The political economy of territorial integration referendums

ABSTRACT

When do political communities support territorial amalgamation in referendums? Existing research has identified heterogeneity between communities as a key hurdle for integration reforms. In this paper, I demonstrate that asymmetries between communities do not necessarily impede political integration. An analysis of 1500 municipal merger referendums held in Swiss municipalities since the new millennium shows that voters’ support for integration is lower when potential merger partners are poorer or larger than their own community. However, the effects of wealth and size differences are interdependent. They reinforce each other’s effect on merger support if they point in the same direction, for example, in a relatively rich and relatively small community, but they can also cancel each other out, for example, in a relatively small, but relatively poor community. These results have important implications for understanding under which conditions integration reforms are politically feasible.

KEYWORDS:

• amalgamation
• referendum
• jurisdictional reform
• polity size
• local government
• voting behaviour
• territorial boundaries
• self-determination

1. INTRODUCTION

How and when political communities¹ decide to tie their fate to, or to separate from, other political communities has triggered a substantive amount of research in fields such as public economics, public administration and political science (Alesina & Spolaore, 2003; Bolton & Roland, 1997; Dahl & Tufte, 1974; Hooghe & Marks, 2016; Laponce, 2004). It has generally been observed that heterogeneity – e.g., cultural differences or economic imbalances – impedes the political integration of jurisdictions and increases communities’ incentives for secession (Gehring & Schneider, 2020; Qvortrup, 2014; Siroky, 2011). Arguably, this is because political communities are reluctant to pool decision-making authority with communities that are distinct from them, since this might hamper the full realization of their own preferences (Alesina & Spolaore, 2003).

In this paper, I build on this general finding and argue that the effects of different kinds of asymmetries and forms of heterogeneity on a community’s willingness to pool decision-making authority are interdependent. They can both compensate or reinforce one another, depending on whether one kind of asymmetry is disadvantageous but the other to the advantage of a community or whether both are (dis)advantageous for the same community. Here, I study the impact of wealth and size differences on voters’ support for territorial integration. Existing studies show that wealth differences indeed matter for political communities’ integration and secession preferences at different levels of governance. At the supranational level, Hobolt (2014) shows that citizens in European Union (EU) member states that are net contributors to the EU budget are more sceptical of EU enlargement because they fear that they would have to pay (even) more to the EU budget as a result of including more, potentially poorer, countries in the EU. At the regional level, Gehring and Schneider (2020) find a robust positive relationship between a region’s relative wealth (compared with the nation-state) on the vote share of secessionist parties (see also Sorens, 2005). Using exogeneous variation in regional wealth, they demonstrate that this relationship is indeed causal. These results are in line with those of Dalle Mulle (2018), who shows that secessionist parties in relatively wealthy regions (Scotland, Catalonia, Flanders and Northern Italy) employ an ‘economic victimization’ discourse by stressing that their regions are being exploited by the rest of the state. More generally, scholars of ethnic conflict and separatism have highlighted the role of economic grievances for secessionist mobilization (Ayres & Saideman, 2000; Cederman et al., 2011; Sorens, 2005). The relative wealth of a political community thus matters for its boundary preferences and higher relative wealth reduces the willingness to share political decision-making authority.

The same applies to the relative size of a political community compared with other communities. The pooling of decision-making authority comes with costs in terms of political control and power (cf. Tatham et al., 2021, p. 610). These costs are more substantial for relatively small communities: their preferences risk to be marginalized in an integrated jurisdiction. Indeed, population size continues to be an important aspect of a political community's power – also in the 21st century (Demeny, 2012). At the supranational level, Hobolt (2014) shows that citizens living in smaller member states – and who fear that small states might lose influence as a result of political integration – are more sceptical of both transferring more decision-making authority to the European level as well as of including new member states. At the regional level, secessionist tendencies are mainly present among political communities or ethnic groups that are excluded from political decision-making on central state matters or that have lost autonomy they previously enjoyed (Germann & Sambanis, 2021; Siroky & Cuffe, 2015). Normally, these communities constitute only a minority of a country’s population (Dalle Mulle, 2018). The relative size of a political community thus seems to be a second important determinant for its boundary preferences.

While existing studies clearly demonstrate the importance of size and wealth differences for communitys' border preferences, they study these factors in isolation. I argue here that asymmetries of a political community's wealth and size with regards to other communities are interdependent: they can offset or reinforce each other. Analysing the outcomes of referendums on municipal mergers in Switzerland since the year 2000 (N = 1542) – in which political communities vote on whether they want to merge with a specific group of other municipalities – yields support for this argument. The negative effect of being wealthier than the merger partners decreases, as the weight of a municipality in terms of its population size in the merger coalition increases. ² Similarly, the positive effect of making up a larger part of a merger coalition’s population decreases as the relative wealth of a community increases.

The article can inform existing research in several ways. First, it contributes to the literature on integration and secession referendums by studying the polity- or group-level factors associated with public support for territorial integration based on a large-N study of comparable referendums. Existing studies that focus on polity- or group-level factors predominantly operationalize territorial reform preferences by focusing on the politicization of territory and borders – e.g., vote shares of secessionist parties or self-determination claims (Germann & Sambanis, 2021; Sorens, 2005). An advantage of the present study is that citizens’ collective decisions in referendums on territory yield a more direct measure of political communities’ preferences that does not hinge on the mobilization by political actors. By demonstrating that concerns over wealth and size asymmetries indeed play an important role for citizens’ vote choice, the article corroborates key propositions of the literature on territorial integration and secession and by showing that these asymmetries can reinforce and compensate each other, the article contributes to existing debates on the role of economic and political grievances (Cederman et al., 2011; Germann & Sambanis, 2021). Sure enough, territorial reforms at the local level cannot be directly compared with supranational integration or secession. First, the stakes for citizens are lower. Local border changes neither affect citizens’ mobility because they are highly permeable, nor the national laws and regulations to which they are subject. Second, local territorial reforms also rarely lead to profound changes in the ethnic, linguistic or religious composition of jurisdictions – as integration and secession dynamics at other scales often do. Yet, the fundamental mechanism – namely that different types of asymmetries are interdependent – might also apply in other contexts and can be tested there.

Second, the article contributes to the growing literature on the local-level causes of municipal mergers by highlighting the role of wealth, alongside size, asymmetries for the successful implementation of a voluntary merger (e.g., Fitjar, 2021; Hanes et al., 2012; Saarimaa & Tukiainen, 2014). Specifically, it also sheds light on citizens’ preferences towards local territorial reforms – a topic that has received little attention so far (but see Beeri & Zaidan, 2021; Strandberg & Lindell, 2020; Strebel & Kübler, 2021).

Third, the article also contributes to research on political behaviour by suggesting that voters collectively take ‘rational’ and informed decisions when voting in referendums (Colombo, 2018) – even when voting at the local level, which is often considered to be ‘second-order’ and less relevant to citizens and where voters might, thus, resort to more expressive rather than rational vote motivations (Gendźwiłł, 2021).

Fourth, recent research on multilevel governance has emphasized the role of mass publics’ preferences in jurisdictional reform processes (Hobolt & De Vries, 2016; Hooghe & Marks, 2016). This article suggests that the concern for self-determination and autonomy indeed is a key motivation for citizens’ preferences regarding territorial integration – also at the local level.

Finally, the article also has practical implications for policymakers who want to amalgamate existing jurisdictions. It suggests that territorial reform coalitions are met with less resistance from affected citizens when all involved jurisdictions can either gain wealth or power.

2. THE ROLE OF WEALTH AND POWER ASYMMETRIES IN REFERENDUMS ON TERRITORY

2.1. Referendums on territory – the local level

The territorial delimitation of a polity is one of the constitutive elements of political jurisdictions. Territorial reforms – and more generally the pooling or separation of political authority between jurisdictions – are frequently decided on in referendums. In a recent article, Mendez and Germann (2018) identify 602 referendum decisions since 1776 that revolve around issues of sovereignty, border changes, and the pooling or separation of political authority. These referendums concern the integration and separation of subnational regions – such as in the Jura case in Switzerland in the 1970s (Laponce, 2004); the secession of regions from states – such as in the Scottish or the Catalan cases (Harguindéguy et al., 2022); and the pooling of authority in and accession to international organizations (Hobolt, 2009) as well as the devolution of authority to subnational regions (Giovannini & Vampa, 2020). Referendums on territorial reforms and authority transfers are thus both relatively widespread and of continued relevance at different territorial scales.

While this research has significantly advanced our understanding of referendums on boundary changes, there are also certain limits to it. It often focuses on the same, high-profile, cases (e.g., Harguindéguy et al., 2022, for a critique); it focuses more on the occurrence and the process of referendums and not on their outcomes (Cortés Rivera, 2020; Laponce, 2004; Qvortrup, 2014); and it is often concerned with explaining individual-level attitudes towards authority transfers and behaviour in referendums (Hierro & Queralt, 2022; Hobolt, 2009; Rodon & Guinjoan, 2018), rather than with studying factors at the level of political communities and polities that impact referendum outcomes. Yet, scholars studying separatism and ethnic conflict have since long emphasized the role of polity- and group-level factors to explain secessionism (Ayres & Saideman, 2000; Cederman et al., 2011; Germann & Sambanis, 2021; Siroky & Cuffe, 2015; Sorens, 2005). At the same time, these studies do not directly tap into mass public preferences for territorial reforms – they attempt to explain political mobilization or violence related to separatism. Hence, such studies might not accurately capture a political community’s collective preferences regarding territorial reforms. Political mobilization might be driven by a small, vocal, minority, or collective preferences might not (yet) be mobilized by political actors. For a quantitative assessment of the polity- or community-level factors that are associated with mass public preferences regarding territorial reforms, a focus on referendum outcomes is, thus, better suited but currently missing.

A territorial scale that has received comparatively little attention in research on referendums on territory and authority transfers is the local one. Yet, the local government landscape of numerous countries across the globe has been undergoing significant change – particularly towards the territorial consolidation of local government structures (Baldersheim & Rose, 2010; Kersting et al., 2009). While in many countries such reforms are implemented top down by national or regional governments (Myksvoll et al., 2022) – with local communities having little say whether and with whom they want to merge – several countries have also experienced voluntary municipal mergers where local communities decide themselves whether and with whom they want to merge. Often, citizens are asked to pronounce themselves on this question in consultative or binding referendums (Folkestad et al., 2021; Karv et al., 2022; Miyazaki, 2014; Strebel, 2019). Since local government mergers – and referendums on them – frequently come in waves, voluntary local government mergers provide a good setting for a quantitative study of polity-level factors that are linked to support for territorial reforms in referendums. Studying territorial amalgamation referendums in the local context also has the additional advantage that these processes take place in the same context and are subject to the same political opportunity structures. Harguindéguy et al. (2021, p. 4), for instance, state that the ‘main difficulty [for a quantitative study] lays in the great diversity of independence referendums’ and of finding comparable polity-level indicators over long time periods. For referendums on local boundary changes, this is not a problem – given that sufficiently large numbers of local governments hold referendums on territorial reforms within the same national context and within a relatively short time period.

In what follows, I discuss the relevance of the two key explanatory factors of this study – the wealth and size differences between polities – for the outcomes of territorial integration referendums at the local level.

2.2. Wealth and power asymmetries

Differences in population size and in wealth are considered important for explaining the outcomes of merger referendums also at the local level. While the relative population size of a jurisdiction compared with other jurisdictions in a territorial reform coalition is positively linked to support for merging, the relative wealth of a jurisdiction is negatively linked to merger support in referendums (Bruns et al., 2015; Fitjar, 2021; Hanes et al., 2012; Miyazaki, 2014; Strebel, 2019).³

The postulated mechanisms for these relationships are straightforward. A jurisdiction’s relative size in a merger coalition is indicative of the political weight it carries in the hypothetical new municipality. In other words, the larger the population share of a municipality, the more it can influence political decisions that are made in the new jurisdiction. That this is a relevant concern for political communities is shown by studies on political representation and public goods distribution after municipal mergers. Jakobsen and Kjaer (2016) as well as Saarimaa and Tukiainen (2016) demonstrate that voters residing in the smaller parts of a newly merged municipality tend to vote more based on a territorial principle. That is, they are more inclined to support candidates from their part of the municipality compared with voters residing in the bigger parts of the new municipality. That this territorial representation has relevant implications is further demonstrated by Harjunen et al. (2021). They show that the smaller and politically less well-represented parts of a new municipality suffer from job losses in local administrations as well as in health and social care services. There are, thus, good reasons for relatively small municipalities to be sceptical of merging. It might mean losing power in political decision-making and ultimately to suffer losses in public goods provision.

Hypothesis 1: The larger a jurisdiction’s population size is compared with the merger coalition, the more likely it is to support a merger.

The rationale for why the relative wealth of a jurisdiction compared with its coalition partners is negatively related to support for merging is equally clear. A jurisdiction that is economically more well-off than its merger partners is likely to suffer adverse economic consequences as a result of a merger (Alesina & Spolaore, 2003; Hanes et al., 2012). Given that public service provision and its financing are aggregated in the new jurisdiction and given that the same standards should apply everywhere, relatively well-off municipalities might either suffer from a decline of public service quality (if public service standards are lowered to the level of the other jurisdictions) or they might have to contribute more financially (to lift public service quality in other municipalities to their level). Indeed, Sørensen (2006) shows that local wealth differences play an important role for Norwegian local councillors’ support or opposition towards a merger.

The relative wealth of a jurisdiction does not only matter for territorial reforms that concern integration. Studies analysing disintegration reforms demonstrate that this logic also works the other way around. Brink (2004) and Lapointe (2018) show that wealthy parts of municipalities are more likely to secede in Sweden and that they are more likely to hold secession referendums in Quebec. At the regional level, numerous studies show that economic grievances explain separatist tendencies (Cederman et al., 2011; Gehring & Schneider, 2020; Sorens, 2005). Moreover, Harguindéguy et al. (2021) show that central states are more likely to allow substates to hold secession referendums if the substate is relatively poorer than the central state. Based on these considerations, I formulate a second hypothesis:

Hypothesis 2: The more well-off a jurisdiction is compared with its merger partners, the less likely it is to accept a merger.

2.3. The interdependence of wealth and power asymmetries

Do the relative wealth and the relative size of a jurisdiction jointly affect the outcomes of territorial amalgamation referendums? Put differently, do wealth and size asymmetries reinforce one another when they are both to the benefit or detriment of the same jurisdiction. And conversely, can they offset one another if they point in opposite directions?

As argued in the previous section, territorial integration leads to the redistribution and pooling of resources and political decision-making authority in a new jurisdiction. The extent to which jurisdictions might win or lose from this reshuffling is associated with their support for integration. Yet, wealth and power losses and gains might not operate in isolation.

Table 1 illustrates four combinations of gains and losses of wealth and power that result from territorial integration and how they should impact merger support. Voters in a relatively rich jurisdiction might be particularly critical of territorial integration if their jurisdiction only accounts for a small part of the merger coalition. They might not only have to share their resources with other jurisdictions, but they also risk being dominated by the other jurisdictions in the merger coalition. Hence, anticipated wealth and power losses might reinforce each other. However, if the same jurisdiction is relatively large, wealth losses should carry less weight in the merger referendum because they are compensated for by possible power gains. On the one hand, such a jurisdiction can likely dominate the new jurisdiction, because it can de facto ‘annex’ smaller jurisdictions and thereby extend its power (Dur & Staal, 2008). On the other hand, the wealth loss suffered by this jurisdiction is mitigated by its relative size: the post-merger jurisdiction will remain relatively wealthy, because a large part of it is. Based on these considerations, I expect that the effect of relative wealth on merger support depends on the relative size of a jurisdiction.

Hypothesis 3.1: The negative effect of relative wealth on merger support decreases as the relative size of a jurisdiction in the merger coalition increases.

Of course, the argument runs both ways. Compared with a jurisdiction that risks losing both wealth and power as a result of territorial consolidation, voters in a jurisdiction which is relatively small but also relatively poor might be less concerned about their power loss because it would be offset by wealth gains. Similarly, compared with a relatively large jurisdiction that would suffer wealth losses as a result of merging, voters in a relatively large jurisdiction that can expect wealth gains should be particularly supportive of merging because they can both extend their power over a larger territory and in addition marginally increase their wealth.

Table 1. Support for territorial integration.

		Power
		Loss	Gain
Wealth	Loss	Low support (Reinforcement)	Medium support (Compensation)
	Gain	Medium support (Compensation)	High support (Reinforcement)

Hypothesis 3.2: The positive effect of relative jurisdiction size on merger support increases as the relative wealth of a jurisdiction in the merger coalition increases.

3. TERRITORIAL REFORMS IN SWITZERLAND

3.1. Swiss municipal merger referendums, 2000–20

To test the proposed hypotheses, I analyse all municipal merger referendum outcomes in Switzerland since the year 2000, the data for which I have collected from official cantonal statistics, newspaper articles and municipalities’ websites. Unlike large-scale reforms in many other countries – e.g., Denmark, Germany or Sweden (Bruns et al., 2015; Hanes et al., 2012; Lassen & Serritzlew, 2011) – municipal mergers in Switzerland are not decided top-down by higher government tiers, but bottom-up by the affected municipalities. Swiss cantons – the intermediate government tier in Switzerland – merely play a facilitating role and provide incentives for municipal mergers.⁴ Kaiser (2014) shows that particularly the financial contributions some cantons grant for merging municipalities effectively function as a necessary condition for municipal mergers to take place in a canton.⁵ Empirically, bottom-up municipal mergers in Switzerland took place in 14 different cantons between 2000 and 2020. While bottom-up municipal mergers are not as widespread as top-down ones, some countries such as Finland, Japan or Norway also recently saw significant merger activity on a voluntary basis that was triggered through higher tier incentives (Folkestad et al., 2021; Miyazaki, 2014; Saarimaa & Tukiainen, 2014).

A key feature of Swiss municipal mergers is that the final, binding, decision on their implementation is taken by the affected citizens at town hall meetings or in popular votes at the ballot box (el-Wakil & Strebel, 2022). Citizens are asked to vote in favour of or against their jurisdiction’s merger with a clearly defined set of other jurisdictions. Between 2000 and 2020, 1542 municipalities held referendums on 443 municipal merger proposals – which is more than half of all Swiss municipalities that had existed in the year 2000 (Figure 1). A merger proposal needs the approval of a majority of voters in each of the involved jurisdictions to be implemented.⁶ A total of 104 merger proposals failed to reach this requirement due to 232 municipalities rejecting the merger. These merger projects were thus not implemented. Overall, the success rate of Swiss municipal merger proposals is quite high, with 85% of all municipalities accepting to merge. This can be attributed to the fact that the specific concerns and preferences of different municipalities are taken into account already in the negotiation over merger contracts to raise their success chances in referendums. Moreover, jurisdictions with highly integration-sceptic communities are probably less likely to participate in a merger process in the first place because local elected representatives who normally negotiate merger proposals will know that a merger proposal will likely not be supported.

Figure 1. Swiss municipalities that voted on a municipal merger project, 2000–20.

Note: Municipal boundaries for year 2000 are shown in white; cantonal boundaries are outlined in black.

Source: Author’s own data on municipal merger referendums; map data: Swiss Federal Office of Topography (swisstopo).

Normally, referendums on municipal mergers are held simultaneously in all municipalities. This prevents a strategic timing of referendums across the involved jurisdictions – for example, by letting those municipalities most favourable to a merger vote first and those least favourable last. Sequencing integration referendums in this way, might lead to a domino effect that puts pressure on reluctant citizens to approve of the reform, since other municipalities have already accepted it (Hobolt, 2009, p. 11).⁷

Referendums on municipal mergers in Switzerland thus offer the unique opportunity to quantitatively study political communities’ support for territorial reforms. Local governments can decide whether or not they want to be involved in municipal merger projects and voters in each jurisdiction can decide in a binding referendum whether their municipality should implement a proposed merger. More generally, Switzerland is a well-suited case to study referendums on territory due to the important role of municipalities in its multilevel system. Swiss municipalities are among the most autonomous in Europe (Ladner et al., 2019). They account for a third of total public spending; they have significant policy and fiscal competences – such as deciding over naturalization, land-use planning and zoning, and local tax rates; and they serve as strong focal points of citizens’ territorial identities (Linder & Mueller, 2021, pp. 74–77). In sum, referendums on municipal mergers in Switzerland thus concern a level of government that matters – both for policymaking and for citizens. Therefore, studying the influence of polity-level factors on referendum outcomes in this context provides insights that are relevant to territorial reforms at other territorial scales as well.

3.2. Measuring relative size and relative wealth

To operationalize the relative size of a municipality, I rely on the population size of a municipality relative to the total of the merger coalition. As already outlined in the argument for Hypothesis 1 above, this indicator serves to capture possible power gains or losses of a jurisdiction. The population size of municipality i in the year of the referendum is divided by the sum of the population of all municipalities involved in a merger coalition, that is, i’s own population plus the one of its merger partner(s) j (1, … , n): $\mathrm{Relative}\mathrm{siz}{\mathrm{e}}_i={\displaystyle \frac{\mathrm{Population}\mathrm{siz}{\mathrm{e}}_i}{\mathrm{Population}\mathrm{siz}{\mathrm{e}}_i+{\sum }_{j=1}^n\mathrm{Population}\mathrm{siz}{\mathrm{e}}_j}\times 100}$Operationalizing relative wealth is less straightforward. What constitutes a jurisdiction’s wealth and how it can best be measured is context dependent. Studies on secession often rely on a region’s gross domestic product (GDP) or its wealth in natural resources (Gehring & Schneider, 2020; Harguindéguy et al., 2021; Sorens, 2005). In her study of citizens’ EU enlargement preferences, Hobolt (2014) uses a dichotomous variable for whether a country is a net contributor or a net beneficiary to the EU. For my purposes, I use a municipality’s federal tax revenues per capita in the year of the merger referendum. Given that the federal tax rate is the same in all Swiss municipalities, federal tax revenues provide a good indication of the tax base that is available to a municipality. For each municipality, the combined revenues from the taxation of individuals and legal entities are divided by a municipality’s population. This indicator is a comparable measure of the resources that can be extracted from a particular community by the local public administration and is thus a good indicator for a community’s collective wealth. Moreover, using the available tax base as an indicator of a community’s wealth is relevant because the tax base directly affects a polity’s leeway to provide public services, and hence one of the core functions it has for its citizens.

As the relevant point of comparison to calculate municipality i’s relative wealth, I take the difference of its per capita tax revenues to the population-weighted mean of its merger partner(s) j. To obtain the latter, I sum up the tax revenues/capita for each merger partner – multiplied by its population size to account for the weight a merger partner carries in the merger coalition – and divide this by the sum of the merger partners’ population: $\mathrm{Relative}\mathrm{wealt}{\mathrm{h}}_i=\mathrm{Tax}\mathrm{revenues}/\mathrm{capit}{\mathrm{a}}_i-{\displaystyle \frac{{\sum }_{j=1}^n\mathrm{Tax}\mathrm{revenues}/\mathrm{capit}{\mathrm{a}}_j\times \mathrm{population}\mathrm{siz}{\mathrm{e}}_j}{{\sum }_{j=1}^n\mathrm{Population}\mathrm{siz}{\mathrm{e}}_j}}$Negative values thus indicate that municipality i is less wealthy than its merger partner(s) j, whereas positive values indicate potential wealth losses as a result of a merger. Accordingly, I expect a lower support for merger proposals at higher values of this variable.

This operationalization of relative wealth via federal tax revenues has two drawbacks which are related to data availability. First, data are currently only available until 2019, newer data have not yet been released. Second, for data protection reasons, data for tax revenues from legal entities is not disclosed for a substantial number of municipalities, particularly small ones. This missing data problem is aggravated by the calculation of the relative wealth measure: if data for only one of the municipalities in a merger coalition are missing, the measure cannot be calculated for any of the municipalities in the merger coalition as this would return incorrect values. To show that the results do not hinge on the exclusion of this substantive number of observations (around 50%, see Table A1 in the supplemental data online), I use an alternative wealth operationalization as a robustness check which allows for the inclusion of these omitted observations (see the results section and Appendix B1 online).

3.3. Methodological approach: estimation and control variables

To test the three hypotheses I focus on two dependent variables: (1) whether or not a municipality accepted a merger (yes–no); and (2) the percentage of votes in favour of a merger.⁸ I use both logistic and linear regression models because the first dependent variable is binary and the second one is continuous. The unit of analysis in the regression models is a municipality that voted on a municipal merger. Since several municipalities in the analysis belong to the same merger coalition, I rely on multilevel regression models to account for the fact that municipalities are nested in merger coalitions and thus their standard errors are correlated. Moreover, municipalities are also nested in cantons. Hence, I introduce cantons as an additional level in the multilevel models. This results in a three-level hierarchical model. Appendix A in the supplemental data online shows random effects analysis of variance (ANOVA), which suggest that this is an appropriate modeling strategy.⁹

The models control for several other municipal-level factors that might impact voters’ decision to accept or reject a merger proposal. First, I include the absolute population size (logarithmic) and tax revenues per capita in addition to the relative one. The reason is that smaller municipalities and municipalities with lower tax revenues might face stronger functional pressures to merge because they lack the necessary resources to carry out their tasks (Steiner, 2003).

In addition to such functional pressures, previous studies have shown that local identity, as well as political ideology play a role in municipal merger processes. While it is difficult to operationalize citizens’ attachment to their municipality at the aggregate level without individual-level data, Strebel (2018) proposes to use population turnover to approximate this concept. A higher population turnover means that a smaller part of the citizens live in a municipality since a long time and residence duration predicts local attachment and involvement at the individual level (Kasarda & Janowitz, 1974). At the communal level, a high population turnover impedes community building and hence renders the development of communal identity more difficult, which in turn might be conducive to merger acceptance.

Political ideology can matter for merger support as well. For three Swiss cantons, Strebel (2019) has shown that municipalities with a higher vote share of right-wing nationalist parties are more likely to reject municipal merger proposals, and in an individual-level analysis of citizens’ perceptions of local cooperation and consolidation reforms in metropolitan areas, Strebel (2022) shows that left-progressives are more and right-wing nationalists less supportive of such reforms. For the present analysis, I thus include the local-level vote share of right-wing nationalist parties as well as of left-progressive parties in the last national elections before the merger referendum.

Moreover, political homogeneity might matter for forging merger coalitions. Bhatti and Hansen (2011) and Bruns et al. (2015) argue that a higher ideological difference between two municipalities makes a merger less likely, because the involved actors fear that their political preferences cannot be realized (see also Sorens, 2005). Therefore, I include a municipality’s difference its merger partners’ population-weighted mean in right-wing nationalist and left-progressive vote shares.

To account for institutional variation and merger legacy, I control for whether the referendum decision was taken in a town hall meeting or at the ballot box (el-Wakil & Strebel, 2022), and whether a municipality has previously merged or attempted to merge. Finally, I control for the municipality types developed by the Swiss Federal Statistical Office, a variable capturing different structural properties of municipalities – e.g., urban centre, agrarian or rural commuter municipality¹⁰ – as well as for time-period fixed effects. Appendix A in the supplemental data online gives an overview of the descriptive statistics for the different indicators.

4. DO THE EFFECTS OF WEALTH AND SIZE DIFFERENCES CONDITION EACH OTHER?

How do potential changes in a political community’s wealth and power impact the outcomes of territorial integration referendums? And are the effects of a community’s relative wealth and relative size on merger support interdependent? Table 2 shows the results of multilevel regression models for the dependent variables merger acceptance and the percentage of yes votes in a merger referendum. Across all specifications there is a significant and substantial positive effect of relative size.¹¹ The higher the percentage of a municipality’s population in a merger coalition, the more likely voters are to accept a merger and the higher the percentage of favourable votes. In the full model without the interaction effect (model 11 in Table 2), a 1 standard deviation (SD) increase in relative size is associated with an increase in the yes vote of 9.7 percentage points. This clearly corroborates Hypothesis 1.

Table 2. Multilevel regression models for merger support.

	Merger acceptance (= 1)						Percentage yes vote
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Relative size	0.009*	0.008^+	0.049***	0.056***	0.058***	0.055***	0.089***	0.089***	0.332***	0.357***	0.357***	0.357***
(% of coalition population)	(0.049)	(0.070)	(0.000)	(0.000)	(0.000)	(0.000)	(0.001)	(0.001)	(0.000)	(0.000)	(0.000)	(0.000)
Relative wealth	−0.281**	−0.636**	−0.922***	−1.198***	−0.881***	−1.197***	−0.802***	−0.782	−1.034*	−4.317***	−2.620***	−4.467***
(Δ tax revenues/capita, CHF1000)	(0.003)	(0.002)	(0.000)	(0.000)	(0.000)	(0.000)	(0.001)	(0.104)	(0.019)	(0.000)	(0.000)	(0.000)
Relative size x		0.007*	0.007*	0.011*		0.010*		−0.001	0.024	0.063**		0.061**
relative wealth		(0.035)	(0.029)	(0.026)		(0.030)		(0.962)	(0.139)	(0.002)		(0.002)
Absolute size/wealth			✓	✓	✓	✓			✓	✓	✓	✓
Municipality controls				✓	✓	✓				✓	✓	✓
Time period FEs					✓	✓					✓	✓
N (municipalities)	792	792	792	732	732	732	731	731	731	696	696	696
N (merger coalitions)	253	253	253	237	237	237	235	235	235	227	227	227
N (cantons)	14	14	14	14	14	14	14	14	14	14	14	14
Log-likelihood	−356	−354	−328	−288	−288	−286	−3165	−3164	−3129	−2945	−2943	−2939
LR χ^2	9.83	14.01	54.51	71.04	69.05	72.36	19.42	19.42	102.27	174.67	181.48	193.23
p > χ^2	0.007	0.003	0.000	0.000	0.000	0.000	0.000	0.005	0.000	0.000	0.000	0.000
AIC	720	718	671	623	628	624	6341	6344	6275	5940	5939	5931
BIC	739	742	708	733	742	743	6369	6376	6316	6054	6057	6054

Note: Cell entries are unstandardized coefficients obtained through -melogit- and -mixed- commands in Stata. p-values are shown in parentheses. ⁺p < 0.1; *p < 0.05; **p < 0.01; ***p < 0.001.

Similarly, I find a consistent and significant negative effect of relative wealth. The higher a jurisdiction’s tax revenues per capita are compared with its merger partners, the less likely voters are to accept a merger and the lower the percentage of yes votes in favour of a merger. A 1 SD increase (about US$2800) in the full model without interaction effect (11) is associated with a decrease of 6.9 percentage points in the yes vote percentage. Again, this clearly corroborates Hypothesis 2.

The empirical evidence is also clear for the interaction between relative size and relative wealth. Hypotheses 3.1 and 3.2 posit positive interaction effects between relative size and relative wealth. The results from models 10 and 12 corroborate this hypothesis, but in models 8 and 9 this effect is not statistically significant. This is partially because absolute size and wealth are not included in the model.¹² When they are included, the standard error of the interaction effect becomes smaller, but the coefficient still does not reach conventional levels of statistical significance. Once municipality-level control variables are included, this is the case.

For the logistic regression models the significance of the coefficient of the interaction effect is meaningless because it depends on the values of all covariates (Berry et al., 2010). Therefore, I calculated second differences of the average marginal effects of relative wealth conditional on relative size, and vice versa. The effect of relative wealth on merger acceptance amounts to −0.16 (p < 0.000) when relative size is 1.18 (−1 SD from the mean) and to −0.04 (p < 0.05) when relative size is 55.7 (+1 SD from the mean). The second difference between these two effects is 0.12 (p < 0.001) and thus highly significant. For relative size, the average marginal effect amounts to 0.001 (n.s.) when relative wealth is −2.78 (−1 SD from the mean) and to 0.012 (p < 0.001) when relative wealth is 2.65 (+1 SD from the mean). The second difference is 0.011 (p < 0.001). The fact that we find highly significant second differences means that there is a statistically significant interaction effect of relative size and relative wealth on merger acceptance.

Figures 2 and 3 display the interaction effects graphically for the results from models 6 and 12. Figure 2 shows the change in the effect size of the tax revenue difference (y-axis) on merger acceptance and percentage yes vote as a function of relative size (x-axis). These figures show that as the relative size of a municipality increases, the effect of relative wealth on merger acceptance and the percentage of yes votes becomes less negative as expected. When a municipality’s population makes up less than 10% of a merger coalition, an increase in the wealth difference of CHF1000 is associated with a 4 percentage point decrease in the percentage of yes votes, whereas the same increase in the wealth difference has no impact on the percentage of yes votes when a municipality makes up 70% of a merger coalition’s population.

Figure 2. Effect of relative wealth on merger support conditional on relative size.

Figure 3. Effect of relative size on merger support conditional on relative wealth.

Similarly, Figure 3 shows the change in the effect of relative size (y-axis) on merger acceptance and percentage yes vote as a function of relative wealth (x-axis). When a municipality has CHF3000 lower tax revenues/capita than the weighted average of its merger partners, a 1 percentage point increase of a municipality’s size in the merger coalition is associated with a 0.17 percentage point increase in the percentage of the yes vote, whereas when the municipality has CHF3000 higher tax revenues, the same change in relative size is associated with a 0.54 percentage point increase in the percentage of the yes vote.

This is clear support for Hypotheses 3.1 and 3.2. It shows that relative wealth and relative size can reinforce each other when they are both to the benefit or detriment of a municipality and they can compensate one another when one benefits and one hurts a municipality. Moreover, Figure 2 suggests that at very high values of relative size, that is, when a municipality accounts for almost the totality of the population of a merger coalition, the negative effect of relative wealth becomes insignificant and thus does not matter anymore for merger support. Relative wealth and relative size thus both reinforce and compensate each other’s effect on merger support. Voters seem to engage in a ‘rational’ calculus and take power and wealth gains and losses jointly into account when voting in a territorial integration referendum.

To further assess the validity of these findings, I conduct several robustness tests. First, I use a different operationalization of the wealth indicator – a municipality’s local tax rate – to test whether the findings merely result from the exclusion of a substantive number of observations due to missing data on tax revenues (see section 3.2). Unlike tax revenues, higher tax rates than those of the merger partners are indicative of lower, not higher, wealth. Therefore, I construct the relative wealth indicator the other way around – subtracting a municipality’s tax rate from the merger partners’ average – so that the results can be interpreted in the same way. Even though local tax rates are not only measuring the economic well-being of a municipality, but also local voters’ tax preferences, I provide results for this measure for two reasons. First, tax rate differences are a very salient issue in Swiss municipal merger decisions (Strebel, 2019). Since local governments in Switzerland can set their own local tax rate, municipalities use local tax rates to compete for wealthy individuals and companies. Hence, ‘voting with one’s feet’ is not uncommon among wealthy Swiss residents (Schaltegger et al., 2011). Municipalities with lower tax rates thus generally have residents with higher income and wealth levels. Second, municipal mergers regularly get rejected at the ballot box, because voters of low-tax municipalities fear that they have to pay higher taxes after a merger – and hence might lose their high net-worth tax payers which would move to another, cheaper, municipality.

The results for this alternative wealth operationalization can be found in Appendix B1 in the supplemental data online. They corroborate the findings reported above. Throughout the different models, there is a negative effect of relative wealth on the two dependent variables. Moreover, the interaction effects exhibit the same pattern as in the main analysis: the interaction effects are positive and highly significant in the linear model, as are the second differences in the logistic regression model.

Second, I use an alternative way of calculating a jurisdiction’s relative size by relying on the percentage of eligible voters in the merger coalition rather than the population share. Since the argument on power loss involves being ‘outvoted’ in the new jurisdiction, the percentage of eligible voters might even more accurately capture this idea. Results for this analysis can be found in Appendix B2 in the supplemental data online. Again, the results clearly corroborate the reported findings.

Third, and in addition to these alternative operationalizations of size and wealth, I estimate a multilevel linear probability model for the dependent variable merger acceptance and logit and ordinary least squares (OLS) models with canton fixed effects and standard errors clustered at the merger coalition. I also restrict the sample to those municipalities that only were involved in a merger project once and to those that voted on the merger on the same day (see Appendix C in the supplemental data online). Also with these alternative model specifications, the results remain robust.

In sum, I find strong support for Hypothesis 1 – that increases in relative jurisdiction size are positively related to support for municipal mergers – and Hypothesis 2 – that increases in relative wealth depress support for territorial integration. Finally, and most importantly, I find robust evidence for an interaction effect between relative wealth and size (Hypotheses 3.1 and 3.2). This suggests that voters are willing to compensate wealth losses with power gains and vice versa, but also that political communities are particularly sceptical of territorial integration if they risk losing both wealth and power.

5. CONCLUSIONS

Under which conditions do political communities support territorial integration in referendums? In this article, I have demonstrated that the relative size and the relative wealth of a jurisdiction compared with its merger partners robustly impact communities’ willingness to integrate or to pool decision-making authority. Moreover, there is clear evidence that the effects of relative size and relative wealth condition each other. Detrimental/beneficial positions with respect to both wealth and size reinforce merger opposition/support, whereas an advantageous position on one, can compensate for a disadvantageous position on the other factor.

Can these results be generalized to other local contexts? There are several reasons for this to be the case. First, Swiss municipalities have significant fiscal and policy autonomy, similar to Scandinavian countries but also to US local governments (Ladner et al., 2019). In these contexts, similar results are likely. More generally, when local governments have some fiscal and some decision-making competences, it is conceivable that considerations over wealth and power matter for political communities’ stance on territorial reforms. Second, scholars find very similar direct effects of relative wealth and relative size on local governments’ and communities’ position towards territorial reforms, in countries as diverse as Sweden (Brink, 2004; Hanes et al., 2012; Sørensen, 2006), Japan (Miyazaki, 2014), Canada (Lapointe, 2018), and Germany (Bruns et al., 2015). Given that the two factors matter in such diverse local contexts, it is well possible to find compensation and reinforcement dynamics there as well.

Do the results also travel beyond the local scale? Here, more caution is advised. There are two factors that limit the generalizability of the findings: (1) what is at stake in a border reform; and (2) cultural (ethnic, linguistic or religious) differences across communities. First, local border changes do not fundamentally impact citizens lives. Before and after a merger, local borders are completely permeable and citizens remain subject to the same national laws and regulations. This is not the case for territorial reforms at other scales, for example, in the case of secession or supranational integration, where the stakes are thus much higher (Bolton & Roland, 1997; Harguindéguy et al., 2021). This also means, however, that at other scales, a loss of wealth and power would be much more consequential for political communities and they might care even more about it. In that sense, local territorial reforms could be seen as a least likely case for wealth and size differences to impact political communities’ territorial reform preferences.

The second limitation to the generalizability of the findings beyond the local scale is that cultural differences between jurisdictions involved in local territorial reforms are often small. Yet, existing research shows that such differences play a key role in referendums on territory (Qvortrup, 2014; Siroky, 2011). In principle there are two ways in which cultural differences between jurisdictions can interact with wealth and size differences. First, they can attenuate them. If cultural differences cross-cut wealth and size asymmetries, it is likely that they trump the role of wealth and size differences, and that voters base their behaviour more on identitarian motives. Second, they can reinforce them. If ethnic, linguistic, or religious differences align with wealth and size asymmetries, communities might follow an ‘us versus them’ logic, and are willing to share power and wealth with their in- but not with their out-group (Mansfield & Mutz, 2013). That such a reinforcement dynamic can indeed be at play is suggested by Lapointe (2018): in a study of local secession referendums in Quebec, he finds that linguistic differences between communities indeed render wealth asymmetries more important. Whether cultural differences act as moderators or as catalysts for wealth and size differences depends on the context and is ultimately an empirical question.

Another limitation of the study arises from the self-selection of municipalities into participating in merger coalitions. Unlike in top-down reforms, where higher government tiers forge merger coalitions, local governments choose whether and with whom to start a merger process themselves in bottom-up reforms (Harjunen et al., 2021). Wealth and size asymmetries likely play a role already at the coalition-formation stage and very heterogeneous coalitions are unlikely to emerge (see also Bhatti & Hansen, 2011). Consequently, the merger coalitions analysed here probably represent a set of cases where wealth and size differences are less pronounced than if they would have been forged top-down. However, this self-selection does not need to be a concern for the validity of the results. It is reasonable to assume that the effects of size and wealth differences on political communities’ support for territorial integration would be even more pronounced when merger coalitions are created exogenously by higher tier governments, because asymmetries within a merger coalition would be bigger and hence of more concern for political communities.

Notwithstanding these limitations, the results have important implications for different fields of study. First, in line with research on territorial reforms and the pooling or separation of political authority, the results confirm that heterogeneity and asymmetries between political communities indeed reduce their acceptance of territorial integration (cf. Alesina & Spolaore, 2003; Cederman et al., 2011; Germann & Sambanis, 2021). However, the study also shows that such asymmetries can amplify and offset each other. Future research should assess to what extent this also applies to other kinds of asymmetries and heterogeneity between political communities. In particular, a promising further avenue of research is to assess how cultural differences across communities interact with wealth and size differences.

Second, the results also generate important insights for research on political behaviour – in particular related to the question of whether citizens can and do take informed decisions on issues in referendums (e.g., Colombo, 2018). The results presented here clearly suggest that they can. Collectively, as a political community, citizens voting in municipal merger referendums make ‘rational’ – and fairly complex – decisions, by weighing up different aspects against each other. What is more, they do so in the local context, where political decisions are often considered to be ‘second-order’ to national decisions and less relevant to voters (Gendźwiłł, 2021). Future research should explore further whether these aggregate-level patterns translate to the individual level. Results on citizens’ support for European integration suggest that this might be the case (Hobolt, 2014).

Finally, the results also confirm a central claim of the seminal post-functionalist theory of governance (Hooghe & Marks, 2016; Tatham et al., 2021). Voters and political communities seem to deeply care about who decides and the influence they can have on political decisions. This is indicated by the strong influence of size differences on support for territorial integration and by the finding that relatively small communities are more likely to reject jurisdictional amalgamation – presumably because they would disproportionately lose influence and power. A central proposition of this theory thus also holds at the local level.

Beyond the scientific community, this study also has practical implications for policymakers that deal with territorial reforms. It points to possibilities for forging ‘win–win’ coalitions between political communities that are to be amalgamated. The analysis suggests that citizens and political communities accept certain losses if they are compensated with adequate gains. Policymakers can make use of these insights in reform processes to reduce public resistance – both in bottom-up and in top-down territorial reforms.

ACKNOWLEDGEMENTS

Previous versions of this paper were presented at the 2021 Annual Meetings of the Swiss and the European Political Science Association. I thank Alice el-Wakil, Matthieu Niederhauser, Janne Tukiainen, the participants at these conferences as well as the two anonymous journal reviewers for very helpful comments and suggestions.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author.

Notes

1 For the purposes of this paper, I equate a political community with what Moore (2015, p. 50) defines as a ‘people’: collective agents that (1) ‘share a conception of themselves as a group’, (2) ‘have the capacity to establish and maintain political institutions’, and (3) ‘have a history of political cooperation together’. This does not require a people to have its own jurisdiction (yet) where it can engage in self-determination. In this article, I discuss political communities that inhabit clearly defined territories.

2 Municipal mergers and territorial reforms more generally can involve two or more existing jurisdictions. To refer to this group of jurisdictions, I use the terms "merger coalition" and "territorial reform coalition" interchangeably.

3 While Miyazaki (2014) finds that the relative wealth of Japanese municipalities is negatively associated with support for mergers in local referendums, he also finds that a higher relative population share is negatively associated with merger support. However, this might be explained by the fact that he does not account for the absolute size of a municipality. The latter is an important indicator for the possibility to benefit from scale economies (Alesina & Spolaore, 2003) or for functional pressures to merge due to the costs of smallness (Strebel, 2018). Larger jurisdictions – which often make up a larger part of the population of a merger coalition – can benefit less from scale economies and hence might see fewer advantages in merging. Not accounting for this factor might thus bias the result of the relative size indicator.

4 A notable exception was a large-scale territorial reform in the Swiss canton of Glarus, where in 2007 cantonal voters decided that the formerly 27 municipalities would be merged into three by 2011 (Hofmann & Rother, 2019).

5 Interestingly, however, Strebel (2018, 2019) shows that the size of the financial incentive does not impact a municipality’s probability to participate or decide in favour of a municipal merger. The financial incentive thus seems to function mainly as a trigger for municipalities to start considering mergers.

6 There are some exceptions to this, however. Eleven mergers were implemented despite one or several municipalities rejecting the proposal. The municipalities accepting the merger implemented it without holding another referendum. In seven cases, municipalities whose voters rejected a merger were forced to implement it by the cantonal parliament. Five of these cases are located in the canton of Ticino, where referendums only have a consultative nature. Here, I consider the referendum outcome and not whether a merger was implemented.

7 For 26 merger projects, the referendums did not take place on the same day, however. Excluding these merger projects from the analysis does not alter the results; see Appendix C3 in the supplemental data online.

8 For 101 municipalities, I was not able to find data on the percentage of voters who voted in favour of a merger – only on the outcome of the referendum (accepted/rejected).

9 The inter-class correlation amounts to 43% and 34% at the level of the merger coalition and to 2.3% and 4.0% at the level of the cantons for the binary and the continuous dependent variable, respectively. Robustness checks with simple logit and ordinary least squares (OLS) models with clustered standard errors and canton fixed effects lead to substantively similar results (see Table C1 in the supplemental data online).

10 For more information on the construction of this typology, see the Federal Statistical Office (German or French only) at https://www.bfs.admin.ch/bfs/de/home/statistiken/querschnittsthemen/raeumliche-analysen/raeumliche-gliederungen/raeumliche-typologien.assetdetail.342284.html (accessed on 5 December 2022).

11 The effect of relative size is much smaller in models 1 and 2 and in 7 and 8. This is because these models do not include the absolute population size. Yet, absolute size has a significant negative effect on both the probability to accept a merger and the yes vote percentage – confirming its postulated role as an indicator for the functional pressures to merge that small municipalities face. When it is not included in the model, part of its effect is picked up by the relative size measure. Since I expect relatively larger municipalities to be more favourable towards mergers, the absence of the absolute size measure depresses the effect size of this variable.

12 See also note 11.