Latin American countries are more homogeneous and can more readily be pooled as long
as proper interaction terms are specified.
The paper proceeds as follows. Section 2 gives a short introduction to the ways regional
differences have been modelled econometrically in the empirical FDI literature. Section
3 reviews the subset of empirical FDI studies that have set out to explain regional differ-
ences in FDI by including regional dummies, by including interactions between regional
dummies and selected explanatory variables to control for heterogeneity, or undertaking
regional studies that assume complete heterogeneity between regions and furthermore al-
low for the inclusion of region-specific variables. Section 4 applies a general-to-specific
analysis of 36 potential FDI determinants in 100 developing countries on an overall as
well as on a regional basis. Finally, Section 5 summarises and concludes.
2 Modelling the Regional Aspect of FDI
The regional aspect of FDI has been approached in many different ways in the empirical
FDI literature. At one extreme, it has been argued that foreign investors think of countries
as being completely independent and homogeneous so that FDI flows can be explained by
the same set of explanatory variables and homogeneous parameters independent of the
countries included in the sample:
FDIn = « + χitβ + uit, (1)
where FDIn is the inflow of FDI to country i (г = 1,..., N) as a share of GDP, N is the
number of developing countries in the sample at time t (t = 1, ...,T), Xit is a vector of
FDI determinants and Uit is an error term. In this case, regional differences in the inflow
of foreign capital can be fully explained by different country characteristics captured by
Xit. The broad FDI literature based on (1) has been reviewed quite frequently but so
far no consensus about the theoretical model or the econometric specification of the FDI
relation has been reached (see Bloningen (2005) for a recent survey).
The inability of (1) to explain the distribution of FDI across countries and regions
has lead some researchers to look for new explanatory variables to be included in Xit
(most notable is the recent inclusion of various risk variables), while others have tested
alternative ways to model FDI. This paper focuses on the latter approach and reviews
empirical FDI studies that allow for regional heterogeneity.
The first group of studies bases the analysis on a panel of countries belonging to
different regions. In general, this group of studies base their empirical FDI specification
on a variant of:
FDIijt = &j + xijtβj + uijt, (2)