homogeneous within a sector across the countries, we re-estimate the specifi-
cation of Table 4 without isb sector by sector. Due to the limited number of
observations, we did not re-estimate for each country but only at the Union
level. Note that, by dropping isb from the list of explanatory variables,
Belgian observations become useful. Hence we include Belgium in this es-
timation. Due to a lack of observations after listwise deletion, we ignore
fishing activities (category b) and mining/quarrying activities (category c).
Table 8 compares the estimated effects of employ and also unemploy across
the NACE Rev.1 major sectors.27 The figures in column i are from the full
sample, while those in columns ii and iii are from those who thought that
immigration would reduce wages and those who did not think so, respectively.
First, we notice that, using the full sample, employers are not particu-
larly more supportive of immigration than the rest in none of the sectors.
However, in the household-activity sector (category p), the estimated effect
of employ is 10%-significantly positive, implying that employers were more
anti-immigration than the rest in that sector. By controlling for the subjec-
tive perception of the wage effect of immigration, we find that the statistical
significance of the positive effect increases for the sector among both those
who expected immigration to decrease wages and those who did not. In
addition, the magnitude is larger among those who did not think that im-
migration would depress wages than those who thought so. What makes
employers in the household-activity sector more anti-immigration than non-
employers? In the sample, we find that all employers in this sector employed
27Due to a lack of space, the full results are not presented. The other estimates are
available from the author upon request.
29