The role of subnational authorities in the implementation of EU directives

Abstract

Studies on the role of regions in the EU policy process concentrate mainly on policy formulation and implementation of regional funds. In this article, we redress this bias by investigating the formal role of subnational authorities in the implementation of EU regulatory policies, specifically in the transposition of directives. Subnational authorities play a secondary, but increasingly important, role in the application of these measures. Their impact is greater on environmental and social policies, as it is also on public contract legislation. More decentralized states display higher levels of subnational involvement but, in these states, regional participation in national policy-making and a high number of regional authorities decrease the likelihood of finding subnational measures of transposition. There is also more subnational involvement in states with territories that have both an elected government as well as special arrangements regulating their relations with the EU. Finally, subnational involvement tends to prolong the process of transposition.

Keywords:

• Compliance
• directives
• federalism
• implementation
• regions
• transposition

ACKNOWLEDGEMENTS

Fabio Franchino would like to thank the Economic and Social Research Council and the Nuffield Foundation for financial support of the project on ‘Compliance with EU Law: Explaining the Transposition of EU Directives’, and the University of Milan for funding under the Programma dell’Università per la Ricerca. The authors received valuable comments on this paper from Alessia Damonte and the participants at the seminar of the Department of Political Science of the Institute of Advanced Studies in Vienna, as well as from two JEPP referees.

Notes

Regionalisation is a process of power decentralisation from the state to a regional authority defined as a ‘territorial entity situated between the local and national levels which has a capacity for authoritative decision making’ (Marks et al. 2008: 113). In this work, the terms region and subnational authority are used interchangeably. Additionally, although the empirical section of this work is centred solely on the European Community pillar of the Treaty on European Union, the terms European Union and EU are used throughout, instead of European Community.

These are the Czech Republic, Finland, Greece, Hungary, Ireland, Italy, Lithuania, the Netherlands, Poland, Romania, Slovakia, Spain and the UK.

Involvement is formal when SNAs directly adopt national measures of implementation. It is informal if SNAs are merely consulted by central state authorities.

See also Rhodes (1986) and, for a recent similar argument related to the divergent transposition performances of unitary (Denmark) and federal (Belgium) states, Bursens (2002).

The results reported by Lampinen and Uusikylä (1998) are more difficult to interpret. The variable ‘political institutions’ is significantly associated with the failure to implement directives. But this is an index composed of 13 factors; it is therefore difficult to discern the impact of regional autonomy, which is one of the building blocks, from that of other components.

CELEX was no longer accessible after this cut-off point. Its content was migrated to the EUR-Lex database, available at http://www.eur-lex.europa.eu. The CELEX manual states that parts of the database are incomplete (e.g., sector 5 on preparatory acts before 1984) but this does not apply to the two sectors from which information was collected (sector 5 on secondary legislation and sector 7 on national implementing measures). For instance, the database lists 341 directives adopted in the five years prior to 1984 and 410 directives in the following five years. A mere difference of 70 directives (14 directives per year) between the period of ‘Eurosclerosis’ (pre-1984) and that of implementation of the single market programme (between 1984 and 1988) does not seem to indicate that the record is incomplete before 1984. See König and Luetgert (2009: 170–2) for a discussion on the reliability of this database.

From an initial random sample of 821 directives, 733 directives display at least one national implementing measure adopted in at least one member state after the date of adoption of the directive. This sample is quite representative of the Eur-Lex database. We have 314 Commission directives (38.6 per cent), while the database lists 872 Commission directives out of a total of 2,358 directives adopted in the same time period (37 per cent); 258 and 262 are agricultural and internal market measures respectively (31.4 and 31.9 per cent) compared to 769 and 914 (32.6 and 38.9 per cent) respectively in the database.

In other words, authorities with a representation score of at least 1,1 in the index developed by Hooghe, Marks and Schakel (2008).

Had the observations gone further back to the 1970s, one might have observed a rise in subnational involvement following the establishment of the Belgian communities and regions. On the other hand, it is difficult to speculate on the consequences of the last two enlargements on subnational involvement. Six of the twelve new member states display some degree of regional authority (Hooghe, Schakel and Marks 2008), but none has joined REGLEG (see Note 10).

REGLEG is an association of 73 regions with legislative power across eight member states (Austria, Belgium, Finland, Germany, Italy, Portugal, Spain and the UK).

Eight French arrêtés préfectoral transposing Council Directive 82/883 are considered national executive measures because they were issued by prefects who are central state representatives in a department or region.

Of course, these dynamics are not at work in more centralized countries where one simply expects less subnational involvement.

In other words, with a representation score of at least 1,1 (Hooghe, Schakel and Marks 2008).

These are Finland (Åland), France (French Guiana, Guadeloupe, Martinique), Germany (Büsingen am Hochrhein, Heligoland), Great Britain (Channel Islands, Isle of Man, Gibraltar), Italy (Campione d’Italia, Livigno), Portugal (Azores, Madeira) and Spain (Canary Islands, Ceuta, Melilla). Mount Athos is also a special territory but its civil governor is appointed by the Greek minister of foreign affairs. Overseas countries and territories are disregarded because they are not part of the EU and their relationship is governed by association agreements.

This analysis of cross-level variation follows Steenbergen and Jones (2002: 224, 231). We opted for a binomial model with a probit link function because subnational involvement reflects an underlying interval variable. Hence, its cumulative distribution is normal.

A table listing the marginal effect of the decentralisation indices and its standard error for each member state is available from the authors.

The marginal effects and confidence intervals have been produced following the code developed by Brambor et al. (2006). We took 10,000 draws from the estimated coefficients and the variance-covariance matrix. In Figure 3, number of SNAs is set at 15, while in Figure 4 shared rule is set at 2.4.

Since legislative measure and subnational measure cannot be observed for directives without national implementing measures, these observations drop out from the analysis.

As post-estimation diagnostic tests, we have employed the link test and inspected the Cox–Snell residuals to assess the appropriateness of the Cox specification. Each covariate was also plotted against the martingale residuals to assess if their functional form was correct. Finally, we used the Schoenfeld residuals to test if the covariates violated the proportional hazards assumption and, following Box-Steffensmeier and Jones (2004: 131–7), we interacted the offending covariates with the natural logarithm of time. The results also hold if these covariates are interacted with time only, rather than with its natural logarithm.