proposals were advanced in the twelve months preceding the Council's decision not to step up
the excessive deficit procedure for Germany and France and close to 50 % in the seventeen
months period leading to the actual reform of the SGP in early 2005. This corresponds to an
average output of 2.7 proposals per month.
In the beginning and up until November 2003, the debate was mostly sustained by economists
from academia. By November 2003 around 60 % of all academics in our samples had already
advanced proposals. Conversely, the bulk of the non-academic economists started to get
involved only after the SGP crisis had become acute, i.e. after November 2003.
At a very early stage, that is at the time of the Maastricht agreement, the academic discussion
on the design of the budgetary framework underpinning EMU was not particularly animated,
though there were some early contributions, for example Buiter et al. (1993). Most attention
had naturally been given to the proper conditions for membership. Among the 101 proposals,
those contributions relating to alternative frameworks, such as market solutions, generally
surfaced at this stage.
As regards the proposed degree of modification to the old SGP, the reform proposals tended
to converge towards a clear majority view as the debate unfolded. In the very early stages,
there was a sharper division of views with a still relatively large share of proposals that did
not see any need for reform on the one side and an also relatively large group favouring
radical changes to the SGP on the other. The gap narrowed over time as an increasing share
came round to the view that there was a need to adjust the existing rules of the SGP. At the
same time that this consensus emerged, questions of political economy started to gain ground
over welfare considerations. This probably reflected the experience of the stalemate between
the Council and the Commission, which highlighted the institutional and credibility problems
of the existing framework.
3.2. Cluster analysis
Cluster analysis is a useful exploratory statistical tool for organising data into groups of
related observations such that those within a specific group are more similar to each other
than they are to those in other groups. For many multivariate data sets, including our 101
reform proposals, clustering is instrumental in providing a meaningful description of the data.
Since no objective criteria for choosing the ‘optimal’ number of clusters exist, we proceeded
on a tentative basis by successively increasing the number of clusters, starting with two. This
-14-