insignificant.)
Finally, Table 9 explores the effect of age on protectionist attitudes in greater detail.
Consistent with our expectations, age squared has a negative coefficient, suggesting that the marginal
effect of age declines as respondents get older. Age remains insignificant in the west, but important in
the east: could it be that the overall correlation between age and attitudes is simply being driven by
the length of time over which east-bloc respondents were exposed to Communist ideology? Two
pieces of evidence suggest not. First, columns 4-6 in Table 9 introduce an interaction term between
skill and age. In the west, the term is negative (and significant at the 12.5% level), suggesting that the
high-skilled become more pro-free trade the olderthey get. If respondents’ human capital increases
with age, as the labor economics literature suggests, then this is precisely what one would expect. On
the other hand, in the east the high-skilled become more protectionist the older they get; which is
consistent with the skills acquired under Communism being largely irrelevant to a modern economy,
and to these individuals’ position in society being threatened by liberalization.28 The second piece of
evidence suggesting that age is influencing attitudes through economic channels is given in Table 8: it
shows that the impact of age on protectionist attitudes is much larger for those in the labor force than
for those not in it.
So much for statistical significance: what about the quantitative impact of these variables on
attitudes towards protection? In order to answer this question, we began by estimating the model in
column 4 ofTable 4, and then set all right hand side variables equal to their median values. Having
done that, we calculated the impact of changing each individual independent variable on the
probabilities that ‘protect’ would take on each of the values 1to5. For binary variables, we
considered the impact of changing the variable from zero to one; for other variables we explored
more complicated changes, as discussed below.
Our point estimates of the coefficients and cutoff points, together with the assumption that
28 Indeed, labor economists have found that the ‘returns to experience obtained under communism
fell during the transition” (Svejnar 1999, p. 2839).
21