approach reveals four relatively stable groups across different types of clustering methods.15
All clusters beyond four contained only a limited number of proposals and hence did not
contain enough information to qualify as a meaningful and distinct group. By contrast, less
than four clusters were not enough to disclose differences across some of the variables
characterising the 101 proposals. The distribution of the characterising variables of the 101
proposals across the four clusters as well as across the characterising variables is presented in
Table 3. It must be stressed that the four clusters do not form perfectly homogenous groups of
economists in terms of the characterising variables. The lines of demarcation are not that
sharp. For some variables the differences across the groups are not statistically significant.
However, the four clusters reveal patterns that give rise to sufficiently distinct and informative
profiles. They mainly serve illustrative purposes.
The four groups identified in our cluster analysis and presented ‘in order of appearance’ can
be characterised as follows.
Disenchanted reformers: The first group, which could be referred to as disenchanted
reformers, includes 31 contributions from professional economists who hold very critical
views about either the general utility or the effectiveness of the SGP or both. As a
consequence, they mostly propose relatively incisive reforms, including major changes to the
existing institutional set-up and/or its replacement by a totally different arrangement.
Part of their reservations about the SGP may be explained by the fact that the group includes a
relatively large share of economists from the US or from non-euro-area countries, hence not
necessarily sharing a passion for the idea of a common currency or the related common fiscal
framework. However, the euro-area economists in the group also hold fairly critical and
sometimes disillusioned views. A further distinguishing feature of the disenchanted reformers
is that they seem to have relatively clear ideas. Specifically, whereas multiple reform
proposals are much more common in the other three groups, more than 90 percent of the
papers presented by the disenchanted reformers advance only one suggestion about how to
'cure' the Pact.
The average date of publication suggests a relatively early entry into the reform debate,
around 13-14 months prior to November 2003. However, the proposals are widely distributed
across time, including a number of very early papers, published well before any signs of SGP
weakness had appeared on the horizon, as well as a number of very late works released after
15 We used chi-square as measure of distance and Ward's method for grouping cases into clusters.
-15-