Introduction
Spatial differences within countries have been linked by many scholars to the recent success of populist and radical right parties in the Western world. In the 2016 US presidential election, Donald Trump’s success emerged in towns and rural areas, while large cities largely supported Hillary Clinton (Monnat and Brown, Reference Monnat and Brown2017). Similarly, in Europe, the populist vote is concentrated in districts characterized by local economic decline, low employment rates, and less educated populations (Dijkstra et al., Reference Dijkstra, Poelman and Rodríguez-Pose2020), and people living in rural areas are more likely to vote for anti-EU parties than those living in cities (de Dominicis et al., Reference de Dominicis, Dijkstra and Pontarollo2020). In the UK, the 2016 Brexit vote revealed a country divided between pro-European metropolitan areas on the one hand and towns and rural areas claiming national sovereignty on the other (Jennings and Stoker, Reference Jennings and Stoker2019). In France, the Yellow Vest protest movement originated in sparsely populated regions, and spatial mobility was central to their revindication (Jetten et al., Reference Jetten, Mols and Preya Selvanathan2020).
In light of these elements, scholars have argued that geographical patterns of political discontent are emerging in the Western world, and the urban–rural axis would be a key dimension of the so-called “geography of discontent” (Dijkstra et al., Reference Dijkstra, Poelman and Rodríguez-Pose2020). According to this idea, the economic inequalities that have emerged within countries have led to growing political discontent in regions characterized by local economic decline. The citizens living in the so-called “places that don’t matter” (Rodríguez-Pose, Reference Rodríguez-Pose2018) would then have started to use the ballot box to take revenge on the political system and the traditional parties.
Our study puts this narrative of the urban–rural dimension of the geography of discontent in Europe to the test. Rather than analyzing the populist and radical right vote, we focus on its roots: political discontent. While votes can be influenced by campaigns and individual candidates, we want to focus on one of the underlying grievances of public opinion. We analyze urban–rural differences in political discontent by using the survey question on satisfaction with how democracy works, a widely used indicator of people’s satisfaction with political institutions (Canache et al., Reference Canache, Mondak and Seligson2001; e.g. Daoust and Nadeau, Reference Daoust and Nadeau2021; Foa et al., Reference Foa, Klassen, Slade, Rand and Collins2020; Lago, Reference Lago2021). It is a single question asking people “On the whole, how satisfied are you with the way democracy works in [country]?”. We want to know whether people living on the outskirts, in small cities and especially in rural areas are less satisfied with the way political institutions work, than people living in large cities. And we also want to describe the historical trends of these differences: has political satisfaction on the outskirts, small cities and rural areas declined over the last two decades compared to big cities? Our analysis contributes to the understanding of the urban–rural dimension of political satisfaction over time and thus to the description of the geography of discontent in Europe.
Some recent studies have identified geographical patterns of place-based resentment in some countries, such as Germany (Arzheimer and Bernemann, Reference Arzheimer and Bernemann2023), the Netherlands (Huijsmans, Reference Huijsmans2023), Canada (Borwein and Lucas, Reference Borwein and Lucas2023), and the US (Munis, Reference Munis2020). However, place-based resentment, which depends on people’s specific attachment to the place where they live and the feeling that it is not getting its fair share, is only one aspect that shapes spatial disparities in political discontent (Arzheimer and Bernemann, Reference Arzheimer and Bernemann2023).
More generally, previous research has found that there is an urban–rural divide in political attitudes among European citizens. This concerns attitudes toward the political system, as well as attitudes toward specific issues such as the welfare state, the police or migration (Kenny and Luca, Reference Kenny and Luca2021; Maxwell, Reference Maxwell2019). Using ESS data from the period 2002–2018, Kenny & Luca (Reference Kenny and Luca2021) show that satisfaction with democracy (SWD) is significantly lower on the outskirts, small cities, and rural areas than in large cities, and Lago (Reference Lago2021) comes to similar conclusions. However, these studies make it difficult to interpret the magnitude of the spatial differences, as they do not provide comparisons with other relevant social cleavages. Moreover, they only present aggregate results at the European level or by European macro-region, thus masking potential heterogeneity between countries. It is not clear whether the urban–rural divide is a significant dimension of political discontent across Europe, or whether it is salient only in some countries. Finally, little is known about the relative trends in political satisfaction in different types of places.
Based on individual-level data from the European Social SurveyFootnote 1 for the period 2002–2020 for 19 countries, our paper analyses levels and trends of political satisfaction, in four types of places: large cities, outskirts of large cities, small cities, and rural areas. We do not attempt to specifically identify place-based discontent, but use SWD as an indicator of people’s overall satisfaction with the political system in different places.
In what follows, we discuss why we would expect to find spatial differences in political satisfaction and outline our main hypotheses. We then present our data and methods and show that urban–rural differences in political satisfaction in Europe are rather small compared to the much larger differences between countries on the one hand, and to other large social divides within countries on the other. Moreover, political satisfaction has not declined in either small cities or rural areas over the last two decades. In the light of our findings, the recent success of populist parties in small towns and rural areas may be a reflection of historically rooted spatial cleavages rather than a symptom of an exceptional outbreak of political discontent in these areas. Thus, there is some important heterogeneity in terms of spatial cleavages in political satisfaction across countries, highlighting the importance of their specific contexts.
Theoretical framework
Satisfaction with democracy as an indicator of political discontent
Satisfaction with how democracy works (hereafter SWD) is a measure of political support (Canache et al., Reference Canache, Mondak and Seligson2001). Following Easton’s (Easton, Reference Easton1965, Reference Easton1975) tripartite model of political support – the political community, the regime, and the authorities – SWD concerns the level of the regime. It then concerns the form of government itself and its formal and informal rules, rather than specific institutions or individuals in power at any given time. Moreover, building on the distinction made by Norris (Reference Norris1999), SWD does not account for the legitimacy of democratic principles at an abstract level, but rather indicates citizens’ support for the regime’s concrete performance. SWD measures people’s assessment of the regime’s effectiveness in delivering goods (Linde and Ekman, Reference Linde and Ekman2003).
Indeed, the economy is a key dimension of people’s satisfaction with the democratic regime (Christmann, Reference Christmann2018; Daoust and Nadeau, Reference Daoust and Nadeau2021; Quaranta and Martini, Reference Quaranta and Martini2016). At the same time, SWD is related to political processes and the institutional context (Norris, Reference Norris2011). At the individual level, SWD is positively influenced by perceptions of government responsiveness (Linde and Peters, Reference Linde and Peters2020) and by citizens’ democratic attitudes such as believing that it matters who they vote for (Ridge, Reference Ridge2022). Research also suggests that, while SWD is more strongly linked to citizens’ perceptions of the state of the economy in poorer countries, political considerations such as feeling represented are particularly relevant in rich countries (Daoust and Nadeau, Reference Daoust and Nadeau2021). Finally, SWD appears to be strongly correlated with several indicators of confidence in political institutions (Canache et al., Reference Canache, Mondak and Seligson2001). Even though scholars have long debated the ultimate meaning of this item (see Canache et al., Reference Canache, Mondak and Seligson2001; Singh and Mayne, Reference Singh and Mayne2023), the above evidence suggests that SWD is a useful and synthetic indicator for examining citizens’ overall political support and, consequently, political discontent.
SWD is also particularly useful for studying spatial inequalities because it is linked to both the material conditions of individuals and the contextual factors that shape their experiences. On the one hand, more affluent people tend to be more satisfied because they have access to high-quality goods, which leads them to evaluate the performance of the democratic regime more positively (Nadeau et al., Reference Nadeau, Daoust and Arel-Bundock2020). On the other hand, previous research has shown that SWD is lower among people who have experienced poor public services and have negative evaluations of local government performance (Weitz-Shapiro, Reference Weitz-Shapiro2008). Differences in levels of SWD across national territories could therefore be found, particularly where there are significant inequalities between regions in terms of the state of the labor market or the provision of public services. Finally, individual and contextual factors of SWD are interrelated, as lower status citizens are also more dependent on state transfers and services, and the lack of individual resources to cope with difficult circumstances makes them more vulnerable to general economic fluctuations (Nadeau et al., Reference Nadeau, Daoust and Arel-Bundock2020).
Our analyses do not investigate the causal mechanisms between contextual conditions and individual SWD, but they build on the demonstrated link between citizens’ SWD and the context in which they live in order to study the urban–rural divide in political satisfaction.
Why would we expect an urban–rural divide in political satisfaction?
The rural–urban divide has been a classic dimension of political cleavages since the emergence of modern states (Lipset and Rokkan, Reference Lipset and Rokkan1967). Since the industrial revolution, cities have been inhabited by workers in manufacturing industries, while rural areas have been dominated by the people engaged in agricultural production, two groups with different interests. Thus, as modernization theory suggests, economic development and demographic growth in cities have been followed by a shift of their inhabitants toward more liberal and tolerant political views, fueled by the large opportunities for socialization and freedom that cities offer (Luca et al., Reference Luca, Terrero-Davila, Stein and Lee2023).
The urban–rural dimension of political polarization is thus a historical reality. However, the recent decline of the industrial economy and the advent of globalization seem to have reignited it. Some cities have been able to jump into the service economy of the globalized era, while other places have not (Moretti, Reference Moretti2012). The agglomeration advantage of the service economy favors large cities, which tend to concentrate resources and employment opportunities. Smaller cities and rural areas have often been left behind. European metropolitan areas, including capitals such as London, Dublin or Warsaw, are well connected in the globalized economy and attract multicultural and dynamic populations, while other areas are increasingly depopulated, especially rural areas (Hurley et al., Reference Hurley, Fernández-Macías, Bisello, Vacas-Soriano and Fana2019). According to the OECD (2020), the average contribution of capital regions to national GDP increased by almost 3 percentage points between 2000 and 2016, reaching 27%. For example, the Ile de France, the Paris region, now accounts for more than 30% of French GDP.
According to a popular narrative, political attitudes would then have diverged between citizens living in cities – which have benefited greatly from global economic growth – and citizens living in suburban communities, post-industrial towns and the rural periphery, where economic opportunities and public services are scarce. These lagging areas have been dubbed “places that don’t matter” (Rodríguez-Pose, Reference Rodríguez-Pose2018). The argument is that people living there have felt left behind by national elites, have become more closed and communitarian (Jennings and Stoker, Reference Jennings and Stoker2017, Reference Jennings and Stoker2019), and have begun to take revenge through the ballot box. The recent uneven economic development between places would then have led to a cultural grievance, with the inhabitants of the “periphery” turning against their governments, which were accused of only looking after the interests of the urban upper-middle classes. The large spatial differences in political satisfaction would have been made visible by the success of anti-system and populist parties in sparsely populated areas, where the so-called “geography of discontent” emerged (Dijkstra et al., Reference Dijkstra, Poelman and Rodríguez-Pose2020).
This idea is supported by qualitative research in the United States, which describes the sense of being “left behind” felt by residents of rural and post-industrial areas (Cramer, Reference Cramer2016; Hochschild, Reference Hochschild2016). And the current urban–rural divide in political attitudes has also been addressed by a number of quantitative studies. As discussed above, several recent studies have pointed to a divide in political attitudes and trust between urban and rural areas in Europe in recent decades (Arzheimer and Bernemann, Reference Arzheimer and Bernemann2023; Huijsmans, Reference Huijsmans2023; Kenny and Luca, Reference Kenny and Luca2021; Lago, Reference Lago2021; Maxwell, Reference Maxwell2019; McKay et al., Reference McKay, Jennings and Stoker2021, Reference McKay, Jennings and Stoker2023; Mitsch et al., Reference Mitsch, Lee and Ralph Morrow2021).
Two main mechanisms have been highlighted. On the one hand, the differences in political attitudes between urban and rural areas are due to the different composition of these places: rural areas have a lower level of education and a larger proportion of the working class, who face harsher living conditions, while the upper and middle classes are more concentrated in cities. The attractiveness of economically successful cities fuels socio-demographic sorting through internal migration. On the other hand, some research argues that it is the spatial context per se that matters, as urban–rural differences in SWD are visible even after controlling for individual characteristics (e.g. Kenny and Luca, Reference Kenny and Luca2021; Lago, Reference Lago2021). According to this view, people living in these areas would express some kind of place-based political resentment (Arzheimer and Bernemann, Reference Arzheimer and Bernemann2023; Huijsmans, Reference Huijsmans2023).
Both mechanisms contribute to the supposed urban–rural divide in political discontent. Based on these arguments, we hypothesize that there are important differences in SWD between urban and rural areas in Europe. We expect these differences to be relevant even when compared to differences between social groups defined, for example, by social class or education. More specifically, even if some heterogeneity in economic and demographic dynamism is likely to characterize places with similar levels of urbanization, we expect to see a hierarchy of political satisfaction between highly urbanized and poorly urbanized places. Our first hypothesis is therefore as follows:
HYPOTHESIS 1: SWD is lower on the outskirts of big cities, and even more so in small cities and rural areas, than in big cities.
Moreover, given the heterogeneous economic and demographic development within countries, we expect these differences to have widened over the last two decades due to the negative trend in political satisfaction in the more peripheral places. Indeed, some scholars have shown that the spatial gap in political attitudes has recently widened in several European countries, such as the Netherlands (Huijsmans et al., Reference Huijsmans, Harteveld, van der Brug and Lancee2021) and England (Jennings and Stoker, Reference Jennings and Stoker2016), especially between residents of large cities and those living in small cities or rural areas. Mitsch and colleagues (2021) analyzed data from the ESS on 18 countries and showed that divergent trends between urban and rural places can also be observed for trust in the political system, although they only focused on the period 2008–2018. We assess recent trends in urban–rural differences in SWD by testing our second hypothesis:
HYPOTHESIS 1: The gap in SWD between big cities, small cities, and rural areas has widened over the last two decades in Europe due to a negative trend in the less urbanized areas.
Data, measures, and methods
Data
We use individual-level survey data from the European Social Survey, which covers the period 2002–2020 with one round every two years. This database has the advantage of including the SWD variable in each round and providing information on the type of place where respondents live, for a large number of European countries.
We base our analyses on all respondents over the age of 18 years and exclude countries that were not consistently observed over the period studied, so that the estimated trends are not biased by an unbalanced sample. This leaves us with 19 European countries, consisting of about 320,000 individuals over the entire period for the descriptive models and, after listwise deletion of cases with missing variables, about 240,000 for the multivariate models (see Table A.1 in the online Appendix for the total number of available observations by country and year).
Valgarðsson and Devine (Reference Valgarðsson and Devine2022) show that measures of SWD vary across data sources. We therefore attempt to increase the robustness of our results by following the practice of identical analysis of parallel data (2021) and reproduce our main analysis using data from the International Social Survey Programme (ISSP) and the European Values Survey (EVS). SWD is available in the ISSP modules on citizenship in 2004Footnote 2 and 2014Footnote 3 for all 19 countries in this study, while the latest round of EVS (fielded between 2017 and 2021)Footnote 4 provides data for 17 of the 19 countries in this study and allows us to use another variable that is very close to SWD: satisfaction with how the political system works in the country. The results of these analyses are presented in the online Appendix.
Measures
Our main dependent variable is satisfaction with how democracy works (SWD). The question is worded as follows: “And on the whole, how satisfied are you with the way democracy works in [country]?”. The response is coded on an 11-point scale, ranging from “0-Extremely dissatisfied” to “10-Extremely satisfied.” We have recoded this variable on a scale of 0-100 in order to facilitate comparison with the results from the other databases and to facilitate interpretation of the results, as the differences between the groups are small.
The average SWD in Europe is around 52 points (on a scale of 0–100) and has remained stable over the last two decades. However, there are notable differences between countries. The average SWD is 72 in Denmark – the country with the highest score – but only 40 in Slovenia – the country with the least satisfied citizens (for more details, see Figure A1 in the Appendix, which shows the average SWD in each country over the whole period, as well as the SWD trends in some countries selected as examples from different European regions and different SWD levels).
Our main independent variable is the type of place where respondents live. The original geographical variable is based on self-assessment and includes five categories: “a big city; suburbs or outskirts of a big city; town or small city; country village; farm or home in the countryside.” As the last category was chosen by only a few people, we have combined the last two categories into “rural areas.” Overall, 17% of respondents in our pooled European sample live in big cities, 13% on the outskirts of big cities, 32% in small cities, and 38% in rural areas. Slovenia is the least urbanized country, with more than 50% of its citizens living in rural areas, compared to only 24% in the UK (see Table A.2 in the Appendix for the distribution of geographical categories in each country under study).
Based on self-assessment, this variable may reflect different interpretations of what a big city and a small city are for different people and in different countries. This is a major limitation of our study, as with any study based on survey data with limited information on respondents’ place of residence, as subjective and objective definitions of places often do not coincide (Nemerever and Rogers, Reference Nemerever and Rogers2021). At the same time, the proportion of respondents self-identifying as rural in our data is similar to Eurostat statistics in most countries (see Table A3 in the Appendix for a comparison of these features). Moreover, an objective measure based on the number of inhabitants in the municipality or on population density would be less suited to the specific urban structure of each country in our sample: “big city” cannot mean the same size in Germany, where three cities – Berlin, Hamburg, and Munich – have more than one million inhabitants and nine others have more than 500,000 inhabitants, and in Switzerland, where no city reaches this size.
Our multivariate analyses include several control variables to account for individual characteristics that are often heterogeneously distributed across places and that may be correlated with SWD. This allows us to roughly disentangle the two mechanisms contributing to urban–rural differences: the composition of places and the direct effect of place-based discontent. We control for demographic characteristics such as gender and age, for being a citizen of the country of residence and for being a member of the dominant ethnic group. Finally, we control for the socio-economic characteristics of the respondents. We include education (less than secondary, secondary or post-secondary, tertiary), income measured by national decilesFootnote 5 , unemployment status, and social class. We measure class using the 16-class Oesch schemeFootnote 6 (Oesch, Reference Oesch2006), which allows us not only to capture the vertical dimension of social hierarchy in detail, but also to distinguish horizontally between categories of citizens who tend to have different political attitudes, such as managers and sociocultural professionals. We also present results using a collapsed version of the same class scheme with only 5 categories, which allows us to better interpret the magnitude of the class effects.
Previous research has shown that SWD is also influenced by the outcome of recent elections: people who voted for the party that won the government tend to be more satisfied with the way democracy works than those who voted for the losing party (Anderson and Guillory, Reference Anderson and Guillory1997; Daoust and Nadeau, Reference Daoust and Nadeau2021; Han and Chang, Reference Han and Chang2016; Singh et al., Reference Singh, Karakoç and Blais2012). This led Lago (Reference Lago2021) to control for the winner/loser status of the party the respondent voted for in the last election in his analyses of urban–rural differences in SWD. However, we believe that adding this variable to the models would introduce an over-control bias. On the one hand, the hypothesis behind the left-behind narrative is that people vote for anti-system parties as a consequence of their political dissatisfaction. Since political discontent is our dependent variable, it would be wrong to control for vote. On the other hand, the party voted for could mediate the causal relationship between place-based political grievances and SWD, because its effect on SWD depends on who won the election. In this view, the vote is a mediating variable and controlling for it would hide part of the correlation between places and SWD.
Models
We run linear regression models with fixed effects for countries and years, as defined by the following equation:
$$\rm y_i = {\rm{\unicode{x03B2}}_1} + {\rm {\unicode{x03B2}}_2} {\rm place_i} + {\rm {\unicode{x03B2}}}_3 \rm year_i +{\rm {\unicode{x03B2}}_5 country_i} + ({\unicode{x03B2}}_4 place_i\ast year_i)+ {\rm {\unicode{x03B2}}_5} controls_i+ \epsilon_i$$
The year fixed effects account for the general time trends in SWD over time, and the country fixed effects make our estimates of urban–rural differences depend only on within-country variation in SWD, excluding baseline differences between countries that may depend on many institutional, cultural and linguistic (question wording) elements. Our models for estimating trends in SWD differences also include the interaction term place*year, which accounts for the different time trends in SWD between our four place categories. We cluster the standard errors by country to address the issue of the potential error correlation within countries, and we include in each model the weights available in the ESS dataset. We also reproduce the main analysis using multilevel models with three levels (individuals, years, countries).
Results
Urban–rural differences in the aggregate European sample
Figure 1 describes the evolution of SWD in the different types of places in Europe. The left-hand plot shows descriptive trends based on simple regression models with country fixed effects. The results have been smoothed using locally weighted scatterplot smoothing (LOWESS) for a better understanding of long-term patterns (the same plot without smoothed trends can be found in Figure A2 in the Appendix). We can see that SWD is significantly higher in big cities than in small cities and even more than in rural areas. This is the case throughout the period. While in the first decade of the 21st century SWD was also lower in suburban areas than in big cities, this difference disappeared in the second decade. SWD increased slowly in big cities and their outskirts for most of the period. It remained stable in small cities and rural areas until 2016 and then increased slightly. So there is no downward trend in the so-called “peripheral areas.” Even in relative terms, small towns and rural areas have not moved away from the big cities: apart from the early 2000s, when all the edges were closer together, the gap in political satisfaction between places has remained constant. The only relevant change is the increase in SWD on the outskirts of large cities over the whole period.
Figure 1. Trends of SWD (0–100) in the different kinds of places in Europe, 2002–2020 (95% confidence intervals).
The observed differences between places confirm our expectations regarding the hierarchy, but they are small. If we compare them with the differences between countries, for example, we see that the urban–rural divide is far from being the first spatial dimension shaping SWD in Europe. On average, there is only a difference of 2.5 points (on a scale of 0–100) between big cities and rural areas over the period studied, while there is a difference of more than 30 points between the more satisfied countries (Denmark and Switzerland) and the less satisfied countries (Slovenia and Hungary). Focusing on the most populous European countries, we see that SWD in France is on average 10 points lower than in Germany and 6 points lower than in the UK. The full list of estimates from the regression models can be found in Table A4 of the Appendix.
When individual characteristics of the respondents are controlled for, the differences between places become even smaller, meaning that they are largely explained by the different socio-demographic composition of people living in different places. The right-hand plot in Figure 1 describes the SWD trends in the different types of place based on multivariate models controlling for the relevant individual characteristics (the same plot without smoothed trends can be found in Figure A2 in the Appendix). In this case, the average difference between big cities and rural areas over the period is only 1.3 points. Still, there is no negative trend in small towns or rural areas.
The relevance of within-country spatial differences in SWD appears to be even lower when compared to the relevance of differences between other social groups, such as social classes or income groups. Figure 2 provides an overview of some between-group differences by comparing the coefficients associated with several socioeconomic variables. We run seven different models, each including country fixed effects and a different socioeconomic variable (indicated by a different color in the figure). We can see that all the socio-demographic variables used are associated with larger differences compared to our geographical indicator and explain more variance in the dependent variable, as indicated by the adjusted R2. For example, the average difference between nationals and non-nationals is 10 points, and people in the fifth quintile of the income distribution are on average 8 points less satisfied than those in the first quintile. These results are almost identical when using a different modeling strategy, that is, multilevel models (the results of this analysis are presented in the Appendix, Figure A3).
Figure 2. Differences of SWD (0–100) between several social groups in Europe, 2002–2020 (95% confidence intervals).
Robustness analyses
To check the robustness of our results, we also repeated these analyses using two other dependent variables that are conceptually close to and highly correlated with SWD (Canache et al., Reference Canache, Mondak and Seligson2001): trust in parliament, trust in politicians. We also repeated these analyses for external political efficacy, that is, the extent to which respondents believe that the political system allows people like them to have a say in what the government does or to influence policy (we averaged these two items). These results are shown in Figure 3 and confirm our previous analyses: for the three variables, the difference between urban and rural places is less than half the size of the differences between the top and bottom social classes or between the top and bottom income quintiles.
Figure 3. Differences of trust in parliament (0–100), trust in politicians (0–100) and external political efficacy (0–100) between several social groups in Europe, 2002–2020 (95% confidence intervals).
We then replicated the analyses of political satisfaction using two other samples, the 2004 and 2014 rounds of the ISSP (using SWD) and the 2017–21 round of the EVS (using satisfaction with the political system). The results are shown in Figure 4 and are again very similar to those for the ESS sample. The average difference between big cities and rural areas is 2.2 (on an identical scale from 0 to 100) according to the ISSP data, and all available socio-economic indicators are associated with larger differences in SWD. Analyses of the EVS data, which provide a measure of satisfaction with the way the political system works, also confirm that unemployment, social class, education, income, and citizenship are much stronger predictors of political satisfaction than the type of place where people live.
Figure 4. Differences of political satisfaction (0–100) between several social groups in Europe, based on ISSP and EVS data (every model includes country fixed effects).
As a further robustness analysis, we looked at spatial differences within income groups. Differences in SWD between places may be very small because they are only relevant for certain groups, namely the most vulnerable, who are less resilient to income circumstances and cannot, for example, compensate for the lack of public services in their area through their personal resources. Indeed, as some research has shown, low-income citizens give more weight to their perception of the economic context in which they live than high-income citizens do when assessing their SWD (Nadeau et al., Reference Nadeau, Daoust and Arel-Bundock2020). We therefore calculated the spatial differences in SWD within each income quintile, controlling only for basic demographic information, age, and gender (see Figure A4 in the Appendix for the full results). In line with our expectations, the geographical differences are slightly larger within the poorest group than within the richest group. However, this variation is small, as even within the lowest quintile of the income distribution, the difference in SWD between urban and rural residents is only 3.5 percentage points. By comparison, the difference in SWD between respondents in the lowest and highest income quintiles is more than 8 percentage points.
These results suggest that spatial differences in SWD may be more visible among the most disadvantaged, but also that socio-economic groups are much more relevant than places in explaining SWD inequalities. We can say that, overall, spatial differences in SWD are small in the aggregate European sample and that the type of place where people live is far from being the most relevant social cleavage of political satisfaction.
Urban–rural differences in single countries
However, the low relevance of spatial disparities in SWD in the aggregate European sample could mask large differences between countries. We then run our models separately for each country in our sample. Figure 5 shows the differences in SWD between big cities and rural areas for each country over the whole period studied (see Figures A5 and A6 in the Appendix for the differences between big cities and the three other types of place in each country, at the descriptive level and net of individual controls), and Figure 6 shows the trends in SWD by place in the six countries with the largest spatial differences (Figures A7 and A8 in the Appendix show the trends in trust in parliament and trust in politicians), confirming the absence of a generalized polarization across places. We can see that for several countries there are no or only weak spatial disparities, while for others the urban–rural divide is relevant.
Figure 5. Differences in SWD between big cities and rural places by country for the period 2002–2020.
Figure 6. Average SWD in the different kinds of places in the countries with the largest urban-rural differences, 2002–2020 (95% confidence intervals).
Unsurprisingly, this is the case in France, where public debate and scientific research have often highlighted an important spatial divide between the few large cities and the so-called “peripheral France” (Guilluy, Reference Guilluy2015). Important differences between the most urbanized and rural French departments have been shown by scholars in terms of depopulation (Oliveau and Doignon, Reference Oliveau and Doignon2016), but also in terms of wealth (Bonnet et al., Reference Bonnet, d’Albis and Sotura2021) or the subjective status of their inhabitants, a measure of people’s perception of their place in the social hierarchy (Vigna, Reference Vigna2023). SWD is on average 5.7 percentage points lower in rural areas than in big cities in France over the period studied. We also observe a negative trend in SWD in rural areas between 2010 and 2016, when the gap with big cities reaches 8 percentage points, comparable to the gap between those with a tertiary education and those with only a secondary education, and larger than the gap between those in the higher-grade service class and unskilled workers (only 6 points).
Strong urban–rural divides in political satisfaction are also observed in the Nordic countries, Slovakia, and Belgium (where the urban–rural divide may reflect the regional divide between the richer and more urbanized Flanders and Brussels-Capital region and the relatively rural and poorer Wallonia). However, there is no negative trend for rural areas in these countries.
Surprisingly, the urban–rural dimension does not seem to be relevant for political satisfaction in the UK, although the Brexit vote in 2016 has been interpreted by some scholars as a symptom of strong spatial differences between the globalized modern cities and the post-industrial towns and peripheral areas (Jennings and Stoker, Reference Jennings and Stoker2019). In some other countries, such as Poland, Spain, the Netherlands and Austria, living in rural areas is practically not associated with lower SWD than living in a big city.
Overall, our results highlight the heterogeneity between countries in terms of spatial cleavages. And, in terms of their evolution, specific stories have unfolded in each country. Explaining this heterogeneity is beyond the scope of this paper, as a variety of factors could be at play. These include, for example, the different levels of urbanization (Lago, Reference Lago2021), and we could also think of several institutional factors, such as the different levels of centralization-federalism or the different electoral systems – majoritarian vs. proportional. Previous literature has shown that less proportional systems are associated with lower levels of SWD (Aarts and Thomassen, Reference Aarts and Thomassen2008), but they could also particularly affect how people living in peripheral areas feel that the political system takes their needs into account. Cultural specificities in the way rural areas are represented in the public debate may also have an impact on the urban–rural gap in SWD. While it would not be possible to take all these factors into account in the context of this study, our analyses are intended to warn, in a more modest way, that it would be misleading to speak of a generalized urban–rural pattern in SWD over the last two decades in Europe.
Conclusion
We analyze spatial differences in political satisfaction in Europe using SWD, a measure of people’s satisfaction with how democracy works in their country. Based on individual survey data from 2002 to 2020 for 19 countries, we analyze the means and trends of SWD in four types of places: big cities, outskirts of big cities, small cities, and rural areas. We also reproduce our main analyses on two other data sources: the ISSP (2004, 2014) and the EVS (2017–21). Building on the literature on “the geography of discontent” (Dijkstra et al., Reference Dijkstra, Poelman and Rodríguez-Pose2020) and on “the places that don’t matter” (Rodríguez-Pose, Reference Rodríguez-Pose2018), we test the hypotheses that SWD is lower in peripheral areas than in big cities and that this difference has increased over the last two decades.
We show that SWD is slightly higher among people living in big cities than in rural areas, with outskirts and small cities in between. This finding goes beyond the simple urban–rural dichotomy and confirms the expected hierarchy between places in terms of political satisfaction. It is also consistent with previous analyses of SWD by Lago (Reference Lago2021) and Kenny & Luca (Reference Kenny and Luca2021) and of political trust (Mitsch et al., Reference Mitsch, Lee and Ralph Morrow2021), both in terms of the hierarchy and the significance of the observed differences. However, these authors rarely discuss the magnitude of these differences. We show that these differences are indeed very small and are negligible compared to the large differences between countries in Europe. SWD is on average only 2.5 percentage points higher in big cities than in rural areas, while the difference between the countries with the highest and lowest scores is more than 30 points.
In terms of mechanisms beyond this difference, then, the urban–rural divide is largely explained by the composition of places, as it becomes even smaller when controlling for individual characteristics. This discredits the idea of an additional direct effect of place on political discontent, e.g., place-based discontent. Most importantly, by comparing several socio-economic indicators, we show that the differences between the four types of place are small compared to the differences between social groups defined by citizenship, employment status, education, social class or income. For example, more than 8 percentage points separate the average SWD of people in the first and last income quintiles. Socioeconomic factors remain the most important predictors of people’s political satisfaction. Our robustness analyses on two other data sources, the ISSP and the EVS, confirm this result.
Finally, our hypothesis on SWD trends should also be rejected. The gap between big cities, small cities, and rural areas has remained broadly stable over the last two decades, while SWD in the outskirts of big cities has grown somewhat faster. The stronger positive trend on the outskirts of big cities could explain the widening of the urban–rural gap highlighted by some previous studies on political trust in the shorter period 2008–2018 (Mitsch et al., Reference Mitsch, Lee and Ralph Morrow2021). Nevertheless, political discontent does not seem to have increased in small cities and rural areas, either at the descriptive level or when controlling for individual socio-economic characteristics.
Our findings challenge the urban–rural dimension of the so-called “geography of discontent” at the European level. The narrative of growing resentment among citizens living in peripheral areas, who feel abandoned by national elites and take revenge through the ballot box, seems to rest on a weak foundation. In many countries, the success of far-right parties in both urban and rural areas may have more to do with the historical spatial divide between progressive cities and conservative rural areas than with any exceptional increase in spatial inequality. The emergence of a new type of political party in Europe may simply have made this divide more visible by representing the conservative and anti-cosmopolitan values of citizens traditionally living in rural areas.
However, there is a great deal of heterogeneity between countries in terms of urban–rural differences. In some of the countries studied, the gap in SWD between big cities and small cities or rural places is more consistent than in others. In France, rural places are also associated with a negative trend between 2010 and 2016. In this country, the urban–rural divide in political satisfaction may indeed have played a role in shaping the recent geography of voting. France’s sparsely populated areas witnessed the rise of the Yellow Vest protest movement in 2018 (Jetten et al., Reference Jetten, Mols and Preya Selvanathan2020) and the electoral success of Marine Le Pen’s far-right party in the 2022 presidential election (Lévy et al., Reference Lévy, Dorian, Sébastian and Justine2022).
Moreover, in other countries other dimensions of spatial inequalities may be more relevant than the urban–rural divide, such as the west-east divide in Germany, regional inequalities in Spain or the north-south divide in Italy. More generally, the so-called “peripheries,” that is, places far from political, economic and cultural centers, do not necessarily correspond to rural areas (de Lange et al., Reference de Lange, van der Brug and Harteveld2023). And the inhabitants of former industrial and densely populated cities experiencing economic decline may also feel abandoned by their national political elites. In sum, geographical cleavages within countries can be linked to the characteristics of their territories, but also to specific historical, economic, and political processes. Explaining this heterogeneity is beyond the scope of this paper and requires further research.
Finally, a major limitation of our analyses is the fact that the urban–rural variable is self-reported. Unfortunately, survey data that allow analyses on subjective indicators rarely provide accurate objective information on the location of respondents. Studies using richer data would help to describe trends in political satisfaction in different places with greater precision.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1755773924000110.
Acknowledgments
This research was carried out within two projects: NCCR LIVES (Overcoming Vulnerability: Life Course Perspectives), funded by the Swiss National Science Foundation (SNF), and SNF project 204489 (Unequal shifts in regions’ employment structure and the rise of cultural grievances). We are grateful for the support.
Funding statement
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Competing interests
None.
Open access
