east (consistent with Blanchflower (2001), who finds that women are less satisfied with democracy);
but age appears only to affect attitudes in the east. Unemployment is weakly associated with
protectionist attitudes in the west, but not at all in the east: attitudes towards being unemployed may
be quite different in a society undergoing rapid structural change than in a typical western economy.
Interestingly, the impact ofbeing rural appears to be almost identical in east and west: this is at odds
with the intuition that western farmers should be more protectionist, since they are less competitive
and more dependent on protection than their eastern counterparts. Of course agriculture is not the
only rural industry, but it is the main one; the data are consistent with other rural characteristics of a
sociological or even cultural nature driving these correlations.
We estimated the model for men and women separately; the results are given in columns 5
and 6. The most notable difference to our mind is that while age appears to matter for men, it does not
matter for women. We were also interested in whether the gender effect was due to the fact that
women are less likely to be in the labor force than men. We therefore estimated the model separately
for those in the labor force and those outside it; the results (given in Table 8) suggest that the gender
effect is not due to lower female labor force participation, since the gender effect is actually stronger
forthose in the labor force than forthose outside it.27
The last two columns in Table 7 address the issue of whether those reporting themselves to be
immobile (within the country) are less influenced by their skill type than those who are mobile; the
reason for the question is that, arguably, the immobile should care more about their sector of
employment (assuming that regions are largely dominated by particular industries). The hypothesis is
rejected in that the interaction between skills and GDP per capita (the test ofHeckscher-Ohlin theory)
is even stronger forthe immobile than forthe mobile. (We also experimented by including interaction
terms between skill and mobility in various models: but these interaction terms are always
27 Moreover, the gender effect remains unchanged when a labor force dummy variable is included in
the model, estimated overthe full sample (results available on request). We counted those in full-time
and part-time work, as well as the unemployed, as being in the labor force. We could also have
included those working less than 15 hours a week, and relatives assisting, but there were too few
respondents in these categories (less than 500 across all countries) forthis to be worthwhile.
20