economic-determinist or reductionist explanation of nationalism is, however, implausible. It seems
safe, therefore, to conclude that nationalist attitudes exercise some autonomous influence and are a
significant factor in the genesis of protectionist policy preferences.
The second result is that, even when cultural or ideological factors are taken into account,
skill matters for policy preferences, and the effect that skill has on those preferences varies across
countries in ways which are consistent with Heckscher-Ohlin theory. Roughly speaking, in countries
with per capita incomes below $12000 the lowest skilled tend to be more in favor of free trade, while
they tend to be more protectionist in countries above that income threshold. Belonging to one of the
higher skill categories only shifts attitudes in a slightly more liberal direction in poor countries; the
effect is a large one in rich countries. Finally, there seems to be a strong negative relationship
between the impact of skills on protectionist attitudes, on the one hand, and income per capita on the
other. That is, high skills are generally associated with a preference for free trade, and this effect is
stronger in richer countries than in poorer countries. Indeed, in some of the poorest countries in our
truncated sample, high skills are, albeit weakly, associated with a preference for protection.
Third, we have found evidence of a pronounced gender gap regarding trade policy
preferences, which is quantitatively important and apparently consistent across countries. We have no
explanation for this phenomenon, though we note that it is consistent with findings regarding the
persistence of gender differences in support for European integration even after controlling for the
effects of a wide range of other variables (Wessels, 1995, 111-114).30
In terms of the economic debates surrounding the determinants of protectionist attitudes, we
have, as stated, found evidence broadly consistent with the Heckscher-Ohlin viewpoint. What we
have not been able to do, given the data we are currently working with, is to compare the impact of
skills on protectionist attitudes with the impact of the sector of employment. Nor have we been able
to follow Scheve and Slaughter in exploring the impact ofhome ownership. We will shortly have
30 Similarly, Carol Graham and Stefano Pettinato find that Latin American men are more likely to
have pro-market attitudes (Graham and Pettinato 2001), and to agree with the proposition that trade is
‘very important’ to their country’s prosperity (Graham and Pettinato forthcoming).
24