Breaking through: how anti-immigrant parties establish themselves and the implications for their study

ABSTRACT

While the sizeable literature on anti-immigrant parties commonly invokes the term “breakthrough” in discussing their electoral fortunes, no existing study has subjected this concept to conceptual or empirical scrutiny. Based on an investigation of 54 anti-immigrant parties in seventeen Western European countries, this paper concludes that these parties indeed tend to establish themselves with rare breakthrough elections rather than by incremental growth, and to stay around once they achieve such success. Moreover, these breakthroughs are more likely when these parties have direct or indirect ties to the party system. On the one hand, these findings confirm an often invoked but so far untested suspicion about the way these parties establish themselves. On the other hand, they have important implications for their future study. They suggest that some elections are more important than others for understanding the success of anti-immigrant parties, and that it is potentially misleading to explain the failure of anti-immigrant politicians in some places as the necessary consequence of country-specific characteristics. Not only can one unpredictable election change everything, the failure of anti-immigrant politicians in the past might also pave the way for like-minded politicians in the future.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Scholars have used several labels to denote this type of party, including extreme right, radical right, and populist right. This paper will use the term “anti-immigrant party”, primarily because the label “right-wing” does not accurately describe the position of many of these parties on socio-economic issues (De Lange 2007).

2 By starting in 1977, this analysis captures the first parliamentary establishment of virtually all modern AIPs in Western Europe, except for the Italian MSI (1948) and the Swiss SD/NA (1967) and RB (1971).

3 For France, I report the share of the vote in the first round, and for Germany the share of the vote for the party list (Zweitstimme). The results have been drawn from each country’s official election website.

4 A related concern is that some of these parties have not run on an anti-immigrant platform for the full duration of their existence. In the subsequent analyses, I am only including the FPÖ after 1986, the BZÖ up until 2009, the FRP in Denmark after 1981, the MSI up until 1995, the FRP in Norway after 1977, and the SVP after 1999.

5 To avoid skewing the distribution to the right artificially, Figure 2 includes the results of defunct parties in the first election in which they no longer participated (Mustillo 2009).

6 We can find two alternative operational definitions of a breakthrough in some of the existing literature: the first time a party enters parliament, or the first time a party manages to be re-elected two times in a row. Neither of these definitions are suited for our investigation, however: the former would make comparisons more difficult because of large cross-national differences in electoral systems, and the latter definition would make the second hypothesis partially tautological.

7 This cut-off of 4 per cent does not only constitute the average of what Norris and Carter consider a threshold of relevance, but it also seems reasonable on empirical grounds, considering that it approximates the standard deviation of the distribution in net changes in the electoral vote illustrated in Figure 2 (3.7 percentage points).

8 Strictly speaking, five other parties do not satisfy the definition either. PS has captured a slightly lower percentage of the vote since breaking through (17.7 and 17.5 per cent, compared to 19.4 per cent in its breakthrough election). VB, RN, LN and the Norwegian FRP all outperformed the result of their breakthrough election in the most recent election, but occasionally dipped below that result in the interim. To consider any of these parties – all of which rank among the three largest in their respective party systems – as unconsolidated, however, seems clearly incorrect.

9 The online appendix reports the results of additional analyses I conducted to validate the robustness of the findings in Table 3. It shows the simple bivariate relationship between embeddedness and breakthrough, as well as the results from models that introduce the two sets of control variables separately to address possible concerns about oversaturation, introduce some of the demand-side variables separately to address possible concerns about multicollinearity, remove control variables with missing values to maximize the number of cases, employ different indicators of a party’s longevity (the number of elections since first parliamentary representation, cf. Mustillo 2009; the number of years since first representation at any level, including subnational and supranational; the number of years since any AIP obtained representation in the country’s parliament; and the number of years since any AIP obtained representation at any level), and operationalize a breakthrough as the largest gain. These analyses produced largely similar results, with three noticeable differences: (1) the effect of being a breakaway party is larger and more significant in most of these analyses; (2) the effect of a history of AIP success, while in the expected direction, is insignificant (Table A6); and (3) the effect of embeddedness becomes insignificant, and the effect of asylum intake significant, when using the alternative operationalization of a breakthrough (Tables A7 and A8), suggesting it might be identifying elections at a later stage in a party’s life cycle.

10 By showing the difference in the expected probability of the lowest and highest value among the middle 50 per cent of cases, the coefficients give directly comparable estimates of the importance of each variable that are unaffected by cases with unusually high or low scores (which especially in nonlinear models might give an inflated impression of the effect size).