more than the annual average) in Catalunya and Baden Württemberg.7 In these last two regions, the
FDI inflows do not target a unique sector all time long, but they alternatively flow to different sectors.
The results obtained mean that either the region (such as Lombardia) displays a strong and
permanent FDI attractiveness in a sector or the region possesses a changing FDI attractiveness over
time (such as Catalunya and Baden-Württemberg). Such an effect can be related to the regional
economic environment and the corresponding changes occurring during the time period of our study.
Regarding the other determinants, the results are different across regions and these observed
differences constitute the most interesting insight of this paper. Recall that we selected 6 sectors
common to the three regions and that we run FDIP, all sectors together,8 on a few determinants for
each region. Our results show that some determinants appear to be statistically significant for some
regions and not for the others. We propose three arguments to explain those differences. First, it is
possible that foreign investment is attracted by a variety of determinants, a few being predominant
(such as GDP per capita) and others less relevant. Therefore, different sets of determinants are
sufficient to attract FDI as long as market size exists in the region. This would confirm GDP per capita
as a sufficient determinant. Second, the FDI performance may be driven by particular determinants
over that period reflecting strengths and weaknesses of each region relative to the endowment in those
determinants. Third, for a given sector, the production of this sector may be of different range or
quality across regions (for instance, luxury and low-range products in the textile sector) and, hence,
investment in that sector may be responsive to different FDI determinants relative to the range.
Let us make some comments on the results for a few determinants:
a) As we mentioned, the dummies and fixed effects are always very significant. It implies that
heterogeneity is an important component in our analysis. By heterogeneity, we mean the
characteristics related to each sector: for instance, risk, entrepreneurial ability 9 and, also,
asymmetric supply or demand shocks that can affect regional economic activity. By doing this,
we control the effects of the regional economic environment.
b) Regarding the indicators of productivity, the results are consistent. The unit labor cost indicator
(ULBV) is significant for Lombardia and Baden-Württemberg and so is the real productivity
measure (RPRODUC). Both indicators are not significant for Catalunya. No obvious
explanation comes to mind. One comment deserves to be mentioned. The economy of
Catalunya, like that of the rest of Spain, has been growing significantly in that period catching
up the EU living standard. This has happened despite a low productivity growth. Possibly, the
market potential in Catalunya, as part of the EU, was attractive enough for foreign investors
regardless of its productivity performance. This does not rule out the possibility that, in the
future, productivity becomes a statistically significant determinant in Catalunya.
c) The regional export performance (EXPORTP) is significant and positive only in Lombardia
and Catalunya. This relationship between foreign investments and export performance may
indicate that those foreign investments have contributed to the export performance, or the
increasing export performance has been a good signal in terms of competitiveness for foreign
investors to favor those destinations. This result is interesting because this relationship does not
7 The Finance and credit sector has been removed from the FDI distribution in Baden-Württemberg.
8 The lack of data does not allow for regressions by sector separately.
9 As argued in Henderson (2003) when using fixed effect at plant level.
19