Income inequality and party (de)polarisation

Abstract

Scholars of US politics report a strong connection between income inequality and party polarisation. This study evaluates this relationship comparatively, and finds that the opposite association holds outside the United States, namely, that increasing inequality depolarises political parties’ economic positions. High levels of inequality change the composition of the core constituencies on the left and depress the participation of low-income voters, leaving left-wing parties with less support when they adopt distinct leftward positions. At the same time, inequality does not create strong incentives for right-wing parties to move further rightward. Therefore, income inequality leads to party systems that are depolarised on economic issues. The results have implications for our understanding of political representation, economic inequality, and political polarisation.

Keywords:

• Income inequality
• party polarisation
• economic policy
• political parties
• turnout
• political participation

Notes

1. These results are consistent with the findings of Iversen and Soskice (2016), who show that inequality can lead to the depolarisation of mass publics in established democracies. This means that both voters and parties depolarise their economic positions when inequality increases. Furthermore, these findings help explain the large variations in redistribution among advanced economies, and the puzzle of why more unequal societies redistribute less (Alesina and Glaeser 2005; Moene and Wallerstein 2001; Perotti 1996). The prediction that increasing inequality is met by redistributive policies (Meltzer and Richard 1981) only occurs if all voters participate and redistributive demands find representation in policy-making (see also Pontusson and Rueda 2010).

2. Adopting a comparative framework, research has shown that the parallel movement of inequality and polarisation is not deterministic outside the United States. For instance, Han (2015) reveals that the responsiveness of political parties to increasing income gaps critically depends upon the permissiveness of electoral systems. Pontusson and Rueda (2010) also underline the importance of low-income voters’ mobilisation.

3. This study follows Pontusson and Rueda (2010) and assumes that low-income voters are more likely to support left-wing economic policies, than the high-income electorate. This assumption is in line with the persistent relevance of class voting described by Chan and Goldthorpe (2007) and by Evans (2000). Furthermore, this assumption is supported by the results of Mueller and Stratmann (2003), who find that polities with higher participation have larger public spending.

4. Pontusson and Rueda (2008: 313) show that ‘Left parties are particularly responsive to wage inequality because their core constituencies consist of voters who derive the lion’s share of their income from dependent employment’. The enduring importance of class in the determination of voting preferences translates into party strategies.

5. Hansford and Gomez (2010) report similar results for US presidential elections.

6. The countries are Australia, Austria, Belgium, Canada, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Latvia, Lithuania, Luxembourg, Mexico, the Netherlands, New Zealand, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Turkey, and the United Kingdom. Time units are election years.

7. In order to count the overall number of parties, the sum of all parties listed in the Comparative Manifesto Project dataset for each election is calculated.

8. The measure exhibits a low degree of right-skewedness. To improve linearity, Online Appendix E also reports the results of the main models using the logged version of the dependent variable.

9. Following Pontusson and Rueda (2008, 2010), this study assumes that low-income voters are distinctly to the left. This assumption builds on extensive research, which highlights the enduring relevance of class voting (see e.g. Chan and Goldthorpe 2007; Elff 2007, 2009; Evans 2000). If a direct test of this assumption is beyond the scope of this research, the test of Hypothesis 1 constitutes an indirect test. If low-income voters were not more likely to vote for the Left, household income inequality should not systematically affect the electoral success of these positions.

10. Since the division is based on quintiles, only elections with five or more parties are sampled.

11. Given the polarising effect of electoral laws (Cox 1990; Dow 2001, 2011; Golder 2003; Han 2015; Hug and Martin 2012), and its positive effect on turnout (Blais and Carty 1990), the models control for electoral system type, with a variable in the Comparative Political Data Set (Armingeon et al. 2014). Since EU membership can constrain the policy platforms chosen by parties (Dorussen and Nanou 2006), membership is included in the model, with data retrieved from the Comparative Political Data Set. Then, following the findings of Kriesi et al. (2006) and Ward et al. (2011), trade openness, unemployment, economic growth are controlled for. Finally, the models control for unionisation levels. Data are retrieved in the Comparative Political Data Set (Armingeon et al. 2014).

12. It is arguable whether reversed causality is a problem, theoretically, because changes in inequality occur at a different pace than changes in party positions – with the second being more responsive to the environment.

13. Among the control covariates, particularly interesting is the result for EU membership. In particular, the results show that being an applicant to the EU strongly depresses the polarisation of political parties on economic issues. This result can indicate that the need to meet the economic and systemic requirements for EU accession push parties to converge on compatible policy positions. These positions do not appear to polarise again once EU accession occurs. The estimated coefficient for EU membership is negative and consequently can show ‘depolarising path dependence’. Further research is however needed to address this finding, and to study the impact of EU accession on the gravitational centre for parties’ economic positions.

14. If a comprehensive analysis of right-wing parties’ behaviour is beyond the scope of this analysis; this study speculates the existence of a contagion effect. When the Left is more centrist, the Right can find electoral incentives in also moderating economic positions not to alienate moderate voters, pushing them towards the left. Further research is needed to evaluate further specific characteristics of the Right that can explain the economic centrism resulting from inequality.

15. See Online Appendix B for a test of the impact of inequality on the most left-wing and most right-wing economic positions.

16. This implication stems from the assumption of a preference of poor voters for more redistributive policy positions. Although there is some scholarly debate, research shows that class is a fundamental variable in shaping voters’ preferences and electoral choices (e.g. Chan and Goldthorpe 2007; Evans 2000).

17. Online Appendix C presents the regression results for the effect of inequality on turnout.

18. Given the division in quintiles, Models 1 to 5 only sample elections with five or more political parties.

19. Lubbers and colleagues (2002) find similar results when studying the determinants of success for extreme right-wing parties.

20. Online Appendix D shows the regression results for the association between turnout and parties’ economic polarisation.