Policy networks, national patterns and Europeanization: Understanding Belgian GMO policy within European framework

Belgium is a founding EU member which appears to be an interesting case study about risk public policies and especially GMO regulation. Belgium has played a pioneering role in the area of R&D on GMO. Several universities and private companies invested their efforts in GMO. Moreover Belgium remained a relatively important area ? considering the size of the country compared to other European states ? for the location of companies. They developed activities in the field of contained use, and even deliberate release. From the eighties to date, the development of the GMO sector has been closely linked to the evolution of the policy design's content. Being an EU member, Belgium must transpose directives (namely 2001/18/EC) and directly apply regulation (especially n°1830/2003). They specify constraints (administrative procedures, labelling and traceability). Through a triangulated methodology (documentary analysis, interviews and reputation approach), we highlight that the explanation of GMO regulation should draw from three approaches. First, a policy sector approach (Marsh and Rhodes 1992) helps identifying the network concerned by the public regulation of GMO. We can qualify it as an ?issue network? (versus ?policy community?). Second, the ?National Patterns Approach? (Richardson 1982, Vogel 1986) allows taking into account the weight of political parties, the federal structure and the role of the administrative agency in charge of implementing the GMO policy. Third, we take into account the Europeanization process (Héritier and Knill 2001). The European impact on Belgian regulation interacts with its domestic characteristics (i.e. a contested interest constellation and a relatively even distribution of powers and resources across opposing actor coalitions). Finally, we can label the type of GMO regulation thanks to ideal-types and compare the Belgian case study with other European countries, US and Canada (Montpetit, Rothmayr and Varone 2007) .