1On November 4, 2018, the New Caledonian citizens voted about their accession to full sovereignty. The result was negative, but two other referendums may be held on the same question. In case of triple No, this territory will retain a unique status within the French Republic, symbolised by a broad field of territorial legislative powers, and by a territorial citizenship whose impacts are crucial, particularly in terms of employment and voting rights. From this point of view, France, which has long been the icon of the unitary and centralized state, would appear to be at the forefront of regional autonomy.

2However, New Caledonia's situation is the result of quite exceptional social, cultural and political tensions. The referendums we are discussing here have been held in Corsica (2003), Guadeloupe (2003), Martinique (2003 and 2010), French Guiana (2010) and Alsace (2013), and they had a much more modest purpose. They did not deal with the independence issue, and it is ambiguous to categorise them as referendums on autonomy. From a normative point of view, they were not aimed at providing territorial legislative and/or fiscal powers. The scope was to merge the regional institution and the departmental one(s) in order to create a ‘single authority’1. From a political point of view, we can talk about a regional empowerment since the competences and resources of one or two departments would have been aggregated with those of the former regional institution. In other terms, regardless of the intention, the project implied an increase in regional administrative autonomy and a further move towards differentiated regionalism.

3Given the regionalist feelings and mobilisations in these territories, we cannot talk about a merely technical-administrative project. These referendums raised general questions about the situation of these territories within the Republic, and the relations between the national and the regional level.

4These consultations were largely unsuccessful, and we cannot made a serious evaluation of the single authority established since 2016 in French Guiana and Martinique. Nevertheless, this sequence of six referendums on the same subject, in the same country and in a relatively short lapse of time, is a worthwhile scientific material. It allows me to address the three main questions of this special issue: that of the regional mobilisation in favour of a referendum; that of the acceptance process from the point of view of the state; and that of the approval of the political project from the point of view of the population. However, it has many other interests.

5First, this is a special kind of project. While municipal mergers are a very common situation2, I do not know any other case of mergers between provincial and regional levels. Merging authorities of various levels is not exceptional3, but this relates to the municipal and provincial levels, and the merger is almost never submitted to a referendum.

6Second, the stakes in the redistribution of local power are higher than at the level of municipal mergers. The loss of one or two departmental authorities would deprive many politicians and bureaucrats of very high positions. In theory, this was likely to strengthen the resistance to these projects.

7Third, these projects have had a national dimension, because they involved a redefinition of the relationship between the state and the territory. The projected single authorities entailed:

8- introducing exceptions to the French unitary principles, according to which institutional structures must be identical as a consequence of the equality of citizens;

9- establishing a new regional institution, with much broader competences and resources than the former;

10- founding a unified, theoretically more legitimate and influential regional representation, and this point was particularly relevant for regionalist organisations.

11Fourth, this sequence provides a new perspective on the state's involvement in local referendums, which is quite ambiguous. On one hand, the question asked and the decision to hold the referendum are defined at the national level; on the other hand, except in the first case, the government did not campaign at all. We can at most talk about partial self-determination, especially when the referendum was not legally binding (in the case of Corsica).

12Fifth, this sequence highlights a little-known function of referendums. Referendums have been conceived above all as essential legitimization tools in specific situations such as the adoption of a Constitution4; as tools used by the elites ‘to change the distribution of power between their respective institutions’,5 as tools ‘for mediation in a conflict’6, as ‘intermittent safety valves against perverse or unresponsive behaviour of representative institutions and politicians’7, etc. Far from it, our series refers to the stances of Dicey, who regarded the referendum as a useful instrument since it gave the people a veto power, in order to prevent abuses of power of the majority party8.

13Finally, this sequence raises the question of the status of direct democracy in France. In a republic traditionally resisting referendums9, it would suggest a significant change in political culture, a willingness to use referendums regularly, particularly to decide institutional issues. Far from it, the results probably led the governments to limit the possibilities for further consultations.

14In sum, this sequence of referendums is a double failure. First, the failure of consultations that could have strengthened differentiated regionalism, in a state that has long symbolised the unitary model. Second, the overall failure of the referendum route in France, which has enjoyed relative enthusiasm during this short period, but which is now likely to be used only in exceptional situations in the coming years.

15To show this, I cannot rely on quantitative analysis, since the number of municipalities is too low in Martinique (34), Guadeloupe (32) and French Guiana (22). Above all, I will focus on the strategies and interactions of the main categories of actors involved: government, regional elite and local elected officials.

1. The Republic against the referendum?

16Even if the use of referendums is not a hallmark of democracy10, the French Republic has been characterised by its reluctance, and the beginnings of the Gaullian period cannot be misleading. From 1963 to 2019, only six national referendums were held in France. However, since the 1990s, the mechanisms sing or imposing the use of local referendums have been developed, and suggest that there may be a real shift.

1.1 The burden of history

17During parliamentary debates under the Third Republic, the opponents of the referendum could rely on the constitutional law of 25 February 1875, according to which ‘legislative power is exercised by two assemblies’. At the local level, several municipalities have held consultations without a legal basis, but hence the government asked the prefects to nullify any deliberations calling a referendum, on the pretext that ‘only the elected representatives of the people have the right to administer'11.

18Actually, referendums were mainly excluded for political reasons. The method was conceived as a weapon against representative democracy, but also against the supremacy of the state, which could be challenged by local referendums.

19During the previous French regimes, direct consultation of the people had been promoted by the unimplemented Republican Constitution of 24 June 1793. This text was adopted by referendum, and gave the people a veto power over the legislation. The consultation of the people was then used by the authoritarian regimes of Napoleon I and III to strengthen their personal power12. That being said, for the notables who were seeking to consolidate the Republic, the referendum was from the outset a repulsive idea.

20At the end of the 19th century, the referendum was also undermined by the fact that it was supported by political actors that were extremely critical of parliamentarianism. On the right were the Boulangists, who seemed to reserve the referendum for the most important political issues13. On the left were the socialists, in particular the Parti ouvrier français, which wanted direct consultation of the people to be one of the essential bases of municipal power14.

21Nonetheless, the ruling Republican parties also excluded referendums in the name of popular incompetence. When in 1890, several right-wing deputies called for the establishment of the communal referendum, the rapporteur replied that the masses are ‘poorly educated’, and ‘vote out of interest and passion rather than reason’. Thus, voters cannot express themselves on ‘municipal issues that are sometimes very complex, very difficult, very sensitive’15.

22In 1958, General de Gaulle's comeback to power seemed to give the French Republic a very different orientation. Nevertheless, the possibility of holding local referendums was developed long after his resignation.

1.2 The development of referendum provisions under the Fifth Republic

23In his Mémoires d’espoir, De Gaulle writes, ‘the French people must decide for itself what is essential to its destiny’. Thus, unlike the previous ones, the Constitution of the Fifth Republic provides that sovereignty can be exercised by the representatives of the people or by referendum. The referendum can therefore be used to revise the Constitution (art. 89), but also to adopt a draft law, particularly with regard to the organization of public authorities (art. 11).

24De Gaulle himself showed a plebiscitary use of the referendum; the people did not only have to decide on the question asked, but also to reiterate its confidence in the Head of state16. Moreover, De Gaulle mainly used the referendum when he did not have a sufficient majority in Parliament to have a free hand. The French people was consulted three times during this phase (1959-1962), but it was no longer consulted between the legislative elections of November 1962, which gave a comfortable majority to De Gaulle, and the referendum of April 1969, when the negative vote precipitated his resignation. Thus, the first utility of the referendum was to bypass Parliament17, and it is easier to understand why it had no legal basis at the local level.

25The gap will gradually disappear. The 1971 law on municipal mergers authorized for the first time the local referendum, at the initiative of elected officials or the prefect. This procedure was used in only about 50 cases18.

26The decentralization process initiated in 1982 seemed to open up broader perspectives. It was based on the ideas of a change in relations between the state, territories and citizens, and the development of local democracy. However, French local representatives remained almost unanimously opposed to any institutionalisation of the local referendum, because it would fundamentally be ‘a matter of local culture’19.

27Thus, it took ten years to give a legal basis to the local referendum, without any binding dimension. First, the law of 6 February 1992 allowed local elected officials the right to organise a consultation. Then, the law of 4 February 1995 allowed citizens to propose a municipal or inter-municipal consultation, by a request signed by one fifth of the voters concerned.

28At the constitutional level, the 2003 revision revealed a strong interest in the development of local referendum provisions. Henceforth, all local authorities can hold a referendum as part of the exercise of their competences, but the binding value requires participation of at least 50% of registered voters.

29Other constitutional provisions have been established for rarer situations and have served as a basis for the six referendums studied here:

- According to article 72-4 of the Constitution, referendum is mandatory to allow an overseas authority to change its constitutional category, and optional for any statutory change.

- According to article 73, referendum is mandatory to merge department and region in the territories ruled by this article.

- According to article 72-1, referendum is optional in the case of the creation or statutory modification of a special status authority.

30The principle was laid down by President Chirac in his speech in Madiana on 11 March 2000: any overseas authority must be able to evolve towards a sui generis status, but the change must be validated by the population, whose vote is therefore binding. This orientation was severely criticised because of the unitary nature of the Republic. In its opinion of 10 October 2002, the Council of state stressed that the choice of a local population thus determined the exercise of sovereignty, which belongs to the whole people or the Parliament. In the case of Corsica, the result was not binding, but the government had assured that it would respect the popular choice. In conclusion, significant provisions were included in the Constitution, but this did not allow us to predict their use.

2. Why these referendums?

31With the exception of Corsica, referendum was always required to validate the merger project, either by law (Alsace) or by the Constitution (Guadeloupe, Martinique, French Guiana). However, juridical requirements are far from being a sufficient factor.

2.1 A complex decision-making process

32According to Altman, there are three main types of referendums: 1/ 'required by Constitutional (or law)', 2/ 'top-down (or from above)', and 3/ 'citizen-initiated (through collection of signatures)'20. Yet, in our cases the decision to hold these referendums proceeded from all three dimensions at once.

33The organization of the referendum may be constitutionally mandatory (arts. 72-4 and 73) or not (art. 72-1). On the other hand, the decision to hold the referendum is always made at the state level, either by the President upon proposal from the government (arts. 72-4 and 73) or by law (art. 72-1), and the decisive question is not the obligation to hold a referendum, but the prior acceptance of the project.

Table 1. Juridical frameworks

	Constitutional basis	Legal basis	Referendum status	Deciding authority
Corsica	Art. 72-1	Act of 10 June 2003 organising a consultation of Corsican voters	Constitutionally and legally optional	Parliament
Guadeloupe	Art. 73		Constitutionally mandatory	Head of state
Martinique/1	Art. 73		Constitutionally mandatory	Head of state
Martinique/2	Art. 73		Constitutionally mandatory	Head of state
Guiana	Art. 73		Constitutionally mandatory	Head of state
Alsace	Art. 72-1	Act of 16 December 2010 on territorial reform	Constitutionally optional Legally mandatory	Government

34There is no obligation on the government to accept, but its choice is influenced by local demand. In the event of weak demand, the risk of very low turnout and rejection would be maximal. This condition is crucial in our six cases, where the demand for merger was predominant among the regional elites.

35The case of overseas authorities in 2003 highlights the government's discretionary power. In Martinique and Guadeloupe, the congress of regional and departmental elected representatives voted largely in favour of the merger. In French Guiana, the congress did the same, but on the one hand the departmental council – where the left had a large majority – then voted against it, and on the other hand the right-wing national government included a Guyanese member who was hostile to this reform. In Corsica, the departmental opposition had not led the government to withdraw the project, while in French Guiana it has been used to justify the status quo.

36In Corsica, the idea of abolishing departmental councils was first advocated by the nationalists, who denounced them as a clientelist tool. In addition, these institutions were elected by majority system, which was very unfavourable to the same nationalists. In 2000, the vast majority (44/51) of the Corsican Assembly supported the idea of a merger, which could only be achieved after a constitutional amendment. Yet, this consensus faded after the 2002 presidential elections. The majority of regional elected representatives (33/51) and parliamentarians (4/6) remained pro-merger, but the departmental councils were opposed. Therefore, the government opted for referendum.

37In Alsace, the advocates of the merger seemed almost hegemonic21. In 2011, regional and departmental elected officials supported it with 101 votes in favour, one against and 19 abstentions. In 2012, 108 elected representatives voted in favour, with five against and nine abstentions.

38Finally, with regard to the 2010 consultations in Martinique and French Guiana, almost all of the regional elite were in favour of change. Nevertheless, change could take two very different orientations. The main issue was not the merger but a change in constitutional category from article 73 to article 74. Here too, a ‘single authority’ would have ruled the territory, but the difference was huge in terms of normative power: only Article 74 allows the exercise of very broad territorial legislative powers, and this perspective was far from unanimous.

2.2 The heterogeneity of pro-referendum coalitions

39In our six cases, the holding of a referendum was conditioned by the political actors who backed the merger project; the pro-status quo normally had no interest in requesting a consultation that they could lose. However, there is not a homogeneous group, which wants a referendum for one reason only. On the one hand, the demand does not only come from local actors; the state can get involved. On the other hand, there are stated and hidden reasons. Strengthening policy coherence is a reason of public interest; strengthening personal power is a reason that can hardly be publicised. Among the motivations, I point five essential ones:

a/ A symbolic objective from the state's point of view: such referendums are supposed to demonstrate the reality of a paradigm shift, and the operationality of the related provisions.

b/ A technical-financial objective from the point of view of both the state and regional elites: merger is supposed to increase the coherence and efficiency of public action, and limit costs.

c/ A regionalist objective, which can be promoted at the regional level but also at the state one: merger is supposed to increase the power and identity of the regional level.

d/ A tactical objective, especially from the point of view of regional and local elites: to maximize political influence if the referendum result is positive.

e/ A tactical objective from the state’s point of view: the choice of referendum can give a positive image of the political authority, and avoid settling a tense regional debate. It would not be based on a conviction, nor on the desire to maximize benefits, but on the opportunity to minimize risks. We can call this a dismissive strategy.

40The heterogeneity of the pro-merger coalitions is even more striking when one looks at the partisan affinities. It would be easy to believe that a government is especially inclined to accept the request when it comes from its political allies. However, except for the decision not to hold a referendum in French Guiana in 2003, these affinities did not play a decisive role. On the one hand, there is never a total congruence between the three territorial levels. In 4 out of 6 cases, the pro-merger regional majority was politically opposed to the government. In Martinique (2003), the latter decided to hold a referendum even though its regional supporters were largely opposed to the merger.

41On the other hand, the different political families rarely show a homogeneous position. Unity of vision is difficult to achieve, particularly because many political and administrative actors have a lot to lose personally in the merger process, especially when they belong to small authorities22.

Table 2. Partisan congruence in referendums23

	Corsica	Guadeloupe	Martinique/1	Martinique/2	Guiana	Alsace
State	Right	Right	Right	Right	Right	Left
Region	Right	Right	Regionalist	Regionalist	Left	Right
Departement (1)	Left (Corse-du-Sud)	Left	Left	Left	Left	Right (Bas-Rhin)
Departement (2)	Left (Haute-Corse)	-	-	-	-	Right (Haut-Rhin)
Regional pro-merger coalition	4/5 Right 2/5 Left 5/5 Regio.	5/5 Right 2/5 Left	4/5 Left 1/5 Right 5/5 Regio.	5/5 Left 3/5 Right 5/5 Regio.	5/5 Left 4/5 Right 3/5 Regio.	4/5 Right 2/5 Left 5/5 Regio.

3. An overall failure

42The six consultations studied here produced mixed, but always negative results from the perspective of reform supporters. I will focus here on the main causes of this failure.

3.1 Negative outcomes and citizen disaffection

43One of the major conclusions of the literature on referendums in Switzerland is the ‘status quo bias’, with women's right to vote as an emblematic example24. The referendum would strengthen the opponents of change.

44The cases studied here are in line with this. In three cases, the No won, narrowly twice, once from afar. In the other three cases, the Yes won, but with very low turnout. In French Guiana and Martinique (2010), the merger process was initiated, then concretized in 2016. On the other hand, in Alsace, three of the four legal conditions (the Yes had to get both more than 50% of the votes cast and 25% of the registered voters in each department) were not fulfilled.

Table 3. The referendum results

	Corsica	Guadeloupe	Martinique/1	Martinique/2	Guiana	Alsace
Vote Yes (on the votes cast)	49.02%	27.02%	49.52%	68.33%	57.49%	57.65% 44.26% (Haut-Rhin) 67.53% (Bas-Rhin)
Turnout	60.52%	50.34%	43.96%	35.78%	27.42%	33.96% 37.18% (Haut-Rhin) 35.11% (Bas-Rhin)
Quorum	No	No	No	No	No	Not reached
Outcome	Status quo	Status quo	Status quo	Merger	Merger	Status quo

45Where the Yes has won, low turnout suggests democratic failure. The use of referendums does not provide the expected level of legitimacy, and the new authorities will be politically weaker.

46In general, participation in referendums is lower than in parliamentary elections, but variations can be very significant.25 This is the case here. In the first three referendums, turnout was close to that of previous presidential and regional elections; in the last three, it was much lower.

Table 4. Turnout in referendums and previous elections26

	Corsica	Guadeloupe	Martinique/1	Martinique/2	Guiana	Alsace
Referendum	60.52% (2003)	50.34% (2003)	43.96% (2003)	35.78% (2010)	27.42% (2010)	33.96% (2013)
Previous presidential elections	67.78% (2002)	38.87% (2002)	45.11% (2002)	65.78% (2007)	63.47% (2007)	80.12% (2012)
Previous regional elections	67.48% (1999)	54.60% (1998)	45.45% (1998)	55.06% (2007)	50.50% (2007)	51.09% (2010)

47According to Kriesi, the major variables of participation are issue complexity and campaign intensity27. For example, in Andalusia, where 56% of the population declared themselves unable to assess the status submitted to the vote in 2007, and where the two former hegemonic parties supported that text, the turnout was 35.9%28. In our six cases, first, the question sounded simple, but the implications of the Yes were difficult to grasp. Second, it is clear that campaign intensity was very unequal.

48The Corsican referendum was the one with the highest turnout. Institutional discussions between the government and Corsican elected officials began at the end of 1999, which strengthened the information and sensitivity of citizens. Above all, the campaign was more intense. On the one hand, this is the only case where the government campaigned in support of the merger. Interior Minister Nicolas Sarkozy has visited Corsica eight times and has therefore strongly personalised the campaign, which was likely to foster traditional political cleavages, and hence turnout. On the other hand, the abolition of the departmental institution could more easily be regarded as a risk, especially for local elected officials, since the municipalities, very numerous and often very small, were very dependent on departmental subsidies29. Thus, the issue salience could have been stronger than elsewhere, both for the local political class and the population, which would have encouraged mobilisation.

3.2 The mistakes of national authorities

49It is difficult to assess the overall factors of failure. On the voters' side, attachment to the old (founded in 1790) departmental institution remains high, especially in rural areas30. However, the failure largely stems from political actors, and first and foremost from national authorities. In theory, since the same authorities proposed the possibility of merging, they should seek to stimulate participation and not to disadvantage the vote Yes. In practice, this was far from being the case.

50Governments have always opted for a short campaign. The referendum date and the question asked were unveiled between two and three months before the polls and this seems to have disadvantaged the Yes side. According to Marquis and Bergman, the quality of voter information is a function of the intensity of the campaign but not its length31. Nonetheless, this conclusion cannot be accepted without considering issue complexity. The impacts of the merger were difficult to assess, as demonstrated by the literature on the subject, whose results are very nuanced and ambivalent32. Thus, when the benefits of a reform are difficult to understand, it seems useful to give the Yes side time to develop its arguments.

51The referendums in Martinique (2010) and French Guiana are a first special case. The official campaign opened only six days before the first referendum, and above all the choice to hold two successive consultations had a significant impact. The question asked on 10 January was more divisive, and therefore likely to generate a higher turnout, but the succession of polls was also likely to lead to a ‘voter fatigue’33, and certainly had a negative effect as well. On 10 January, participation had reached 55.32% in Martinique and 48.16% in French Guiana; on 24 January, it was about 20 points lower in both cases. The result was not in question, but the legitimacy of the new single authority was weakened.

52Alsace is a second special case, because it is the only one where the law imposed restrictive conditions. In each department, the Yes vote had to get a majority of the votes cast accounting for at least 25% of the registered voters. The last condition, called 'approval quorum', protects the status quo even more than the 'participation quorum'34, and it has not been fulfilled in any department.

53Finally, Corsica is a third special case, because it is the only one where the government has been strongly involved in favour of reform. It is likely that this involvement had the main effect of stimulating opponents, especially since the concomitant pension reform was unpopular35. In EU integration referendums, voter behaviour is driven more by European concerns than by national ones, but the government popularity does not lose any influence36. In our tight referendums with unclear stakes, this last factor was probably decisive.

3.3 The inability of the Yes side

54Although it was the most representative in any case, the Yes side was unable to prevent major defections, and unable to convince voters. Thus, where at least one survey was held, the final result was lower for the Yes vote.

55Of course, polls can be wrong, and ‘the most volatile referendum campaigns will be those in which there is little partisan, issue or ideological basis on which voters might tend to form an opinion easily. […]. Voters cannot be expected to have well-formed opinions on a issue that has not previously been a subject of public debate in other political arenas’37. Our referendums are in line with this situation, which favours volatility, but the trend was always in favour of the No side.

Table 5. The collapse of the Yes vote

	Corsica 2003	Guadeloupe 2003	Martinique 2003	Alsace 2013
Poll N-6	62%
Poll N-5	55%
Poll N-4	54%
Poll N-3	54%
Poll N-2	55%	63%	77%
Poll N-1	51%	39%	70%
Last poll (N)	55%	30%	59%	71% (Haut-Rhin) 78% (Bas-Rhin)
Final results	49.02%	27.02%	49.52%	44.26% (Haut-Rhin) 67.53% (Bas-Rhin)

56On the one hand, the parties committed to the Yes side have almost all experienced significant defections. Even when the Yes leader is not publicly challenged, his alleged supporters may refrain from campaigning or supporting the No, because of their conviction and/or local interests. However, in front of an aggressive ‘No’ campaign, the only response of the Yes side has often been to personalise the issue around the main promoter of the reform: Interior Minister Sarkozy for Corsica, or Regional Presidents Michaux-Chevry and Richert for Guadeloupe and Alsace. Nonetheless, this personification can encourage apathy out of comfort, or the repositioning of actors previously committed to the project, to avoid that the leader gains a too strong political position.

57In Corsica, all the right-wing regional leaders supported the Yes, and the government mobilised in this direction; yet, many right-wing departmental and municipal elected officials voted for the No38. In Alsace, President Richert was unable to mobilise effectively his party, particularly in the Haut-Rhin. Even worse, the mayor of Colmar, the capital of the Haut-Rhin, initially cautious, became more and more strongly opposed to the merger. Thus, the Yes vote obtained 10.26% fewer votes than Richert's list at the 2010 regional elections.

58The regional opposition is necessarily more inclined to criticize the majority's choices. In Guadeloupe, the socialists, who dominated the General Council and strongly supported the principle of merger, denounced the governmental project, abstained from voting at the congress of elected representatives in order to allow the holding of the referendum, and led the No campaign.

59On the other hand, the Yes side failed to convince voters, and here too the personalisation strategy had mainly negative effects. When the project seems essentially linked to a person, voters are driven more by general political factors, and less by the question asked. For some voters, referendum is an outlet for dissatisfaction39, and in our cases, the judicial sentences of L. Michaux-Chevry and the unpopular pension reform promoted by N. Sarkozy probably favoured the No vote.

4. What lessons can be learned?

60Among the key lessons of these six consultations, I mainly focus on the lack of political congruence between the elites and the population, the absence of satisfying political debates around the project, and the significant victory of localisms over regionalisms.

4.1 Non-congruence between elites and population

61For a long time, we know that the population do not necessarily share the almost unanimous positioning of the elites40. Despite this, in Switzerland, the people most often adopt the same position as the elites. Over the period 1947-1995, the people followed the government’s vote recommendation in 77% of cases41, over the period 1995-2010 in 80.7%42.

62Our six referendums are a far cry from that43, especially since there was a double lack of congruence. Since the majority of the most representative elected representatives supported the change, we can talk about an inability to convince both of the project benefits, and of the issue-saliency (which is decisive for turnout)44.

63Under the Fifth Republic, national referendums are often characterised by the accessibility of the question and/or the high salience of the issue. Moreover, the only referendum lost by De Gaulle was the one most likely to disturb the French local system, and therefore to arouse the mistrust of local elected officials, including on De Gaulle’s political side. In sum, regarding our referendums, the Yes side was from the beginning in a situation of disadvantage. The merger project disturbed the same local system, and did not meet strong social demand.

4.2 Low quality debates

64The quality of democratic debate, while difficult to assess, is an essential issue since ‘one dimension of legitimacy is the quality, accessibility, accuracy and truthfulness of information circulating in the public sphere’45. The more incomplete or falsified the information, the more difficult the formation of voters’ preference, and this can have decisive impacts.

65The consultations we are studying were announced late by the government, and voters may have felt an attempt to force the decision, which probably had a negative influence on the vote Yes. This rapidity was all the more inappropriate in the case of a reform whose benefits did not seem prominent, and whose effects were uncertain.

66When the campaign is short, the voter has less time to gather information, and this seems to reinforce the No, which theoretically has an ‘asymmetric advantage’. ‘No campaigners do not necessarily have to make a coherent or persuasive case against a proposal. Risk aversion can be a powerful basis on which to appeal some voters in and of itself, and the No side of a referendum campaign usually (not always) hold a monopoly on this emotion’46. Practically, the theme of fear dominated our six campaigns. The most symbolic example was the leaflet distributed in Martinique in 2003, according to which the Yes vote was a commitment to independence, to the ‘return to colonialism’, to additional taxes, to a ‘single and dictatorial power’, etc., and entailed the loss of social benefits, of purchasing power, of 40% of civil servants' positions, and so forth47.

67While not so extreme, No supporters have always used similar and unprovable statements. For example, according to the socialist leader of Guadeloupe, the single authority would imply ‘the hegemony of a tyrannical majority’, ‘a plan to dismiss contract workers and freeze competitions for a long time’ in the civil service, and even ‘the impossibility of running freely in elections’48. Similarly, according to the communist departmental secretary of the Haut-Rhin, the supporters of the single Alsatian authority wanted to break up French unity, reduce public services in order to leave more room for the private sector, and curtail social rights through an alignment with the Swiss and German models49.

68Finally, one of the main arguments of the No side was the loss of political proximity due to the abolition of the departmental council, especially where the region was to merge with two departments – in Alsace and Corsica – because this meant more losers among the elites, and more fears of regional centralisation. Consequently, in the case of Corsica the government had proposed two ‘territorial councils’ – without legal personality –, which would have been in charge of ‘implementing the policies of the single authority’. In Alsace, the congress of elected officials proposed that the Alsatian Assembly would have been located in Strasbourg (Bas-Rhin), and the executive council in Colmar (Haut-Rhin).

69Nevertheless, it was mainly in the departmental capitals that this concern had a powerful impact. In Bastia, the capital of Haute-Corse, the No won by more than 70%; in Colmar, the capital of Haut-Rhin, the No won by 67%. In the small municipalities allegedly the most institutionally fragile, and therefore the most likely to vote No, other factors shaped the voters’ choice.

4.3 Localisms won

70In theory, for voters wishing to strengthen the regional level, the only possible option was to vote Yes. However, the prospect of merger did not satisfy the most convinced regionalists. In Corsica and Alsace in particular, many abstained or did not campaign50.

71Above all, the actors opposed to regionalist logic had very valuable assets. On the one hand, the attachment to departmental councils, which appeared almost two centuries before the regional ones, could not be minimized. On the other hand, the actors who feared a loss of their status could not be ignored, especially since they were probably the most committed campaigners. Finally, the links between the municipalities and the departmental councils could not be underestimated. In a system based on intermediation and arrangement, departmental councils played a crucial role vis-à-vis the state51. Since the decentralization of 1982, they have become essential and autonomous bodies for allocating resources. Thus, all the municipalities satisfied with this system of intermediation and allocation, largely based on personal ties, were likely to favour the No.

72Since the financial dependence of municipalities on the departmental council is typically a function of their own resources, it could be expected that the smallest municipalities - and therefore the most deprived - would be the most likely to vote No. In this sense, in Norway, hostility to mergers was significantly higher in small municipalities52. However, our six cases do not confirm this hypothesis. Either the correlation is not significant or the impact is nil. In Corsica, the No won because of the opposition of the two largest municipalities.

73Consequently, localism has defeated regionalism, but localist behaviour is not exclusive to small municipalities, and is not synonymous with No vote. In Corsica, the majority of small municipalities (less than 300 inhabitants) voted Yes, but this cannot be conceived only through the preferences formation. This result is largely due to the personal attachment of many voters to their mayor or Member of Parliament. According to Paul Giacobbi, MP and President of the departmental council of Haute-Corse, the No would have largely won if not for these attachments53.

5. Mainly national impacts?

74Drawing common lessons from these six consultations is all the more difficult because we have too short experience with implemented mergers. My main point is that the impacts of these referendums have usually been low or moderate at the regional level; conversely, this sequence may have had a more significant impact at the national level.

5.1 Regional impacts: small winners, small losers?

75At the regional level, it is too early to draw conclusions from the merger in Martinique and Guiana, but the consequences of these referendums were very diverse. Above all, the institutional impacts were much more salient than the impacts on the party systems.

Table 6. Approximation of the referendums impacts

	Corsica	Guadeloupe	Martinique/1	Martinique/2	Guiana	Alsace
Institutional impacts	Low	High	Very Low	Uncertain	Uncertain	Very high
Party system impacts	Moderate	Low	Moderate	Moderate	Low	Low

76In Guadeloupe, the referendum had a strong impact, since the broad victory of the No side legitimised the status quo. In Alsace, it was even stronger because of the government's subsequent choices. The Alsace Region was integrated in 2015 into the new Grand Est Region, which also includes the former Lorraine and Champagne-Ardenne Regions. If the 2013 referendum had been successful, it would have been almost impossible to impose this integration two years later. Thus, the 2013 vote allowed the dilution of Alsatian regional status.

77It is especially noticeable, as with European integration referendums54, that the people’s choices are easily reconsidered:

- In Martinique, elected officials were very easily able to reopen the institutional debate, less than five years after the short defeat of 2003, and justify the organisation of a second referendum on the same theme in 2010.

- In Corsica, the merger project, rejected by the population in 2003, was again supported by the Corsican Assembly from 2014, and approved by the government in 2015. The single authority was set up on 1st January 2018 without a referendum. The Corsican Assembly initially voted in favour of a second consultation, but agreed with the government, which officially dismissed it for reasons of speed.

- In Alsace, integration into the Grand Est region has caused very negative reactions55. From 2017, the presidents of the two departmental councils requested the merger of their institutions to establish a special-status authority, and the government tried to satisfy them. The law of August 2, 2019 provides for the creation of the ‘European authority of Alsace’ in 2021, without any new referendum. Nonetheless, it will be a single department with a few specific competences, and still integrated in the Grand Est region.

78At the level of party systems, the specific impacts of these referendums have been low, or salient only in the short term. In Corsica, nationalists and the right-wing regional majority were weakened, but the left, which had a large majority in the polls, was unable to rule the region after the 2004 territorial elections. The internal divisions of the left and right parties seem to have generated a zero-sum game. For the nationalists, which had a great interest in the merger project, four years after this failure they began a progression that made them the dominant political force in Corsica.

79In Guadeloupe and Martinique, after the 2003 referendums, the left remained in control in the departmental councils. On the other hand, the regional majorities had contrasting fortunes. In Martinique, the independentist president was strengthened by the 2004 elections, rising in the first round from 24.57% to 37.28% of the votes, particularly to the detriment of the right. On the other hand, in Guadeloupe the left has seized regional power and retained it until today. Nevertheless, President Michaux-Chevry's defeat was widely favoured by her judicial convictions in 2002, for favouritism in the award of public markets, forgery, and breach of trust. Despite this, she still won 37.62% of the votes in the first round.

80In French Guiana and Martinique, the 2010 double consultation only initially had a negative impact on regional majorities. In Martinique, the independentists defeat seems to be mainly due to the failure of the 10 January referendum, which was about their main priority, i.e. a change in constitutional status. The left, victorious in the 2010 regional elections, supported the merger as much as the independentists. In a second step, the independentists won the first elections to the single authority in 2015. As a result, the left, which formerly controlled the departmental council, is now the loser.

81In French Guiana, the first post-referendum regional elections, won by the right, seemed to mean a deep shift. However, here too, this victory was probably favoured by the No success in the 10 January referendum, and it did not have any lasting consequences. Since 2016, the left has held a hegemonic position in the new single authority, in which the right has no elected representatives.

82Finally, in Alsace, the right is still dominant, and the Socialist Party collapse in 2015 is not mainly due to its divisions with regard to the referendum; the socialist government's choice to integrate Alsace into the Grand Est Region was surely a much more powerful factor. Similarly, the progression of the radical right, opposed to the merger, seems mainly correlated to national factors56.

5.2 The national consequences: a dark future for the referendum in France?

83According to Altman, the referendum should be conceived as a democratic safety valve when veto actors are unable to agree57. The referendums imposed for mergers of local authorities in France have been designed in the opposite way. According to the parliamentary works for the 2003 constitutional revision, the purpose of these referendums was to provide a veto power to the population. While national referendums are a tool to legitimize change, these referendums have been framed as an obstacle to change. The point is not to note that the practice of referendums favours the status quo, as in Switzerland, but to assign this function to referendums. In our cases, this approach had powerful effects.

84Our six cases sound like referendums where there was almost nothing to win for the governments that decide on their organisation. At the same time, these were not just local referendums, because they seemed likely to call into question the government's credibility.

85In the end, they did not have any significant impact on the government, but their overall failure have probably encouraged national political actors to be even more cautious about referendum, and to back regressive provisions regarding local democracy. In order to impose a redrawing of the regional map, in front of reluctant elected officials and populations, the law of 16 January 2015 took a step backwards in referendum status in France. According to the government, the urgent need to redraw the regional map justified the abolition of the referendum requirement for merging regions, merging departments, and merging regional and departmental authorities. Both the left-wing majority and the right-wing opposition supported this amendment.

86Thus, the sequence of referendums studied here is far more of a ‘symbolic innovation’58 than a change in French political culture. Since the decentralization process initiated in 1982, the state has changed its relationship with the territories, given much greater powers and resources to local elected officials, and broken with the principle of institutional uniformity, without consulting the people. With the failure of this sequence, it is now likely that the merger referendum will be used exceptionally, when a project involves a significant political risk for the government (such as a ‘hands-washing mechanism’)59, or when the regional elites are almost unanimous.

Conclusions

87The sequence of referendums studied here is a failure on several levels:

- Failure of referendum promoters, who failed in four out of six cases, and were never able to defend their project efficiently.

- Failure in terms of democratic participation, which has been average for the first three consultations, and very low for the last three.

- Failure in terms of democratic deliberation, with campaigns dominated by the theme of fear and the spread of untruthful information.

- Failure of regionalists, unable to promote successfully an alternative vision against a very well established local system, in which the departmental institution is essential.

- Failure of the state authority, which found itself trapped in the provisions it had instituted, and which was unable to achieve reforms that it considered rather positive.

- Failure of the referendum tool in France, whose use will probably be even more restricted in the coming years.

88Despite the provisions introduced since the early 1990s, which strongly increase the referendum opportunities, France had remained reluctant to use it. Moreover, the new provisions hardly concealed this old reluctance. These polls are called ‘consultations’, not referendums, and the quorum requirements gave additional blocking power to minority actors. Thus, France is one of the European countries that most rarely uses the local referendum60, which is essentially ‘a communication tool for the benefit of French elected representatives, who use it to consolidate their legitimacy’61.

89A public demand exists. 72 percent of the French say they are in favour of the citizen-initiated referendum, and 70 percent consider that ‘there should be referendums on most important issues’62. However, the demand for referendums is generally higher among those opposed to mainstream parties63, which incites the latter to rule out the possibility.

90Usually a government decides to hold a referendum when the rules require it, when the political situation seems to impose it, to legitimize its choice, or to optimize its political position. On the other hand, these French provisions have been designed to give veto power to the population, thereby constraining all reformist intentions, both local and national. At the metropolitan level, this veto has already been abolished. Yet, for the overseas, it is constitutionalised, and the current government is not proposing its suppression.

91Of course, the sequence of referendums studied here did not mean a clear positioning in favour of a regionalization of political power, or the empowerment of regional power in France. Nonetheless, its failure is expected to discourage such endeavours, particularly where the use of referendums is mandatory, or may seem useful in order to decide a divisive issue, as in Corsica.

Notes

Pour citer cet article