Credibility, Complexity and Uncertainty: Explaining the Institutional Independence of 29 EU Agencies

Abstract

The delegation of policy-making tasks to EU agencies and their remarkable growth in number over the past two decades mark a striking new development in the EU's institutional make-up. While most of the nascent literature on the EU's ‘agencification’ addresses the conditions for agency creation and the implications of agency governance from the perspective of democratic accountability, there is a lack of empirical research systematically scrutinising the institutional structure and degree of formal-institutional independence of these agencies. This article offers a comprehensive empirical assessment and measure of the variation in institutional independence displayed by the entire set of 29 EU agencies operating under the EU's three pillars and tests hypotheses explaining variation in formal independence among agencies.

View correction statement:

Corrigendum

Acknowledgements

This article is part of the project ‘Agency Governance and its Challenges to the EU System of Representation’, jointly led by the two authors at the University of Mannheim's Mannheim Centre for European Social Research (MZES). The project is affiliated to the EU Integrated Project Reconstituting Democracy in Europe (RECON) and financed by ARENA – Centre for European Studies at the University of Oslo.

We want to thank Anja Becky for excellent research assistance in gathering the data on agencies' formal institutional make-up. Moreover, we would like to thank Mark A. Pollack and the participants of the EUSA 2009 panel on ‘Principals, Agents and EU institutions’ as well as the reviewers for excellent comments on the article.

Notes

1. In this context, we cannot address the question why the tasks fulfilled by agencies have been delegated to these agencies and not to the Commission (Kelemen 2002: 94; Majone 2001). Obviously, the question whether this is the result of members states' reluctance to delegate further competences to the Commission (Kelemen 2002: 95) or due to the alleged politicisation of the Commission whereby its ‘trustee’-quality is called into question (Thomson 2008; Wonka 2007) is an interesting one, yet we cannot address it here.

2. Krapohl (2004), for instance, follows this path by taking into account differences in the institutional design of two EU agencies to infer variation in the effectiveness of regulatory outputs.

3. The policy credibility argument advanced here is based on the assumption that policy-makers have time-inconsistent preferences in policies of economic planning and regulation (Kydland and Prescott 1977). In order to credibly commit themselves to a course of efficient regulatory policy-making, policy-makers have to ‘bind’ themselves institutionally, for example, by delegating regulatory competencies to independent agencies. The important point is that whenever legislators are tempted to forgo the benefits of regulatory policy-making for short-term political interests, the agency acts differently: Not as an ‘agent’ pursuing the same interests as the principal, but as a ‘fiduciary’ (Majone 2001).

4. We wish to thank one of our referees for pointing this out to us.

5. The application of this perspective and reasoning to the EU has been criticised for neglecting the political and material costs that can come with EU regulations (Follesdal and Hix 2006) and for not appreciating the political quality of (expert) information leading to particular regulatory rules (Shapiro 1997).

6. Yataganas (2001) lists all agencies established in the EU until 1999 and offers a typology to categorise the different agencies. Kelemen's categorisation covers agencies established until 2004. The categorisation of the two authors diverges for three agencies. These and the more recently established agencies were categorised by the authors.

7. To check the robustness of our results, we re-specified our independence index, subsuming the variable tapping an agency's formal mandate (V1) under A2 thereby purging A1 from the index (cf. Appendix 1). Obviously, this results in some changes in agencies' independence values and consequently their ordering. When estimating the models in Table 2 with this alternative, the results for models 1, 3 and 6 remain basically the same, both as regards coefficients and standard errors. However, while the size of the coefficients of models 2 and 5 also remain the same with an alternative specification of our independence index, ‘social regulation’ (model 2) and ‘regulation’ (model 5) turn out to be significant. As expected, the results are sensitive to the specification of our index. Yet our robustness check demonstrates that the findings confirm the general picture according to which agencies operating in the field of regulation are more independent than informational and executive agencies. Moreover, the robustness check demonstrates that our index does not systematically privilege the relationships captured by our hypotheses.