Falling behind whom? Economic geographies of right-wing populism in Europe

ABSTRACTExisting studies suggest that right-wing populist parties (RWPPs) appeal to people in communities that have fallen behind in material terms. However, it remains open which benchmark communities apply as they become politically discontented. We argue that the structure of territorial inequalities influences the benchmarks used by people in regions falling behind. Panel data regressions using subnational election results in EU states from 1990 to 2018 reveal a sharp contrast between the economic geographies of right-wing populism in core and peripheral EU member states. We find a strong association between falling behind the richest region of the country and RWPP support within core EU countries, while in peripheral EU states falling behind the EU core is associated with regional support for RWPPs. This suggests that RWPP voters in peripheral countries cue on how they are faring relative to the EU core, while RWPP supporters in core countries cue on how they are faring relative to dynamic regions of their own country. Our analysis also shows that increased manufacturing employment reinforces the effect of falling behind the richest region in core EU member states, while we find no strong evidence that regional economic stagnation is important to the electoral performance of RWPPs.KEYWORDS: Europegeographyinequalityright-wing populism AcknowledgementsAn earlier version of the paper was presented at the Annual Meeting of the American Political Science Association (Montreal) in September 2022 and at a workshop at Copenhagen Business School in October 2022. We thank the participants in both events for constructive feedback.Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 While SD’s national vote share increased by 2.5 points, its Stockholm vote share increased by less than one percentage point. Statistics Sweden, https://www.scb.se/en/finding-statistics/statistics-by-subject-area/democracy/general-elections/general-elections-results/.2 Dropping the three smallest member states (Luxembourg, Cyprus and Malta) and losing some additional observations for lack of data on independent variables, our analysis is restricted to 1,053 regional units in 25 countries. Regional units are NUTS 3 regions for 19 countries and NUTS 2 regions for 6 countries (Belgium, Ireland, Netherlands, Poland, Slovenia and the UK). Countries that have joined the EU since 1990 enter the dataset the year they obtained the status of an ‘accession country.’ The EU-NED dataset and codebook are available at: https://dataverse.harvard.edu/file.xhtml?fileId=6157990&version=1.13 We consider ‘right-wing’ to be interchangeable with ‘far Right’ and ‘radical Right.’ While many recent studies posit common determinants of left-wing and right-wing populism, Gonthier (Citation2023) as well as Burgoon et al. (Citation2019) emphasise differences in the motivations of individuals who support left-wing and right-wing populist parties.4 It goes without saying that there is an important (inter-)subjective component to social status, but social status should not be conflated with the conflicts over cultural values emphasised by Inglehart and Norris (Citation2019). In our understanding, a ‘materialist account’ does not imply that ‘objective conditions’ suffice to explain support for right-wing populist parties. Notable contributions to the literature that link right-wing populism to divergence in regional economic trajectories include Hobolt (Citation2016), McNamara (Citation2017), Rodríguez-Pose (Citation2018), Essletzbichler et al. (Citation2018), Carreras et al. (Citation2019), Djikstra et al. (Citation2019), Schraff (Citation2019) and Adler and Ansell (Citation2020). See Chou et al. (Citation2022) for a broader discussion of the ‘localist turn’ in the study of populism.5 All parties that we identify as ‘right-wing populist’ are also coded by PopuList as ‘Eurosceptic.’ See Djikstra et al. (Citation2019) for a useful discussion of the overlap between populism and Euroskepticism.6 As documented by Hense and Schäfer (Citation2022), perceptions of not having any political voice are closely associated with voting for RWPPs across European democracies. See Lipps and Schraff (Citation2021) on the effects of regional inequality on trust in national political institutions and EU institutions.7 Other studies that identify differences in public attitudes across the divide between rich and poor EU member states include De Vries (Citation2018), Vasilopoulou and Talving (Citation2020), and Mayne and Katsanidou (Citation2022).8 Analyzing party programmes presented in Land elections, León and Scantamburlo (Citation2022) provide a fascinating case study of how the AfD balances efforts to mobilise regional grievances with appeals to German national identity.9 Our argumentation also draws inspiration from Scase (Citation1977). Building on Runciman (Citation1966), Scase shows convincingly that Swedish manual workers are much more likely to compare themselves to (upper) middle-class individuals than their British counterparts. He attributes the difference between his two samples to the structure of national union movements.10 Note that our conceptualisation of the core-periphery distinction is fundamentally economic and thus different from similar distinctions by students of European integration (e.g., Schimmelfennig, Citation2016). Like the empirical analysis that follows, Table 1 excludes the three smallest EU member states. Luxembourg clearly belongs in the EU core by virtue of its GDP per capita ($131,511 in 2021) as well as being a founding member of the EU. The UK features as a core EU member state because our analysis pertains to the period from 1990 to 2018. Note also that all core member states were richer than all peripheral member states already in 1990.11 ARDECO estimates of regional GDP per capita take purchasing power into account. Source: https://knowledge4policy.ec.europa.eu/territorial/ardeco-database_en.12 The rise of regional inequality stands in sharp contrast to stability of personal (‘vertical’) income inequality in the EU core since the financial crisis of 2007–2008: averaging across the ten countries, the top 10 per cent share of pre-tax national income increased by less than 1 per cent from 2006 to 2021 (https://wid.world/). The Swedish case illustrates divergent trends in vertical and horizontal inequalities as well as regional variation in right-wing populist support. As defined by Statistics Sweden, the percentage of adults ‘at risk of poverty’ was 14.2 per cent in metropolitan areas and 14.3 per cent in smaller towns and rural areas in 2010. By 2020, the figure for metropolitan areas had dropped to 11.8 per cent while the figure for smaller towns and rural areas had increased to 20.3 per cent (https://www.statistikdatabasen.scb.se/pxweb/en/ssd/START__LE__LE0101/). Over the same period, the overall Gini coefficient for post-tax national income declined from .31 to .29 (https://wid.world/).13 We prefer a fixed effects specification for the region and year levels rather than random effects as standard error estimates are more conservative. Moreover, there currently are no readily available estimation procedures that allow quasi-binomial link functions in multi-level GLM, as well as a lack of tools to calculate clustered standard errors.14 Our measures of the aforementioned control variables are also based on ARDECO data (https://knowledge4policy.ec.europa.eu/territorial/ardeco-database_en). See Appendix 1 for descriptive statistics on all variables included in our analysis.15 Note that there are 11 core countries and 14 peripheral ones, and still the N of the core-country sample is 4,888 and the N of the periphery-country sample 2,588. This is due to the fact the data for richer countries covers a longer time period and that richer countries tend to have more regions (e.g., Germany has around 400 NUTS3 regions).16 The interaction effects reported in Table 4 are estimated using standardised variables and a double-demeaned estimator (Giesselmann & Schmidt-Catran, Citation2022).17 The non-linear patterns at high values of GDP growth in the left-hand panel of Figure 7 result from statistical extrapolation, with very few observations driving these results (see rug plot below the marginal effects).Additional informationFundingDominik Schraff’s work on this paper was supported by an Ambizione Grant from the Swiss National Science Foundation (grant number 186002). Jonas Pontusson’s contribution was funded by the European Research Council under the European Union’s Horizon 2020 research and innovation program (Advanced grant number 741538).Notes on contributorsDominik SchraffDominik Schraff is associate professor of political science at Aalborg University, Denmark.Jonas PontussonJonas Pontusson is professor of political science at the University of Geneva, Switzerland.