Who liberalizes high-skilled immigration policy and when? Partisanship and the timing of policy liberalization in 19 European states

ABSTRACT

This article investigates the timing of HSI policy liberalization in Europe by focusing on the role of government partisanship in overcoming diverging pressures for and against greater openness. Using event history analysis to analyse changes in HSI policy in 19 European states from 1999 to 2011, this article finds that having right-leaning governments significantly accelerated HSI policy liberalization, while having left-leaning governments delayed it. The findings suggest that skilled immigration is not driven only by labour market dynamics but also by government partisanship. Moreover, the article sheds light on why right-wing parties, in contrast to their usual tendency to champion restrictive immigration policies, may pursue policy liberalization in the case of HSI.

KEYWORDS:

• High-skilled immigration
• labour migration policy
• policy change
• political economy
• party studies
• comparative politics

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 However, precise instruments and definitions of who constitutes an HSI differ across states (Parsons et al. 2014).

2 Authors’ own calculation based on DIOC: Database on Immigrants in OECD and non-OECD Countries (2000/01–2015/16 data).

3 Alternatively, parties can find common ground by establishing “strange bedfellow” cross-party coalitions (see Zolberg 1999; Berg and Spehar 2013).

4 The highest category, a “major change”, applies if the change affected an entire migrant group, e.g. all labour migrants (de Haas, Natter, and Vezzoli 2015, 15).

5 For example, France limited in 2011 the number of occupations qualifying for the shortage list. This decision was reversed in 2012 (DEMIG 2015).

6 While Ireland in 2009, the Netherlands in 2011 and Portugal in 2012 implemented restrictive changes, they were usually confined to corrective “fine-tuning”, i.e., the changes were non-fundamental or effectively continued existing legislation (de Haas, Natter, and Vezzoli 2015).

7 Where data evidenced missingness between two recorded time-points, I linearly extrapolated the values.

8 The choice of 2007 rather than 2009, the year the Card was passed, is to avoid a simultaneity bias and to capture the fact that policy diffusion effects may have already taken place during the negotiations phase (see Menz 2009).

9 However, the unstandardized coefficient cannot be directly compared in terms of its effect size to other variables’ coefficients.