19
countries on their distance from the median EU country along two dimensions: type of national
production system and extent of welfare redistribution. This is defensible but crude.
12
We also
do not rule out the possibility that the effect of types of capitalism is blurred by poor
information—or information processing—on the part of citizens.
Political-economic factors do not reduce party-level variance very much, and they
explain only an additional 1.3 percent of variance at the individual level. Interactive variables
for Stolper-Samuelson effects have the correct signs, but only Professional/Manager*GNI is
weakly significant. Three other individual-level political-economic variables are significant in
our analysis. The strongest is National Economic Prospects. A national pessimist at the 5
th
percentile is estimated to be 6.5 percent less supportive of European integration than a national
optimist at the 95
th
percentile, holding other independent variables at their means (Figure 1).
Pessimism about one’s personal economic prospects (at the 5
th
percentile) depresses support by
an estimated 3.8 percent, compared to optimism at the 95
th
percentile. Finally, support for
European integration is on average 3.1 percent higher among students than among respondents
who left school before the age of 16, holding other independent variables at their means.
The more individuals benefit economically from trade liberalization in Europe, the more
likely they are to support European integration. So support is generally stronger among highly
educated individuals and professionals or managers in relatively wealthy countries. Subjective
and socio-tropic (i.e. national) conceptions of benefit also have the expected effect. Support for
European integration is higher among those who are happy with their economic prospects and
those of their country, and is higher in countries that are net beneficiaries from the EU, and
higher in political economies that are closest to the EU median. Together, these variables
account for 15.0 percent of total variance, which is line in with previous studies. The surprise is
that they are overshadowed by identity.
Convention