Institutional Dynamics of Switzerland’s Differentiated Integration

Abstract

This chapter focuses on the day-to-day dynamics of the sectoral agreements. This issue is topical, because the functioning of the sectoral agreements has been subject to criticism by the EU for several years but has not been analysed empirically. Jenni draws on institutionalist arguments found in neo-functionalist integration theories, conceiving of the sectoral agreements as incomplete contracts. The central claim is that the less sectoral agreements are ambiguous with regard to their relation to EU law, and the lower the obligational incompleteness, the higher the probability that they are revised. This argument is corroborated in bivariate and multivariate regression analyses which reveal the important role of Mixed Committees and the dynamic character of the Schengen association. Interestingly, beside these legal obligations, the substantive closeness to EU law is also correlated to more frequent agreement revisions.

Abstract

Switzerland is a challenging case for diplomats and researchers of European integration alike. The main reason is that the country has refused to subordinate itself to European institutions while nevertheless participating in a considerable number of EU policies. Since 2008, the Council of the European Union has repeatedly stated that Switzerland’s sectoral approach has reached its limits. In particular, the Council has criticised the incorrect implementation of several agreements (FMPA, FTA) and the static character of the market access agreements that put in danger the homogeneity of legislation in the Single Market. In the terms introduced in Chap. 2, the Council criticises the incorrect substantive extension of EU rules to Switzerland and states that this is partially related to the lack of a mechanism for revision and enforcement of the agreements. As a solution, the Council calls for institutional rules that would ensure that Switzerland continuously adopts new EU legislation in the areas of the agreements as well as independent surveillance and enforcement of the agreements (Council of the European Union 2012, 2010, 2008). In the terms introduced in Chap. 2, the Council calls for more legal integration. Apparently, the European diplomats assume that stronger legal integration would lead to more coherent substantive integration.

Notes

• 1. Supranationalist integration theory stands in the tradition of neo-functionalist reasoning as developed by Ernst Haas. I use the term supranationalism throughout this chapter because I focus on the aspects of the theory that explain the significance of formal rules and the role of actors in developing integration with a day-to-day focus. I use the newer term supranationalism rather than neo-functionalism because I do not focus on spill-over arguments which were important for the original argument (cf. Leuffen et al. 2013: 64 ff.).

• 2. I use the term incorporation of EU rules (EU rule incorporation) to refer to the specific Swiss way of adopting EU rules into domestic legislation. The term transposition is widely used to describe the implementation of EU directives by member states and thus refers to a formally regulated and sanctionable process. This is an inadequate description of the Swiss way of incorporating EU rules. I thus use the term transposition only in order to refer to a process which is equivalent to the EU-internal process.

• 3. Because the total revision only entered into force in 2011, it does not appear in the data set, which covers all federal laws and sectoral agreements that entered into force until and including 2010.

• 4. Two agreements are not administered by a Mixed Committee: the Agreement on Pensions and the Agreement on Taxation of Savings (Thürer et al. 2007). The reasons for the lack of a Mixed Committee are different. In the case of the agreement on pensions, there is no need for a Mixed Committee because the agreement does not rely on EU law. The lack of a Mixed Committee in the taxation of savings agreement is more interesting, as this agreement builds directly on the respective EU directive.

• 5. Technically, every amendment to be included in the Schengen agreement has the form of a diplomatic exchange of letters between the European Commission and the Federal Council. The dynamic provisions do not contain any delegation norm that would allow the government to adopt these exchanges of letters in its own right (Good 2010).

• 6. The dynamic provisions in the new agreement on customs security are slightly different. The procedure to adopt new legislation is less clearly defined than in the Schengen and Dublin agreements, and the EU has only the right to take compensatory measures in case Switzerland does not transpose new legislation (Epiney et al. 2012).

• 7. Negotiation dynamics are, among other factors, the subject of the analysis presented in Chap. 4.

• 8. The three categories are used, for example, by Astrid Epiney et al. (2012), Thürer et al. (2007) and Tobler (2008). Although these scholars share an understanding of what agreement belongs to which category, they do not define and use the categories in a way social scientists use variables. The operationalisation of the variables is thus a result of my own research but clearly inspired by the work of these scholars.

• 9. The two agreements in the data set with dynamic provisions are the Schengen and Dublin association agreements. Technically, the updates of the Schengen agreement are all diplomatic exchanges of letters; thus new treaties of international law with an SR number separate from the original Schengen treaty. From the point of view of their significance, however, these exchanges of letters are not new treaties: Their only purpose is to introduce changes in the original agreement, and they are never updated. For the purpose of the present analysis, I thus count them as revisions of the Schengen association agreement.

• 10. The following agreements were once or more totally or partially revised in the period between 1990 and 2010: Insurance (SR 0.961.1), Schengen Association (SR 0.362.31), Land Transport (SR 0.740.72), Frontex (SR 0.362.312), Free Trade (SR 0.632.401), Cooperation with EURATOM (SR 0.420.513.1), Protocol No. 2 on Proceeded Agricultural Goods (SR 0.632.401.2), Protocol No. 3 on “Originating Products” (SR 0.632.401.3), Protocol No. 4 on Special Provisions regarding Ireland (SR 0.632.401.4), Protocol No. 5 on import of products requiring compulsory stockpiling (SR 0.632.401.5), Statistics (SR 0.431.026.81), Simplification of Formalities in Trade in Goods (SR 0.631.242.03), Agreement regarding Protocol No. 2 (SR 0.632.401.22), Trade with Agricultural Goods (SR 0.916.026.81), Mutual Recognition of Conformity Assessments (SR 0.946.526.81), Free Movement of Persons (SR 0.142.112.681), Watch Industry (SR 0.632.290.131), MEDIA (SR 0.784.405.226.8), Air Transport (SR 0.748.127.192.68).

• 11. The dependent variable was recoded in a way that the count variable (number of revisions per agreement-year pair) became a binary variable (agreement was revised in a given year one or several times, yes or no), because most agreements were revised only once or twice a year anyway (see Table 3.1). Using a count variable would thus attribute too much influence to the variable values of the few observations with very high numbers of revisions per agreement and year. The data structure of agreement-year pairs resembles a panel structure, but estimation techniques accounting for that structure are not necessary because the independent variables apart from the control variables vary only between agreements but not over time. The main research interest is to explain the variation between agreements.

• 12. A test based on the Bayesian information criteria indicates that the control variables added in Model A+ do not significantly improve the model fit (reported as AIC and BIC, smaller numbers would indicate better model fit, see e.g., Long and Freese 2001). A Wald test hints at the opposite. Test results are reported in the last three rows of Table 3.3.

• 13. The results of the same regression analysis with the original data (without the independent variable dynamic agreement) are reported in Table 3.9. In this analysis, the correlation of Mixed Committee decisions with references to EU law is statistically not significant.

• 14. The bootstrap technique could not be applied because several covariate patterns are rare. Therefore, the tests of significance have to be interpreted with care as they may be too optimistic.

• 15. The Wald test could not be performed. With only 19 clusters (19 sectoral agreements), there are not enough degrees of freedom in the model to estimate the probability that all coefficients are simultaneously zero.

• 16. The present analysis focuses on the question of whether domestic rule transpositions are used as compensatory measures in areas with sectoral agreements. The hypotheses to be tested provide no arguments about the quality of such transpositions (full or partial adaptations). The same independent variables were also tested using a multinomial logit model, distinguishing between full and partial adaptations, compatible reforms, and reforms without relation to EU law. A test showed that compatible reforms cannot be distinguished from reforms without relation. A model using only full and partial adaptations and other reforms confirmed the positive correlation of harmonisation agreements with full and partial adaptations (results not reported).

• 17. Domestic and international legislation is categorised separately in the Classified Compilation of Federal Legislation. The sub-chapters used to categorise both forms of legislation are similar but not completely equal. Table 3.10 shows how the sub-chapters were merged.

• 18. Unfortunately, for 15 more federal law reforms, the rule transposition variables could not be coded because of unavailable coding sources.

• 19. The missing value is due to missing information in the legal text itself with regard to whether or not a referendum was held. Normally, one can find this information in the last article of a law containing the provisions about the entry into force. For the coding of the referendum variable, this article was used instead of the chronology of popular votes on the website of the federal administration, because this chronology does not allow in every case to assign the votes unequivocally to the affected federal laws.