VoxEU Column Politics and economics Europe's nations and regions EU policies

The rise of extreme politics in a federation

Polarisation, populism, and extremism are on the rise on both sides of the Atlantic. This column focuses on the role of policies in multi-level federations (such as the EU) in partially explaining the rise of extreme political parties. An analysis of differences in vote shares between European and national parliamentary elections suggests that support for extreme politicians is highest in countries with the largest gains and losses from federal policies. Eurosceptic parties, which are very protective of national interests, win higher shares of the EU vote in core and periphery countries, whilst the opposite is true for countries in the middle.

This column is a lead commentary in the VoxEU debate on "Populism"

Polarisation, populism, and extremism are on the rise on both sides of the Atlantic. Since the societal and economic costs of such political instability can be large, recent contributions to VoxEU have tackled the drivers of the process, as well a possible solutions (Albanese et al. 2020, Colantone et al. 2019b, Eichengreen 2019, Guriev 2019, Rodrik 2019, Tabellini 2019). So far, explanations for political extremism have focused on the economic and cultural insecurity of voters (Albanese et al. 2019, Colantone et al. 2019a, Guiso et al. 2017, Guriev 2018). However, existing research has understudied the role of federalism. In this column, we show that taking into account the federalism aspect offers not just a complementary explanation, but also potential solutions.

heterogeneity can foster extreme politics

In a federal setting, lower-level jurisdictions (regions) are inevitably affected by policies introduced at the highest (federal) level. For example, if migrant workers pay local taxes, attractive regions will be ‘winners’ of any federal policy supporting the free movement of workers. Regions facing labour outflows will be ‘losers’. Migration has indeed widened regional disparities (Goldin et al. 2018). As shown in the left panel of Figure 1, Germany and the UK have seen the highest net within-EU migration inflows, whilst Romania, Poland, and Portugal are in the opposite camp (Fries-Tersch et al. 2018).1 Secondly, the EU periphery countries -- which have lower credit ratings, as shown in the central panel of Figure 1 -- would clearly gain from the issuance of common Eurobonds and debt mutualisation.2 Similar divides apply to the net transfers of EU cohesion and agricultural funding (see the right panel of Figure 1).

What is crucial in these and other examples (e.g. environmental and energy policies) is that certain regions are -- or are perceived as -- winners or losers because of persistent spatial heterogeneity, which at best only changes in the medium to long run. Such underlying fundamentals are usually economic and geographical – related to the divide between ‘rust belt’ or rural areas on the one side, and successful urban agglomerations on the other. They could also be related to regionally concentrated natural resources or other endowments such as social and human capital.3

Figure 1 Migration, credit rating, and net transfers in the EU

Notes: The 28 countries are divided into 3 groups: 10 Green, 9 Gray, and 9 Red. Red and Green countries are at each extreme of the spectrum, while Grey refers to countries in the middle. The map on the left ranks countries based on workers' mobility (net inflow - Green highest). The central map ranks countries based on their S&P credit rating (Green highest). The map on the right ranks countries based on EU net-transfers as a % of GNI (Red - highest).

Once a salient federal policy becomes a focal point in voters’ minds, marking their region as winning or losing from being part of the federation, voting behaviour can change drastically. In Daniele et al. (2020), we explain how the winners want to vote for federal politicians promising to reinforce the federal policy, and the losers prefer the opposite. What both groups have in common, however, is that they are delegating extremely protective representatives to manipulate federal negotiations in their region’s favour. In that sense, neither the winning nor the losing regions gain much and are trapped in a kind of prisoner's dilemma: they keep each other in check once inside the coalition, serving as each other's counterweight.

This outcome is entirely due to the strategic motive to improve welfare in one’s own region. Unlike theories that rely on a shift in self-categorisation into an identity – which would then be centred on the regional identity dimension4 – our argument is based on the belief that the federal policy works in favour of a set of regions, and against some others. As a result, voters are willing to incur the ideological, reputational, and efficiency costs of electing a tough negotiator with more extreme preferences than their own.5

Interestingly, this strategic motive persists even when voters are not sure their representative will be part of the federal coalition. Winners outside the coalition can then always rely on winners inside the coalition to defend their mutual interests, and the same goes for losers. The usual incentive to elect a centrist, more accommodating representative to improve chances of making it into the coalition, therefore disappears.

Support for ‘extremely protective’ parties in the EU

This theory is directly relevant for the world’s largest ongoing federal experiment, the EU, where it is easy to identify ‘winners’ and ‘losers’, and the ‘extremely protective’ political parties. Using the net contributions of each member state to the overall EU budget, we cluster EU member states in three intuitive groups (marked in the right panel of Figure 1): main contributors (green), main recipients (red), and those in between (grey).6 For instance, net transfers received from the EU represented 3.53% of GNI for Lithuania, 2.9% for Bulgaria, and 2.11% for Poland (in 2000-2015). The Netherlands and Germany were the main net contributors. Since the benefits from the EU Single Market are not necessarily well understood by the general public, we argue that the three clusters based on net transfers capture winning/losing perceptions of EU policies quite well.7 Indeed, the same demarcations have emerged during the Eurocrisis in 2012-2015, and again in the handling of the COVID-19 crisis.8

To identify political parties focusing on the extreme protectionism, we use the categorisation of Eurosceptic parties from Algan et al. (2017). Such parties usually campaign on a platform of protecting the interests of the member state in question (Colantone et al. 2019a). Think for example of Syriza (Greece), Podemos (Spain), AfD (Germany), Front National (France), or Lega (Italy). We then analyse party performances from 1990 onwards and compare European Parliament elections to national elections. Eurosceptic parties are more successful in the European elections relative to their performance in the national elections. Confirming our predictions, the relationship between benefitting from EU policies and support for Eurosceptics is U-shaped: ‘winning’ and ‘losing’ member states are more likely to vote for the Eurosceptic parties than the states in the ‘middle’.

Figure 2 Average performance of Eurosceptic and moderate parties

Notes: Each dot represents the average performance of Eurosceptic (left panel) and Moderate (right panel) parties in a country. On the vertical axis, we report the differences in the share of votes between European and national elections. On the horizontal axis, we show net contributions to the EU. We consider five-year windows: whenever necessary, we consider parties’ average performance across all elections within the window.

Figure 2 illustrates this pattern: the left panel shows that Eurosceptic parties do better in European than in national elections, but only in winning (receiving) or losing (contributing) member states, producing a U-shaped relationship (quadratic term significant at p=0.002). Instead, there is no such effect for moderate parties (see the right panel of Figure 2).

At the first sight, those results are compatible with the so-called ‘second-order’ conjecture, suggesting that voters take European elections less seriously and use them to express discontent towards local governments. In Daniele et al. (2020) we run several robustness checks to show that such conjecture is not supported by our data, in particular, with an online survey that we administered in Finland, France, and Italy right after the last European parliamentary elections.

These voting patterns could lead to tensions and even to the eventual break-up of the federation. What can be done? One obvious solution would be to weaken the regional ties of federal politicians, who would then have the incentive to design policies benefiting the entire federation, and not just their own countries or regions. In the EU context, this might imply the creation of a pan-European constituency to elect some EU deputies and/or the requirement of a minimum number of countries where a party should be forced to run. This would turn the EU into a mixed electoral system, similarly to other federal entities as Mexico, Germany, Italy or South Africa, in which both proportional and majoritarian electoral systems coexist.

Implementing this solution in the EU would certainly not be easy. Among other things, it would require politicians campaigning in different countries and different languages. The voters may lose interest if they cannot clearly identify with certain parties. However, on the plus side, it would put debates on the opportunities and challenges of the union squarely in the spotlight, which can only contribute to the creation of a shared political sphere any democracy relies on.


Acemoglu, D, G Egorov and K Sonin (2013), “A Political Theory of Populism”, The Quarterly Journal of Economics 128(2): 771–805.

Albanese, G, G Barone and G de Blasio (2019), “Populist Voting and Losers’ Discontent: Does Redistribution Matter?”, Marco Fanno Working Papers – 239.

Albanese, G, G Barone and G de Blasio (2020), “Regional redistribution and populist voting”,, 04 February.

Algan, Y, S Guriev, E Papaioannou and E Passari (2017), “The European trust crisis and the rise of populism”, Brookings Papers on Economic Activity: 309–400.

Colantone, I and P Stanig (2019a), “The Surge of Economic Nationalism in Western Europe”, Journal of Economic Perspectives 33(4): 128–51.

Colantone, I and P Stanig (2019b), “Heterogeneous drivers of heterogeneous populism”,, 10 December.

Daniele, G, A Piolatto and W Sas (2020), “Does the Winner Take It All? Redistributive Policies and Political Extremism”, Barcelona GSE Working Paper 1157.

Eichengreen, B (2019), “Two faces of populism”,, 29 October.

Fries-Tersch, E, T Tugran, A Markowska and M Jones (2018), “2018 Annual Report on intra-EU Labour Mobility”, European Commission.

Gennaioli, N and G Tabellini (2018), “Identity, Beliefs, and Political Conflict”, CEPR Discussion Paper 13390.

Goldin, I and B Nabarro (2018), “Losing it: The economics and politics of migration”,, 24 October.

Guiso, L, H Herrera, M Morelli and T Sonno (2017), “Demand and supply of populism”, CEPR Discussion Paper 11871.

Guriev, S (2018), “Economic drivers of populism”, AEA Papers and Proceedings 108: 200–203.

Guriev, S (2019), “Moderator’s introduction to the Vox debate on populism”,, 29 October.

Rodrik, D (2019), “Many forms of populism”,, 29 October.

Tabellini, G (2019), “The rise of populism”,, 29 October.


1 Migration data were extracted from Table 30 in Fries-Tersch et al. (2018). We computed the ‘net inflow stock’ for each country as the difference between the number of EU foreign citizens registered as resident and the number own citizens registered as resident in any other EU country. Only people aged between 20 and 64 were considered. Twelve EU countries have positive net inflows. Germany and the UK together account for 67% of the positive inflow. Out of the remaining 16 countries with negative inflows, Romania, Poland, and Portugal account for 69% of the total.

2 For the panel of Figure 1, S&P rating data were taken from  (updated on 12 April 2020). It refers to the Long Term Sovereign Rating. The green group includes ratings from AAA to AA. The grey group includes ratings from AA- to A, while the red group includes ratings from A- to BB-.

3 The same winner/loser reasoning also holds at the national level, e.g. in Belgium, Canada, Italy, Spain, and Germany.

4 In Gennaioli and Tabellini (2018) for example, a rise in immigration or increased exposure to imports can create a new divide along the dimension of culture or nationalism, changing group identification. Cultural factors can then become a catalyst for conflict, boosting polarisation.

5 The motive of voters is hence not to select honest politicians out of a pool of potentially corrupt candidates, as in Acemoglu et al. (2013) where politicians offer platforms left (or right) of the median voter to signal trustworthiness. In our model, the shift away from the federal centre is chosen by the median voter strategically in order to stake out a better bargaining position at the federal level.

6 Net transfers are computed from the EU budget as % of GNI in the period 2000-2015 for each EU country.

7 We also replicate our analysis using Eurobarometer survey data, grouping countries according to reported beliefs of benefiting from EU membership; the results are similar.

8 The current situation in Europe may even be seen as a result of the voting behaviour we describe: southern member states are in favour of radical EU stimulus and mutualising incurred debts to cushion the COVID19 shock, whilst northern ‘frugal' countries are slow to accept these proposals.

1,055 Reads