10% of Italy. Hence, in Table 1, the U.K. is the country with the highest share of skilled
immigrants, both considering the OECD and non-OECD immigrants’ populations, Italy is the
country with the lowest share of skilled among the non-OECD immigrants and Germany is the
country with the lowest share of skilled among the OECD immigrants.
V. Key findings
The results of the estimations on the four databases share several regularities. The static
Models (Models 1 to 4 and 6 to 9 in Tables 2, 4 and 5, and 1 to 5 and 7 to 11 in Table 3) will be
considered first. Most coefficients of the base gravity equations are as expected. They are positive
and significant for the total GDP variables (TGDP) and negative for the difference in countries
GDP (sq_GDPDiff), both in the inward and outward FDI equations of the four countries (Tables 2
to 4). This confirms the expectation on the prevalence of the horizontal model of FDI. The
coefficients of the differences in the per-capita GDP, PCGDPDiff, are less homogeneous. For three
countries, Germany, Italy and France, results are not robust to the different specifications, while
they are positive and significant in the U.K. outward equations (Table 5, Models 1 to 4). This seems
to indicate that the U.K. multinationals follow both models of international expansion, horizontal
and vertical. The coefficient of the PCGDPDiff variable is positive and significant in the inward
equations of France, Germany and Italy, suggesting the existence of some vertical forms of
offshoring among developed countries.
Differently from expected, distance (DIST) is not an impediment to the outward FDI of three
of our four countries, U.K., Germany and France. This especially holds for the U.K., where the
coefficient of the distance variable is positive in Models 1 to 4 of Table 5. At the opposite, it is
negative and significant for the Italian outward FDI, in Models 1 to 5 of Table 3. In the regressions
corresponding to Germany and France, the signs of the coefficients turn from negative in Model 1
to positive in Model 2 (Tables 2 and 4), suggesting that, as expected, the presence of immigrants
lowers the distance costs of investing abroad. The sign of the distance variable is instead negative in
the inward regressions of the four countries, in all specifications. This lack of symmetry between
the inward and outward FDI seems to indicate a wider geographic expansion of the investments of
our four countries into the world markets relatively to those received from abroad. An exception is
Italy, where the sign of the coefficient is negative in both the inward and the outward equations.
The differing sign and significance of the coefficient of the OPENNESS variable between
the regressions and specifications shows an heterogeneity of relations between our four countries’
FDI and the partner economies’ openness to the international trade. The sign is positive, significant
and robust to all the specifications only in the Germany’ outward FDI equations (Table 2, Models 1
13