The interpretation of legal fit we propose differs substantially from the idea advanced by Dimitrova and Rhinard (2005) Dimitrova, A. and Rhinard, M. 2005. The power of norms in the transposition of EU directives. European Integration online Papers (EIoP), 16(9) [Google Scholar] because we define ‘norms’ in legal terms and not as the prevailing value orientations in society.

In addition, some authors have conceptualized misfit primarily in institutional terms, focusing on the compatibility of directives with national-level institutions (Giuliani 2003 Giuliani, M. 2003. “Europeanization in comparative perspective: institutional fit and national adaptation”. In The Politics of Europeanization, Edited by: Featherstone, K. and Radaelli, C. M. Oxford: Oxford University Press. [Crossref] , [Google Scholar]).

For example, in the case of the transposition of the tobacco directive in Spain, EURLEX reports that Spain completed the transposition in December 2005 (five months after the deadline) by means of the law LEY 28/2005, de 26 de diciembre, de medidas sanitarias frente al tabaquismo y reguladora de la venta, el suministro, el consumo y la publicidad de los productos del tabaco entering into force on 1 January 2006. In reality, however, compliance was not achieved until 13 January 2007 with amendments of the Spanish legislation following an infringement procedure pursued against Spain by the European Commission. Relying only on the EURLEX database we would have underestimated the transposition delay in Spain by more than a year.

To compare our index with Franchino's it is helpful to use the following formalization: let N_i be the total number of major provisions in a directive i, O_i ^j the total number of major provisions in directive i attributing obligations to actor j with some discretion, and C_i ^j the number of provisions attributing obligations to j without discretion. Actors can be the member states (ms) or the Commission (com). Since we only focus on member states, we did not add the suffix ms in the main text. For a directive i, N_i  = O_i ^ms +C_i ^ms +O_i ^com +C_i ^com . Focusing on member states, Franchino's delegation index is defined as O_i ^ms/N_i , which is equivalent to the discretion index of Thomson et al. Our index, using this notation, would read as d_i ^ms = O_i ^ms /(O_i ^ms +C_i ^ms ), which differs with regard to the denominator. In addition, Franchino's discretion ratio for member states is defined as (O_i ^ms/N_i )-[(O_i ^ms/N_i )*(L_i ^ms /12)], where L_i ^ms is the number of different types of constraint imposed on member states by the European legislator in directive i. Note that L_i ^j /12 is Franchino's constraint ratio.

These judgements are based on interviews with Commission officials on the directives researched in this article. Following our method, we computed the following discretion scores: 0.09 for the directive on tobacco advertisement, 0.43 for the directive on the resale right, 0.44 for the directive on the long-term status of third-country nationals, and 0.52 for the directive on protection from vibration.

For example, the resale right directive was transposed in Belgium by a law and by a royal order that concentrates on enforcement, practical aspects and administrative matters. Since more than one legal act was used, and at least one of the acts is a law, the misfit in the Belgium case was ranked as ‘high’. The same directive was transposed in Ireland with a (new) government regulation. Since the subject matter of the directive had not been regulated in Ireland before, the transposition of the directive did not have to supplant any existing legislation. Given that only one lower-level legal instrument was used in the transposition, the legal misfit was considered ‘small’.

The effective deadline in the cases of Bulgaria and Romania has been taken to be 1 January 2007 – the date of accession of these countries.

In addition, Denmark, Ireland, and the United Kingdom have opted out of the directive on third-country nationals.

Furthermore, we look into the ‘worlds of compliance’ hypothesis. With the ‘world of dead letters’ as a reference category, only the ‘world of law observance’ (Sweden, Denmark, and Finland) is significantly different from the rest. No differences in performance in regard to the ‘world of domestic politics’ (p = 0.61) and to the ‘world of neglect’ (p = 0.38) can be noted.

In a separate model (not reported here), we use the individual variables part of the measure of legal fit instead of the composite measure. The individual variables are all in the expected direction although not statistically significant. This finding highlights the added value of combining several indicators in order to create a better measure of legal fit.

Testing for the effect of federalism (by using a dummy variable for federal countries) reveals that, although federal countries use more time to transpose than unitary states, the effect is not statistically significant (p = 0.34). Similarly, using the index of regional autonomy developed by Gary Marks, Liesbet Hooghe, and Arjan Schakel (Marks et al. 2008 Marks, G., Hooghe, L. and Schakel, A. 2008. Regional authority in 42 democracies, 1950–2006: a measure and five hypotheses. Regional and Federal Studies, 18(2–3): 111–302. [Taylor & Francis Online] , [Google Scholar]) we find that countries with a higher degree of regionalism transpose slower, but the effect is not significant (p = 0.37).