intense, less segmental sides, etc.)
(Chapter 7). As tensions declined, decentralization certainly did not result in ethno-territorial
escalation. However, the question remained: Is the pacification I observed statistically
related to the five decentralization reforms under study? Did decentralization pacify ethno-
territorial tensions (as H1 states)? Was it a factor of relevance or just an institutional bystander?
This chapter answered this question. Using Poisson and logistic regression analyses, it exposed
whether decentralization is significantly related to the frequency of ethno-territorial conflict and
to the chance that cabinet conflicts are ethno-territorial in nature. The latter indicates whether
the factors under study can be considered determinants of ethno-territorial conflict per se.
Being triggered by insights provided in previous studies, the analyses control for the role of
several other (potential) determinants of ethno-territorial conflict.
The results consistently confirm the link between sub-state autonomy and federal stability
(H1). Decentralization is significantly associated with decreasing frequencies of ethno-
territorial conflicts (fewer monthly clashes), lower odds that cabinet conflicts are ethno-
territorial in nature, decreasing odds that ethno-territorial issues are at stake when coalition
partners clash, and decreasing odds that coalition partners engage in the kind of segmental
trench warfare that involves a front(s) of all Flemish and/or Francophone parties. These findings
hold when controlling for other factors of relevance (notably, the latter result only holds when
doing so). As such, the results add to the empirical case against the paradox thesis (H3).
Decentralization did not only come with declining levels of conflict (Chapter 8), it is also
statistically related to this decline. Institutionalizing divisions did not prevent politicians
from transcending them. On the contrary. At least at the level of central-level cabinet
conflicts, decentralization seems to have been a pacifying rather than an escalating force.
Of course, decentralization is only one piece of an extensive puzzle. To use one of the many
catch-phrases that brighten up the literature: it is not a panacea (e.g. Bermeo, 2002; Keil &
Anderson, 2018; Kymlicka, 2001a; Seymour & Gagnon, 2012). Going beyond institutions, the
analyses also examine the relevance of issue saliency, regionalist parties’ strength, their cabinet
participation, and the proximity of federal elections. For all factors, except for the strength
of regionalist parties, the analyses echo the expectations (at least to some extent). These
variables are all positively related to ethno-territorial conflict and more specifically, to the
chances that Flemish and Francophone ‘blocks’ stand against each other when cabinet conflicts
arise (confirming H6, H7, and H8). They are thus linked to the changing nature of clashes. But
204
only regionalist cabinet participation is also related to the frequency of ethno-territorial
conflicts. Hence, conflict levels do not increase when regionalist party scores are higher, when
decentralization is more salient in the build-up towards elections, or when elections come
nearer. As such, the electoral concerns and incentives under study do not appear to be an
important driving force of ethno-territorial tensions. The perspective of oncoming elections
does not lead to growing levels of political aggressiveness or increasingly clear ethnic
outbidding dynamics (Rabushka & Shepsle, 1972).
Providing strong support for H6, the presence of regionalist parties in federal cabinets yields
the most outspoken results. It triples the odds that cabinet conflicts are fought along linguistic
lines. The inclusion of regionalist parties in the federal coalition is also associated with an
expected increase of 70%–71% in the monthly number of ethno-territorial conflicts.
Contrarily, their electoral strength seems to be rather irrelevant. This is striking.
Regionalist parties stress ethnic divisions, push territorial issues to the top of the political
agenda, and can trigger dynamics of ethnic outbidding (Horowitz, 2000; Meguid, 2005; Müller-
Rommel, 1998; Rabushka & Shepsle, 1972). Their strength was thus expected to be a conflict-
fuelling force. However, the empirical evidence does not corroborate the expectation. None of
the models provided any evidence for such a relation. Perhaps, this non-effect is partly due to
the fact that the regionalist parties’ electoral peak was already reached before the starting point
of this work, namely by the early 1970s (Delwit, 2011a; Deschouwer, 2009; van Haute & Pilet,
2006). Being in decline, with the exception of the radical right VB, their strength might not
have reflected the kind of pressure needed to truly light the fuse of conflict. By 1979 (the first
year covered by the data used here), the regionalist threat might have already been perceived as
being under control in the eyes of the traditional parties. Whatever the case, the results
demonstrate that (at least in the period under study) it is not the electoral strength of
regionalist parties that fuels ethno-territorial tensions but the choice to let them govern.
Perhaps, this conclusion would fail to hold in recent times due to the N-VA’s electoral success
and its participation in a federal coalition that avoided ethno-territorial debates (Benoît Rihoux
et al., 2016, p. 30).
These results highlight the importance of the institutional setting while showing that the actors
at play, the regionalist and electoral pressure they experience, and the saliency of ethno-
territorial debates deserve our attention too. They also demonstrate the fruitfulness of
recognizing the heterogeneity of ethno-territorial conflicts: conflict on ethno-territorial
issues seems to have other driving forces than conflict along segmental lines.
The story doesn’t end here. In evaluating the paradox thesis, attention should not only be
devoted to the frequency or nature of conflicts. It is also important to study how politicians
handle or solve conflicts once they arise (H2 and H4). After all, decentralization might have
205
led to declining levels of conflict (Chapters 8 and 9), but perhaps, it made it harder to solve the
ones that still arise (H4). Several authors have raised concerns about the (long-term) effects of
state reform on conflict resolution dynamics at the central level, arguing for instance that it
makes it harder to find consensus (e.g. Caluwaerts & Reuchamps, 2015). Providing a
longitudinal overview of conflict outcomes, Chapter 10 assesses this logic.
206
207
Chapter 10.
Conflict outcomes: Decentralization did not burden conflict
resolution
142
To truly evaluate the paradox thesis, it is not only important to assess the prevalence of cabinet
conflicts. It is also relevant to study how politicians handle or solve conflicts once they arise.
Indeed, while decentralization might have led to declining levels of conflict, these very reforms
might have made it harder to solve the clashes that still exist. But what does conflict-solving in
power-sharing arrangements look like in practice? This chapter sheds light on the outcomes of
conflicts in the fifteen power-sharing cabinets covered by the yearbook data (1979-2006).
Specifically, it assesses (1) how politicians (fail to) resolve conflicts, and (2) how this evolved
against the background of decentralization. Apart from providing a general overview of the
prevalence of outcome types, I also discern different dimensions of functionality. In light of the
paradox vs. pacification debate, the most relevant one is the consensus dimension. The results
show that only a minority of clashes are settled in consensus. However, in this respect as in
other dimensions, there is great stability over time. Decentralization did not make it harder to
solve ethno-territorial clashes. Nor did it impair the general conflict resolution capacity of
power-sharing cabinets. As such, the pacification noted in earlier chapters did not come at the
expense of federal functionality. If anything, central cabinets are quite consistently (if not
comfortably) numb.
1. Aggregated results: how are ethno-territorial conflicts solved?
How do ethno-territorial conflicts in consociational power-sharing-cabinets end? In
general, three outcome categories dominate the picture (Figure 23). First and most notably,
a large set of clashes simply blows over without any observed consequences (40%). Perhaps,
142
Conflicts that might not have been between coalition partners (cases of doubt) are not included in the
analyses. Contrarily, and given the partitocratic nature of the Belgian polity (Baudewyns et al., 2022; De
Winter & Dumont, 2006), cases are included when there is doubt about the applicability of the relevance
criterion. The Belgian partitocracy comes with very low chances of purely individual actions by political
loners. Some of these analyses and results are expected to be presented in an article that is under review at
the time of publication (Vandenberghe, 2022d).
208
the large number of conflicts without any notable results whatsoever (blow over) might be the
result of the significant share of low-intensity clashes (29%). These are conflicts that only
involve forms of e.g. sneering or short-lived bickering, but no forms of threats, blockings,
resignations, or the like. As shown in Appendix 16, the proportion of least-intense clashes that
simply blow over is in fact far larger than that of the conflicts with higher intensity levels (over
54%, as opposed to 32-33%). However, these numbers also show that many (about one in three)
intense clashes blow over too.
Figure 23. Outcomes of ethno-territorial cabinet conflicts (yearbook data; N= 96)
Second, a significant share of ethno-territorial conflicts is settled with a compromise (24%).
One example is the 1979 clash about the Francophone public school (kindergarten) in the
facility community of Linkebeek (c5). Minister of Education J. Ramaekers (Flemish social
democrats, BSP) decided to suspend the funding of the school (also retroactively) because some
of the parents of the children who signed up did not live in Linkebeek or one of the other facility
communities surrounding Brussels (which goes against the 1963 language law). The decision
was strongly condemned by the Francophone parties and the regionalist FDF threatened with
retaliation (aiming at the Dutch educational system in Brussels). To keep the school running, a
non-profit organization was established and funded by i.a. the Francophone social democrats
(PS; 300.000 BF) and the FDF (500.000 BF). Right before the end of the school year and after
long negotiations, the government found an agreement on the issue: the ‘unlawful’ children
could end their year but would not be allowed to sign up in the consecutive school year (except
Compromise
24%
Impos. (non-segm.)
5%
Impos. (segm.)
4%
Resignation member
3%
Resignation party
1%
Resignation cabinet
4%
Residual
2%
Blow over
40%
Procedural
17%
209
for two children whose parents were teachers). From July on, the government resumed the
funding.
Third, almost one in five clashes ends with a procedural decision (17%), such as the 1979
agreement to settle a fierce clash on state reform by establishing a working group that would
discuss the fate of Brussels (c22, a conflict discussed in Chapter 7). A similar approach was
used in the summer of 1990, when Prime Minister Wilfried Martens announced the decision to
establish a vague working schedule on state reform and install a working group consisting of
staffers of different cabinet members (c231). The announcement was an explicit attempt to
accommodate the impatient Flemish nationalist coalition party (VU), which threatened to leave
the cabinet if no progress was made on the state reform issue.
Meanwhile, an equally large share of ethno-territorial conflicts ends with overt forms of
hostility (imposition or resignation; 17%). This is at odds with the idea(l) that functional
power-sharing should facilitate elite cooperation and joint decision-making (McCulloch &
Vandeginste, 2019, p. 1178; McEvoy & Aboultaif, 2022). Almost ten per cent of all ethno-
territorial clashes end with politicians imposing their will upon their coalition partner(s).
Interestingly, segmental impositions (4%), i.e. situations in which the Flemish parties impose
their will upon the Francophones (or vice versa) are roughly as frequent as non-segmental
impositions (5%). Given the majority position of the Flemings, this is rather remarkable. Insofar
as it existed, l’Etat Belgo-Flamand is clearly a thing of the past. Finally and unsurprisingly,
only a rather small set of clashes ends with the resignation of a cabinet member (3%), a coalition
party (1%), or the cabinet as a whole (4%).
What do these results tell us from a functionality perspective (Table 15)? Recall that
different functionality dimensions are discerned. After all, different functionality conceptions
exist and as such, functionality is partly in the eye of the beholder. Attempting to provide a
more aggregated picture too, I first present the total functionality level (TFL; 0-500); that is the
sum of all percentages in the bottom row of Table 15. The score for the entire period under
study (TFL= 317) indicates a medium level of functionality. On a more fine-grained level and
unsurprisingly, the results show that different dimensions of functionality indeed provide a
different picture: the lower the bar, the greater the level of functionality. Politicians
manage to settle almost all conflicts while avoiding the resignation of the cabinet (96%) or its
constitutive members (92%). But in policy terms, many conflicts are fought in vain. Often,
politicians fail to find a substantive solution or deliver any palpable result at all. Only one in
two clashes actually results in a decision (53%) and only part of these provide some kind of
substantive answer to the issue at hand (35% of all ethno-territorial conflict outcomes). From a
consociational point of view, this need not be problematic. Failing to decide or leaving clashes
substantially unresolved is not per se at odds with the consociational premise of accommodative
210
elite behaviour. On the contrary. Such situations can be very confrontational (e.g. a cabinet
resignation), but they can also be indicative of a very accommodative spirit; a willingness to
overcome divisions and choose political stability over policy victory. Agreeing to disagree, for
instance through procedural decisions, is part of the traditional consociational toolbox (Lijphart,
1968c). Such instances can be the product of the kind of ‘willingness not to be the final winner’
that is key to the functioning of power-sharing systems (Deschouwer & Temmerman, 2012, p.
501). To consociational scholars, it is arguably not the decision or substantive dimension that
matters but the consensus dimension. This is also the most relevant dimension for the debate
on the pacifying or escalating effect of decentralization (H2 and H4). But even if functional
conflict resolution is equated to consensus decision-making, the picture is still rather
bleak. Only a minority of clashes are actually settled in consensus (41%), that is with a
procedural decision or compromise. Three in five ethno-territorial conflicts end without explicit
consensus.
Interestingly, as Appendix 17 shows, these findings on ethno-territorial conflicts are not
strikingly different from the results for other cabinet conflicts. As such, ethno-territorial
conflict resolution does not appear to be harder than the settlement of other clashes.
Table 15. Outcomes of ethno-territorial conflicts by type and functionality dimension:
percentage of all ethno-territorial conflict outcomes (N= 96)
Conflict
outcome type
Cabinet
survival
Cabinet
member/party
survival
Decision
Substantive
Consensus
Procedural
16.67%
16.67%
16.67%
16.67%
16.67%
Compromise
23.96%
23.96%
23.96%
23.96%
23.96%
Imposition
9.38%
9.38%
9.38%
9.38/9.38%
9.38%
Resignation
4.17/8.34%
0.00/8.34%
1.04/8.34%
1.04/8.34%
0.00/8.34%
Blow over
39.58%
39.58%
39.58%
39.58%
39.58%
Residual
2.08/2.08%
2.08/2.08%
2.08/2.08%
1.04/2.08%
0.00/2.08%
Total % of
conflicts
95.84%
91.67%
53.13%
35.42%
40.63%
211
2. Evolution of (ethno-territorial) conflict outcomes against the
background of decentralization
How did ethno-territorial conflict management evolve over time and (most importantly)
against the background of decentralization? Did the pacification discussed above come at the
expense of functionality? In exploring these trends, the conflict outcome data are presented on
different levels. To begin with, at the yearly level, no clear-cut decrease in functionality is
observed (Figure 24). Rather, various fluctuations are observed both in the yearly TFL scores
and in all individual dimensions of functionality. Interestingly, the least functional periods were
the ones marked by the highest numbers of ethno-territorial conflicts (cf. Chapter 7), namely
1979-81, 1986-87, and 1990-91. The exceptionally dysfunctional year of 2003 forms an
exception to this apparent relation (but in this year, only one ethno-territorial conflict was
noted). More generally, conflict frequency and conflict resolution functionality are not just
two sides of the same coin. They clearly follow different trends, and the process of ethno-
territorial pacification described in Chapter 7 did not come with growing levels of functionality.
Of course, this yearly picture only tells us so much. Especially given the rather low and strongly
varying number of yearly clashes (X= 3.43; SD= 2.41), it might be relevant to consider trends
at a more aggregated level.
In this respect, the periods between the different state reforms can be discerned. For each of
these periods, Figure 25 shows the percentage of ethno-territorial conflict outcomes and the
percentage of outcomes of other clashes that meets a certain functionality dimension. Four
remarkable observations deserve some reflection.
First, neither ethno-territorial nor other conflicts are increasingly harder or easier to
accommodate. There is no general increase or decrease in conflict resolution functionality.
Rather and once again, strong fluctuations are noted on all dimensions (although far less on the
survival dimensions, which remain rather stable). This indicates that decentralization did not
pacify tensions in the short term while making the process of (segmental) conflict resolution
harder in the long run, as is sometimes hypothesized (Caluwaerts & Reuchamps, 2015). Conflict
resolution in power-sharing cabinets is not increasingly hard. Especially for non-ethno-
territorial conflicts, total functionality levels are rather stable throughout time (292-344/500).
212
0
10
20
30
40
50
60
70
80
90
100
Cabinet survival
Figure 24. Outcomes of ethno-territorial cabinet conflicts in Belgium by year: total functionality level
(0-500) and percentage of outcomes meeting different functionality dimensions (1979-2006; N= 96)
0
10
20
30
40
50
60
70
80
90
100
Cabinet member/party survival
0
10
20
30
40
50
60
70
80
90
100
Decision
0
10
20
30
40
50
60
70
80
90
100
Substantial
0
10
20
30
40
50
60
70
80
90
100
Consensus
0
50
100
150
200
250
300
350
400
450
500
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
Total functionality level
Note: As discussed in Chapter 6 of this work, the total functionality level (TFL) provides the aggregated score of the five
functionality dimensions used in this study (cabinet survival, cabinet member/party survival, decision, substantive, consensus).
213
0
20
40
60
80
100
Before
SR2 SR2 -
SR3 SR3 -
SR4 SR4 -
SR5 After
SR5
Cabinet member/party survival
ETC non
0
20
40
60
80
100
Before
SR2 SR2 -
SR3 SR3 -
SR4 SR4 -
SR5 After
SR5
Cabinet survival
ETC non
0
20
40
60
80
100
Before
SR2 SR2 -
SR3 SR3 -
SR4 SR4 -
SR5 After
SR5
Decision
ETC non
0
20
40
60
80
100
Before
SR2 SR2 -
SR3 SR3 -
SR4 SR4 -
SR5 After
SR5
Substantive
ETC non
0
20
40
60
80
100
Before
SR2 SR2 -
SR3 SR3 -
SR4 SR4 -
SR5 After
SR5
Consensus
ETC non
Figure 25. Outcomes of ethno-territorial (N= 96) vs. other cabinet conflicts (N= 229) in Belgium by state
reform period: total functionality level (0-500) and percentage of outcomes meeting different
functionality dimensions (1979-2006)
0
100
200
300
400
500
Before
SR2 SR2 -
SR3 SR3 -
SR4 SR4 -
SR5 After
SR5
Total functionality level
ETC non
Note: As discussed in Chapter 6 of this work, the total functionality level (TFL) provides the aggregated score of the five
functionality dimensions used in this work (cabinet survival, cabinet member/party survival, decision, substantive, consensus).
214
Second and related, ethno-territorial clashes are certainly not always harder to settle than
other conflicts and in this respect, there is no growing disparity between ethno-territorial
and non-ethno-territorial. Rather, there appears to be a trend of leapfrogging on most
dimensions of functionality, with ethno-territorial conflict resolution being more functional in
one period but less so in the next, or vice versa. This is somewhat surprising. After all, ethno-
territorial clashes have an infamous reputation for being particularly hard to settle. And as
discussed in Chapter 5, this reputation is not unjustified (at least from a theoretical perspective).
The fact that this does not translate into the results presented here is quite puzzling and calls for
further research.
Third, on all indicators under study, the period between the third state reform and the
fourth state reform (1989-1993) was a low point in terms of the functionality of ethno-
territorial conflict resolution. As former Prime Minister Wilfried Martens himself recalls in his
memoirs, his last years as a Prime Minister (1988-1992) are usually evaluated as years of
lethargy, a lack of vision, and an unusually high number of failures (Martens, 2006, p. 465).
The results appear to validate this perception. At first sight, the troublesome nature of this
period might seem to be surprising. After all, and as discussed in Chapter 3, the third state
reform came with major compromises on Brussels, the municipalities with facilities
(pacification law), and the regionalization of the national economic sectors. As such, it provided
a long-sought solution for one of the most thorny issues that burdened political negotiations and
triggered numerous clashes for decades. However, the Martens VIII cabinet had not yet found
an agreement on the third phase of the state reform (Witte, 2009d, p. 385), which touched upon
contentious issues such as the shifting of the residual competences towards the sub-states, the
reform of parliament, the establishment of directly elected sub-Flemish and Walloon
parliaments, and the provision of treaty-making power for the sub-states. Most ethno-territorial
clashes within the troublesome period between state reform three and four were about this third
phase, which was highly important for the Flemish regionalist coalition party (Volksunie). Two
other clashes revolved around weapon export, which was strongly contested by Flemish parties
but considered key for the Walloon economic development by many Francophone politicians
(most notably the social-democrat Parti Socialiste). As Chapter 8 exposed, the troublesome
nature of this period does not only reflect in conflict resolution dynamics. It also manifested
itself in the sheer number of ethno-territorial conflicts (both in absolute and relative terms).
Fourth, this most dysfunctional period was followed by the most functional one (the
period between the fourth and the fifth state reform). On four out of five dimensions, the
1993-2001 years can be considered the golden era of functionality – at least when it comes to
ethno-territorial conflict resolution. For at least two reasons, this should not come as a surprise.
To begin with, this period largely overlaps with the premiership of Jean-Luc Dehaene (1992-
215
1999), a Flemish Christian democrat who enjoyed respect in both segments and who is widely
remembered for his ability to settle conflicts with creative and often complex solutions (hence
his nickname, ‘the plumber’). Additionally, the fourth state reform (1993) settled many of the
tense issues discussed above (see Chapter 3). As such, it is often considered the most
fundamental state reform of all. Amongst others, it established directly elected Flemish and
Walloon parliaments, constitutionally recognized Belgium as a ‘federal state’ (Art. 1), provided
sub-states with additional competences, and included a major reform of parliament (Alen, 1995;
Witte, 2009c, pp. 386-387). Although the data do not show any general trend of growing
functionality against the background of decentralization, the remarkable shift in functionality
levels before and after the fourth state reform does appear to illustrate how targeted forms of
decentralization might facilitate consociational power-sharing.
3. Intermediate conclusions
Decentralization is sometimes blamed for burdening the process of conflict
resolution at the central policy level (H4). Accordingly, the pacification noted in Chapters 8
and 9 might have come at the expense of the functionality of the federal level. Putting this
concern to the test, this chapter uses the yearbook data (1979-2006) to expose how conflicts in
power-sharing cabinets are solved and how this evolved against the background of consecutive
state reforms.
What does conflict-solving look like? Most ethno-territorial clashes either blow over, result in
compromises, or lead to the kind of procedural decisions that fail to tackle the problem
substantively. Almost one in five of them ends with some kind of overt hostility (impositions
and resignations). Next to mapping the raw reality of conflict resolution, I also discerned
different dimensions of functionality – ranging from the survival of the cabinet to the
presence of consensus decisions. Different indicators yield different conclusions. Nearly all
ethno-territorial clashes are settled without resignations, but only half of them actually result in
a decision and less than four in ten conflicts are settled substantively (with a decision on the
nub of the problem). If the functionality of power-sharing institutions refers to the absence of
“regular legislative logjams or decision-making vacuums” (McCulloch, 2021, p. 5; see also
McCulloch, 2022), then Belgian cabinets are surely dysfunctional. Adding to the rather grim
picture of politicians’ conflict-solving behaviour, only a small minority (41%) of ethno-
territorial conflicts actually end in some kind of consensus (a procedural decision or
compromise). Accordingly, at least on some indicators, the situation does resemble a three-
legged race. Tying the Flemish and Francophone segments together and compelling them
to cooperate prevents the minority from being regularly ignored, but once tensions arise,
it often makes it difficult to get things done and overcome immobilism. Surely, however,
216
there is no need for over-exaggeration. Belgian cabinets are not the kind of zombie power-
sharing executives described by Nagle (2020), i.e. creatures that are “resistant to practically any
reform”. They still managed to make (substantive) decisions – even in conflict situations and
in a very unfavourable context (that is presumably) full of gridlock incentives and mechanisms.
Although power-sharing cabinets in Belgium can hardly be called alive and kicking, they aren’t
‘walking dead’ either.
Interestingly, conflict resolution did not become increasingly hard over time and against the
background of consecutive decentralization reforms. As such, decentralization does not make
it harder to buy off peace, as is sometimes argued (Caluwaerts & Reuchamps, 2015). However,
it didn’t make it easier either. As such, both H2 and H4 are refuted. Insofar as consociational
power-sharing is prone to immobilism and confrontational behaviour, it seems to be quite
consistently -if not comfortably- numb. This holds for both ethno-territorial conflicts and other
conflicts alike. So, decentralization did not pacify tensions at the expense of the general
functionality of the central policy level, nor did it make it harder to solve the ethno-
territorial tensions that still exist.
Looking back, we see that conflict levels declined (Chapter 8), this decline clearly relates to
decentralization (Chapter 9), and the remaining conflicts are less problematic (Chapter 7) and
not harder to settle (Chapter 10). Having taken the Belgian canary to the power-sharing coal
mine, I resurface with a clear conclusion: the paradox thesis is disproved on all levels.
217
Part IV.
Conclusion
218
219
Chapter 11.
Conclusion: no power-sharing paradox
The functioning and survival of Belgium rely on the power-sharing of its two main
segments (Flemish and Francophone politicians). Often, this collaboration proves to be
particularly hard, resulting in long-lasting gridlocks and traumatizing clashes. Debates about
the governability of Belgium and the feasibility of power-sharing often target the countries’
institutional architecture. More specifically, there is strong disagreement on the impact of
the decentralization reforms that turned the formerly unitary state into a highly
decentralized federation. According to one view, these far-reaching reforms are (largely) to
blame for the existing tensions. The antithesis of this perspective is that, in the search for lasting
stability, decentralization is our best shot. As such, contemporary debates in and on Belgium
mirror the long-lasting academic debate about the effects of decentralization (Amoretti &
Bermeo, 2004; L. Anderson, 2013; Bermeo, 2002; Brancati, 2006, 2009; Erk & Anderson,
2010b; Horowitz, 2000, 2007; Keil & Anderson, 2018; Swenden, 2013b; Watts, 2015; Wolff,
2013). While it is often promoted as a tool for ethno-territorial conflict management, many
scholars consider it to be ineffective at best and counterproductive at worst. This view is echoed
in the so-called paradox of federalism (Erk & Anderson, 2010b): rather than mitigating
conflict, decentralization might exacerbate tensions and undermine stability.
While the debate on the repercussions of decentralization is far from settled, its empirical
basis is not in proportion to its importance. Our knowledge about the ethno-territorial clashes
that burden power-sharing is still very limited and the existing literature is marked by several
lacunas. Most studies focus on the so-called sub-state challenge: does decentralization mitigate
the regionalist or even separatist threat from the periphery? Few studies have focussed on the
effects of decentralization on the power-sharing challenge: does decentralization burden or
facilitate inter-segmental power-sharing? Existing indicators of power-sharing tensions are
rather rough and typically focus on the birth and death of power-sharing cabinets (formation
duration, cabinet duration, etc.). What lacks is a fine-grained view of the day-to-day functioning
of these cabinets, the tensions that mark them, and the way this is impacted by decentralization.
Advancing upon the important work already done, this doctoral dissertation aims to address
these lacunas. To assess whether decentralization ‘fuels the fire or dampens the flames’
(Brancati, 2006), I shed light on four decades of conflicts in power-sharing executives, their
morphology, prevalence, evolution, and some of their determinants. In total, over 1,000
cabinet conflicts were exposed by hand-coding 1,743 pages of yearbooks (Res Publica) and
9,547 news agency articles (Belga). Using these datasets, two questions are tackled: How did
220
the ethno-territorial conflict in Belgian power-sharing cabinets evolve (RQ1)? And how did
decentralization impact this conflict (RQ2)? Studying Belgium allows me to provide a solid
evaluation of the paradox thesis and yield insights that are relevant well beyond its borders.
After all, its mishmash of institutional and historical features makes it a theoretical powder keg;
a least-likely case for pacification (Eckstein, 1975). If decentralization fuels tensions, this
should be highly apparent in Belgium. If not, this provides a serious blow to the paradox thesis.
1. Main results
1.1. Ethno-territorial conflict: a multifaceted burden
Of course, not all politics is ethno-territorial. Most clashes in the four decades under
study are not the result of the ethno-territorial cleavage. But the pertinence of ethno-territorial
conflicts should not be underestimated. They are a major burden to segmental power-sharing
and their explosive reputation is clearly deserved (Chapter 7). Between 17% and 29% of all
studied conflicts in Belgian cabinets (1979-2018) are ethno-territorial in nature. On average,
politicians get involved in such a conflict every two to three-and-a-half months. These numbers
help understand why, throughout the decades, vast amounts of political energy have been
dedicated to pacification attempts.
The detailed dissection of ethno-territorial conflicts provided in this work resulted in some
unexpected findings (Chapter 7). For one, tensions on ethno-territorial issues are far more
common than conflicts between segmental sides; i.e. conflicts in which all Flemings (or
Francophones) actually act ‘as one’. Indeed, segmental power-sharing as such (the fact that
both sides must govern together) seems to be far less explosive than the many issues and
discussions that are inherent to governing a divided state: language laws and their applications,
state reform debates, etc. Belgian politicians rarely engage in the kind of segmental trench
warfare that puts all Flemish against all Francophone coalition parties (1.5% to 7.3% of all
conflicts). Conflict sides strongly vary over time, and when political battles arise, Flemish and
Francophone coalition partners often stand aside. Most tensions (roughly 60%) are either fought
within segments (e.g. amongst Flemings) or characterized by segmentally mixed sides.
Analysing their day-to-day behaviour shows that the segments are no political monoliths.
However, in many cases (about one in four cabinet conflicts), fights erupt between some
Flemings and some Francophones. As such, Flemings and Francophones often clash but in
doing so, they rarely act ‘as one’ (i.e. as homogeneous blocks).
The results also show how the very nature of ethno-territorial conflict evolved (Chapter 7).
221
Particularly in recent times, ethno-territorial tensions less commonly revolve around linguistic
issues or state reform itself (the usual suspects) and more often concern distributive questions
like the spreading of climate efforts, night-time flights, or the (imbalanced compliance to)
federal quota for doctors and dentists. Apart from this thematic transformation, the ethno-
territorial poltergeist also got a friendlier face. The longitudinal analyses demonstrate that
ethno-territorial conflicts became less intense, are less commonly fought along segmental lines,
and are less commonly full-fledged conflicts (clashes in which segmental blocks clash on ethno-
territorial issues). Of course, these findings come with certain nuances, but the general picture
is solid.
These results clearly show that Belgian politics is not just the sum of two irreconcilable
segments, as is sometimes noted. Moreover, insofar as it exists, segmental trench warfare is in
decline. Arguably, the idea that the opposite is true might be fueled by the fact that roughly a
quarter of clashes are fought between some (not all) of the Flemish and some (not all) of the
Francophone coalition parties (29%-23% of all conflicts). As such, about four in ten cabinet
conflicts are fought between Flemish and Francophone actors. Adding to this, parties often
claim to speak on behalf of their segment, while the media has a recorded tendency to use
segmental frames and depict statements of some politicians as the views of their segments at
large (Sinardet, 2008, 2012).
In a way, these conclusions provide new input for the debate on the transformation of
Belgium into a federation with four or even ten sub-states (based on the provinces and
Brussels) (e.g. De Grauwe & Van Parijs, 2021).
143
These pleas are often rooted in the idea that
the bipolarity of the current system nourishes tensions. For several reasons, the effectiveness of
such projects is questionable – especially when the aim is to pacify tensions at the federal level.
In this respect, a distinction can be made between behavioural bipolarity and the perception of
bipolarity. At the federal level, the former is already quite rare. The results presented here show
that bipolar clashes whereby Flemish and/or Francophone parties act ‘as one’ are rather scarce
(and increasingly so). Bipolarity appears to be far stronger in perception. But attempts to breach
this perception by restructuring the number of sub-states will presumably be futile if they are
not combined with a reform of the federal level. As long as the bipolar structure at the federal
143
Some scholars have presented a similar plea for a four-unit Belgian federation based on the concern that
the current situation in which competences are scattered across three communities and three regions leaves
little room for coherent policy-making (e.g. Van Parijs, 2018).
222
policy level remains intact (e.g. the parity in the Council of Ministers, the two language groups
in parliament and their veto rights, and most importantly the accumulation of two party systems
at the federal level), the perception of two irreconcilable blocks is likely to persist.
1.2. Pacification over time
How did the ethno-territorial conflict in Belgian power-sharing cabinets evolve (RQ1)?
I studied the frequencies of ethno-territorial conflicts and different sub-types (e.g. intense
conflicts, full-fledged ethno-territorial conflicts) from different angles: comparisons of cabinets
and yearly absolute and relative frequencies (Chapter 8). The results are as clear as they are
surprising. Decentralization did not come with growing levels of ethno-territorial conflict.
Strong fluctuations are noted over time and in line with the idea that politicians let tensions rise
before settling them with a (non-)agreement (Deschouwer, 2002, p. 79; 2006, p. 904), intense
conflicts even follow a strikingly clear cyclical pattern. More generally, insofar as any long-
term tendency can be discerned, it is that of a tentative decline of conflict frequencies. Since
the 2011 agreement on the sixth state reform, this pacification even resulted in an unprecedented
period of lasting stability. Meanwhile, the ethno-territorial conflict gradually became a side-
product of confrontational politics, whereas it used to be the predominant source of distress
itself for decades (until the mid-1990s, ethno-territorial clashes sometimes accounted for (over)
half of the yearly conflicts in power-sharing executives). As decentralization cannot be blamed
for an escalation that did not occur, this in itself debunks the paradox thesis.
1.3. Decentralization relates to pacification
Is the pacification I observed statistically related to the five state reforms under study?
Was decentralization a factor of relevance or just an institutional bystander? To assess this, I
ran a variety of multivariate Poisson and logistic regression analyses (Chapter 9). These tested
whether the degree of decentralization significantly relates to the frequency of ethno-territorial
conflict and to the chance that cabinet conflicts are ethno-territorial in nature. The latter
indicates whether the factors under study can be considered determinants of ethno-territorial
conflict per se. The analyses also control for several other factors of (presumed) relevance.
The results clearly confirm the relationship between decentralization and pacification. It
not only relates to decreasing frequencies of ethno-territorial conflicts (less monthly clashes).
The decentralization degree is also linked to lower odds that conflicts are ethno-territorial in
nature, decreasing odds that ethno-territorial issues are at stake when politicians fight, and
decreasing odds that coalition partners engage in the kind of segmental trench warfare in which
the Flemish and/or Francophone parties act ‘as one’. All in all, the empirical link between sub-
223
state autonomy and federal stability is compelling. Institutionalizing divisions does not seem to
deepen them, nor does it prevent politicians from transcending them. Similarly, it doesn’t result
in an inflation of ethno-territorial debates. The analyses presented throughout this work show
that the opposite is true. If anything, decentralization seems to provide a more fruitful context
for accommodative power-sharing.
Of course, it is not a panacea (e.g. Bermeo, 2002; Keil & Anderson, 2018; Kymlicka, 2001a;
Seymour & Gagnon, 2012). Many other factors are at play, some of which were included
in the analyses. Specifically, I accounted for the (potential) role of (1) the electoral saliency of
decentralization, (2) the proximity of federal elections, (3) the electoral strength of regionalist
parties, and (4) their cabinet participation. The results indicate that while the institutional game
board is important, the actors at play within a system matter too. With one exception (regionalist
party strength), all variables positively relate to ethno-territorial conflict. More specifically,
they are associated with higher chances that Flemings and/or Francophones act ‘as a block’
when conflicts arise (conflict along segmental lines). Regionalist parties’ cabinet participation
is also positively related to the frequency of ethno-territorial conflicts. This should not come as
a surprise. These parties are known for stressing ethnic divisions, pushing territorial issues to
the top of the political agenda, and triggering dynamics of ethnic outbidding (Horowitz, 2000;
Meguid, 2005; Müller-Rommel, 1998; Rabushka & Shepsle, 1972). But interestingly, it is thus
not the electoral strength of these parties that seems to fuel tensions, but the choice to let them
govern. The participation of regionalists in power-sharing executives triples the odds that
cabinet conflicts are fought along linguistic lines and is associated with an expected 70%-71%
increase in the monthly number of ethno-territorial conflicts. Contrarily, conflict levels do not
increase when decentralization is more salient in the build-up towards elections, when
regionalist parties have a greater share of seats in parliament, or when federal elections come
nearer. These results add major nuances to the idea that ethno-territorial conflict is a (mere)
derivate of electoral incentives and concerns. The electoral pressures under study do not
appear to be an important driving force behind (the quantity of) conflicts.
1.4. Decentralization did not burden conflict resolution
Decentralization is sometimes blamed for burdening the process of conflict
resolution at the central policy level. According to Caluwaerts and Reuchamps (2015, p. 277),
to give just one example, decentralization “undermines the problem-solving capacity of the
other power-sharing mechanisms”. Reflecting on Belgium’s recent history, they argue that
“granting segmental autonomy effectively accommodates political conflicts that are currently
on the agenda, but renders the process of intersegmental conflict accommodation increasingly
more difficult in the long run.” As such, the abovementioned pacification at the federal level
224
might have come at the expense of its functionality. The yearbook data (1979-2006) allowed
me to put this concern to the test. Specifically, I exposed how conflicts in power-sharing
cabinets are solved and how this evolved against the background of decentralization (Chapter
10).
To begin with, I presented a (perhaps) surprising picture of (ethno-territorial) conflict
resolution in Belgian cabinets. While 40% of all ethno-territorial conflicts blow over without
any palpable result whatsoever, almost one in five clashes are settled with the kind of procedural
decisions that fail to tackle the problem substantively: the establishment of a committee, the
decision to continue the debate at a later stage, the choice to gain legal advice or (time-
consuming) input from stakeholders, etc. In governing Belgium, kicking the can is an often-
used technique. Such procedural decisions are as frequent as clashes that end in overt hostility
(with an imposition or resignation), and clearly, accommodative conflict-solving is not the
standard: only one in four clashes results in a compromise; i.e. a substantive agreement on the
nub of the problem.
Apart from mapping the raw outcomes of conflicts, I also discerned different dimensions
of conflict-solving functionality – ranging from the survival of the cabinet to the presence of
consensus decisions. Of course, functionality means different things to different people. It is
partly in the eye of the beholder and different indicators yield different conclusions. Political
elites manage to settle nearly all ethno-territorial conflicts without resignations, but only half
of their fights actually result in a decision and even fewer conflicts are settled with a substantive
decision (one on the nub of the problem). Only a small minority of four in ten ethno-territorial
conflicts actually ends in consensus. All in all, the picture is thus rather grim. It would be an
exaggeration to depict Belgian cabinets as the kind of zombie power-sharing executives
described by Nagle (2020, p. 137), i.e. spiritless creatures that are “resistant to practically any
reform”. They are not walking dead, but they do stumble through life. Indeed, if the
functionality of power-sharing institutions refers to the absence of “regular legislative logjams
or decision-making vacuums” (McCulloch, 2021, p. 5; see also McCulloch, 2022), then Belgian
cabinets are surely dysfunctional. At least in terms of conflict resolution, the situation seems
to resemble a three-legged race. Tying the Flemish and Francophone segments together and
obliging them to cooperate prevents the minority from being regularly surpassed. But when
tensions between partners arise, it is often difficult to go forward, get things done, and overcome
immobilism.
Next, the analyses of trends over time show that conflict resolution did not become increasingly
hard. From a distance, consociational power-sharing seems to be rather consistently -if not
comfortably- numb. Dysfunctionality was not the price we paid for pacification.
Decentralization did not facilitate conflict resolution, but it also didn’t make it harder. As such,
225
hollowing the centre (L. Hooghe, 2004) did not pacify tensions at the expense of the
conflict-solving functionality of the federal government. When it comes to conflict and
conflict resolution, there is no zero-sum game between stability and functionality.
1.5. The paradox of pacification
The conclusions presented here disprove the paradox thesis on all levels. There are
fewer ethno-territorial clashes than in the past, the ones that still arise have a friendlier face, the
pacification trends are statistically related to decentralization, and hollowing out the central
level did not burden the federal conflict resolution process in the long run. From a theoretical
perspective, this is extremely puzzling. Belgium is home to a cocktail of risk factors, which
makes it a theoretical powder keg; a least-likely case for pacification (Eckstein, 1975). It is
widely seen as a disintegrative model, burdened by “dynamics that remove the basis for
consociational governance” and “mechanisms that gradually destroy collaborative and cohesive
capacity” (Popelier, 2021, p. 110). Against this background, the real paradox of Belgian
federalism is not its failure but its apparent historical success; the fact that it appears to have
pacified tensions in a context that is bound for escalation. What accounts for this remarkable
gap between the theory and the results? I present four tentative reflections.
First, and most evidently given the scope of this work, attention should be drawn to the many
ways in which decentralization can de-escalate tensions. As discussed elaboratively in
Chapter 5, it can (amongst others) satisfy autonomy demands, reduce minorities’ fear of being
marginalized or overruled (e.g. by the Flemish majority), allow segments to voice concerns and
grievances through legitimate channels (e.g. sub-state parliaments), provide an important
symbolical recognition of their existence and relevance, (as such) strengthen the loyalty of sub-
state groups towards the state, reduce the number of issues that needs to be addressed, lift the
very need for power-sharing between segments with different or even diverging policy views,
increase the visibility of intra-segmental divisions, and allow for the kind of proxy warfare in
which central coalition partners attack one another via sub-state actors and institutions (Keil &
Anderson, 2018). There are thus plenty of reasons why the pacifying role of decentralization
should not really surprise. But it is not the only element that can help us make sense of the
conflict declines I observed.
Second, and contradicting many conventional analyses, the findings indicate that bipolar
federalism does not preclude the presence of ‘tension-reducing coalitions’, i.e. alliances
that shift across negotiations, as such avoiding the animosity between two permanent
adversaries from building up (Elazar, 1987, p. 244). In Belgium, conflicts sides frequently vary,
segments rarely act as a front and often clash amongst one another. Opponents in one clash are
226
allies in other conflicts. There is no permanent enemy. Bipolar federalism does not imply that
all politics is bipolar, nor that segments usually act as homogeneous entities. With an effective
number of parties between 6.8 and 9.1, and notoriously large coalitions of 4–6 parties (1978-
2018; Döring & Manow, 2020), the Belgian case nicely illustrates that bipolar federations can
leave plenty of room for shifting coalitions across negotiations. Of course, the stabilizing impact
of such dynamics is at least partly impeded by the burden of perceptions. Both in public and
political debates, the perception of two irreconcilable blocks is quite firmly established.
Third, the results clearly prove that Belgian politicians do have incentives to be prudent.
Incentives “are the key to accommodation in difficult conditions” (Horowitz, 2002, p. 24). But
the troublesome combination of a list-PR system and parties that are split along segmental lines
is often thought to provide no pre-electoral incentives for cooperative behaviour (Horowitz,
1991, 2002). Rather, they presumably face strong pressures to engage in the kind of
confrontational competition that can deteriorate into ‘ethnic outbidding’ dynamics: an inflation
of claims and a radicalization of stances (Rabushka & Shepsle, 1972). This work demonstrates
that even in a highly unfavourable context, this need not be the case. Parties are not doomed to
confrontational strategies. Many incentives for accommodative behaviour exist – both for vote-
, office-, and policy-seeking parties (Müller & Strøm, 1999). First, from a vote-seeking
perspective, there are solid reasons to be accommodative. Similar to the way mainstream parties
can react to the emergence of niche parties by taking a contrasting position (an adversarial
strategy; Meguid, 2005, 2008), politicians can react to confrontational parties (or parties that
are depicted as such) by presenting themselves as responsible, cooperative, and constructive
actors. In this respect, the pleas in favour of recentralization or a federal electoral district in
Belgium (as well as the tonality of those pleas) are indicative of such attempts to highlight
differences with so-called trouble-seekers (most notably the Flemish nationalist parties).
Second and third, from a policy-seeking and office-seeking perspective too, there are incentives
for accommodative behaviour. In consociational Belgium, both goals (cabinet participation and
policy impact) require segmental cooperation. As such, the ambition of parties to impact policy
or join cabinets can act as an important counterbalance to the potential urge to be too harsh
towards ‘the other side’. Rationales like these help explain why prominent politicians have
explicitly denounced ethno-territorial confrontation. One example is the very recent charge of
Prime Minister De Croo vis-à-vis the Flemish nationalists. During the party conference of the
Flemish liberals (Open VLD) in the build-up towards the 2024 general elections, he stated:
‘We must dare to challenge the logic of division of the nationalists and separatists.
The wolves in sheep's clothing who have only one agenda: to shut down our country
and split it. […] A divided country can never be a strong country. I will state things
even clearer: a politician who does not believe in our country, will not make it
227
better. Behind the call for a new major state reform, there is only one strategy and
only one goal: to once again shut down our country for 500 days or more. It is the
strategy of putrefaction and immobilism. Of political gridlock and navel-gazing.
This while we need the exact opposite.’ (De Croo, 2023)
But perhaps, and as this example highlights, the most fundamental incentive of all is the fear of
what might happen when confrontational behaviour prevails. The (potential) repercussions of
conflict can act as a strong deterrent. As such, the puzzling pacification I found might also be
the result of the first self-denying prophecy described by Arend Lijphart (1968c; Ch. 12;
1969b, p. 246).
144
In his early work on consociational democracy and the lack of cross-cutting
cleavages, Lijphart already vividly described how the presence of great risks of instability and
tensions can be an important factor of stability. Substantial theoretical risks can be recognized
in advance and if political leaders are well aware of the lurking dangers, their behaviour will be
shaped accordingly. Confrontational demarches will be avoided and the search for conflict-
defusing procedures and reforms will gain track. The greater the danger, the greater the pressure
to avoid it. As such, the doom scenarios of societal violence, disintegration, or regime-
threatening gridlocks might never materialize. At least in part, a similar logic might have been
at play in Belgium. Most Belgian politicians are susceptible to such incentives, as the majority
of them are loyal to the Belgian state and its survival (e.g. Dodeigne et al., 2016; Sinardet et al.,
2013). Having historical and more recent experiences in mind, and being informed about
escalations abroad, they are well aware of the flammable nature of segmental divisions and the
potential repercussions of confrontational attitudes.
145
Decades ago, scholars already pointed
towards the fear of long-lasting gridlocks and their economic repercussions when making sense
of the lower levels of cabinet turnover since the 1980s (Dumont & De Winter, 1999, pp. 48-
49). Similar pressures might account for (part of the) pacification noted here.
These pressures still exist, because the federal policy level is still ‘too big to fail’. In this
respect, I concur with the idea that decentralization emptied the federal level to such an extent
that the cost of non-agreement was reduced to problematic levels (Caluwaerts & Reuchamps,
144
The second self-denying prophecy refers to the deliberate transition from a cartel democracy to a
centripetal democracy (Lijphart, 1969b, p. 246).
145
In several instances, ethno-territorial protests in Belgium risked turning ugly too. Violence was not always
avoided (Delcorps, 2012; L. Hooghe, 2004).
228
2015). It is often claimed that, given that the sub-states can keep the country running, the failure
to reach agreements at the federal level might no longer result in the kind of generalized policy
paralyses that often pushes politicians towards compromise (see also L. Hooghe, 2004). Surely,
decentralization decreased the relevance and significance of the federal policy level. But it
hasn’t ‘emptied’ this level to such an extent that federal gridlocks are now irrelevant or
perceived as such. The costs of non-agreement are lowered, but still significant. After all, and
even after the sixth state reform, the federal government still accounts for a significant share of
the total public revenue and expenditure (Decoster & Sas, 2013). Also, key competences still
pertain to the federal level, including the bulk of social security, defence, justice, key taxes (e.g.
personal income tax, corporate taxation), crucial aspects of labour market policy (e.g. labour
law, wage policy), and residual competences (art. 35 of the Constitution). Due to the latter, new
challenges like cyber security, drones, artificial intelligence, or medical-ethical issues almost
automatically fall under the authority of the federal government, the result of which is ‘an
invisible state reform’; a silent strengthening of the central policy level (see De Standaard,
2019; Reybrouck & Sottiaux, 2019). Finally, politicians still adhere to the traditional logic of
package deals. As different dossiers are usually linked in the search for a general agreement,
even smaller clashes can lead to large-scale immobilism. In sum, and providing strong
incentives for accommodative behaviour, the risk of a generalized policy paralysis still hangs
over Belgian politicians like a sword of Damocles.
1.6. The paradox of centralism
In light of the results, it might be tempting to consider ethno-territorial tensions to be a
thing of the past; a beast that has been tamed. As such, some might even question the need to
keep the complex range of tension-reducing mechanisms in place. After all, many of these
tools -including the decentralization reforms- have added to the complexity of Belgium. Using
the words of Deschouwer (2013a), they turned it into a labyrinth state. Of course, the search for
transparency, accountability, and efficiency is important. But stability should not be taken for
granted. Caution is advised. Thoughtlessly reversing the decentralization process because there
is some federal stability is like cutting loose (some cords of) your parachute because you’re no
longer in free fall. It risks reviving the very tensions that were gradually defused during
decades of hard-fought piecemeal engineering. In recent years, political support for the
recentralization of certain sub-state competences has been growing (Dodeigne et al., 2016;
Dodeigne et al., 2022; Dodeigne et al., 2015; Pascolo et al., 2021; Sinardet et al., 2013).
Amongst others, such reforms are often presented as pacification attempts; as ways to facilitate
decision-making on policy domains and issues that give rise to frequent tensions between sub-
states. But especially when troublesome issues are targeted, such reforms come with a risk of
federal re-escalation. Of course, the results presented here are far too nuanced to expect a one-
229
on-one relationship. Not every policy transfer to the federal level would act like a missile
exploding upon impact. But at least in some cases, an ill-considered recentralization of
conflict-prone domains might well result in the recentralization of the tensions they
trigger.
In this respect, an important question is which conflicts are most troublesome: inter-
governmental clashes (on decentralized policies) or disputes within the federal government (on
central policies)? For at least one reason, central-level clashes are more disruptive. After all,
and in contrast to disputes at the federal level, conflicts between sub-states can only give rise
to single policy paralyses. As noted by Deschouwer (2006, p. 908):
“The crucial difference is, however, the default option when intergovernmental
negotiations fail: a single policy paralysis. Only the topic on which no agreement
can be reached remains unsolved. Regional and federal governments can go on
with other matters belonging to their area of competence.”
In contrast, at the federal level, where all debates are hog-tied, there is a solid risk of a
generalized policy paralysis; a deadlock of the cabinet as a whole that poses a threat to the
country’s stability as a whole. As such, transferring the hot potato to the federal government
might result in tensions that are far more severe.
2. Limitations and future research
Of course, this work comes with several limitations. Just like decentralization, it is only
one piece of an extensive puzzle. The following paragraphs cover several shortcomings and the
room they leave for future analyses.
To begin with, this analysis of Belgium calls for a comparative successor. The approach
presented here could be applied to assess conflict dynamics in other divided polities. Doing so
would shed new light on an old debate. Is power-sharing in Belgium actually more explosive
than in other divided states, as its cocktail of risk factors indicates? Why are some states more
prone to everyday clashes (or their pacification) than others? Is conflict resolution harder in
consociational systems? Are power-sharing dynamics primarily a derivate of institutional
factors? Or a case-specific result of historical dynamics and path dependency? Does the nature
of decentralization matter? For example, the strength of secessionism has been shown to depend
upon the static or dynamic nature of autonomy (whether it is introduced as an unchangeable
settlement or rather expands throughout consecutive reform packages) (Lecours, 2021). Does
the same hold for ethno-territorial conflict levels? And what other or moderating factors are at
230
play?
Second, this is not an encompassing evaluation of federalism or “the” paradox of
federalism. All scholars studying the effects of decentralization or federalism more generally
face the same problem: “it is impossible to establish a firm set of criteria by which to judge
success and failure” (Burgess, 2006, p. 269). As a matter of fact, and as I stressed from the very
beginning of this work, there is no such thing as “the” paradox of federalism. Numerous
indicators exist, providing different results and yielding varying conclusions. Attention can be
devoted to violent or non-violent conflicts, power-sharing dynamics or the sub-state challenge,
etc. This work focuses on the underexplored impact of decentralization on power-sharing
dynamics and more specifically, on conflicts in power-sharing executives. It provided a brief
overview of other indicators (Chapter 4), showing amongst others that the formation of power-
sharing executives became increasingly hard, while their lifespan increased (e.g. Dandoy &
Terrière, 2021; De Winter & Dumont, 2014, 2021; Dumont & De Winter, 1999; Dumont et al.,
2001). The results presented in this work add to this picture by demonstrating that the period
between their birth and death became more stable. Arguably, this was one of the most important
ambitions of the decentralization process. As such, I chose to evaluate its successfulness by
checking whether the initial goals were met.
Similarly, the focus of this work is on dynamics at the federal level. Potentially, as sub-state
responsibilities increased, part of the tensions that used to burden the federal government turned
into intergovernmental clashes between different sub-states or policy levels. Whether this
potential transfer of tensions from the centre to the sub-states actually took place is yet to be
assessed. Doing so could add significantly to our evaluation of the paradox thesis. However, as
stated above, such a decentralization of conflicts would arguably have a net-positive effect in
terms of stability, as intergovernmental tensions are arguably less disruptive. In contrast to
federal tensions, they only involve single policy paralyses.
More generally, even the focus on conflict resolution is only one of many potential
perspectives and of course, conflict-solving is not always problem-solving. Political conflicts
are often settled without actually providing a lasting solution for the underlying problem.
Sometimes, conflict settlements even add to it. This is painfully illustrated by the well-known
technique to ‘buy off peace’ (Caluwaerts & Reuchamps, 2015, 2020; L. Hooghe, 2004), which
added to the problematic sustainability of Belgium’s public finances (Cornille et al., 2022) (to
give just one example). I studied pacification, not its cost.
Other limitations of this work relate to the data and analyses I present. To begin with, the
need to combine two datasets, one based on yearbooks (1979-2006) and one based on digital
news archives (1995-2018) is evidently sub-optimal. Likewise, the sources used here fail to
231
provide sight on certain periods. Most notably, I do not shed light on the unitary era preceding
the 1970 state reform. My analyses start in 1979. While the period covered suffices to study the
paradox thesis, it would still be interesting to see what segmental power-sharing looked like
before the introduction of any decentralization and in the absence of consociational devices.
Related, conflict outcomes are only mapped in one dataset, thus leaving us blind to their
evolution after 2006. Additional data-collection efforts would be (very) labour-intensive but
particularly valuable. Likewise, extending the explanatory analyses to cover the post-2006
period is recommended. That being said, it seems unlikely that such efforts would challenge
the conclusions presented here. After all, my analyses show that the most troublesome periods
in terms of ethno-territorial politics occurred in the past. In recent years, and after the sixth state
reform more specifically, tensions were at an all-time low. As such, the chances that an analysis
of post-2006 data would yield fundamentally different results (a confirmation of the paradox)
seem rather slim.
Talking about conflict endings, my focus was (only) on the way politicians defused the
bombs; the way they turned a conflictual situation into a pacified one. Of course, conflicts
might remain unresolved in the short run, but there might be solutions in the long run. An
imposition on one moment might be compensated by a broader compromise negotiated at a
later time. Issues like the status of Brussels, to give just one major example, have triggered
numerous clashes. But in the end (the late 1980s), consensus was found and the Brussels-Capital
region was established. Such a long-term perspective was beyond the scope of this work, but in
evaluating the relationship between decentralization on conflict resolution dynamics,
assessments like these would be particularly valuable.
Additionally, the results of the explanatory analyses are generally quite weak. Decentralization
is not a stairway to heaven. It matters, but clearly, its role should not be overestimated.
Several elements might account for this. First, there is the distorting role of path dependency
and lagged effects (dynamics that burden all studies of conflict and cooperation). Any effects
of decentralization might take some time to materialize, especially because decentralization
deals are typically products of conflictual periods of negotiations and clashes (at least on state
reform itself, as my analyses show). These tensions might continue to burden the atmosphere
for some time. When the battle is over, the wounds still need to heal. Gaining insight into such
dynamics of lasting distrust, for instance through elite interviews, would be highly interesting.
Second, there is an enormous plethora of potential factors at play. Some of these were covered
in my analyses, but much remains to be explored. In this respect, future studies can draw
inspiration from the broader literature on the determinants of cabinet stability, which addresses
the role of e.g. coalition agreements’ length and content (Klüver, Bäck, & Krauss, 2023; Krauss,
2018; Moury & Timmermans, 2013) or the degree of ideological differentiation between
232
coalition parties (Jones, 2001; Laver, 1974; Müller et al., 2008, pp. 21-23; Sartori, 1976;
Warwick, 1979, 1992). Third, each state reform in Belgium is “a contradictory combination of
strong autonomist and strong unitarian tendencies” (L. Hooghe, 1993, p. 55). Negotiations on
these reforms include numerous parties with varying wishes, goals, and incentives.
Unsurprisingly, each resulting compromise is a political patchwork. As former Prime Minister
Gaston Eyskens, a Flemish Christian democrat (CVP), often stated: ‘A dromedary is a horse
that walked through the cabinet council.’ (G. Eyskens, 1993, p. 697) We cannot blame
dromedaries for failing to perform like horses.
233
234
235
References
Aalberg, T., Strömbäck, J., & de Vreese, C. H. (2012). The framing of politics as strategy and
game: A review of concepts, operationalizations and key findings. Journalism, 13(2),
162-178.
Abendschön, S., & Steinmetz, S. (2014). The gender gap in voting revisited: Women's party
preferences in a European context. Social Politics, 21(2), 315-344.
Abts, K., Dalle Mulle, E., & Laermans, R. (2019). Beyond issue diversification: N-VA and the
communitarisation of political, economic and cultural conflicts in Belgium. West
European Politics, 42(4), 848-872.
Abts, K., Dmitriy, P., & Marc, S. (2012). The federal elections in Belgium, June 2010. Electoral
Studies, 31(2), 448-452.
Adams, J., Clark, M., Ezrow, L., & Glasgow, G. (2006). Are niche parties fundamentally
different from mainstream parties? The causes and the electoral consequences of
Western European parties' policy shifts, 1976–1998. American journal of political
science, 50(3), 513-529.
Alen, A. (1995). Belgium - The Belgian Federal State after the Fourth State Reform. European
Public Law, 1(1), 8-15.
Alesina, A., & Spolaore, E. (1997). On the number and size of nations. The Quarterly Journal
of Economics, 112(4), 1027-1056.
Alesina, A., Spolaore, E., & Wacziarg, R. (2000). Economic integration and political
disintegration. American Economic Review, 90(5), 1276-1296.
Algemene Directie Statistiek - Statistics Belgium. (n.d.-a). Huwelijken. Retrieved from
https://statbel.fgov.be/nl/themas/bevolking/partnerschap/huwelijken#figures
Algemene Directie Statistiek - Statistics Belgium. (n.d.-b). Werkgelegenheid en werkloosheid
- meer cijfers. Retrieved from https://statbel.fgov.be/nl/themas/werk-
opleiding/arbeidsmarkt/werkgelegenheid-en-werkloosheid/plus
Almond, G. A. (1956). Comparative Political Systems. The Journal of politics, 18(3), 391-409.
Alonso, S. (2012). Challenging the State: Devolution and the Battle for Partisan Credibility: A
Comparison of Belgium, Italy, Spain, and the United Kingdom. Oxford: Oxford
University Press.
Amoretti, U. M., & Bermeo, N. (Eds.). (2004). Federalism and Territorial Cleavages. London:
236
The Johns Hopkins University Press.
Anderson, B. (1983). Imagined Communities: Reflections on the Origin and Spread of
Nationalism. London: Verso.
Anderson, L. (2013). Federal Solutions to Ethnic Problems: Accommodating Diversity. New
York: Routledge.
Anderson, L. (2016). Ethnofederalism and the management of ethnic conflict: Assessing the
alternatives. Publius: The journal of Federalism, 46(1), 1-24.
Andeweg, R. B. (2000). Consociational Democracy. Annual Review of Political Science, 3,
509-536.
Andeweg, R. B., & Timmermans, A. (2008). Conflict Management in Coalition Government.
In K. Strøm, W. C. Müller, & T. Bergman (Eds.), Cabinets and Coalition Bargaining:
The Democratic Life Cycle in Western Europe (pp. 269-300). New York: Oxford
University Press.
Artés, J. (2013). Do Spanish politicians keep their promises? Party Politics, 19(1), 143-158.
Asbrock, F., & Van Hiel, A. (2017). An Insiders' Outside Perspective on the Flemish-Walloon
Conflict: The Role of Identification and Disidentification for the German-Speaking
Minority. Psychologica Belgica, 57(3), 115-131. doi:10.5334/pb.347
Bachrach, P., & Baratz, M. S. (1963). Decisions and nondecisions: An analytical framework.
American political science review, 57(3), 632-642.
Bahtić-Kunrath, B. (2011). Of veto players and entity-voting: institutional gridlock in the
Bosnian reform process. Nationalities Papers, 39(6), 899-923.
Bakke, K. M., & Wibbels, E. (2006). Diversity, disparity, and civil conflict in federal states.
World Politics, 59(1), 1-50.
Bakker, R., De Vries, C., Edwards, E., Hooghe, L., Jolly, S., Marks, G., . . . Vachudova, M. A.
(2015). Measuring party positions in Europe: The Chapel Hill expert survey trend file,
1999–2010. Party Politics, 21(1), 143-152.
Bakvis, H. (1987). Alternative Models of Governance: Federalism, Consociationalism, and
Corporatism. In H. Bakvis & W. M. Chandler (Eds.), Federalism and the Role of the
State (pp. 279-305). Toronto: University of Toronto Press.
Bara, J. (2005). A question of trust: Implementing party manifestos. Parliamentary Affairs,
58(3), 585-599.
237
Barry, B. (1975a). The Consociational Model and Its Dangers. European Journal of Political
Research, 3(4), 393-412.
Barry, B. (1975b). Political Accommodation and Consociational Democracy. British journal of
political science, 5(4), 477-505.
Baudewyns, P., Brans, M., Reuchamps, M., Rihoux, B., & Van Ingelgom, V. (Eds.). (2022).
The winter of democracy: Partitocracy in Belgium. Louvain-la-Neuve: Presses
universitaires de Louvain.
Baudewyns, P., Dandoy, R., & Reuchamps, M. (2015). The success of the regionalist parties in
the 2014 elections in Belgium. Regional & Federal Studies, 25(1), 91-102.
Baumgartner, F. R., Breunig, C., & Grossman, E. (Eds.). (2019). Comparative Policy Agendas:
Theory, Tools, Data. Oxford: Oxford University Press.
Béland, D., & Lecours, A. (2016). The 2014 Scottish Referendum and the Nationalism-Social
Policy Nexus. Canadian Political Science Review, 10(1), 1-30.
Béland, D., & Lecours, A. (2018). Federalism, policy change, and social security in Belgium:
Explaining the decentralization of family allowances in the Sixth State Reform. Journal
of European Social Policy, 28(1), 55-69.
Belmont, K., Mainwaring, S., & Reynolds, A. (2002). Introduction: Institutional Design,
Conflict Management, and Democracy. In A. Reynolds (Ed.), The architecture of
democracy: Constitutional design, conflict management, and democracy (pp. 1-11).
New York: Oxford University Press.
Benz, A. (2011). Escaping Joint-Decision Traps. In G. Falkner (Ed.), The EU's Decision Traps:
Comparing Policies (pp. 199-216). Oxford: Oxford University Press.
Bergman, T., Gerber, E. R., Kastner, S., & Nyblade, B. (2008). The Empirical Study of Cabinet
Governance. In K. Strøm, W. C. Müller, & T. Bergman (Eds.), Cabinets and Coalition
Bargaining: The Democratic Life Cycle in Western Europe (pp. 85-122). Oxford:
Oxford University Press.
Bergman, T. e., Bäck, H. e., & Hellström, J. e. (Eds.). (2021). Coalition governance in western
Europe. Oxford: Oxford university press.
Bermeo, N. (2002). A New Look at Federalism: The Import of Institutions. Journal of
Democracy, 13(2), 96-110.
Biard, B. (2022). Les partis frères en Belgique : les relations entre le MR et l’Open VLD.
Courrier hebdomadaire du CRISP, 2536-2537(11), 5-92. doi:10.3917/cris.2536.0005
238
Billiet, J. B. (2011). Flanders and Wallonia, Right versus Left: is this Real? In B. De Wever
(Ed.), Right-Wing Flanders, Left-Wing Wallonia? Is This So? If So, Why? And is it a
Problem? (pp. 11-24). E-book: Re-Bel initiative - Rethinking Belgium's institutions in
the European context.
Billiet, J. B., Abts, K., & Swyngedouw, M. (2015). Right-wing Flanders and left-wing
Wallonia? Public opinions on redistribution, the welfare state and immigrants. In B.
Rihoux, V. Van Ingelgom, & S. Defacqz (Eds.), La légitimité de la science politique
(pp. 81–100). Louvain-la-Neuve: Presses universitaires de Louvain.
Billiet, J. B., Maddens, B., & Frognier, A.-P. (2006). Does Belgium (Still) Exist? Differences
in Political Culture between Flemings and Walloons. West European Politics, 29(5),
912-932.
Billig, M. (1995). Banal nationalism. London: Sage.
Billig, M. (2017). Banal Nationalism and the Imagining of Politics. In M. Skey & M. Antonsich
(Eds.), Everyday Nationhood: Theorising Culture, Identity and Belonging after Banal
Nationalism (pp. 307-321). London: Palgrave Macmillan UK.
Bischof, D. (2017). Towards a renewal of the niche party concept: Parties, market shares and
condensed offers. Party Politics, 23(3), 220-235.
Blaise, P., de Coorebyter, V., & Faniel, J. (2007). Les résultats des élections fédérales du 10
juin 2007. Courrier hebdomadaire du CRISP, 19641965(19), 7-102.
Blero, B. (2015). La scission de la circonscription électorale de Bruxelles-Hal-Vilvorde.
Courrier hebdomadaire du CRISP, 2279(34), 5-54. doi:10.3917/cris.2279.0005
Blöchliger, H., & King, D. (2007). Less than you thought: The fiscal autonomy of sub-central
governments. OECD Economic Studies, 2006/2.
doi:https://doi.org/10.1787/eco_studies-v2006-art12-en
Blöchliger, H., & Rabesona, J. (2009). The Fiscal Autonomy of Sub-Central Governments.
OECD Working Papers on Fiscal Federalism, No. 9.
doi:https://doi.org/10.1787/5k97b111wb0t-en
Blommaert, J. (2011). The long language-ideological debate in Belgium. Journal of
Multicultural Discourses 6(3), 241–256.
Blondel, J., & Müller-Rommel, F. (Eds.). (1993). Governing Together: The Extent and Limits
of Joint Decision-Making in Western European Cabinets. London: Palgrave Macmillan.
Bogaards, M. (1998). The favourable factors for consociational democracy: A review.
European Journal of Political Research, 33, 475-496.
239
Bogaards, M. (2000). The uneasy relationship between empirical and normative types in
consociational theory. Journal of Theoretical Politics, 12(4), 395-423.
Bogaards, M. (2006). Democracy and power-sharing in multinational states: Thematic
introduction. International Journal on Multicultural Societies, 8(2), 119-126.
Bolleyer, N. (2012). New party organization in Western Europe: Of party hierarchies,
stratarchies and federations. Party Politics, 18(3), 315-336.
Borghetto, E., & Belchior, A. M. (2020). Party Manifestos, Opposition and Media as
Determinants of the Cabinet Agenda. Political Studies, 68(1), 37-53.
doi:10.1177/0032321718820738
Borghetto, E., & Carammia, M. (2014). Party priorities, government formation and the making
of the executive agenda. In The Challenge of Coalition Government (pp. 54-75):
Routledge.
Bottel, M., Mazzoleni, O., & Pallaver, G. (2016). Are government regionalist parties
mainstream parties? A comparison of the Südtiroler Volkspartei, the Union Valdôtaine,
and the Lega dei Ticinesi. Contemporary Italian Politics, 8(2), 160-179.
Bouteca, N., & Devos, C. (2014). The Process of Rescaling Interests in the Belgian Context:
The Impact of Regional Governmental Strength. Territory, Politics, Governance, 2(3),
287-302.
Bouteca, N., Devos, C., Moens, P., Ossenblok, K., & Terrière, L. (2018). Belgisch federalisme.
Gent: Academia press.
Bouvier, J. (2012). Local government in Belgium. In A.-M. Moreno (Ed.), Local government
in the Member States of the European Union: a comparative legal perspective (pp. 45-
67). Madrid: National Institute of Public Administration (INAP).
Brancati, D. (2006). Decentralization: Fueling the Fire or Dampening the Flames of Ethnic
Conflict and Secessionism? International Organization, 60(3), 651-685.
Brancati, D. (2008). The origins and strengths of regional parties. British journal of political
science, 38(1), 135-159.
Brancati, D. (2009). Peace by Design: Managing intrastate conflict through decentralization.
Oxford: Oxford University Press.
Brans, M., De Winter, L., & Swenden, W. (Eds.). (2009). The Politics of Belgium: Institutions
and policy under bipolar and centrifugal federalism. New York: Routledge.
Brassinne de La Buissière, J. (2005). La régionalisation des lois communale et provinciale et
240
de la législation connexe. Courrier hebdomadaire, 7-69.
Breuilly, J. (1982). Nationalism and the State. Manchester: Manchester University Press.
Breuning, M., & Ishiyama, J. T. (1998). The rhetoric of nationalism: rhetorical strategies of the
Volksunie and Vlaams Blok in Belgium, 1991-1995. Political Communication, 15(1),
5-26.
Brinton, C. (1965). The anatomy of revolution. New York: Vintage Books.
Browne, E. C., Gleiber, D. W., & Mashoba, C. S. (1984). Evaluating conflict of interest theory:
Western European cabinet coalitions, 1945–80. British journal of political science,
14(1), 1-32.
Budge, I., & Farlie, D. (1983). Explaining and predicting elections: Issue effects and party
strategies in twenty-three democracies. London: George Allen & Unwin.
Budge, I., Klingemann, H.-D., Volkens, A., Bara, J., & Tanenbaum, E. (2001). Mapping policy
preferences: estimates for parties, electors, and governments, 1945-1998 (Vol. 1).
Oxford: Oxford University Press.
Budge, I., & Pennings, P. (2007). Missing the message and shooting the messenger: Benoit and
Laver’s ‘response’. Electoral Studies, 26(1), 136–141.
Buelens, J., & Van Dyck, R. (1998). Regionalist parties in French-speaking Belgium: the
Rassemblement Wallon and the Front Démocratique des Francophones. In L. De Winter
& H. Türsan (Eds.), Regionalist Parties in Western Europe (pp. 51-69). London:
Routledge.
Burgess, M. (2006). Comparative Federalism: Theory and practice. New York: Routledge.
Buyst, E. (2011). Continuity and change in regional disparities in Belgium during the twentieth
century. Journal of Historical Geography, 37, 329-337.
Cagiao y Conde, J., & Gagnon, A.-G. (Eds.). (2021). Federalism and Secession. Bruxelles: PIE-
Peter Lang SA Éditions Scientifiques Internationales.
Caluwaerts, D., & Reuchamps, M. (2015). Combining Federalism with Consociationalism: Is
Belgian Consociational Federalism Digging its Own Grave? Ethnopolitics, 14(3), 277-
295.
Caluwaerts, D., & Reuchamps, M. (2020). Still Consociational? Belgian Democracy, 50 Years
After ‘The Politics of Accommodation’. Politics of the Low Countries, 2(1), 28-50.
Carty, R. K. (2004). Parties as Franchise Systems: The Stratarchical Organizational Imperative.
241
Party Politics, 10(1), 5-24. doi:10.1177/1354068804039118
Celis, K., & Van Daele, J. (Eds.). (2011). Harry Van Velthoven. Waarheen met België? Van
taalstrijd tot communautaire conflicten. Een selectie uit 35 jaar wetenschappelijk
onderzoek. Brussel: Academic and Scientific Publishing.
Coffé, H., Cioroianu, I., & Vandenberghe, M. (2019). Social Media Networks of Political
Candidates in Belgium’s Multilevel Elections. Paper presented at the ECPR Conference
'Advances in the Empirical and Theoretical Study of Parliaments', Konstanz, Germany.
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and psychological
measurement, 20(1), 37-46.
Cordell, K., & Wolff, S. (Eds.). (2016). The routledge handbook of ethnic conflict. London:
Routledge.
Cornille, D., Deroose, M., Godefroid, H., Melyn, W., Stinglhamber, P., & Van Parys, S. (2022).
How sustainable are the finances of the federal government, the regions and the
communities in Belgium? NBB Economic Review(19).
Craeybeckx, J. (2009). From the Great War to the Great Depression. In E. Witte, J. Craeybeckx,
& A. Meynen (Eds.), Political History of Belgium: From 1830 onwards (pp. 143-181).
Brussels: Academic and Scientific Publishing.
CRISP. (1961). 2. Les positions socialistes Wallonnes (1944-1960). Courrier hebdomadaire du
CRISP, 92(2), 4-I. doi:10.3917/cris.092.0004
CRISP. (2023). Gouvernements nationaux et fédéraux depuis 1944. Retrieved from
https://www.crisp.be/documents-politiques/gouvernements/gouvernements-nationaux-
federaux/
Daalder, H. (1983). The Comparative Study of Parties and Party Systems: An Overview. In H.
Daalder & P. Mair (Eds.), Western European Party Systems (pp. 1-28). London: Sage.
Dahl, R. A. (1966). Some explanations. In R. A. Dahl (Ed.), Political oppositions in western
democracies. London: Yale University Press.
Dalle Mulle, E. (2016). New trends in justifications for national self-determination: evidence
from Scotland and Flanders. Ethnopolitics, 15(2), 211-229.
Dalle Mulle, E. (2018). The Nationalism of the Rich: Discourses and strategies of separatist
parties in Catalonia, Flanders, Northern Italy and Scotland. London: Routledge.
Dalle Mulle, E., & Serrano, I. (2019). Between a principled and a consequentialist logic: Theory
and practice of secession in Catalonia and Scotland. Nations and nationalism, 25(2),
242
630-651.
Dalton, R. J. (2008). The quantity and the quality of party systems: Party system polarization,
its measurement, and its consequences. Comparative political studies, 41(7), 899-920.
Damgaard, E. (2008). Cabinet Termination. In K. Strøm, W. C. Müller, & T. Bergman (Eds.),
Cabinets and Coalition Bargaining: The Democratic Life Cycle in Western Europe (pp.
301-326). Oxford: Oxford University Press.
Dandoy, R. (2014). Regionalist parties and immigration in belgium. In E. Hepburn & R. Zapata-
Barrero (Eds.), The Politics of Immigration in Multi-Level States: Governance and
Political Parties (pp. 200-222). London: Palgrave Macmillan.
Dandoy, R., & Baudewyns, P. (2005). The preservation of social security as a national function
in the Belgian federal state. The territorial politics of welfare, 148-167.
Dandoy, R., & De Decker, N. (2009). Peut-on encore parler de ‘partis frères’ en Belgique? In
J.-B. Pilet, J.-M. De Waele, & S. Jaumain (Eds.), L’absence de partis nationaux:
menace ou opportunité (pp. 19-35). Bruxelles: Editions de l'Université de Bruxelles.
Dandoy, R., & Joly, J. (2018). Party system change in Belgium: From stability to
fragmentation? In M. Lisi (Ed.), Party system change, the European crisis and the state
of democracy (pp. 24-45). New York: Routledge.
Dandoy, R., & Terrière, L. (2021). Caretaker Governments in Belgium The new normal? In D.
Caluwaerts & M. Reuchamps (Eds.), Belgian Exceptionalism Belgian Politics between
Realism and Surrealism (pp. 123-137). New York: Routledge.
Dardanelli, P., Kincaid, J., Fenna, A., Kaiser, A., Lecours, A., Singh, A., . . . Vogel, S. (2019).
De/Centralisation Dataset 1790-2010.[Data Collection]. Colchester, Essex: UK Data
Service.
De Bruycker, P. (1989). Bruxelles dans la réforme de l'Etat. Courrier hebdomadaire du CRISP,
1230-1231(5-6), 1-61. doi:10.3917/cris.1230.0001
De Clercq, E. (2023). Deliberatieve democratie bij de hervorming van het Belgische
federalisme. Het dialoogplatform als eerste bescheiden poging. In K. Reybrouck, S.
Sottiaux, & W. Pas (Eds.), Inspiratie voor de staatshervorming (pp. 9-50). Brussel:
Lefebvre Sarrut Belgium NV.
De Croo, A. (2023). Toespraak Alexander De Croo - Groot Liberaal Congres. Retrieved from
https://www.openvld.be/toespraak_alexander_de_croo_groot_liberaal_congres
De Croo, A., & Magnette, P. (2020). Regeerakkoord - 30 september 2020. Brussel.
243
De Grauwe, P., & Van Parijs, P. (2021, November 24). Ostbelgien: op weg naar een volwaardig
gewest? [Opinie]. De Standaard.
(2007, September 7). De keien van de Wetstraat (Season 1, Episode 1) [Television series
episode]. In De keien van de Wetstraat; VRT.
De Metsenaere, M. (1998). Taalgrens - Talentellingen. In R. de Schryver, B. De Wever, G.
Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie van
de Vlaamse Beweging (pp. 2950-2956). Tielt: Lannoo.
De Metsenaere, M., Deslé, E., Detant, A., Mares, A., Parmentier, S., & Sieben, L. (1998).
Brussel. In R. de Schryver, B. De Wever, G. Durnez, L. Gevers, P. van Hees, & M. De
Metsenaere (Eds.), Nieuwe Encyclopedie van de Vlaamse Beweging (pp. 622-652).
Tielt: Lannoo.
De Smedt, J., Hooghe, M., & Walgrave, S. (2011). Meest actieve parlementsleden nauwelijks
in nieuws : Vlaamse parlementsleden in de Vlaamse televisiejournaals (juni 2009 - juni
2011) (ENA Nieuwsmonitor) (ENA Ed.). Antwerpen: Universiteit Antwerpen.
De Standaard (2019). Stefan Sottiaux (KU Leuven) over de (herziening van de) grondwet: ‘Eén
minister van Klimaat? Een absurd idee’. Retrieved from
https://www.standaard.be/cnt/dmf20190324_04277945
de Vet, B. (2019). Between party and parliament: the roles of parliamentary party group
leaders in partitocratic Belgium. (PhD thesis). Ghent University, Ghent.
De Wever, B. (1994). Greep naar de macht. Vlaams-nationalisme en Nieuwe Orde: het VNV
1933-1945. Tielt: Lannoo.
De Wever, B. (1998). Vlaamse Militanten Orde - Vlaamse Militanten Organisatie. In R. de
Schryver, B. De Wever, G. Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.),
Nieuwe Encyclopedie van de Vlaamse Beweging (pp. 3465-3467). Tielt: Lannoo.
De Wever, B. (Ed.) (2011). Right-wing Flanders, left-wing Wallonia? Is this so? If so, why?
And is it a problem? E-book: Re-Bel initiative - Rethinking Belgium's institutions in the
European context.
De Wever, B., & Seberechts, F. (1998). Vlaams-nationalistische partijen. In R. de Schryver, B.
De Wever, G. Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe
Encyclopedie van de Vlaamse Beweging (pp. 3334-3361). Tielt: Lannoo.
De Wever, B., Verdoodt, F.-J., & Vrints, A. (2019). Flemish patriots and the construction of
the nation: how the Flemish nation ceased to be'small' (Nise Essays 4). Antwerp:
Peristyle.
244
De Wever, B., & Vrints, A. (2008). Vlaams-nationalisme: Natievorming, ideologie en politieke
stroming. In L. Sanders & C. Devos (Eds.), Politieke ideologieën in Vlaanderen (pp.
319-379). Antwerpen: Standaard Uitgeverij.
De Winter, L. (1998a). Conclusion: a comparative analysis of the electoral, office and policy
success of ethnoregionalist parties. In L. De Winter & H. Türsan (Eds.), Regionalist
Parties in Western Europe (pp. 204-247). London: Routledge.
De Winter, L. (1998b). The Volksunie and the dilemma between policy success and electoral
survival in Flanders. In L. De Winter & H. Türsan (Eds.), Regionalist Parties in Western
Europe. London: Routledge.
De Winter, L. (2006). Multi-level party Competition and Coordination in Belgium. In D. Hough
& C. Jeffery (Eds.), Devolution and electoral politics (pp. 76-95). Manchester:
Manchester University Press.
De Winter, L., & Baudewyns, P. (2009). Belgium: Towards the Breakdown of a Nation-State
in the Heart of Europe? Nationalism and ethnic politics, 15(3-4), 280-304.
De Winter, L., & Dumont, P. (2006). Do Belgian parties undermine the democratic chain of
delegation? West European Politics, 29(5), 957-976.
De Winter, L., & Dumont, P. (2014). Regeringsvorming in België: mission impossible? In C.
Devos (Ed.), België #2014: Een politieke geschiedenis van morgen (pp. 241-271). Gent:
Borgerhoff & Lamberigts.
De Winter, L., & Dumont, P. (2021). Coalition Formation in Belgium: From exceptional
complexity to regime breakdown? In D. Caluwaerts & M. Reuchamps (Eds.), Belgian
exceptionalism: Belgian politics between realism and surrealism (pp. 108-122). New
York: Routledge.
De Winter, L., Gomez, M., & Lynch, P. (2018). Ethnoregionalist parties. In K. Detterbeck &
E. Hepburn (Eds.), Handbook of Territorial Politics. Cheltenham: Edward Elgar
Publishing.
De Winter, L., Swyngedouw, M., & Dumont, P. (2009). Party System (s) and Electoral
Behaviour in Belgium: From Stability to Balkanisation. In M. Brans, L. De Winter, &
W. Swenden (Eds.), The Politics of Belgium (pp. 71-94). London: Routledge.
De Winter, L., & Türsan, H. (Eds.). (1998). Regionalist parties in Western Europe. London:
Routledge.
De Winter, L. (2007). La recherche sur les identités ethno-territoriales en Belgique. [Research
on Ethno-Territorial Identities in Belgium]. Revue internationale de politique
comparée, 14(4), 575-595. doi:10.3917/ripc.144.0575
245
Decoster, A., & Sas, W. (2011). De bijzondere financieringswet voor dummies. Flemosi
discussion paper 4.
Decoster, A., & Sas, W. (2012). De nieuwe financieringswet: anders, maar ook beter? Leuvense
Economische Standpunten 2012/137. Leuven: Centrum voor Economische Studiën,
K.U. Leuven.
Decoster, A., & Sas, W. (2013). De nieuwe bijzondere financieringswet van de 6de
staatshervorming: werden de beloften ingelost? FLEMOSI discussion paper 28.
Dedecker, N. (2011). Le Mouvement réformateur (MR). In P. Delwit, J.-B. Pilet, & E. van
Haute (Eds.), Les partis politiques en Belgique (pp. 145-162). Bruxelles: Les editions
de l’université de Bruxelles.
Delcorps, V. (2012). La violence communautaire en Belgique. La Révue Générale, 10(8-9), 19-
40.
Delcorps, V., Dujardin, V., & Maingain, O. (2014). FDF: 50 ans d'engagement politique.
Bruxelles: Editions Racine.
Delwit, P. (1998). Le parti et le gouvernement. Les socialistes de la libération aux Golden
Sixties. In H. Le Paige & P. Delwit (Eds.), Les Socialistes et le pouvoir: gouverner pour
réformer? (pp. 221-258). Bruxelles: Éditions Labor.
Delwit, P. (2011a). Elections et gouvernements en Belgique depuis 1945. In P. Delwit, J.-B.
Pilet, & E. van Haute (Eds.), Les partis politiques en Belgique (pp. 319-352). Brussels:
Editions de l’Université de Bruxelles.
Delwit, P. (2011b). Partis et systèmes de partis en Belgique en perspective. In P. Delwit, J.-B.
Pilet, & E. van Haute (Eds.), Les partis politiques en Belgique (pp. 7-34). Brussels:
Editions de l’Université de Bruxelles.
Delwit, P. (2021a). Le Parti du travail de Belgique – Partij van de Arbeid (PTB-PVDA)
Redéploiement de la gauche radicale en Belgique. In P. Delwit (Ed.), Les partis
politiques en Belgique (pp. 349-375). Bruxelles: Editions de l'Université de Bruxelles.
Delwit, P. (2021b). Le Parti socialiste (PS). Les défis d’une social-démocratie en déclin. In P.
Delwit & E. van Haute (Eds.), Les partis politiques en Belgique (pp. 225-251). Brussels:
Editions de l’Université de Bruxelles.
Delwit, P. (2022). The Labor Party of Belgium (PTB-PVDA): A Modern Radical Left Party?
Frontiers in Political Science, 4, 1-12.
Delwit, P., & van Haute, E. (2002). L’implosion et la fin d’un parti: la Volksunie. L'Année
sociale 2001 / Institut de Sociologie, Université libre de Bruxelles, 13-24.
246
Delwit, P., & van Haute, E. (Eds.). (2021). Les partis politiques en Belgique. Brussels: Editions
de l'Université de Bruxelles.
Deneckere, G., De Paepe, T., De Wever, B., & Vanthemsche, G. (2014). Een geschiedenis van
België. Gent: Academia Press.
Deprez, K., & Vos, L. (Eds.). (1998). Nationalism in Belgium: Shifting Identities 1780-1995.
Houndmills, Basingstoke: Macmillan Press LTD.
Deprez, K., & Vos, L. (Eds.). (1999). Nationalisme in België: Identiteiten in beweging 1780-
2000. Antwerpen: Houtekiet.
Deschouwer, K. (1994). The termination of coalitions in Belgium. Res Publica, 36(1), 43-55.
Deschouwer, K. (1999-2000). Comprendre le nationalisme flamand. Fédéralisme
Régionalisme, 1 - Nationalisme et démocratie. Retrieved from
https://popups.uliege.be/1374-3864/index.php?id=285
Deschouwer, K. (2002). Falling Apart Together. The Changing Nature of Belgian
Consociationalism. Acta Politica, 37, 68-85.
Deschouwer, K. (2004). From consociation to federation. In K. R. Luther & K. Deschouwer
(Eds.), Party elites in divided societies: Political parties in consociational democracy
(pp. 93-127). New York: Routledge.
Deschouwer, K. (2005). The unintended consequences of consociational federalism: the case
of Belgium. In I. O'Flynn & D. Russell (Eds.), Power Sharing: New Challenges for
Divided Societies (pp. 92-106). London: Pluto Press.
Deschouwer, K. (2006). And the peace goes on? Consociational democracy and Belgian politics
in the twenty-first century. West European Politics, 29(5), 895-911.
Deschouwer, K. (2009). The Rise and Fall of the Belgian Regionalist Parties. Regional and
Federal Studies, 19(4-5), 559-577.
Deschouwer, K. (2012). The Politics of Belgium: Governing a Divided Society (2nd ed.).
Basingstoke: Palgrave-Macmillan.
Deschouwer, K. (2013a). The Belgian Federation: A labyrinth state. In J. Loughlin, J. Kincaid,
& W. Swenden (Eds.), Routledge Handbook of Regionalism & Federalism (pp. 211-
222). New York: Routledge.
Deschouwer, K. (2013b). Party Strategies, Voter Demands and Territorial Reform in Belgium.
West European Politics, 36(2), 338-358.
247
Deschouwer, K., De Winter, L., Reuchamps, M., Sinardet, D., & Dodeigne, J. (2015).
Communautaire overtuigingen en stemgedrag. In K. Deschouwer, P. Delwit, M.
Hooghe, P. Baudewyns, & S. Walgrave (Eds.), De kiezer ontcijferd: over stemgedrag
en stemmotivaties (pp. 151-167). Leuven: LannooCampus.
Deschouwer, K., Pilet, J.-B., & van Haute, E. (2017). Party families in a split party system. In
K. Deschouwer (Ed.), Mind the Gap. Political Participation and Representation in
Belgium (pp. 91-112). London: Rowman & Littlefield.
Deschouwer, K., & Sinardet, D. (2010). Identiteiten, communautaire standpunten en
stemgedrag. In K. Deschouwer, P. Delwit, M. Hooghe, & S. Walgrave (Eds.), De
stemmen van het volk: een analyse van het kiesgedrag in Vlaanderen en Wallonië op 7
juni 2009 (pp. 75-98). Brussel: VUBPRESS.
Deschouwer, K., & Temmerman, M. (2012). Elite behaviour and elite communication in a
divided society: the Belgian federal coalition formation of 2007. Journal of Language
and Politics, 11(4), 500-520.
Deutsch, K. W. (1953). Nationalism and Social Communication: An Inquiry into the
Foundations of Nationality. Cambridge: Technology Press.
Devos, C., & Bouteca, N. (2008). Brussel-Halle-Vilvoorde: Voeren van de 21ste eeuw.
Fédéralisme Régionalisme, 8(1), 1-17.
Devos, C., & Sinardet, D. (2012). Governing without a government: The Belgian experiment.
Governance, 25(2), 167-171.
Dewachter, W. (1996). La Belgique d’aujourd’hui comme société politique. In A. Dieckhoff
(Ed.), La Belgique: la force de la désunion (pp. 105-142). Brussels: Editions Complexe.
Deweerdt, M. (1979). Overzicht van het Belgisch politiek gebeuren in 1978. Res Publica, 12(2),
185-207.
Deweerdt, M. (1998a). Overzicht van het Belgische politiek gebeuren in 1997. Res Publica,
40(3-4), 311-375.
Deweerdt, M. (1998b). Taalgrens - De grondwettelijke vastlegging van de taalgrens en verdere
betwistingen. In R. de Schryver, B. De Wever, G. Durnez, L. Gevers, P. van Hees, &
M. De Metsenaere (Eds.), Nieuwe Encyclopedie van de Vlaamse Beweging (pp. 2960-
2962). Tielt: Lannoo.
Deweerdt, M. (1999). Overzicht van het Belgische politiek gebeuren in 1998. Res Publica,
41(2-3), 163-238.
Deweerdt, M. (2021, September, 29). [Personal communication (e-mail)].
248
Dodeigne, J., Gramme, P., Reuchamps, M., & Sinardet, D. (2016). Beyond linguistic and party
homogeneity: Determinants of Belgian MPs’ preferences on federalism and state
reform. Party Politics, 22(4), 427-439. doi:10.1177/1354068814550429
Dodeigne, J., Niessen, C., Reuchamps, M., & Sinardet, D. (2022). Identités et préférences des
parlementaires envers le fédéralisme belge à l’aube d’une septième réforme de l’État.
Courrier hebdomadaire du CRISP, 2532-2533(7-8), 5-62.
Dodeigne, J., Reuchamps, M., & Sinardet, D. (2015). Identités, préférences et attitudes des
parlementaires envers le fédéralisme belge après la sixième réforme de l’État. Courrier
hebdomadaire du CRISP, 2278(33), 5-52. doi:10.3917/cris.2278.0005
Döring, H., & Manow, P. (2020). Parliaments and governments database (ParlGov):
Information on parties, elections and cabinets in modern democracies.
Duchacek, I. D. (1988). Dyadic Federations and Confederations. Publius: The journal of
Federalism, 18(2), 5-31.
Dumont, P., & De Winter, L. (1999). La formation et le maintien des gouvernements (1946-
1999). Courrier hebdomadaire du CRISP, 1664(39), 1-59.
Dumont, P., De Winter, L., & Dandoy, R. (2001). Démissions gouvernementales et
performances électorales des majorités sortantes (1946-1999). Courrier hebdomadaire
du CRISP, 1722(17), 5-51.
Duval, D., & Pétry, F. (2020). Citizens’ evaluations of campaign pledge fulfillment in Canada.
Party Politics, 26(4), 437-447.
Eckstein, H. (1966). Division and Cohesion in Democracy: A Study of Norway. Princeton:
Princeton University Press.
Eckstein, H. (1975). Case study and theory in political science. In F. I. Greenstein & N. W.
Polsby (Eds.), Handbook of political science. Political science: Scope and theory (pp.
94-137). Reading, Mass.: Addison-Wesley.
Elazar, D. J. (1976). Federalism vs. Decentralization: The Drift from Authenticity. Publius: The
journal of Federalism, 6(4), 9-19.
Elazar, D. J. (1987). Exploring federalism. Tuscaloosa: University of Alabama Press.
Elchardus, M., & Pelleriaux, K. (1998). De polis verdeeld. Hoe de kiezers links en rechts
herdefiniëren. In M. Swyngedouw, J. B. Billiet, A. Carton, & R. Beerten (Eds.), De
(on)redelijke kiezer. Onderzoek naar de politieke opvattingen van Vlamingen.
Verkiezingen van 21 mei 1995. Leuven: Acco.
249
Elias, A., & Tronconi, F. (2011). From protest to power: Autonomist parties and the challenges
of representation. Vienna: Braumüller.
English, R. (2003). Armed struggle: The history of the IRA. New York: Oxford University
Press.
Erk, J., & Anderson, L. M. (2010a). The Paradox of Federalism: Does self-rule accommodate
or exacerbate ethnic divisions? In J. Erk & L. M. Anderson (Eds.), The Paradox of
Federalism: Does self-rule accommodate or exacerbate ethnic divisions? (pp. 1-12).
New York: Routledge.
Erk, J., & Anderson, L. M. (Eds.). (2010b). The Paradox of Federalism: Does self-rule
accommodate or exacerbate ethnic divisions? New York: Routledge.
Eyskens, G. (1993). Gaston Eyskens: de memoires (J. Smits Ed.). Tielt: Lannoo.
Eyskens, M. (2014). Mark Eyskens - Citaten en quotes. Retrieved from
https://www.eyskens.com/styled-3/styled-3/citaten.html
Ezrow, L. (2008). Research note: On the inverse relationship between votes and proximity for
niche parties. European Journal of Political Research, 47(2), 206-220.
Fabre, E. (2009). Belgian federalism in a comparative perspective. Vives discussion paper no.
5. Leuven: Vives.
Federal Public Services Home Affairs, I. (2019). Results in figures. Retrieved from
https://verkiezingen2019.belgium.be/en/results-figures?el=CK&id=CKR00000
Fernandes, J. M., Leston-Bandeira, C., & Schwemmer, C. (2018). Election proximity and
representation focus in party-constrained environments. Party Politics, 24(6), 674-685.
Ferriter, D. (2019). The border: The legacy of a century of Anglo-Irish politics. London: Profile
Books.
FOD Binnenlandse Zaken, I. (2020). Bevolkingscijfers per provincie en per gemeente op 1
januari 2020. Retrieved from
https://www.ibz.rrn.fgov.be/fileadmin/user_upload/fr/pop/statistiques/population-
bevolking-20200101.pdf
Froio, C., Bevan, S., & Jennings, W. (2017). Party mandates and the politics of attention:Party
platforms, public priorities and the policy agenda in Britain. Party Politics, 23(6), 692-
703. doi:10.1177/1354068815625228
Gagnon, A.-G. (2021). Multinational federalism: challenges, shortcomings and promises.
Regional & Federal Studies, 31(1), 99-114.
250
Gellner, E. (2008). Nations and Nationalism (2nd ed.). New York: Cornell University Press.
Gemenis, K. (2013). What to Do (and Not to Do) with the Comparative Manifestos Project
Data. Political Studies, 61, 3-23.
Gevers, L., & Vos, L. (1998). Studentenbeweging - Leuven. In R. de Schryver, B. De Wever,
G. Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie
van de Vlaamse Beweging (pp. 2902-2917). Tielt: Lannoo.
Ghai, Y. (2000). Autonomy as a strategy for diffusing conflict. In P. C. Stern & D. Druckman
(Eds.), International conflict resolution after the Cold War (pp. 483-530). Washington,
D.C.: National academy press.
Giger, N. (2009). Towards a modern gender gap in Europe?: A comparative analysis of voting
behavior in 12 countries. The Social Science Journal, 46(3), 474-492.
Goossens, J. (2017). Electoral Reforms in Belgium's Sixth State Reform: Historic Split of
Electoral Constituency BHV, Reform of the Senate, and Coincident Elections. Election
Law Journal: Rules, Politics, and Policy, 16(2), 316-324.
Goossens, J., & Cannoot, P. (2015). Belgian federalism after the sixth state reform. Perspectives
on Federalism, 7(2), 29-55.
Govaert, S. (2002). La Volksunie. Du déclin à la dispartion (1993-2001). Courrier
hebdomadaire du CRISP, 1748(3), 5-44.
Gray, A. M., & Birrell, D. (2012). Coalition government in Northern Ireland: social policy and
the lowest common denominator thesis. Social policy and society, 11(1), 15-25.
Guelke, A. (2012). Politics in Deeply Divided Societies. Cambridge: Polity Press.
Guibernau i Berdún, M. M. (1996). Nationalisms: the nation-state and nationalism in the
twentieth century (1st print., repr. ed.): Oxford : Polity press.
Guibernau, M. (2001). Regionalism. In A. S. Leoussi (Ed.), Encyclopaedia of nationalism (pp.
263-265). London: Transaction publishers.
Guibernau, M. (2006). National identity, devolution and secession in Canada, Britain and Spain.
Nations and nationalism, 12(1), 51-76.
Gurr, T. R. (1970). Why men rebel (2nd print. ed.). Princeton (N.J.): Princeton university press.
Gurr, T. R. (1993). Minorities at Risk: A Global View of Ethnopolitical Conflicts. Washington:
United States Institute of Peace Press.
251
Gutmann, A. (Ed.) (1994). Multiculturalism: examining the politics of recognition. Princeton
(N.J.): Princeton university press.
Haagdorens, L., & Wouters, N. (1998). Mars(en) op Brussel. In R. de Schryver, B. De Wever,
G. Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie
van de Vlaamse Beweging (pp. 2004-2006). Tielt: Lannoo.
Habermas, J. (1991). The structural transformation of the public sphere: An inquiry into a
category of bourgeois society. Cambridge: MIT press.
Hale, H. E. (2000). The parade of sovereignties: Testing theories of secession in the Soviet
setting. British journal of political science, 30(1), 31-56.
Hale, H. E. (2004). Divided We Stand: Institutional Sources of Ethnofederal State Survival and
Collapse. World Politics, 56(2), 165-193.
Hepburn, E. (2013). New challenges for stateless nationalist and regionalist parties. Oxon:
Routledge.
Heylen, W., & Van Hecke, S. (2008). Regeringen die niet regeren. Het malgoverno van de
Belgische politiek (1977-1981). Leuven: Uitgeverij LannooCampus.
Hirschman, A. O. (1970). Exit, voice, and loyalty: Responses to decline in firms, organizations,
and states. Cambridge: Harvard university press.
Hobsbawm, E. J. (1990). Nations and Nationalism since 1780. Cambridge: Cambridge
University Press.
Hoflack, K. (2021). De premier. n.d.: Borgerhoff & Lamberigts.
Hooghe, L. (1993). Belgium: From Regionalism to Federalism. Regional Policy and Politics,
3(1), 44-68.
Hooghe, L. (2004). Belgium: Hollowing the Center. In U. M. Amoretti & N. Bermeo (Eds.),
Federalism and Territorial Cleavages (pp. 55–92). Baltimore: Johns Hopkins
University Press.
Hooghe, L., Marks, G., Schakel, A. H., Osterkatz, S. C., Niedzwiecki, S., & Shair-Rosenfield,
S. (2016). Measuring regional authority: A postfunctionalist theory of governance,
Volume I. Oxford: Oxford University Press.
Hooghe, L., Marks, G. N., & Schakel, A. H. (2010). The rise of regional authority: A
comparative study of 42 democracies. London: Routledge.
Hooghe, M. (2012). The political crisis in Belgium (2007–2011): a federal system without
252
federal loyalty. Representation, 48(1), 131-138.
Hooghe, M., & Stiers, D. (2022). Nationalist identities, preferences and voting for sub‐state
nationalist parties in the Flemish region of Belgium. Nations and nationalism, 28(4),
1356-1378.
Horowitz, D. L. (1985). Ethnic Groups in Conflict (2nd ed.). Berkeley: University of California
Press.
Horowitz, D. L. (1991). A Democratic South Africa? Constitutional Engineering in a Divided
Society. Berkeley: University of California Press.
Horowitz, D. L. (1993). The Challenge of Ethnic Conflict: Democracy in Divided Societies.
Journal of Democracy, 4(4), 18-38.
Horowitz, D. L. (2000). Ethnic Groups in Conflict (2nd ed.). Berkeley: University of California
Press.
Horowitz, D. L. (2002). Constitutional Design: Proposals versus Processes. In A. Reynolds
(Ed.), The architecture of democracy: Constitutional design, conflict management, and
democracy (pp. 15-36). New York: Oxford University Press.
Horowitz, D. L. (2003). The cracked foundations of the right to secede. Journal of Democracy,
14(2), 5-17.
Horowitz, D. L. (2004). The alternative vote and interethnic moderation: A reply to Fraenkel
and Grofman. Public choice, 121(3), 507-517.
Horowitz, D. L. (2007). The many uses of federalism. Drake Law Review, 55(4), 953-966.
Horowitz, D. L. (2014). Ethnic Power Sharing: Three Big Problems. Journal of Democracy,
24(2), 5-20.
Hroch, M. (2015). European Nations: Explaining Their Formation. London: Verso.
Huyse, L. (1970). Passiviteit, pacificatie en verzuiling in de Belgische politiek: Een
sociologische studie. Antwerpen: Standaard Wetenschappelijke Uitgeverij.
Huyse, L. (1981). Political Conflict in Bicultural Belgium. In A. Lijphart (Ed.), Conflict and
Coexistence in Belgium. The Dynamics of a Culturally Divided Society (pp. 107-126).
Berkeley: Institute of International Studies, University of California.
Huyse, L., & Dhondt, S. (1991). Onverwerkt verleden. Collaboratie en repressie in België,
1942-1952. Leuven: Kritak.
253
Huysseune, M. (2017). The flemish paradox: The hegemony of pro-independence parties in a
region largely indifferent towards independence. Ethnopolitics, 16(4), 352-368.
Ichijo, A., & Uzelac, G. (Eds.). (2005). When is the nation? Towards an understanding of
theories of nationalism. New York: Routledge.
Inglehart, R. (1977). The silent revolution: Changing values and political styles among Western
publics. Princeton: Princeton University Press.
Inglehart, R. (1997). Modernization and postmodernization: Cultural, economic, and political
change in 43 societies. Princeton: Princeton University Press.
Inglehart, R., & Norris, P. (2000). The developmental theory of the gender gap: Women’s and
men’s voting behavior in global perspective. International Political Science Review,
21(4), 441-463.
Inglehart, R., & Norris, P. (2003). Rising tide: Gender equality and cultural change around the
world. Cambridge: Cambridge University Press.
IPSOS. (2023). De Grote Peiling - Le Grand Baromètre Retrieved from
https://graphicvalley.hln.be/grotepeiling/grotepeiling032023.pdf
ISPO. (2022). Belgian National Elections Study (BNES), 1991-2019. Retrieved from
https://soc.kuleuven.be/ceso/ispo/projects/copy_of_the-transformation-of-the-socio-
economic-left-2013-right-cleavage
Jacobs, D., & Swyngedouw, M. (2003). Territorial and Non-Territorial Federalism in Belgium:
Reform of the Brussels Capital Region, 2001. Regional & Federal Studies, 13(2), 127-
139.
Jans, M. T. (2001). Leveled domestic polities: Comparing institutional reform and
ethnonational conflicts in Canada and Belgium (1960-1989). Res Publica, 43(1), 37-58.
Janssens, R. (2018). BRIO-taalbarometer 4: De talen van Brussel. Retrieved from
https://www.briobrussel.be/node/14763
Janssens, R. (2019). Language policies versus language practices: A new language conflict? In
G. Caliendo, R. Janssens, S. Slembrouck, & P. van Avermaert (Eds.), Urban
Multilingualism in Europe: Bridging the Gap between Language Policies and
Language Practices (pp. 13-35). Boston: De Gruyter.
Jones, D. R. (2001). Party polarization and legislative gridlock. Political Research Quarterly,
54(1), 125-141.
Justaert, M. (2019, March 25). 'Eén minister van Klimaat? Een absurd idee'. De Standaard, p.
254
12.
JVD. (1989, 11 May 1989). "Begroting 90 wordt dé test op sociaal-ekonomisch vlak". De Tijd,
p. 3.
Katz, R. S., & Mair, P. (1993). The evolution of party organizations in Europe: The three faces
of party organization. American Review of Politics, 14, 593-618.
Keil, S., & Anderson, P. (2018). Decentralization as a tool for conflict resolution. In K.
Detterbeck & E. Hepburn (Eds.), Handbook of Territorial Politics (pp. 89-106).
Cheltenham: Edward Elgar.
Kesteloot, C. (1993). Mouvement wallon et identité nationale. Courrier hebdomadaire du
CRISP, 1392(7), 1-48.
Kesteloot, C. (1998a). The Growth of the Walloon Movement. In K. Deprez & L. Vos (Eds.),
Nationalism in Belgium. Shifting Identities, 1780–1995 (pp. 139-152). Houndmills:
Macmillan.
Kesteloot, C. (1998b). Voeren. In R. de Schryver, B. De Wever, G. Durnez, L. Gevers, P. van
Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie van de Vlaamse Beweging (pp.
3521-3526). Tielt: Lannoo.
Kesteloot, C. (1998c). Waalse Beweging. In R. de Schryver, B. De Wever, G. Durnez, L.
Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie van de Vlaamse
Beweging (pp. 3635-3651). Tielt: Lannoo.
Kincaid, J. (2011). Introduction: Alternative Models, Constitutional Foundations, and
Institutional Features of Federal Governance. In J. Kincaid (Ed.), Federalism (pp. vii-
x). London: Sage.
Klüver, H., Bäck, H., & Krauss, S. (2023). Coalition Agreements as Control Devices: Coalition
Governance in Western and Eastern Europe. Oxford: Oxford University Press.
Koppen, J., Distelmans, B., & Janssens, R. (2002). Taalfaciliteiten in de Rand:
Ontwikkelingslijnen, conflictgebieden en taalpraktijk. Brusselse Thema's 9. Brussel:
VUBPress.
Kossmann, E. H. (1986). De Lage Landen 1780-1980 : twee eeuwen Nederland en België. Deel
2: 1914-1980. Amsterdam: Elsevier.
Krauss, S. (2018). Stability through control? The influence of coalition agreements on the
stability of coalition cabinets. West European Politics, 41(6), 1282-1304.
Kymlicka, W. (1998). Is federalism a viable alternative to secession? In P. B. Lehning (Ed.),
255
Theories of Secession (pp. 109-148). London: Routledge.
Kymlicka, W. (2001a). Minority Nationalism and Multination Federalism. In W. Kymlicka
(Ed.), Politics in the Vernacular: Nationalism, Multiculturalism, and Citizenship (pp.
91-119). Oxford: Oxford University Press.
Kymlicka, W. (2001b). Politics in the Vernacular: Nationalism, Multiculturalism, and
Citizenship. Oxford: Oxford University Press.
Lake, D. A., & Rothchild, D. (1996). Containing Fear: The Origins and Management of Ethnic
Conflict. International security, 21(2), 41-75.
Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical
data. biometrics, 33(1), 159-174.
Lanneau, C. (2015). L’action du FDF dans les Régions et Communautés (1971-2014). I.
Parlement de la Communauté française et Parlement flamand. Courrier hebdomadaire
du CRISP, 2258-2259(13), 5-61. doi:10.3917/cris.2258.0005
Laver, M. (1974). Dynamic factors in government coalition formation. European Journal of
Political Research, 2(3), 259-270.
Laver, M., & Schofield, N. (1990). Multiparty government: The politics of coalition in Europe.
Oxford: Oxford university press.
Lecours, A. (2021). Nationalism, secessionism, and autonomy. Oxford: Oxford University
Press.
Lengauer, G., Esser, F., & Berganza, R. (2012). Negativity in political news: A review of
concepts, operationalizations and key findings. Journalism, 13(2), 179-202.
Lernout, G. (1998). Petitionnement. In R. de Schryver, B. De Wever, G. Durnez, L. Gevers, P.
van Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie van de Vlaamse Beweging
(pp. 2464-2466). Tielt: Lannoo.
Levy, J. S. (2008). Case studies: Types, designs, and logics of inference. Conflict management
and peace science, 25(1), 1-18.
Lijphart, A. (1968a). The Politics of Accommodation: Pluralism and Democracy in the
Netherlands. Berkeley: University of California Press.
Lijphart, A. (1968b). Typologies of democratic systems. Comparative political studies, 1(1), 3-
44.
Lijphart, A. (1968c). Verzuiling, pacificatie en kentering in de Nederlandse politiek.
256
Amsterdam: De Bussy.
Lijphart, A. (1969a). Consociational Democracy. World Politics, 21(2), 207-225.
Lijphart, A. (1969b). Kentering in de Nederlandse politiek. Acta Politica, 4(3), 231-247.
Lijphart, A. (1977a). Democracy in Plural Societies: A Comparative Exploration. New Haven:
Yale University Press.
Lijphart, A. (1977b). Majority rule versus democracy in deeply divided societies. Politikon:
South African Journal of Political Studies, 4(2), 113-126.
Lijphart, A. (1979). Consociation and Federation: Conceptual and Empirical Links. Canadian
Journal of Political Science/Revue canadienne de science politique, 12(3), 499-515.
Lijphart, A. (1980). Federal, confederal, and consociational options for the South African plural
society. In R. Rotberg & J. Barratt (Eds.), Conflict and Compromise in South Africa (pp.
51-75). Lexington, MA: Lexington Books.
Lijphart, A. (1981a). Consociational theory: Problems and prospects. A reply. Comparative
politics, 13(3), 355-360.
Lijphart, A. (1981b). Introduction: The Belgian Example of Cultural Coexistence in
Comparative Perspective. In A. Lijphart (Ed.), Conflict and Coexistence in Belgium.
The Dynamics of a Culturally Divided Society (pp. 1-12). Berkeley: Institute of
International Studies, University of California.
Lijphart, A. (1985a). Non-Majoritarian Democracy: A Comparison of Federal and
Consociational Theories. Publius: The journal of Federalism, 15(2), 3-15.
Lijphart, A. (1985b). Power-sharing in South-Africa. Berkeley (Calif.): University of California
Berkeley; Institute of international studies.
Lijphart, A. (1989). Democratic political systems: Types, cases, causes, and consequences.
Journal of Theoretical Politics, 1(1), 33-48.
Lijphart, A. (2000). Definitions, evidence, and policy: a response to Matthijs Bogaards' critique.
Journal of Theoretical Politics, 12(4), 425-431.
Lijphart, A. (2002). The Wave of Power-Sharing Democracy. In A. Reynolds (Ed.), The
architecture of democracy: Constitutional design, conflict management, and democracy
(pp. 37-54). New York: Oxford University Press.
Lijphart, A. (2004). Constitutional Design for Divided Societies. Journal of Democracy, 15(2),
96-109.
257
Lijphart, A. (2008). Thinking about Democracy: Power Sharing and Majority Rule in Theory
and Practice. New York: Routledge.
Lijphart, A. (Ed.) (1981c). Conflict and Coexistence in Belgium. The Dynamics of a Culturally
Divided Society. Berkeley: Institute of International Studies, University of California.
Linz, J. J., & Stepan, A. (1992). Political Identities and Electoral Sequences: Spain, the Soviet
Union, and Yugoslavia. Daedalus, 121(2), 123-139.
Lipset, S. M., & Rokkan, S. (1967). Party Systems and Voter Alignments: Cross-National
Perspectives. New York: Free press.
Louwerse, T. (2012). Mechanisms of issue congruence: The democratic party mandate. West
European Politics, 35(6), 1249-1271.
Louwerse, T., & Van Aelst, P. (2013). The exceptional Belgian case? Government formation
duration in comparative perspective. Paper presented at the The State of the Federation
(Vereniging voor Politieke Wetenschappen, Association Belge Francophone de Science
Politique), Louvain-la-Neuve.
Luebbert, G. M. (1984). A theory of government formation. Comparative political studies,
17(2), 229-264.
Luebbert, G. M. (1986). Comparative Democracy: Policymaking and Governing Coalitions in
Europe and Israel. New York: Columbia University Press.
Lustick, I. (1979). Stability in Deeply Divided Societies: Consociationalism versus Control.
World Politics, 31(3), 325-344.
Lustick, I. (1997). Lijphart, Lakatos, and consociationalism. World Politics, 50(1), 88-117.
Luther, K. R., & Deschouwer, K. (Eds.). (2004). Party elites in divided societies: Political
parties in consociational democracy. New York: Routledge.
Maddens, B., Beerten, R., & Billiet, J. (1994). O dierbaar België? Het natiebewustzijn van
Vlamingen en Walen. Leuven: ISPO/SOI.
Mansbridge, J. (2003). Rethinking representation. American political science review, 97(4),
515-528.
Marangoni, F., & Vercesi, M. (2015). The government and its hard decisions: how conflict is
managed within the coalition. In N. Conti & F. Marangoni (Eds.), The Challenge of
Coalition Government: The Italian Case (pp. 17-35). New York: Routledge.
Marks, G., Hooghe, L., & Schakel, A. H. (2008). Patterns of regional authority. Regional and
258
Federal Studies, 18(2-3), 167-181.
Martens, W. (2006). De memoires: luctor et emergo. Tielt: Lannoo.
Martínez‐Herrera, E. (2002). From nation‐building to building identification with political
communities: Consequences of political decentralisation in Spain, the Basque Country,
Catalonia and Galicia, 1978–2001. European Journal of Political Research, 41(4), 421-
453.
Massetti, E., & Schakel, A. H. (2013). Ideology matters: Why decentralisation has a
differentiated effect on regionalist parties' fortunes in Western democracies. European
Journal of Political Research, 52(6), 797-821.
Massetti, E., & Schakel, A. H. (2015). From class to region: How regionalist parties link (and
subsume) left-right into centre-periphery politics. Party Politics, 21(6), 866-886.
Massetti, E., & Schakel, A. H. (2016). Between autonomy and secession: Decentralization and
regionalist party ideological radicalism. Party Politics, 22(1), 59-79.
Massetti, E., & Schakel, A. H. (2017). Decentralisation reforms and regionalist parties’
strength: accommodation, empowerment or both? Political Studies, 65(2), 432-451.
Matthieß, T. (2020). Retrospective pledge voting: A comparative study of the electoral
consequences of government parties’ pledge fulfilment. European Journal of Political
Research, 59(4), 774-796.
Mazzoleni, M. (2009). The saliency of regionalization in party systems: A comparative analysis
of regional decentralization in party manifestos. Party Politics, 15(2), 199-218.
Mazzoleni, O., & Mueller, S. (2016). Regionalist Parties in Western Europe. New York:
Routledge.
McCulloch, A. (2009). Seeking Stability Amid Deep Division: Consociationalism and
Centripetalism in Comparative Perspective. (PhD Thesis). Queen’s University,
Kingston, Ontario.
McCulloch, A. (2014). Power-Sharing and Political Stability in Deeply Divided Societies. New
York: Routledge.
McCulloch, A. (2021). Introduction: Power-Sharing in Europe—From Adoptability to End-
Ability. In S. Keil & A. McCulloch (Eds.), Power-Sharing in Europe: Past Practice,
Present Cases, and Future Directions (pp. 1-18). Cham: Palgrave Macmillan.
McCulloch, A. (2022). Consociational Executives: Power-Sharing Governments Between
Inclusion and Functionality. In F. Requejo & M. Sanjaume-Calvet (Eds.), Defensive
259
Federalism (pp. 22-45). London: Routledge.
McCulloch, A., & Vandeginste, S. (2019). Veto power and power-sharing: insights from
Burundi (2000–2018). Democratization, 26(7), 1176-1193.
McEvoy, J. (2015). Power-sharing executives: Governing in Bosnia, Macedonia, and Northern
Ireland. Philadelphia: University of Pennsylvania Press.
McEvoy, J., & Aboultaif, E. W. (2022). Power-sharing challenges: From weak adoptability to
dysfunction in Iraq. Ethnopolitics, 21(3), 238-257.
McGarry, J., & O'Leary, B. (2009). Must pluri-national federations fail? Ethnopolitics, 8(1), 5-
25.
McGarry, J., & O’Leary, B. (1993a). Introduction: The macro-political regulation of ethnic
conflict. In J. McGarry & B. O’Leary (Eds.), The Politics of Ethnic Conflict Regulation
(pp. 1-40). London: Routledge.
McGarry, J., & O’Leary, B. (1994). The Political Regulation of National and Ethnic Conflict.
Parliamentary Affairs, 47, 94-115.
McGarry, J., & O’Leary, B. (Eds.). (1993b). The politics of ethnic conflict regulation: Case
studies of protracted ethnic conflicts. London: Routledge.
McKittrick, D., Kelters, S., Feeney, B., Thornton, C., & McVea, D. (1999). Lost lives: The
stories of the men, women and children who died as a result of the Northern Ireland
troubles. Edinburgh: Mainstream Publishing.
McKittrick, D., & McVea, D. (2012). Making sense of the troubles: The story of the conflict in
Northern Ireland. London: Viking.
McLaughlin, E. (2005). Governance and social policy in Northern Ireland (1999-2004): the
devolution years and postscript. In M. Powell, L. Bauld, & K. Clarke (Eds.), Social
Policy Review 17 (pp. 107-124). Bristol: Policy Press.
Meguid, B. M. (2005). Competition between unequals: The role of mainstream party strategy
in niche party success. American political science review, 99(3), 347-359.
Meguid, B. M. (2008). Party competition between unequals: strategies and electoral fortunes
in Western Europe. Cambridge: Cambridge University Press.
Meguid, B. M. (2009). Institutional change as strategy: The role of decentralization in party
competition. Paper presented at the APSA 2009 Toronto Meeting.
Meguid, B. M. (2022). Adaptation or inflexibility? Niche party responsiveness to policy
260
competition, with evidence from regionalist parties. European Journal of Political
Research, 62(2), 355-376.
Meyer, T. M., & Miller, B. (2015). The niche party concept and its measurement. Party Politics,
21(2), 259-271.
Meynen, A. (2009). Economic and Social Policy Since the 1950s. In E. Witte, J. Craeybeckx,
& A. Meynen (Eds.), Political History of Belgium: From 1830 onwards (pp. 271-359).
Brussels: Academic and Scientific Publishing.
Midi, H., Sarkar, S. K., & Rana, S. (2010). Collinearity diagnostics of binary logistic regression
model. Journal of interdisciplinary mathematics, 13(3), 253-267.
Mosselmans, J. (2009). The Brussels-Halle-Vilvoorde Question: A Linguistic Trap. European
Public Law, 15(1), 5-16.
Moury, C., & Timmermans, A. (2013). Inter-Party Conflict Management in Coalition
Governments: Analyzing the Role of Coalition Agreements in Belgium, Germany, Italy
and the Netherlands. Politics and Governance, 1(2), 117-131.
Mouton, O., & Vanpeteghem, B. (2003). Numero uno: Guy Verhofstadts weg naar de top. Tielt:
Lannoo.
Müller-Rommel, F. (1998). Ethnoregionalist parties in Western Europe: theoretical
considerations and framework of analysis. In L. De Winter & H. Türsan (Eds.),
Regionalist Parties in Western Europe (pp. 17-27). London: Routledge.
Müller, W. C., Bergman, T., & Strøm, K. (2008). Coalition Theory and Cabinet Governance:
An Introduction. In K. Strøm, W. C. Müller, & T. Bergman (Eds.), Cabinets and
Coalition Bargaining: The Democratic Life Cycle in Western Europe (pp. 1-50).
Oxford: Oxford University Press.
Müller, W. C., & Strøm, K. (1999). Policy, office, or votes? How political parties in Western
Europe make hard decisions. Cambridge: Cambridge university press.
Müller, W. C., & Strøm, K. (2000). Coalition Governments in Western Europe: An
Introduction. In W. C. Müller & K. Strøm (Eds.), Coalition Governments in Western
Europe (pp. 1-31). Oxford: Oxford university press.
Murtagh, C. (2021). Toward Inclusive Power-Sharing in Northern Ireland: Two Steps Forward,
One Step Back. In S. Keil & A. McCulloch (Eds.), Power-Sharing in Europe: Past
Practice, Present Cases, and Future Directions (pp. 141-169). E-book: Palgrave
MacMillan.
Nagle, J. (2020). Consociationalism is Dead! Long Live Zombie Power‐Sharing! Studies in
261
Ethnicity and Nationalism, 20(2), 137-144.
Naurin, E. (2011). Election promises, party behaviour and voter perceptions. London: Palgrave
Macmillan.
Naurin, E., Thomson, R., Cotta, M., & Russo, F. (2020). Research Handbook on Political
Representation. In M. Cotta & F. Russo (Eds.), Chapter 23: The fulfilment of election
pledges (pp. 289-300). Cheltenham: Edward Elgar Publishing Limited.
Newman, S. (1995). Losing the electoral battles and winning the policy wars: Ethnoregional
conflict in Belgium. Nationalism and ethnic politics, 1(4), 44-72.
Niessen, C., Reuchamps, M., Stjepanović, D., & Habra, A. (2020). When Have Dyadic
Federations Succeeded and When Have They Failed? A Comparative Analysis of
Bipolar Federalism Around the World. In Federalism and National Diversity in the 21st
Century (pp. 41-72). London: Palgrave Macmillan.
Nordlinger, E. A. (1972). Conflict Regulation in Divided Societies. Cambridge, MA: Center for
International Affairs, Harvard University.
Nousiainen, J. (1993). Decision-making, policy content and conflict resolution in Western
European cabinets. In J. Blondel, F. Müller-Rommel, p. 259-, & MacMillan. (Eds.),
Governing Together: The Extent and Limits of Joint Decision-Making in Western
European Cabinets (pp. 259-282). New York: St Martin's press.
O'Neill, K. (2003). Decentralization as an electoral strategy. Comparative political studies,
36(9), 1068-1091.
O’Leary, B. (2005). Debating consociational politics: Normative and explanatory arguments.
In S. Noel (Ed.), From power sharing to democracy: Post-conflict institutions in
ethnically divided societies (pp. 3-43). Montreal: McGill-Queen’s University Press.
Pascolo, L., Vermassen, D., Reuchamps, M., & Caluwaerts, D. (2021). The changing dynamics
of Belgian federalism: Is there a reversal of the paradox of federalism? In D. Caluwaerts
& M. Reuchamps (Eds.), Belgian Exceptionalism. Belgian Politics between Realism
and Surrealism (pp. 138-152). New York: Routledge.
Pauwels, T. (2011). Le Vlaams Belang. In P. Delwit, J.-B. Pilet, & E. van Haute (Eds.), Les
partis politiques en Belgique (pp. 219-234). Brussels: Editions de l’Université de
Bruxelles.
Peiffer, Q., & Sautois, J. (2014). L’autonomie constitutive après la sixième réforme de l’État.
In J. Sautois & M. Uyttendaele (Eds.), La sixième réforme de l'État (2012-2013):
Tournant historique ou soubresaut ordinaire? (pp. 103-144). Limal: Anthemis.
262
Peters, B. G. (2006). Consociationalism, corruption and chocolate: Belgian exceptionalism.
West European Politics, 29(5), 1079-1092.
Petrocik, J. R. (1996). Issue ownership in presidential elections, with a 1980 case study.
American journal of political science, 825-850.
Pilet, J.-B. (2005). Le Rassemblement wallon (RW) et le Front démocratique des francphones
(FDF): des partis victimes de leurs succès? In P. Delwit (Ed.), Les partis régionalistes
en Europe. Des acteurs en dévéloppement? (pp. 265-282). Brussels: Editions de
l’Université de Bruxelles.
Pilet, J.-B., & van Haute, E. (2008). The federal elections in Belgium, June 2007. Electoral
Studies, 27(3), 547-550.
Pitkin, H. F. (1967). The concept of representation (Vol. 75). Los Angeles: University of
California Press.
Politics of the Low Countries. (n.d.). Retrieved from
https://www.elevenjournals.com/tijdschrift/PLC/detail
Polk, J., Rovny, J., Bakker, R., Edwards, E., Hooghe, L., Jolly, S., . . . Zilovic, M. (2017).
Explaining the salience of anti-elitism and reducing political corruption for political
parties in Europe with the 2014 Chapel Hill Expert Survey data. Research & Politics,
4(1), 1-9.
Popelier, P. (2019). Asymmetry and Complexity as a Device for Multinational Conflict
Management. A Country Study of Constitutional Asymmetry in Belgium. In P. Popelier
& M. Sahadžić (Eds.), Constitutional Asymmetry in Multinational Federalism:
Managing Multinationalism in Multi-tiered Systems (pp. 17-45). E-book: Palgrave
Macmillan.
Popelier, P. (2021). Power-sharing in Belgium: The disintegrative model. In S. Keil & A.
McCulloch (Eds.), Power-Sharing in Europe: Past Practice, Present Cases, and Future
Directions (pp. 89-114). London: Palgrave MacMillan.
Popelier, P., & Lemmens, K. (2015). The constitution of Belgium : a contextual analysis.
Oxford: Hart.
Posen, B. R. (1993). The security dilemma and ethnic conflict. Survival, 35(1), 27-47.
Quévit, M. (1978). Les Causes du Déclin Wallon. Brussels: Ed. Vie Ouvrière.
Rabushka, A., & Shepsle, K. A. (1972). Politics in Plural Societies: A Theory of Democratic
Instability. Columbus, Ohio: Charles E. Merrill Publishing Company.
263
Reilly, B. (1997). Preferential voting and political engineering: A comparative study. Journal
of Commonwealth & Comparative Politics, 35(1), 1-19.
Reilly, B. (2001). Democracy in Divided Societies: Electoral Engineering for Conflict
Management. Cambridge: Cambridge University Press.
Reilly, B. (2002). Electoral Systems for Divided Societies. Journal of Democracy, 13(2), 156-
170.
Reuchamps, M. (2013). The Current Challenges on the Belgian Federalism and the Sixth
Reform of the State. In A. López Basaguren & L. Escajedo San Epifanio (Eds.), The
Ways of Federalism in Western Countries and the Horizons of Territorial Autonomy in
Spain: Volume 1 (pp. 375-392). Berlin, Heidelberg: Springer.
Reuchamps, M. (2017). Belgium: The Short Story of a Long History of (In)Stability? [50
Shades of Federalism]. Retrieved from
http://50shadesoffederalism.com/policies/linguistic-diversity-plurinational-states/
Reuchamps, M., Boerjan, H., Niessen, C., & Randour, F. (2021). More or less regional
autonomy? A qualitative analysis of citizen arguments towards (de) centralization in
Belgium. Comparative European Politics, 19(2), 225-247.
Reuchamps, M., Kavadias, D., & Deschouwer, K. (2014). Drawing Belgium: using mental
maps to measure territorial conflict. Territory, Politics, Governance, 2(1), 30-51.
Reuchamps, M., Sinardet, D., Dodeigne, J., & Caluwaerts, D. (2017). Reforming Belgium’s
federalism: Comparing the views of MPs and voters. Government and Opposition,
52(3), 460-482.
Reybrouck, K., & Sottiaux, S. (2019). De federale bevoegdheden: Antwerpen: Intersentia.
Reynolds, A. (Ed.) (2002). The Architecture of Democracy: Constitutional Design, Conflict
Management, and Democracy. New York: Oxford University Press.
Reynolds, A., Reilly, B., & Council, N. R. (2000). Electoral systems and conflict in divided
societies. In P. C. Stern & D. Druckman (Eds.), International Conflict Resolution After
the Cold War (pp. 420-482). Washington, DC: The National Academies Press.
Rihoux, B., Vandeleene, A., De Winter, L., Baudewyns, P., & Deruette, S. (2015). Belgium.
European Journal of Political Research - Political Data Yearbook, 54, 33-43.
Rihoux, B., Vandeleene, A., De Winter, L., Baudewyns, P., & Deruette, S. (2016). Belgium.
European Journal of Political Research - Political Data Yearbook, 55(1), 30-35.
Rihoux, B., Vandeleene, A., De Winter, L., Baudewyns, P., & Deruette, S. (2020). Belgium :
264
political developments and data in 2019. European Journal of Political Research -
Political Data Yearbook, 59(1), 34-55. doi:https://doi.org/10.1111/2047-8852.12303
Rihoux, B., Vandeleene, A., De Winter, L., Baudewyns, P., & Deruette, S. (2021). Belgium :
political developments and data in 2020. European Journal of Political Research -
Political Data Yearbook, 60(1), 31–48. doi:https://doi.org/10.1111/2047-8852.12316
Riker, W. H. (1964). Federalism: origin, operation, significance. Boston: Little Brown.
Riker, W. H. (1975). Federalism. In F. I. Greenstein & N. W. Polsby (Eds.), Handbook of
Political Science, Volume 5: Governmental Institutions and Processes. Reading, Mass.:
Addison-Wesley.
Rochtus, D. (2012). The Rebirth of Flemish Nationalism: Assessing the Impact of N‐VA
Chairman Bart De Wever's Charisma. Studies in Ethnicity and Nationalism, 12(2), 268-
285.
Rodden, J. (2004). Comparative Federalism and Decentralization: On Meaning and
Measurement. Comparative politics, 36(4), 481-500.
Roeder, P. G., & Rothchild, D. (2005). Dilemmas of State-Building in Divided Societies. In P.
G. Roeder & D. Rothchild (Eds.), In Sustainable Peace: Power and Democracy After
Civil Wars (pp. 1-26). Ithaca, NY: Cornell University Press.
Romainville, C. (2015). Dynamics of Belgian Plurinational Federalism: A Small State Under
Pressure. BC Int'l & Comp. L. Rev., 38, 225.
Rose, R., & Wessels, B. (2019). Money, sex and broken promises: Politicians’ bad behaviour
reduces trust. Parliamentary Affairs, 72(3), 481-500.
Rose, W. (2000). The security dilemma and ethnic conflict: Some new hypotheses. Security
Studies, 9(4), 1-51.
Rothchild, D., & Hartzell, C. A. (1999). Security in deeply divided societies: The role of
territorial autonomy. Nationalism and ethnic politics, 5(3-4), 254-271.
Rudolph, J. R., & Thompson, R. J. (1985). Ethnoterritorial movements and the policy process:
Accommodating nationalist demands in the developed world. Comparative politics,
17(3), 291-311.
Ruys, M. (1998). Van den Brande, Luc. In R. de Schryver, B. De Wever, G. Durnez, L. Gevers,
P. van Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie van de Vlaamse
Beweging (pp. 591-593). Tielt: Lannoo.
Ruys, M., & Garré, G. (1998). Craeybeckx, Lode (eigenlijk François F.L.). In R. de Schryver,
265
B. De Wever, G. Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe
Encyclopedie van de Vlaamse Beweging (pp. 817-818). Tielt: Lannoo.
Saalfeld, T. (2008). Institutions, chance and choices: The dynamics of cabinet survival in the
parliamentary democracies of Western Europe (1945-99). In K. Strøm, W. C. Müller,
& T. Bergman (Eds.), Cabinets and Coalition Bargaining: The Democratic Life Cycle
in Western Europe (pp. 327-268). Oxford: Oxford University Press.
Saeys, F. (2007). Statuut, organisatie en financiering van de openbare televisieomroep in
Vlaanderen. In Publieke televisie in Vlaanderen: Een geschiedenis (pp. 23-51). Gent:
Academia Press.
Sartori, G. (1976). Parties and party systems. New York: Cambridge University Press.
Sautois, J. (2014). La Région de Bruxelles-Capitale: Chronique de la naissance d'une région à
part. Administration publique: Revue du droit public et des sciences administratives,
37(2), 108-158.
Sautois, J., & Uyttendaele, M. (2013). La sixième réforme de l'État (2012-2013): Tournant
historique ou soubresaut ordinaire?(Belgique). Limal: Anthemis.
Scharpf, F. W. (1988). The joint‐decision trap: Lessons from German federalism and European
integration. Public administration, 66(3), 239-278.
Scharpf, F. W. (2006). The joint‐decision trap revisited. JCMS: Journal of Common Market
Studies, 44(4), 845-864.
Schneider, A. (2003). Decentralization: Conceptualization and measurement. Studies in
comparative international development, 38, 32-56.
Schofield, N. (1987). Stability of coalition governments in Western Europe: 1945–1986.
European Journal of Political Economy, 3(4), 555-591.
Senelle, R., & van de Velde, E. (1998). Staatshervorming. In R. de Schryver, B. De Wever, G.
Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie van
de Vlaamse Beweging (pp. 2825-2839). Tielt: Lannoo.
Seymour, M., & Gagnon, A.-G. (2012). Introduction: Multinational Federalism: Questions and
Queries. In M. Seymour & A.-G. Gagnon (Eds.), Multinational Federalism: Problems
and Prospects (pp. 1-19). Basingstoke: Palgrave Macmillan.
Shair-Rosenfield, S., Schakel, A. H., Niedzwiecki, S., Marks, G., Hooghe, L., & Chapman-
Osterkatz, S. (2021). Language difference and regional authority. Regional & Federal
Studies, 31(1), 73-97.
266
Sijstermans, J. (2021). The Vlaams Belang: A mass party of the 21st century. Politics and
Governance, 9(4), 275-285.
Sijstermans, J., & Brown Swan, C. (2022). Shades of solidarity: Comparing Scottish and
Flemish responses to Catalonia. Regional & Federal Studies, 32(2), 161-181.
doi:10.1080/13597566.2021.1881064
Sinardet, D. (2007). Wederzijdse mediarepresentaties van de nationale "andere": Vlamingen,
Franstaligen en het Belgische federale samenlevingsmodel. (PhD thesis). University of
Antwerp, Antwerp.
Sinardet, D. (2008). Vlaamse en Franstalige media over Franstaligen en Vlamingen
Wederzijdse representaties van de ‘andere’ in politieke debatprogramma’s. In H. Van
den Bulck & A. Dhoest (Eds.), Media, Cultuur, Identiteit: Actueel onderzoek naar
media en maatschappij (pp. 51-69). Gent: Academia Press.
Sinardet, D. (2010). From consociational consciousness to majoritarian myth: Consociational
democracy, multi-level politics and the Belgian case of Brussels-Halle-Vilvoorde. Acta
Politica, 45(3), 346-369.
Sinardet, D. (2011). Le fédéralisme consociatif belge: vecteur d’instabilité? Pouvoirs(1), 21-
35.
Sinardet, D. (2012). Is There a Belgian Public Sphere? What the Case of a Federal Multilingual
Country Can Contribute to the Debate on Transnational Public Spheres, and Vice Versa.
In M. Seymour & A.-G. Gagnon (Eds.), Multinational Federalism: Problems and
Prospects (pp. 172-202). London: Palgrave Macmillan.
Sinardet, D., Dodeigne, J., & Reuchamps, M. (2013). Beyond the Myth of Unanimity. Opinions
of Belgian MPs on Federalism and the Sixth Reform of the Belgian State. In W.
Vermeersch (Ed.), Belgian Society and Politics 2013: As Ever, in Between Elections
(pp. 71-83). Ghent: Gerrit Kreveld Foundation.
Sisk, T. D. (1996). Power sharing and international mediation in ethnic conflicts. Washington,
DC: US Institute of Peace Press.
Smooha, S., & Hanf, T. (1992). The Diverse Modes of Conflict-Regulation in Deeply Divided
Societies. International Journal of Comparative Sociology, 33(1-2), 26-47.
Sorens, J. (2009). The partisan logic of decentralization in Europe. Regional & Federal Studies,
19(2), 255-272.
Stepan, A. C. (1999). Federalism and Democracy: Beyond the U.S. Model. Journal of
Democracy, 10(4), 19-34.
267
Stokes, D. E. (1963). Spatial Models of Party Competition. American political science review,
57(2), 368-377.
Strøm, K., Müller, W. C., & Bergman, T. (Eds.). (2008). Cabinets and Coalition Bargaining:
The Democratic Life Cycle in Western Europe. Oxford: Oxford University Press.
Studlar, D. T., McAllister, I., & Hayes, B. C. (1998). Explaining the gender gap in voting: A
cross-national analysis. Social Science Quarterly, 79(4), 779-798.
Swenden, W. (2006). Federalism and Regionalism in Western Europe: A Comparative and
Thematic Analysis. Houndmills, Basingstoke: Palgrave Macmillan.
Swenden, W. (2013a). Rebooting Territorial Pluralism: Belgium and the Crisis of Governability
2007-2011. In J. McGarry & R. Simeon (Eds.), Assessing Territorial Pluralism (pp.
196-219). Vancouver: University of British Columbia Press.
Swenden, W. (2013b). Territorial strategies for managing plurinational states. In J. Loughlin,
J. Kincaid, & W. Swenden (Eds.), Routledge Handbook of Regionalism & Federalism
(pp. 61-75). New York: Routledge.
Swenden, W., Brans, M., & De Winter, L. (2009). Introduction: The Politics of Belgium:
Institutions and policy under bipolar and centrifugal federalism. In M. Brans, L. De
Winter, & W. Swenden (Eds.), The Politics of Belgium: Institutions and policy under
bipolar and centrifugal federalism (pp. 1-11). New York: Routledge.
Swenden, W., & Jans, M. T. (2009). 'Will it Stay or Will it Go?' Federalism and the
Sustainability of Belgium. In M. Brans, L. De Winter, & W. Swenden (Eds.), The
Politics of Belgium: Institutions and policy under bipolar and centrifugal federalism
(pp. 15-32). New York: Routledge.
Swyngedouw, M., Abts, K., Baute, S., Galle, J., & Meuleman, B. (2015). Het communautaire
in de verkiezingen van 25 mei 2014: analyse op basis van de postelectorale
verkiezingsonderzoeken 1991-2014. Onderzoeksverslag CeSO/ISPO. Leuven: Centrum
voor Sociologisch Onderzoek (CeSo)/Instituut voor Sociaal en Politiek
Opinieonderzoek (ISPO).
Swyngedouw, M., Abts, K., & Galle, J. (2014). Vlamingen en de communautaire kwestie. In
K. Abts, M. Swyngedouw, J. Billiet, & B. Meuleman (Eds.), Vlaanderen kiest. Trends
in stemgedrag en opvattingen over politiek, staatshervorming en kerk (pp. 219-245).
Tielt: Lannoo Campus.
Swyngedouw, M., & Rink, N. (2008). Hoe Vlaams-Belgischgezind zijn de Vlamingen? Een
analyse op basis van het postelectorale verkiezingsonderzoek 2007 (Onderzoeksverslag
CeSO/ISPO). Leuven: Centrum voor Sociologisch Onderzoek (CeSo)/Instituut voor
Sociaal en Politiek Opinieonderzoek (ISPO).
268
Taylor, C., & Gutmann, A. (1992). Multiculturalism and the politics of recognition: an essay.
Princeton (N.J.): Princeton university press.
Terrière, L. (2021). ‘With respect for the core business’: on regionalist parties and their minister
portfolios in government. Territory, Politics, Governance, 1-20.
doi:10.1080/21622671.2021.1948914
Terrière, L., & Bouteca, N. (2020). With respect for the core business: the impact of party
ideology on the odds of government participation among regionalist parties. ACTA
POLITOLOGICA (PRAHA), 12(3), 75-100.
Terrière, L., & Bouteca, N. (2021). Regionalist Parties in Central Government: A Shift in
Ministerial Portfolio Preferences? Nationalities Papers, 1-15.
Thijs, T. (2019, 1 May 2019). Vlaams Belang komt op in alle Waalse provincies: ‘Franstaligen
kapen ook stemmen weg in Vlaanderen’, Online article. De Morgen.
Thijssen, P., Sinardet, D., & Dandoy, R. (2015). Maken de contacten de Belgen? De relatie
tussen Vlaams-Waalse contacten, identiteiten en houdingen over staatshervorming. In
K. Deschouwer, P. Delwit, M. Hooghe, P. Baudewyns, & S. Walgrave (Eds.), De kiezer
ontcijferd: over stemgedrag en stemmotivaties (pp. 130-150). Leuven: LannooCampus.
Thomson, R. (2001). The programme to policy linkage: The fulfilment of election pledges on
socio-economic policy in the Netherlands, 1986–1998. European Journal of Political
Research, 40(2), 171-197.
Thomson, R. (2011). Citizens’ evaluations of the fulfillment of election pledges: Evidence from
Ireland. The Journal of politics, 73(1), 187-201.
Thomson, R., & Brandenburg, H. (2019). Trust and Citizens’ Evaluations of Promise Keeping
by Governing Parties. Political Studies, 67(1), 249-266.
doi:10.1177/0032321718764177
Thomson, R., Royed, T., Naurin, E., Artés, J., Costello, R., Ennser‐Jedenastik, L., . . . Pétry, F.
(2017). The fulfillment of parties’ election pledges: A comparative study on the impact
of power sharing. American journal of political science, 61(3), 527-542.
Tierney, S. (2009). Federalism in a Unitary State: A paradox too far? Regional & Federal
Studies, 19(2), 237-253.
Tilly, C. (1975). Revolutions and collective violence. In F. I. Greenstein & N. W. Polsby (Eds.),
Handbook of political science (Vol. 3). Reading, Mass.: Addison-Wesley.
Timmermans, A. (2006). Standing apart and sitting together: Enforcing coalition agreements in
multiparty systems. European Journal of Political Research, 45(2), 263-283.
269
Timmermans, A., & Moury, C. (2006). Coalition Governance in Belgium and The Netherlands:
Rising Government Stability Against All Electoral Odds. Acta Politica, 41, 389-407.
Toubeau, S., & Wagner, M. (2015). Explaining party positions on decentralization. British
journal of political science, 45(1), 97-119.
Treisman, D. (2002). Defining and measuring decentralization: a global perspective.
Unpublished manuscript. Retrieved from
<http://www.sscnet.ucla.edu/polisci/faculty/treisman/Papers/defin.pdf>
Treisman, D. (2007). The architecture of government: Rethinking political decentralization.
New York: Cambridge University Press.
Tsebelis, G. (1990). Elite interaction and constitution building in consociational democracies.
Journal of Theoretical Politics, 2(1), 5-29.
Türsan, H. (1998). Introduction: ethnoregionalist parties as ethnic entrepreneurs. In L. De
Winter & H. Türsan (Eds.), Regionalist Parties in Western Europe (pp. 1-16). London:
Routledge.
United Nations, OHCHR (2017). UN independent expert urges Spanish Government to reverse
decision on Catalan autonomy [Press release]. Retrieved from
https://www.ohchr.org/en/press-releases/2017/10/un-independent-expert-urges-
spanish-government-reverse-decision-catalan
Valcke, T., Reynaert, H., & Steyvers, K. (2011). Belgium. In H. Heinelt & X. Bertrana (Eds.),
The second tier of local government in Europe : provinces, counties, départements and
landkreise in comparison (pp. 27-55). London: Routledge.
Van Bosstraeten, P.-J. (2008). De splitsing van de Belgische Socialistische Partij. Twee
verklarende documenten. WT. Tijdschrift over de geschiedenis van de Vlaamse
beweging, 67(2), 99-114.
Van Dam, D. (1998). Histoire de mouvement Wallon. In M. Martiniello & M. Swyngedouw
(Eds.), Où Va la Belgique? Les Soubresauts d'une Petite Démocratie Européenne (pp.
73–84). Paris: L'Harmattan.
Van de Voorde, A. (2003). Mark Eyskens: politicus-professor tussen woord en daad: een
biographie. Tielt: Lannoo Uitgeverij.
Van den Wijngaert, M. (2011). Van een unitair naar een federaal België. In M. Van den
Wijngaert (Ed.), Van een unitair naar een federaal België: 40 jaar beleidsvorming in
gemeenschappen en gewesten (1971-2011) (pp. 19-38). Brussel: Academic and
Scientific Publishers nv.
270
Van Gorp, B., & Sinardet, D. (2017). The role of news media in the policy process in Belgium.
In M. Brans & D. Aubin (Eds.), Policy Analysis in Belgium (Vol. 10, pp. 235-253).
Bristol: Policy Press.
van Haute, E. (2005). La Volksunie (VU) : triomphe des idées, défaite du parti. In P. Delwit
(Ed.), Les partis régionalistes en Europe. Des acteurs en dévéloppement? Bruxelles:
Editions de l’Université de Bruxelles.
van Haute, E. (2011). Volksunie, Nieuw-Vlaams Alliantie, Spirit, Vlaams-Progressief. In P.
Delwit, J.-B. Pilet, & E. van Haute (Eds.), Les partis politiques en Belgique (pp. 201-
218). Bruxelles: Les editions de l’université de Bruxelles.
van Haute, E. (2016). Regionalist parties in Belgium (N-VA, FDF): A renewed success? In O.
Mazzoleni & S. Mueller (Eds.), Regionalist Parties in Western Europe (pp. 98-118).
London: Routledge.
van Haute, E. (2021). La Nieuw-Vlaamse Alliantie (N-VA). In P. Delwit & E. van Haute (Eds.),
Les partis politiques en Belgique (pp. 113-136). Bruxelles: Ed. de l'Université de
Bruxelles.
van Haute, E., & Pilet, J.-B. (2006). Regionalist parties in Belgium (VU, RW, FDF): Victims
of their own success? Regional & Federal Studies, 16(3), 297-313.
Van Parijs, P. (2018). Belgium. Een utopie voor onze tijd. Kalmthout: Polis.
Van Schendelen, M. (1985). Consociational Democracy: The Views of Arend Lijphart and
Collected Criticisms. The Political Science Reviewer, 15, 143.
Van Velthoven, H. (2005). De opbouw van een Brusselse identiteit. Naar een Brussels Gewest-
Gemeenschap?/The construction of a Brussels identity. Towards a Brussels Region-
Community? (summary). WT. Tijdschrift over de geschiedenis van de Vlaamse
beweging, 64, 58-79. doi:https://doi.org/10.21825/wt.v64i.12712
Van Velthoven, H. (2011). Waarheen met België? Vijftien jaar later. Een stand van zaken in
2011. In K. Celis & J. van Daele (Eds.), Waarheen met België? Van taalstrijd tot
communautaire conflicten: een selectie uit 35 jaar wetenschappelijk onderzoek (pp.
319-359). Brussel: ASP (Academic & Scientific Publishers).
Van Velthoven, H. (2019). Bevriende vijanden: Hoe de Belgische socialisten uit elkaar
groeiden. Kalmthout: Polis.
Vandenberghe, M. (2022a). Cabinet conflicts in Belgium (1979-2018) - Extended version.
Mendeley Data. Available at: https://data.mendeley.com/datasets/zvyt86jffd/1
Vandenberghe, M. (2022b). Mapping Cabinet Conflicts and Conflict Features. Refined
271
Definitions, Coding Instructions and Results from Belgium (1995-2018). Politics of the
Low Countries, 4(1), 116-128.
Vandenberghe, M. (2022c). Power-sharing and the Paradox of Federalism: Federalization and
the Evolution of Ethno-territorial Conflict in the Case of Belgium (1979-2018).
Ethnopolitics, Online first. doi:10.1080/17449057.2022.2045828
Vandenberghe, M. (2022d). A three-legged race: Assessing the functionality of consociational
power-sharing with cabinet conflict-resolution data from Belgium (1979-2006).
Manuscript submitted for publication.
Vandenberghe, M. (2023). Pacification through decentralization: An explanatory analysis of
ethno-territorial cabinet conflict in Belgium (1979-2006). Manuscript submitted for
publication.
Vandeweyer, L., Huyse, L., Hoflack, K., & Ruys, M. (1998). Repressie. In R. de Schryver, B.
De Wever, G. Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe
Encyclopedie van de Vlaamse Beweging (pp. 2584-2602). Tielt: Lannoo.
Velaers, J., & Van Goethem, H. (2001). Leopold III: De Koning, het Land, de Oorlog. Tielt:
Lannoo.
Verba, S. (1965). Organizational membership and democratic consensus. The Journal of
politics, 27(3), 467-497.
Vercesi, M. (2016). Coalition Politics and Inter‐Party Conflict Management: A Theoretical
Framework. Politics & Policy, 44(2), 168-219.
Verleden, F. (2009). Splitting the difference: the radical approach of the Belgian parties. In W.
Swenden & B. Maddens (Eds.), Territorial party politics in Western Europe (pp. 145-
166). London: Palgrave Macmillan.
Vlaams Belang, V. (2023, April 25). "De N-VA is zo groot en machtig geworden, maar heeft
nooit het communautaire op de agenda durven te plaatsen. Wij zullen diezelfde fout
nooit maken." @tomvangrieken #Momentum [Twitter]. Retrieved April 25, 2023 from
https://twitter.com/vlbelang/status/1650790254360637440?cxt=HHwWgMC93a6_5O
gtAAAA
Vliegenthart, R., Boomgaarden, H. G., & Boumans, J. W. (2011). Changes in political news
coverage: Personalization, conflict and negativity in British and Dutch newspapers. In
K. Brants & K. Voltmer (Eds.), Political communication in postmodern democracy (pp.
92-110). Basingstoke: Palgrave Macmillan.
Vliegenthart, R., Walgrave, S., & Zicha, B. (2013). How preferences, information and
institutions interactively drive agenda‐setting: Questions in the Belgian parliament,
272
1993–2000. European Journal of Political Research, 52(3), 390-418.
Volkens, A., Bara, J., & Budge, I. (2009). Data quality in content analysis. The case of the
comparative manifestos project. Historical Social Research/Historische
Sozialforschung, 234-251.
Volkens, A., Burst, T., Krause, W., Lehmann, P., Matthieß, T., Regel, S., . . . Zehnter, L.
(2021a). The Manifesto Data Collection. Manifesto Project (MRG/CMP/MARPOR).
Version 2021a.
Volkens, A., Burst, T., Krause, W., Lehmann, P., Matthieß, T., Regel, S., . . . Zehnter, L.
(2021b). The Manifesto Project Dataset - Codebook. Manifesto Project (MRG / CMP /
MARPOR). Version 2021a. Berlin: Wissenschaftszentrum Berlin für Sozialforschung
(WZB).
Volkens, A., Krause, W., Lehmann, P., Matthieß, T., Merz, N., Regel, S., & Weßels, B. (2019).
The Manifesto Data Collection. Manifesto Project (MRG/CMP/MARPOR). Version
2019b.
Vuye, H., & Wouters, V. (2017). Sleutels tot ontgrendeling : uitdagingen aan de Vlaamse
meerderheid: Deurne: Uitgeverij Doorbraak.
Wagner, M. (2012). Defining and measuring niche parties. Party Politics, 18(6), 845-864.
Walgrave, S., & De Swert, K. (2005). De ‘Kanseliersbonus’ Revisited: Over de Dominante
Aanwezigheid van de Regering en de Premier in het TV-nieuws. In M. Hooghe, K. de
Swert, & S. Walgrave (Eds.), Nieuws op televisie: televisiejournaals als venster op de
wereld (pp. 79-98). Leuven: Acco.
Walgrave, S., Joly, J., & Sevenans, J. (2019). The Belgian Agendas Project. In F. R.
Baumgartner, C. Breunig, & E. Grossman (Eds.), Comparative Policy Agendas: Theory,
Tools, Data (pp. 57-66). Oxford: Oxford University Press.
Walgrave, S., Joly, J., Van Assche, T., & Zicha, B. (2009). Comparative Agendas Project:
Dutch version of the Belgian codebook II (Version 13/11/2009) English translation.
Retrieved from
https://comparativeagendas.s3.amazonaws.com/codebookfiles/BelgianCodebook_Engl
ishTranslation.pdf
Walgrave, S., Lefevere, J., van Erkel, P., Jennart, I., Rihoux, B., Van Ingelgom, V., . . .
Deschouwer, K. (2019). Vlamingen en Walen stemden voor verschillende partijen maar
verschillen minder van mening over het beleid dat ze willen. Nota op basis van de
RepResent-studie. EOS RepResent consortium.
Walgrave, S., Zicha, B., Hardy, A., Joly, J., & Van Assche, T. (2014). Strong Devolution but
273
No Increasing Issue Divergence: Evolving Issue Priorities of the Belgian Political
Parties, 1987–2010. In C. Green-Pedersen & S. Walgrave (Eds.), Agenda Setting,
Policies, and Political Systems: A Comparative Approach (pp. 145-163). Aldershot:
University of Chicago Press.
Walsh, D. (2018). Territorial Self-Government as a Conflict Management Tool. Basingstoke:
Palgrave Macmillan.
Warrens, M. J. (2015). Five ways to look at Cohen's kappa. Journal of Psychology &
Psychotherapy, 5(4), 1.
Warwick, P. (1979). The durability of coalition governments in parliamentary democracies.
Comparative political studies, 11(4), 465-498.
Warwick, P. (1992). Ideological diversity and government survival in Western European
parliamentary democracies. Comparative political studies, 25(3), 332-361.
Warwick, P. (1994). Government Survival in Parliamentary Democracies. Cambridge:
Cambridge University Press.
Warwick, P., & Easton, S. T. (1992). The cabinet stability controversy: New perspectives on a
classic problem. American journal of political science, 122-146.
Watts, R. L. (2008). Comparing Federal Systems (3rd ed.). Montreal: McGillQueens University
Press.
Watts, R. L. (2013). Typologies of federalism. In J. Loughlin, J. Kincaid, & W. Swenden (Eds.),
Routledge Handbook of Regionalism & Federalism (pp. 19-33). New York: Routledge.
Watts, R. L. (2015). Can Federal Political Systems Accommodate National Minorities? In E.
M. Belser, A. Fang-Bär, N. Massüger, & R. Oleschak Pillai (Eds.), States Falling Apart?
Secessionist and Autonomy Movements in Europe. Bern: Stämpfli Verlag.
Wauters, B. (2005). Divisions within an Ethno-regional Party: the Volksunie in Belgium.
Regional and Federal Studies, 15(3), 329-352.
Wheare, K. C. (1963). Federal government. Oxford: Oxford University Press.
White, J. K. (1992). EE Schattschneider and the responsible party model. PS: Political Science
& Politics, 25(2), 167-171.
Wilford, R. (2010). Northern Ireland: The politics of constraint. Parliamentary Affairs, 63(1),
134-155.
Wils, L. (1998). Vlaemsche Commissie. In R. de Schryver, B. De Wever, G. Durnez, L. Gevers,
274
P. van Hees, & M. De Metsenaere (Eds.), Nieuwe Encyclopedie van de Vlaamse
Beweging (pp. 3504-3506). Tielt: Lannoo.
Wils, L. (2009). Van de Belgische naar de Vlaamse natie: Een geschiedenis van de Vlaamse
Beweging. Leuven: Acco.
Wils, L. (2020). Op zoek naar een natie. Het ontstaan van Vlaanderen binnen België.
Kalmthout: Polis.
Witte, E. (1990). Twintig jaar politieke strijd rond de abortuswetgeving in België (1970-1990).
Res Publica, 32(4).
Witte, E. (1993a). Language and territoriality. A summary of developments in Belgium.
International Journal on Group Rights, 1, 203-223.
Witte, E. (1993b). Language and territoriality. A summary of developments in Belgium.
International Journal on Minority and Group Rights, 1(3), 203-223.
Witte, E. (2009a). Breakthrough of a Liberal Constitutional State (1830-1848). In E. Witte, J.
Craeybeckx, & A. Meynen (Eds.), Political History of Belgium: From 1830 onwards
(pp. 19-59). Brussels: Academic and Scientific Publishing.
Witte, E. (2009b). The Expansion of Democracy (1885-1918). In E. Witte, J. Craeybeckx, &
A. Meynen (Eds.), Political History of Belgium: From 1830 onwards (pp. 97-141).
Brussels: Academic and Scientific Publishing.
Witte, E. (2009c). Increasing tension between the communities and the creation of a federalised
Belgium. In E. Witte, J. Craeybeckx, & A. Meynen (Eds.), Political History of Belgium:
From 1830 onwards (pp. 361-392). Brussels: Academic and Scientific Publishing.
Witte, E. (2009d). Inreasing Tension Between the Communities and the Creation of a
Federalised Belgium. In E. Witte, J. Craeybeckx, & A. Meynen (Eds.), Political History
of Belgium: From 1830 onwards (pp. 361-391). Brussels: Academic and Scientific
Publishing.
Witte, E. (2009e). Restoration and Renewal (1944-1950). In E. Witte, J. Craeybeckx, & A.
Meynen (Eds.), Political History of Belgium: From 1830 onwards (pp. 221-269).
Brussels: Academic and Scientific Publishing.
Witte, E., Craeybeckx, J., & Meynen, A. (2009). Political History of Belgium: From 1830
onwards. Brussels: Academic and Scientific Publishing.
Witte, E., & Van Velthoven, H. (1998). Taalpolitiek en -wetgeving. In R. de Schryver, B. De
Wever, G. Durnez, L. Gevers, P. van Hees, & M. De Metsenaere (Eds.), Nieuwe
Encyclopedie van de Vlaamse Beweging (pp. 2994-3043). Tielt: Lannoo.
275
Witte, E., & Van Velthoven, H. (2010). Strijden om taal: de Belgische taalkwestie in historisch
perspectief. Kapellen: Uitgeverij Pelckmans.
Wolff, S. (2005). Electoral Systems Design and Power-Sharing Regimes. In I. O'Flynn & D.
Russell (Eds.), Power Sharing: New Challenges for Divided Societies (pp. 59-74).
London: Pluto Press.
Wolff, S. (2013). Conflict management in divided societies: The many uses of territorial self-
governance. International Journal on Minority and Group Rights, 20(1), 27-50.
Wolff, S., & Yakinthou, C. (2012). Conflict Management in Divided Societies: Theories and
Practice. New York: Routledge.
Wolfsfeld, G. (2011). Making Sense of Media and Politics: Five Principles in Political
Communication. New York: Routledge.
Zeitzoff, T. (2017). How social media is changing conflict. Journal of Conflict Resolution,
61(9), 1970-1991.
Zhuravskaya, E., Petrova, M., & Enikolopov, R. (2020). Political effects of the internet and
social media. Annual review of economics, 12, 415-438.
276
277
Appendices
Appendix 1. Belgian cabinets under study
Cabinet
Start
Resignation
Composition
Vanden Boeynants II
20/10/1978
18/12/1978
BSP-PSB · CVP · PSC · VU · FDF
Martens I
03/04/1979
16/01/1980
CVP · PS · BSP · PSC · FDF
Martens II
23/01/1980
09/04/1980
CVP · PS · BSP · PSC
Martens III
18/05/1980
07/10/1980
CVP · PS · BSP · PSC · PVV · PRL
Martens IV
22/10/1980
02/04/1981
CVP · PS · SP · PSC
M. Eyskens
06/04/1981
21/09/1981
CVP · PS · SP · PSC
Martens V
17/12/1981
14/10/1985
CVP · PVV · PRL · PSC
Martens VI
28/11/1985
19/10/1987
CVP · PVV · PRL · PSC
Martens VII
21/10/1987
14/12/1987
CVP · PRL · PVV · PSC
Martens VIII
09/05/1988
29/09/1991
CVP · PS · SP · PSC · VU
Martens IX
29/09/1991
25/11/1991
CVP · PS · SP · PSC
Dehaene I
07/03/1992
21/05/1995
CVP · PS · SP · PSC
Dehaene II
23/06/1995
16/06/1999
CVP · PS · SP · PSC
Verhofstadt I
12/07/1999
19/05/2003
VLD · PS · PRL · SP · Ecolo · Agalev
Verhofstadt II
12/07/2003
11/06/2007
VLD · PS · MR · SP.a-Spirit
Verhofstadt III
21/12/2007
20/03/2008
CD&V · MR · PS · Open VLD · CDH
Leterme I
20/03/2008
30/12/2008
CD&V · MR · PS · Open VLD · CDH
Van Rompuy
30/12/2008
25/11/2009
CD&V · MR · PS · Open VLD · CDH
Leterme II
25/11/2009
13/06/2010
CD&V · MR · PS · Open VLD · CDH
Di Rupo
06/12/2011
25/05/2014
PS · CD&V · MR · SP.a · Open VLD · CDH
Michel I
11/10/2014
9/12/2018
N-VA · MR · CD&V · Open VLD
Sources: De Winter and Dumont (2014); Dumont and De Winter (1999). Any deviations from these works are based on my
own considerations (e.g. Leterme II started in 2009, not 2008). The FDF is no longer considered to be a distinct coalition party
after the establishment of the liberal MR (Mouvement Réformateur) of which it became part in 2002.
278
Appendix 2. Length (pages), source, and authors of the Dutch ‘overviews of Belgian political events’ in the
Res Publica yearbooks (1979-2006)
Year
Length
Res Publica
Author
1979
22
1980/3
Deweerdt, M.
1980
28
1981/2-3
Deweerdt, M.
1981
40
1982/2
Deweerdt, M.
1982
44
1983/2-3
Deweerdt, M.
1983
60
1984/4
Deweerdt, M.
1984
46
1985/2-3
Deweerdt, M.
1985
55
1986/3
Deweerdt, M.
1986
73
1987/3
Deweerdt, M.
1987
65
1988/2-3
Deweerdt, M.
1988
65
1989/3
Deweerdt, M.
1989
38
1990/2-3
Deweerdt, M.
1990
21
1991/3-4
D’Haese, R. & Van den Driessche, P.
1991
58
1992/3-4
Deweerdt, M. & Falter, R.
1992
47
1993/3-4
Deweerdt, M. & Falter, R.
1993
36
1994/3-4
Deweerdt, M.
1994
32
1995/3-4
Deweerdt, M.
1995
25
1996/3-4
Platel, M.
1996
54
1997/4
Deweerdt, M.
1997
65
1998/3-4
Deweerdt, M.
1998
77
1999/2-3
Deweerdt, M.
1999
81
2000/2-3
Deweerdt, M.
2000
101
2001/2-3
Deweerdt, M.
2001
98
2002/2-3
Deweerdt, M.
2002
105
2003/2-3
Deweerdt, M.
2003
116
2004/2-3
Deweerdt, M.
2004
111
2005/2-3
Deweerdt, M.
2005
104
2006/2-3
Deweerdt, M.
2006
95
2007/2-3
Deweerdt, M.
Total
1762
28 chapters
5 authors
279
Appendix 3. Self-rule scores with and without provinces (using the RAI data and methodology)
Year
Sub-states
Sub-states and
provinces(a)
Absolute
difference
Proportional
difference
1979
4.41
14.41
10.00
3.27
1980
7.19
17.19
10.00
2.39
1981
7.19
17.19
10.00
2.39
1982
7.19
17.19
10.00
2.39
1983
7.19
17.19
10.00
2.39
1984
7.20
17.20
10.00
2.39
1985
7.20
17.20
10.00
2.39
1986
7.20
17.20
10.00
2.39
1987
7.20
17.20
10.00
2.39
1988
7.20
17.20
10.00
2.39
1989
11.19
21.19
10.00
1.89
1990
11.19
21.19
10.00
1.89
1991
11.19
21.19
10.00
1.89
1992
11.19
21.19
10.00
1.89
1993
11.19
21.19
10.00
1.89
1994
11.19
21.19
10.00
1.89
1995
13.08
22.07
8.99
1.69
1996
13.08
22.07
8.99
1.69
1997
13.08
22.07
8.99
1.69
1998
13.08
22.07
8.99
1.69
1999
13.08
22.07
8.99
1.69
2000
13.08
22.07
8.99
1.69
2001
13.08
22.07
8.99
1.69
2002
13.97
22.97
8.99
1.64
2003
13.97
22.97
8.99
1.64
2004
13.97
22.97
8.99
1.64
2005
13.98
22.97
8.99
1.64
2006
13.98
22.97
8.99
1.64
Average
10.63
20.20
9.57
2.01
Note 1: The period shown equals the period covered by the analyses that use these scores (explanatory analyses; 1979-2006).
Note 2: Measurement of yearly scores: sum of the multiplications of self-rule scores of selected entities with their population
weight in Belgium. Sub-states include: the Flemish community (regionid= 402), the German-speaking community (404), the
Brussels-Capital Region (406), its predecessor (the Brussels Agglomeration, 407), and two peculiar categories that avoid
distortion in the peculiar Belgian context: the ‘combined authority’ for the French community and Walloon region (498) and
the ‘additional authority’ that grasps the authority of the Flemish and French communities over Dutch- and French-speaking
residents in the Brussels-Capital Region (499).
(a) This calculation is included in the RAI_country dataset (self-rule score for Belgium)
280
Appendix 4 Cabinet conflicts on ethno-territorial issues: yearly percentage of all cabinet conflicts (1979-
2018)
0
10
20
30
40
50
60
70
80
90
100
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Yearbook data News archive data
Decentralization deal
281
Appendix 5. Cabinet conflicts by sides: yearly percentage of all cabinet conflicts (yearbook data: 1979-2006; news
archive data: 2007-2018).
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Inter-segmental (full; one or two sides) Inter-segmental (partly) Mix or intra-segmental Unknown
Note: Michel I (2014–2018) only included one Francophone party (MR). Not all inter-segmental clashes in this
cabinet are considered conflicts ‘along segmental lines’ (and as such, coded as ethno-territorial conflicts). To
be conflicts ‘along segmental lines.
282
Appendix 6. Ethno-territorial cabinet conflict: yearly absolute frequency (1979-2018) (axes not adapted to
compensate for the quantitative difference between both datasets)
0
2
4
6
8
10
12
14
16
18
20
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Yearbook data News archive data
Decentralization deal
283
Appendix 7. Ethno-territorial cabinet conflict frequency: yearly equivalent by cabinet (1979-2018) (axes
not adapted to compensate for the quantitative difference between both datasets)
0
2
4
6
8
10
12
14
16
18
Martens I
Martens II
Martens III
Martens IV
M. Eyskens
Martens V
Martens VI
Martens VII
Martens VIII
Martens IX
Dehaene I
Dehaene II
Verhofstadt I
Verhofstadt II
Verhofstadt III
Leterme I
Van Rompuy
Leterme II
Di Rupo
Michel I
Yearbook data News archive data
Note: Including the periods in which the coalitions were caretaker cabinets. Vanden Boeynants
II not included in the analysis of yearbook data and Dehaene I not included in the analysis of
news archive data (as these datasets do not cover the full lifespan of these cabinets). Michel I
cabinet: end date equals resignation date, as this marks the end of the coding effort.
284
Appendix 8. Cabinet duration, number of ethno-territorial cabinet conflicts, and yearly equivalent of
ethno-territorial cabinet conflicts by cabinet (yearbook data)
Cabinet
Cabinet
duration
(days)
Number of
ethno-
territorial
cabinet
conflicts
Yearly
equivalent of
ethno-
territorial
cabinet
conflicts
Martens I
295
9
11.14
Martens II
116
3
9.44
Martens III
157
1
2.33
Martens IV
166
0
0.00
M. Eyskens
255
2
2.86
Martens V
1442
13
3.29
Martens VI
692
14
7.38
Martens VII
201
1
1.82
Martens VIII
1238
16
4.72
Martens IX
160
2
4.56
Dehaene I
1203
4
1.21
Dehaene II
1480
13
3.21
Verhofstadt I
1461
5
1.25
Total
8866
83
3.95
Note: Including the periods in which the coalitions were caretaker
cabinets. Vanden Boeynants II not included in the analysis of
yearbook data (as the datasets does not cover the full lifespan of
this cabinet).
285
Appendix 9. Cabinet duration, number of ethno-territorial cabinet conflicts, and yearly equivalent of
ethno-territorial cabinet conflicts by cabinet (news archive data)
Cabinet
Cabinet duration
(days)
Number of ethno-
territorial cabinet
conflicts
Yearly equivalent
of ethno-territorial
cabinet conflicts
Dehaene II
1480
34
8.39
Verhofstadt I
1461
23
5.75
Verhofstadt II
1623
33
7.42
Verhofstadt III
90
4
16.22
Leterme I
285
12
15.37
Van Rompuy
330
14
15.49
Leterme II
741
20
9.85
Di Rupo
1040
12
4.21
Michel I
1520
17
4.08
Total
8570
169
7.20
Note: Including the periods in which the coalitions were caretaker cabinets.
Dehaene I not included in the analysis (as the datasets do not cover the full
lifespan of this cabinet). Michel I cabinet: end date equals resignation date, as
this marks the end of the coding effort.
286
Appendix 10. Intense ethno-territorial cabinet conflicts (0 excluded): yearly absolute frequency (1979-
2018) (axes not adapted to compensate for the quantitative difference between both datasets)
0
2
4
6
8
10
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
News archive data Yearbook data
Decentralization deal
287
Appendix 11. Intense ethno-territorial cabinet conflict frequency: yearly equivalent by cabinet
(1979-2018) (axes not adapted to compensate for the quantitative difference between both datasets)
0
2
4
6
8
10
Yearbook data News archive data
Note: Including the periods in which the coalitions were caretaker cabinets. Vanden Boeynants
II not included in the analysis of yearbook data and Dehaene I not included in the analysis of
news archive data (as these datasets do not cover the full lives of these cabinets). Michel I
cabinet: end date equals resignation date, as this marks the end of the coding effort.
288
Appendix 12. Absolute frequency of cabinet conflicts on ethno-territorial issues, between segmental sides,
and of each type of ethno-territorial conflict (1979–2018)(a)
Year
Conflicts on
ethno-territorial
issue
Conflicts between
segmental sides
Thematic ETCs
Factionalized
ETCs
Full-fledged ETCs
YB
NwA
YB
NwA
YB
NwA
YB
NwA
YB
NwA
1979
11
8
3
0
8
1980
4
2
3
1
1
1981
2
2
0
0
2
1982
1
1
0
0
1
1983
5
4
1
0
4
1984
3
2
1
0
2
1985
3
2
2
1
1
1986
8
0
8
0
0
1987
7
1
6
0
1
1988
4
1
3
0
1
1989
2
2
1
1
1
1990
5
0
5
0
0
1991
3
4
1
2
2
1992
1
1
1
1
0
1993
0
0
1
1
1
1994
2
2
1
0
1
1995
0
3
0
0
1
3
3
0
0
0
1996
2
4
1
3
1
2
3
1
0
2
1997
1
10
3
4
0
7
0
1
4
3
1998
1
5
3
4
2
4
0
3
2
1
1999
4
11
4
7
1
4
0
0
1
7
2000
4
2
2
1
1
1
1
0
0
1
2001
2
2
1
0
0
2
1
0
0
0
2002
1
12
1
1
0
12
0
1
2
0
2003
0
6
1
3
0
3
1
0
2
3
2004
2
11
2
4
0
7
0
0
0
4
2005
2
3
3
3
3
0
0
0
8
3
2006
0
6
0
3
3
3
1
0
1
3
2007
9
0
9
0
0
2008
15
6
10
1
5
2009
15
2
13
0
2
2010
16
2
14
0
2
2011
2
1
2
1
0
2012
3
0
3
0
0
2013
4
0
4
0
0
289
2014
6
0
6
0
0
2015
3
0
3
0
0
2016
7
0
7
0
0
2017
3
1
3
1
0
2018
1
1
1
1
0
Total
80
159
53
46
43
123
16
10
37
36
(a) YB= yearbook data (1979-2006); NwA= news archive data (1995-2018)
290
Appendix 13. Yearly equivalent of cabinet conflicts on ethno-territorial issues, between segmental sides,
and of each type of ethno-territorial conflict by cabinet (1979–2018)(a)
Cabinet
Conflicts on
ethno-territorial
issue
Conflicts between
segmental sides
Thematic ETCs
Factionalized
ETCs
Full-fledged ETCs
YB
NwA
YB
NwA
YB
NwA
YB
NwA
YB
NwA
Martens I
11.14
8.66
2.47
0.00
8.66
Martens II
6.29
3.15
6.29
3.15
0.00
Martens III
2.32
0.00
2.32
0.00
0.00
Martens IV
0.00
0.00
0.00
0.00
0.00
M. Eyskens
2.86
2.86
0.00
0.00
2.86
Martens V
3.04
2.28
1.01
0.25
2.02
Martens VI
7.38
0.53
6.86
0.00
0.53
Martens VII
1.82
0.00
1.82
0.00
0.00
Martens VIII
3.83
1.77
2.95
0.88
0.88
Martens IX
2.28
4.56
0.00
2.28
2.28
Dehaene I
0.91
0.61
0.61
0.30
0.30
Dehaene II
1.73
7.15
2.47
4.19
0.74
4.19
1.48
1.23
0.99
2.96
Verhofstadt I
2.00
5.50
1.50
1.25
1.00
4.50
0.50
0.25
1.00
1.00
Verhofstadt II
7.42
2.47
4.95
0.00
2.47
Verhofstadt III
16.22
0.00
16.22
0.00
0.00
Leterme I
14.09
7.68
7.68
1.28
6.40
Van Rompuy
15.48
2.21
13.27
0.00
2.21
Leterme II
9.36
1.48
8.37
0.49
0.99
Di Rupo
4.21
0.00
4.21
0.00
0.00
Michel I
3.60
0.48
3.60
0.48
0.00
Total
3.01
6.77
1.89
1.96
1.73
5.24
0.62
0.43
1.28
1.53
(a) YB= yearbook data (1979-2006); NwA= news archive data (1995-2018)
Note: the calculation includes the periods in which the coalitions were caretaker cabinets to measure cabinet duration.
Vanden Boeynants II not included in the analysis of yearbook data and Dehaene I not included in the analysis of news archive
data (as these datasets do not cover the full lifespan of these cabinets). Michel I cabinet: end date equals resignation date,
as this marks the end of the coding effort.
291
Appendix 14. Poisson regression with monthly number of ethno-territorial conflicts as dependent variable
and VB (Vlaams Blok/Vlaams Belang) as a non-regionalist party (incidence rate ratios and Std. Errors)
Model 1a
Model 2b
Model 3b
Decentralization
0.87 ***
(0.03)
0.86 *
(0.07)
0.80 **
(0.09)
Regionalist party strength
(% of seats)
0.96
(0.06)
0.91
(0.08)
Regionalist party in coalition (dummy)
1.72 *
(0.24)
1.71 *
(0.25)
Decentralization saliency
0.99
(0.12)
Election proximity
1.02 *
(0.01)
Number of non-ETCs
1.02
(0.11)
Number of ETCs in previous month
1.20
(0.18)
N
336
336
335
AIC
429.45
428.21
429.49
BIC
437.08
443.48
460.00
Pseudo R²
0.06
0.08
0.10
*** p < 0.001; ** p < 0.01; * p < 0.05
Notes: intercepts not reported; model 3b: VIF = 6.75 (decentralization) and 6.73 (regionalist party
strength)
292
Appendix 15. Logistic regression models with VB (Vlaams Blok/Vlaams Belang) as a non-regionalist party
DV: Ethno-territorial
Model 4a
Model 5b
Model 6b
Decentralization
0.86 ***
(0.79 – 0.92)
0.80 **
(0.68 – 0.94)
0.71 **
(0.58 – 0.86)
Regionalist party strength
(% of seats)
0.91
(0.79 – 1.05)
0.89
(0.75 – 1.07)
Regionalist party in coalition
(dummy)
1.93 *
(1.01 – 3.68)
2.17 *
(1.11 – 4.23)
Decentralization saliency
1.21
(0.90 – 1.63)
Election proximity
1.02
(01.00 – 1.05)
N
322
322
322
R² Tjur
0.050
0.068
0.084
DV: Segmental side(s)
Model 7a
Model 8b
Model 9b
Decentralization
0.93
(0.85 – 1.02)
0.99
(0.82 – 1.21)
0.90
(0.70 – 1.18)
Regionalist party strength
(% of seats)
1.03
(0.87 – 1.23)
1.19
(0.93 – 1.57)
Regionalist party in coalition
(dummy)
2.01
(0.95 – 4.16)
2.83 *
(1.24 – 6.43)
Decentralization saliency
1.77 **
(1.23 – 2.57)
Election proximity
1.02
(0.99 – 1.05)
N
322
322
322
R² Tjur
0.008
0.021
0.060
DV: Ethno-terr. issue
Model 10a
Model 11b
Model 12b
Decentralization
0.80 ***
(0.74 – 0.87)
0.75 **
(0.63 – 0.88)
0.71 **
(0.57 – 0.88)
Regionalist party strength
(% of seats)
0.91
(0.78 – 1.05)
0.88
(0.73 – 1.07)
Regionalist party in coalition
(dummy)
1.79
(0.90 – 3.51)
1.84
(0.92 – 3.66)
Decentralization saliency
1.03
(0.73 – 1.42)
Election proximity
1.01
(0.98 – 1.04)
N
319
319
319
R² Tjur
0.090
0.106
0.108
*** p < 0.001; ** p < 0.01; * p < 0.05
293
Notes: intercepts not reported; model 6b: VIF = 6.38 (decentralization) and 6.65 (regionalist party strength); model
9b: VIF = 7.98 (decentralization) and 10.25 (regionalist party strength); model 12b: VIF = 6.16 (decentralization) and
5.92 (regionalist party strength)
294
Appendix 16. Outcomes of ethno-territorial conflicts: frequency by conflict intensity level (1979-2006; N=
95)
Outcome
Intensity level
0
1
2
Total
N
%
N
%
N
%
N
%
Compromise
9
32.14%
8
22.22%
6
19.35%
23
24.21%
Imposition (non-segmental)
0
0.00%
4
11.11%
1
3.23%
5
5.26%
Imposition (segmental)
0
0.00%
4
11.11%
0
0.00%
4
4.21%
Resignation (member)
0
0.00%
0
0.00%
3
9.68%
3
3.16%
Resignation (party)
0
0.00%
0
0.00%
1
3.23%
1
1.05%
Resignation (cabinet)
0
0.00%
0
0.00%
4
12.90%
4
4.21%
Residual
1
3.57%
0
0.00%
1
3.23%
2
2.11%
Blow over
15
53.57%
12
33.33%
10
32.26%
37
38.95%
Procedural
3
10.71%
8
22.22%
5
16.13%
16
16.84%
Total
28
100.00
%
36
100.00
%
31
100.00
%
95
100.00
%
295
Appendix 17. Outcomes of ethno-territorial (ETCs) vs. other conflicts (non-ETCs) by outcome type and
applicability of functionality dimensions
Conflict
outcome
type
Functionality dimension
Cabinet
survival
Cabinet
members’ and
parties’
survival
Decision
Substantive
Consensus
Non-
ETC
ETC
Non-
ETC
ETC
Non-
ETC
ETC
Non-ETC
ETC
Non-
ETC
ETC
Procedural
7%
17%
7%
17%
7%
17%
7%
17%
7%
17%
Compromise
24%
24%
24%
24%
24%
24%
24%
24%
24%
24%
Imposition
15%
9%
15%
9%
15%
9%
15/15%
9/9%
15%
9%
Resignation
1/3%
4/8%
0/3%
0/8%
0/3%
1/8%
0/3%
1/8%
0/3%
0/8%
Blow over
50%
40%
50%
40%
50%
40%
50%
40%
50%
40%
Residual
1/1%
2/2%
1/1%
2/2%
1/1%
2/2%
1/1%
1/2%
0/1%
0/2%
Total % of
conflicts
98%
96%
97%
92%
47%
53%
40%
35%
31%
41%
Note: Blue background: outcome type meets dimension. White: dimension not met. Light blue: dimension might be
met (and is met in the reported number of cases).
The functioning of Belgium relies on the power-sharing of Flemish and Francophone
politicians. As this collaboration often proves to be particularly hard, the governability
of Belgium and the feasibility of power-sharing is regularly questioned. In this respect,
observers and scholars typically point towards the impact of the decentralization reforms
that transformed Belgium into a peculiar federation. Decentralization is often promoted
as a tool for conflict management, but according to the ‘paradox of federalism’, it might
exacerbate tensions and undermine stability. As the debate goes on, a solid empirical
view on the ethno-territorial clashes that burden everyday power-sharing is yet to be
presented. This doctoral dissertation aims to address this lacuna. It sheds unprecedented
light on over 1,000 cabinet conflicts in Belgium (1979-2018). Studying their
characteristics, evolutions, outcomes, and several conflict determinants, two questions
are tackled. (1) How did the ethno-territorial conflict in Belgian power-sharing cabinets
evolve? (2) And how did decentralization impact this conflict? The results disprove
the paradox thesis on all levels. There are fewer ethno-territorial clashes than in the
past, the ones that still arise have a friendlier face, the pacification trends are statistically
related to decentralization, and hollowing out the central level did not burden the process
of conflict resolution in the long run. In making sense of these results, the author
provides some counterweight to conventional views on bipolar federalism, the political
incentives in split party systems, and the idea that the Belgian federal level has been
emptied to a problematic degree.
