Table 4 presents the benchmark result for the Union as well as each
member state except Belgium.23 The variable of our main interest is employ.
We find that the status of being an employer is not statistically significant
at the Union level. However, it is so for four countries. Employers were less
likely to prefer immigration restriction than the others only in Austria: the
probability of being restrictive is lower for employers than the rest by about
.2. For the Netherlands, Sweden and the United Kingdom, the estimated
probability difference is positive, although the size is small ranging from .065
in Sweden to .080 in the UK.
[Table 4 about here]
Turning to the other variables, the estimated marginal effect of isb also
has different signs across the countries. It is significantly positive for Ger-
many, Greece and Luxembourg and is negative for Finland and the UK.
However, it is significant not only statistically but also in size only for Lux-
embourg where an increase in the sectoral employability of foreign workers
by .1 is related to an increase in the probability of being restrictive approx-
imately by .02.Themarginaleffect of isb is not statistically significant at
the Union level.
The estimated effect of unemploy suggests that the unemployed were in
general more likely to support immigration restriction than the rest: the
probability of being anti-immigration is higher for the unemployed than the
variables should be used in modelling the variation, for different sets of variables passed
the significance test.
23 In the EU probit, we include 13 country dummies, although the corresponding esti-
mates are not reported in the following tables due to a lack of space.
16