Path-dependent or dynamic? Cantonal integration policies between regional citizenship traditions and right populist party politics

Abstract

Immigrant integration policies are commonly assumed to be shaped by traditional understandings of citizenship and belonging. Accordingly, such policy frameworks are considered path-dependent and thus stable over time. The recent surge of radical right anti-immigrant parties in many Western European countries challenges this stability-thesis. Several scholars contend that this change in the party political landscape caused an illiberal turn, resulting in more restrictive integration policies. This paper tests whether integration policies are indeed challenged by radical right parties, or whether they remain path-dependent policy frameworks, using subnational comparative analysis. The results of our cross-sectional regression analysis over twenty six Swiss cantons reveal that cantonal integration policies reflect clear patterns of regional citizenship culture, suggesting that they remain stable policy frameworks. By contrast, the considerable electoral success of the right populist Swiss People's Party (SVP) over the last two decades did not directly result in more restrictive cantonal integration policies.

Keywords:

• Integration policy
• regional citizenship traditions
• immigrant sceptic voting behaviour
• right populist parties
• subnational comparison
• Swiss cantons

Notes

1. Continuous data (percentage of French-speaking population) did not yield significantly different results. Therefore, the dummy solution was retained.

2. Considering sample restrictions, it is not surprising that the correlations were less stable in the survey data (Pearson's r min. 0.18, max. 0.69***, based on SELECTS '95, '99 and '03). For the variable ‘immigrant sceptic attitude’, the mean of the three years has been used. The poll data on the other hand reveal very strong and highly significant correlations with the voting data from 2004 (Pearson's r > 0.8***).

3. For all the regression models, influential cases were identified with cooks-d statistics. An elimination of influential cases (cooks d > 4/n) only slightly fostered already significant results. Therefore, all the models are based on the full sample of available cantons.