such a model).16 Our data set allows us to explore whether skill has a differential impact on attitudes
across countries, and thus allows for a cleaner test of the Heckscher-Ohlin predictions. Finally, our
data allow us to test these ‘economic’ relationships while controlling for the effect of variations along
two dimensions of nationalist ideology.
4. The determinants of attitudes towards protection
We begin by looking at some crude country-level correlations. Table 3 takes the country
means of seven variables (given in Table 2), and calculates the cross-country correlations between
these variables. While this ignores the vast range of variation in attitudes within countries, the data
are nonetheless instructive. Richer countries have higher skill levels, and higher rates of internal
mobility, than poorer countries; they also show more patriotism, and less chauvinism. There is a
strong positive correlation (+0.691) between protectionism and chauvinism, but only a weak
correlation between protectionism and patriotism. Skill, mobility and income per capita are all
associated across countries with pro-free-trade sentiments, rather than with protectionism.
However, our interest is in the determinants of protectionism at the individual level and Table
4 provides some exploratory analysis. In all cases, the dependent variable is ‘protect’, which as
already mentioned is an ordered variable running from 1to5. We therefore used ordered probit
methods in estimating our relationships.17 In each case, there is assumed to be a latent variable,
PROTECT*, related to the independent variables as in equation (1):
PROTECTij* = αi + β1SKILLj + β2SKILLj*GDPCAPi + β3Xij + εij (3)
There are also 4 cutoff points, μ1-μ4, such that protect takes the value 1 if PROTECT* lies below μ1, 2
16 For example, it might be the case that better-educated people everywhere are more flexible and
able to cope with the rigors of the market; or even that they are more likely to understand the
intellectual case for free trade.
17 For an introduction, see Greene (2000), Chapter 19.
13