Citizens’ attitudes towards local autonomy and inter‑local cooperation: evidence from Western Europe

Recent decades have seen a strengthening of local autonomy in many European states. At the same time, local governance capacities were strengthened through intermunicipal structures and territorial consolidation to prepare local authorities for additional tasks. In this paper, we assess what citizens think about strengthening local autonomy and strengthening inter-local cooperation. We argue that citizens’ attitudes towards local autonomy and inter-local cooperation are a function of their behavioural, emotional and ideological connection to the local. Using data from a population-based survey in eight West European metropolitan areas in France, Germany, Switzerland and the UK, we show that local autonomy and inter-local cooperation supporters have divergent ideological positions concerning the allocation of political authority across state scales. Inter-local cooperation is supported by voters of new progressive left parties but opposed by right-wing nationalist partisans— who in contrast favour local autonomy. This suggests that the demarcation–integra-tion divide which structures citizens’ attitudes towards international integration also matters for subnational and local governance reforms.


Introduction
In recent decades, many countries have decentralized authority from the national level to regional tiers and local governments (Hooghe et al. 2010;Ladner et al. 2019). To be successful, local governments need the capacity to actually carry out the new responsibilities they receive from decentralization and devolution. However, existing politico-administrative structures at the local level are not always conducive to such capacities. Some governance problems extend beyond the boundaries of existing local governments and local governments sometimes lack the organizational capacities to deal with such problems (Goldsmith and Page 2010). To strengthen these problem-solving capacities, many countries promote and conduct territorial or organizational reforms of their local government systems. Vertical reforms of decentralizing and devolving political authority are thus often accompanied by horizontal reforms of local governance structures (Van Houwelingen 2018).
In this paper, we investigate citizens' attitudes towards this nexus and assess mass support for local autonomy and inter-local cooperation. We argue that those in favour of strengthening local autonomy, i.e. vertical reforms, and those open to inter-local cooperation, i.e. horizontal reforms, are not necessarily the same people. Drawing on existing research, we assume that citizens' support for vertical or horizontal reforms is a function of their behavioural and emotional connection to the 'local' (Lidström and Schaap 2018). Moreover, we posit that citizens' ideological position regarding the scalar organization of the state has a differentiated impact on their support for local autonomy and inter-local cooperation (see Strebel 2019).
We examine this argument with survey data on 5000 respondents in eight West European metropolitan areas in France, Germany, Switzerland, and the UK. The results suggest that mass support for vertical and for horizontal reforms predominantly hinges on political ideologies concerning demarcation and integration. Citizens thus seem to perceive local governance reforms that involve the reallocation of political authority in similar ways that they perceive multilevel governance reforms at other territorial scales.

Attitudes towards local autonomy and inter-local cooperation
Subnational governance in Europe has changed substantively since World War II. Political authority has been decentralized from the national to the regional and to the local level (see Goldsmith and Page 2010;Hooghe et al. 2010). At the same time, many countries have engaged in local governance reforms-such as the consolidation of the local government landscape or the strengthening of intermunicipal cooperation (Hulst and van Montfort 2007;Baldersheim and Rose 2010). These two processes are intertwined, as Van Houwelingen (2018, 198) observes: Administrative reform and the transfer of responsibilities from the national or regional government to local governments is often accompanied or preceded by a process of municipal amalgamations.
More particularly, we expect that many citizens prefer either vertical autonomy or horizontal inter-local cooperation. This expectation ties in with the long-running debate on the organization of local government in city-regions (see Savitch and Vogel 2009). For a long time, this debate pitted (neo-)progressive advocacy of governmental consolidation and centralization against celebrations of local autonomy and competition rooted in public choice theory. Emerging in the 1990s, a third perspective in this debate emphasizes the role of inter-local governmental cooperation as a functional equivalent to centralized government. But while inter-local cooperation allows capturing scale economies in public service delivery, it factually hampers local autonomy as it increases horizontal interdependencies and reduces local democratic control. Local autonomy is thus not only challenged by formal centralization, but also by functional intergovernmentalism.
Based on existing research, we expect that citizens' attitudes towards local autonomy and inter-local cooperation hinge on their behaviour in a local context and on their emotional attachment to that local context. Moreover, we expect that citizens' preferences for local governance reforms are also a function of their general political ideology regarding the 'politics of scale'. These three sets of factors capture different dimensions of citizens' relation to the local. In what follows, we present three sets of hypotheses that explain these assumptions.

Local ties
The first hypothesis revolves around individuals' local ties. Local ties refer to individuals' behavioural connection with a local community-for example their residence tenure, or their involvement in local associations. The extant literature on this topic argues that local ties are crucial for a person's social integration in a local community. Persons with strong local ties tend to be more integrated in a municipality and hence attribute more value to the local community than persons with weak local ties (Bühlmann 2012). The strength of citizens' local ties and their integration in a local community might in turn affect their attitudes towards local autonomy and inter-local cooperation. We expect that citizens with strong local ties are more supportive of local autonomy, but less supportive of inter-local cooperation.

H 1a
The stronger citizens' local ties, the more they support local autonomy.

H 1b
The stronger citizens' local ties, the less they support inter-local cooperation.
The rationale behind this is that local autonomy emphasizes the importance of the local community to which persons with strong local ties are attached. By contrast, inter-local cooperation means that political authority must be shared with other communities than the one these persons have established strong local ties with. Consequently, they view inter-local cooperation more critically. Existing research supports this argument: Owens and Sumner (2018) found that homeowners and persons with longer local residence tenure are less supportive of sharing local tax revenues with other local governments in their region than tenants and persons with short residence tenure.

Place attachment
The second hypothesis concerns individuals' emotional attachment to their place of residence. Local attachment is an important explanatory variable for political attitudes and behaviour (Manzo and Perkins 2006). Collignon and Sajuria (2018) show that persons with strong local identities attribute more importance to the local rootedness of political candidates and they are more likely to vote for local independent candidates (Otjes 2018). In a review article on place attachment, Lewicka (2011) reports that people with stronger local attachment are more likely to mobilize against unwanted planning decisions and Bonaiuto et al. (2002) demonstrate how locals are more attached to their place and hold more negative attitudes towards the creation of national protected areas by national governments in their region than non-locals-arguably because such a project is perceived as an external intrusion in local affairs by higher government tiers and activates local attachment. Finally, regionalism and federalism scholars exploring citizens' attitudes towards decentralization have shown that preferences for strengthening regional authority are more pronounced among those with stronger regional attachment (Cole and Baudewyns 2004;Henderson et al. 2013).
We argue that local attachment is also linked to attitudes towards local autonomy and inter-local cooperation. We expect that those with a preference for the local as compared to other spatial scales are more supportive of local autonomy and more skeptical of inter-local cooperation. Increasing local autonomy means attributing more importance to the local level, which people with a preference for the local scale strongly care about. At the same time, inter-local cooperation means sharing local political authority, which diminishes the independent political control of a local community over its resources and policies.

H 2a
The more citizens feel attached to the local scale as opposed to other scales, the more they support local autonomy.

H 2b
The more citizens feel attached to the local scale as opposed to other scales, the less they support inter-local cooperation.
In his study on attitudes towards metropolitan governance in two Swedish cityregions, Eklund (2018) provides evidence for the second hypothesis. He shows that those who feel mostly at home in their municipality support the status quo and are more critical of metropolitan reform.

Political ideology and state scales
Our third hypothesis concerns ideological drivers of citizens' attitudes towards the transformation of local governance. The strengthening of both local autonomy and inter-local cooperation relates to the allocation of political authority across different scales. But territorial scales of the state are not primordial: they are socially constructed and politically contested (Delaney and Leitner 1997;Brenner 2004). In recent decades, this issue has been politicized at the international level along the GAL-TAN conflict (Hooghe et al. 2002). Supporters of green/alternative/liberal (GAL) parties have a cosmopolitan orientation and thus support the reallocation of political authority from national to supranational institutions, while supporters of traditionalist/authoritarian/nationalist (TAN) parties have a more communitarian orientation and are opposed to shifting authority beyond the nation-state (Kriesi et al. 2008;De Wilde et al. 2019). Recently, scholars have argued that the allocation of political authority at the subnational level mobilizes voters along the same conflict. Both Rösel (2017) and Fitzgerald (2018) show that TAN parties can successfully tap into protest mobilization against reallocations of political authority at the local level, i.e. municipal mergers or inter-municipal cooperation arrangements. In a similar vein, Strebel (2019) shows that the local vote share of TAN parties is negatively correlated with the success of municipal merger projects in local referendums-presumably because voters of these parties are more concerned with local self-determination and sovereignty. These findings suggest that the GAL-TAN divide is also relevant for attitudes towards local autonomy and inter-local cooperation. TAN partisans can be assumed to support local autonomy-because this means allocating authority to more immediate communities-whereas they oppose inter-local cooperation, given that this means sharing authority with other communities.
What about GAL parties? Do their supporters just mirror the reactions of TAN voters to political authority shifts, i.e. oppose local autonomy and support interlocal cooperation? It might not be that simple. Left parties indeed tend to advocate centralization, mainly because re-distributive policies and 'big government' are easier to fund and organize at higher territorial scales (Peterson 1981;Lowery 2000). However, especially green and alternative parties also tend to value decentralization, grassroots democracy, as well as empowerment of local communities (Stavrakakis 1997, 275). While GAL supporters might thus be internally divided on the question of local autonomy, we can expect them to support interlocal cooperation. The latter can indeed be seen as mutually beneficial interactions among autonomous but connected and interdependent communities-much like international cooperation-that not only foster pareto-efficient responses to supra-local problems, but also allow addressing issues of spatial equity and redistribution (Kübler and Rochat 2019). We can thus formulate the following hypotheses on how political ideologies, captured by the GAL-TAN divide, influence attitudes towards local autonomy and inter-local cooperation:

Case selection and data
For the empirical test of our hypotheses, we use data from the 'Democratic Governance and Citizenship Regional Survey' (NCCR Democracy 2016). The survey's aim was to gauge citizens' perceptions of democratic legitimacy beyond the state in eight West European metropolitan areas in France, Germany, Switzerland and the UK . 1 The metropolitan context represents a hard case to test our argument. In metropolitan areas, the mismatch between politico-administrative and functional structures is particularly pronounced and citizens witness the consequences of potential deficiencies in local governance structures in their everyday lives, for example in transportation or spatial planning. Therefore, we would expect that if the desire for local autonomy would coincide with the desire for inter-local cooperation, this would most likely be the case in metropolitan areas-due to the heightened problem pressure in these contexts. If we do not find a positive association of these attitudes here, we can be rather confident that local autonomy and interlocal cooperation support constitute two distinct dimensions in other contexts too.
The four countries and the eight metropolitan areas represent a diverse set of cases. Table 1 presents the case selection, as well as the distribution of respondents. For each city-region, the sample was stratified in order to reflect the population distribution between the centre city and the suburbs of the metropolitan area. 2 We have selected two federal and two unitary countries with different levels of local autonomy as measured by the Local Autonomy Index (Ladner et al. 2019). The metropolitan areas were selected based on their metropolitan governance structure-i.e. whether they have a metropolitan government-like institution or not (Lefèvre 1998). In addition, we have selected the capital city and a second major metropolitan area in each country. Both a country's and a metropolitan area's institutional structure as well as the city-region's role in the country's political system can influence citizens' perceptions of local autonomy and cooperation. For example, citizens in a country where autonomy of local governments is high might be more favourable towards local autonomy, because they are already familiar with decentralized structures. Similarly, residents of metropolitan areas with a metropolitan government might have different views on local cooperation, based on their experience with existing institutions. In sum, this diverse selection of cases allows us to test our hypotheses in a variety of different contexts.

Operationalization
To operationalize citizens' attitudes towards local autonomy and local cooperation, the two dependent variables in our study, we rely on six items from the survey. The three items to operationalize citizens' attitudes towards local autonomy stem from a question about respondents' opinion on whether (a) things would work better if local or if national governments would take important decisions, (b) local or national politicians know better what citizens want, and (c) local or national governments should determine tax rates. The respondents could place themselves on a scale ranging from 0 (preference for local level) to 10 (preference for national level). These three items tap into important dimensions of local autonomy, such as local decisionmaking authority, local representation, and fiscal autonomy (see Ladner et al. 2019). We recode the three items so that low values mean preference for the national and high values preference for the local level.
We equally rely on three items to measure citizens' attitudes towards cooperation between local governments. Citizens were asked whether they think that (a) their local officials should take the interests of other municipalities in the same region into account before making decisions, (b) local governments in the same region should help each other financially in the event of fiscal difficulties and (c) intermunicipal cooperation is a good way to deal with public problems in their metropolitan area. Again, these three items capture crucial dimensions of inter-local cooperation such as interest negotiation, fiscal equalization, and joint problem-solving (see e.g. Hulst and van Montfort 2007;Sellers et al. 2017). 3 Figure 1 shows the mean values for the six items across the eight metropolitan areas. While it is evident that the extent of agreement differs across metropolitan areas, we can also see that the rank order of the six items is surprisingly similar. With respect to local autonomy, respondents in (almost) all eight metropolitan areas show most support for the statement that local representatives are better suited to represent citizens' interest, closely followed by the statement that things would work better if more decisions were taken locally. The statement that local governments should be able to determine their own tax rates receives the least support across all metropolitan areas. With respect to inter-local cooperation, a difference between the two federal and the two unitary countries emerges. Intermunicipal cooperation as a way of dealing with metropolitan problems is the item that is most positively evaluated in the two federal countries (i.e. Switzerland and Germany), followed by the consideration of other local governments' interests, whereas it is the opposite in the two unitary countries (i.e. France and the UK). In all eight metropolitan areas, equalization payments as a form of inter-local cooperation and spatial equity receive the least support. This ties in with Henderson and Wyn Jones' finding (2021) that economic solidarity across British regions is the most contested dimension of their 'subjective unionism' concept.
To test whether the six items indeed capture two different attitudinal dimensions, we conducted an exploratory factor analysis. 4 Table 2 shows that the six items load on two distinct factors with which the respective items are strongly associated. This result is remarkable, particularly for the inter-local cooperation dimension which involves rather distinct items that involve both interest negotiations as well as redistribution. 5 It suggests that we are indeed dealing with two Table 2 Exploratory factor analysis of citizen's attitudes towards local autonomy and inter-local cooperation Exploratory factor analysis with oblique promax rotation in Stata (-factor, pcf-oblique, promax-); N = 3982

Local autonomy
Inter-local cooperation distinct latent dimensions which are meaningful to citizens in the eight metropolitan areas. 6 Figure 2 shows how the normalized factor scores for the two dimensions relate to each other. The two factors are negatively correlated which suggests that respondents supportive of local autonomy indeed tend to be more critical of inter-local cooperation and vice versa. Yet, the two dimensions are not mutually exclusive: there is a substantial number of respondents who support both local autonomy and inter-local cooperation. This latter group of respondents ties in with the argument of 'polycentrists' in the metropolitan governance debate, who tend to present intergovernmental cooperation as centralization cum local autonomy (see Wright et al. 2011).
To operationalize local ties, we use four different items that figure prominently in the existing literature: residence duration, homeownership, commuting, and membership in local associations. 7 Each of these indicators captures a different aspect of local ties. After some initial years of residing in a certain place, people can be Fig. 2 Scatterplot of local autonomy and inter-local cooperation 6 A concern of comparative research is equivalence, i.e. whether a set of indicators measures the same thing across different contexts (Van Deth 1998). The same item-factor relation across different contexts is an indicator for equivalence. To assess whether we find the same item-factor structure across all eight metropolitan areas, we conduct separate factor analysis for each metropolitan area. Table C.1 in Online Appendix C shows that the six items load on the same two factors in all eight metropolitan areas. 7 Question wordings for the various items can be found in Table B.1 in Online Appendix B. considered knowing a local community sufficiently to start feeling 'rooted', 8 whereas homeowners have a financial tie with a certain place and hence have an interest in influencing the future of this place (Bühlmann 2012). Commuters, by contrast, are spatially less tied to their place of residence than less mobile persons, and hence might develop an interest in inter-local issues (Lidström 2018). Finally, members of local associations have a strong social tie with their place of residence place and hence have an interest in influencing its future (Fitzgerald 2018). 9 Place attachment is operationalized via a question asking respondents to indicate their level of attachment to different territorial scales-from local to global. The answers are used to construct a rank-ordered indicator of local attachment. Higher levels indicate that individuals feel more attached to the local as opposed to other territorial scales. 10 To operationalize respondents' preferences for GAL or TAN parties, we code whether respondents report to feel close to no or another party, a TAN party, or a GAL party. 11 In addition to these independent variables, we also include local context characteristics in our model. Perceptions of local autonomy and local cooperation might also depend on local experiences. Citizens in economically fragile municipalities might see cooperation as a way of coping with functional pressures. We use median income and unemployment rates at the municipality level to control for this. Similarly, citizens in peripheral and small municipalities might hold different perceptions of autonomy and cooperation. They might be more critical of cooperation, because they might fear to be overpowered by large and central municipalities in joint governance institutions. Finally, we also control for the role of socio-demographic attributes (age, gender, education, and income). 12 8 We dichotomize the indicator at more or less than 3 years of residence, since existing literature shows that behavioural changes concerning the local level come about approximately after this duration of residence (Magre et al. 2014). 9 We have tested whether these four indicators can be collapsed into a local ties index using a Mokken scaling procedure on the dichotomized items. This is not the case and suggests that the four items capture different aspects of local ties (see also correlation matrix in Table C.3). 10 Here, it is important to make a distinction between absolute attachment and relative attachment to communities at different scales. Existing research shows that citizens generally tend to be 'community identifiers' or 'non-identifiers'. If someone feels emotionally attached to the local community, this person is also more likely to be attached to communities at other spatial scales. Community identifiers believe 'in the personal fulfillment of personal needs through collective action' (Davidson and Cotter 1993, 60), whereas those who do not identify with communities tend to be more politically disaffected (Anderson 2010). For local autonomy and local cooperation, we might thus expect that community identifiers are opposed to local autonomy-since this essentially means dividing relevant communities-whereas they support local cooperation since this indicates collective action. For the case at hand, we are not so much interested in the general difference between identifiers and non-identifiers but rather in their relative preference for the local community over communities at other spatial scales (see also Henderson and Wyn Jones 2021). 11 GAL and TAN parties were coded based on the 2014 Chapel Hill Expert Survey (Polk et al. 2017). Figure B.1 in the Online Appendix shows the salience of and the position on GAL-TAN issues for political parties in the four countries. Coded as GAL parties (upper-left corner) are those with salience > .6 and position < .2; TAN parties (upper-right corner) are those with salience > .6 and position > .8. 12 Descriptive statistics for all variables can be found in Table C.4 in Online Appendix C.

Estimation
Due to the hierarchical nature of our data-respondents are nested in municipalities and in metropolitan areas-we use linear multilevel regression models with random intercepts (Hox 2010). Respondents constitute the lower level in our analysis, whereas municipalities constitute the higher level. Given the limited number of metropolitan areas, we cannot incorporate a third level in our multilevel model. Therefore, we include metropolitan-area fixed effects to account for variation in the dependent variables across metropolitan areas. Random effects ANOVAs with municipalities, metropolitan areas, or countries as level-2 show that variation (intraclass correlation) in the dependent variable exists mostly at the level of the municipality, and less so at the level of the metropolitan area or the country (see Table  C.5 in Online Appendix C). This suggests that two respondents living in the same municipality are more alike than two respondents living in different municipalities, but that differences between respondents from different metropolitan areas and countries are not necessarily more pronounced than differences between respondents from the same metropolitan area or country. Respondents with missing values for dependent or independent variables are deleted listwise from the regression models.

Local autonomy versus inter-local cooperation
The three hypotheses formulated in Sect. 2 postulate that citizens with stronger local ties, stronger local identification and a preference for TAN parties favour local autonomy, while those with weaker local ties, weak local identification, and a preference for GAL parties favour local cooperation.
To ease interpretation, we use coefficient plots to illustrate the findings for our hypotheses (Fig. 3). The presented coefficients are taken from the full model (see Table C.6 and C.7 in Online Appendix C). Figure 3 shows the results for the three sets of hypotheses. The results for the local ties' indicators do not support our hypotheses. We do not find longer term residents and local association members to be more supportive of local autonomy and, contrary to our expectations, homeowners are less and not more supportive of local autonomy. The only result that is in line with our expectation for local ties is that frequent commuters are less supportive of local autonomy than less mobile individuals. With respect to support for inter-local cooperation, we do not find any differences depending on residence duration, homeownership, or commuting frequency. Against our hypothesis, we find that local association members are more and not less supportive of inter-local cooperation. This difference amounts to .04 on the inter-local cooperation scale. In the light of these results, we reject hypothesis H 1a and H 1b .
Turning to the second set of hypotheses we do not find that citizens with a stronger relative attachment to the local level are more supportive of local autonomy or less supportive of local cooperation. While there is a tendency for those with higher relative local attachment to support local autonomy and oppose inter-local cooperation, these coefficients are not statistically significant at the 95% threshold. 13 For the third set of hypotheses we find, first, that TAN supporters are indeed more favourable of local autonomy and more opposed to local cooperation, lending support to both hypotheses H 3a and H 3c . The substantive effects are also not negligible. TAN supporters rank .05 points higher on the local autonomy scale, and .06 points lower on the local cooperation scale than the baseline group of people supporting no or other parties.
By contrast, we do not find GAL supporters to be less supportive of local autonomy than the baseline which is in line with H 3b . If we take a more detailed look at the results for the individual local autonomy items, an interesting pattern emerges. GAL supporters-like TAN supporters-believe that local interest representation is superior to national interest representation ( Figure C.2), but they are opposed to the idea of local governments collecting their own taxes. This is perfectly in line Fig. 3 Main results 13 If we include measures of absolute attachment to the local and the national scale in the model, we find those with a stronger relative local attachment to be more in favour of local autonomy and more opposed to local cooperation which is in line with hypothesis H 2a and H 2b (see Figure C.4). This might be the case because the inclusion of absolute attachment indicators filters out a relevant distinction that is captured by attachment questions, namely between respondents who generally feel attached to communities and those that are not attached (see argument for using relative attachment in endnote 10 in the operationalization section). When we account for this in our model, we find that relative attachment to the local scale indeed matters for autonomy and cooperation support in the predicted ways.
with green and left ideology as outlined in the theoretical argument. Local interest representation emphasizes the grassroots component whereas national tax collection represents support for 'big government', equality, and redistribution. For inter-local cooperation, we also find the expected pattern: GAL party supporters, are more in favour of local cooperation than the baseline group-corroborating H 3d . This effect is also rather substantive. GAL partisans are placed .034 points higher on the local cooperation scale. The main difference between TAN and GAL partisans thus seems to lie in their position on local cooperation. This mirrors the situation at the national level, where these two groups are positioned on opposite ends of the demarcation-integration cleavage with respect to international cooperation and integration (Kriesi et al. 2008;Hooghe and Marks 2018).
With respect to the control variables-socio-demographics as well as local context indicators-we find that older and wealthier persons are more supportive of strengthening local autonomy, as well as those living in small and economically well-off municipalities. In addition, more educated individuals tend to be more supportive of local cooperation (for full regression models, see Table C.6 and C.7 in Online Appendix C). Moreover, we find intercept differences across countries: French respondents are generally more supportive of strengthening local autonomy than Swiss and British respondents, whereas British respondents tend to be more in favour of inter-local cooperation than respondents in the other three countries.
We run several tests to assess the robustness of our results (see Online Appendix C). First, we assess whether existing supralocal institutions in the eight metropolitan areas are related to citizens' support for local autonomy and inter-local cooperation. We do not find differences between metropolitan areas with or without a metropolitan government-like structure, nor between municipalities that are and are not part of the jurisdiction of a metropolitan government. Yet, residents of capital city-regions tend to be more supportive of inter-local cooperation (see Figure C.1 in Online Appendix C).
A second concern one might have is that relative local attachment mediates the effects of the local ties' indicators. Homeownership, residence duration, and commuting have been found to be linked to local attachment in previous research (Bühlmann 2012). To assess whether this is a reason for concern, we have estimated separate regression models for each of the three set of hypotheses, as well as a baseline model including only control variables. The results suggest that the local attachment variable does not mediate the effects of the indicators for local ties. 14 Finally, the effect of political ideology might simply be an artefact of multilevel government-opposition dynamics. Instead of fundamental beliefs about scale politics, partisanship might solely matter because 'your' party is in power at the local but not at the national level and hence you prefer local autonomy whereas you prefer centralization if your party is in power at the national level. We test this possibility with relative local trust as a proxy for local versus national government support. Including this indicator in our analysis does not alter the results for the GAL-TAN variable (see Online Appendix C for a more detailed explanation).
These various robustness checks support our main results. Citizens indeed have differentiated-and sometimes opposed-views on local autonomy and inter-local cooperation (Table 3).While especially local ties do not seem to play out in the expected ways and while local attachment is significantly linked neither to support for local autonomy nor for inter-local cooperation, we find clear evidence for a strong role of political ideology. Supporters of right-wing nationalist parties support local autonomy but oppose inter-local cooperation. Supporters of green, alternative and left parties are divided on the issue of local autonomy but support inter-local cooperation. This suggests that citizens' beliefs about the politics of scale are an important factor for understanding their attitudes towards multilevel governance reforms.

Conclusion
In this paper we have assessed the attitudes of citizens living in eight West European metropolitan areas towards two of the currently most common tendencies in the transformation of local governance: strengthening (vertical) local autonomy and increasing (horizontal) inter-local cooperation. We hypothesized that these attitudes are shaped by individuals' relation to the local-captured by behavioral (local ties) and emotional (local attachment) as well as by ideological factors (operationalized through the GAL-TAN divide).
The empirical analysis of respondents from French, German, Swiss, and British metropolitan areas suggests that citizens' attitudes towards local autonomy and towards inter-local cooperation constitute two separate latent dimensions that are captured by distinct items. The results of our analysis also show that the attitudes towards these two dimensions of local governance are related to similar factors. More precisely, these attitudes are neither strongly related to respondent's ties to their local community, nor to their emotional attachment to local places. Instead, political ideology is key: supporters of TAN parties are more favourable of vertical local autonomy than supporters of GAL parties. The latter, in contrast, are clearly less critical of horizontal inter-local cooperation.
Emphasizing the ideological component of attitudes towards the transformation of local governance, our findings thus buttress the argument that the rescaling of state territories is subject to political contestation not only at the supra-national, but also at subnational levels of government. Our analysis shows that scale politics at the metropolitan level entail concerns for self-determination and sovereignty mobilized by TAN parties as well as aspects of cooperation and interdependence mobilized by GAL parties.
More generally, our study suggests that there is a link between national demarcation and localism. Following Heinisch and Marent (2018), we can argue that nationalist right-wing populist parties engage in 'territorial claims making' that not only politicizes supranational integration, but also has consequences for scale politics at the subnational level. Indeed, TAN parties mobilize voters with demarcationist positions not only in debates on national or international authority (Kriesi et al. 2008;Hooghe and Marks 2018;De Wilde et al. 2019)-they adopt the same positions with respect to local and national authority (Fitzgerald 2018;Strebel 2019). Our paper adds to the further understanding of the multi-level nature of the demarcation-integration divide, as it suggests that this divide is also structuring citizens' attitudes with respect to state rescaling at the metropolitan level: cosmopolitans favour upscaling and integration while communitarians favour downscaling and demarcation.
Michael A. Strebel is post-doctoral researcher at the Institute of Political Studies at the University of Lausanne. His research interests include public opinion, multilevel governance, territorial reforms, local elites, and democratic legitimacy. He recently published work in the European Journal of Political Research and in Local Government Studies. Daniel Kübler is professor at the Department of Political Science and codirects the Centre for Democracy Studies at the University of Zurich. His research interests include urban politics and policy, multilevel governance and democracy, representative bureaucracy, as well as Swiss politics. Recent work has appeared in the Journal of Public Administration Research and Theory and in the European Journal of Political Research.