2 Related literature
The role of both distance and size has been extensively investigated in the
empirical literature on multinationals (henceforth, MNEs). A positive relation
is found between market size and FDI inflows. Wheeler and Mody (1992),
for instance, show that capital expenditures by US MNEs in a given market
increase more than proportionally with market size. Furthermore, FDI flows
are more intense among countries with similar market sizes (Markusen and
Maskus, 1999; Carr, Markusen and Maskus, 2000). As to the empirical relation
between economic distance and FDI flows, the picture is somehow less clear-
cut. A first approach is to relate measures of FDI (stocks or flows) to measures
of economic distance between source and host countries. As reported in Shatz
and Venables (2001), controlling for market size, FDIs from most major sources
(US, EU and Japan) tend to fall with the distance to the host country, as
in gravity-type empirical analyses of trade volumes.5 However, this approach
cannot help to understand whether FDI becomes more or less likely than trade
in serving foreign markets as economic distance rises. Brainard (1997) analyses
at the sectoral level the share of exports over total US sales (including also US
affiliate sales) in each foreign market. Trade costs are found to affect positively
this measure of FDI activity versus trade. A firm-level analysis on Swedish
MNEs with a similar aim is found Ekholm (1998), who investigates first the
decision to serve a foreign market via exports or via FDI, and subsequently
the pattern of Swedish exports over total Swedish sales. In her results distance
is negatively related to the decision to undertake FDI in a given market, but,
once the decision is taken, the share of affiliates sales on total Swedish sales
rises with distance, as found in Brainard (1997). These contributions focus on
horizontal FDIs. Other studies are targeted to international investments of a
vertical type. Shatz (1999) considers exports of affiliates of US MNEs located
in different developing countries. He finds a positive relation between exports
directed to the US and measures of transport costs and trade openness.
As to the non-monotonic relation between FDI and distance for large mar-
kets reported in Figure 1, the existing literature suggests possible explanations
that rely on the complex interplay between trade and transport costs, market
size, factor endowments and the extent and composition of FDI (horizontal vs
vertical). In particular, the theory of MNEs points out that the relation between
the likelihood of FDI and distance depends crucially on the type of FDI consid-
ered. Horizontal FDIs, aimed at selling in the local market are more likely the
higher transport costs are (see, e.g., Horstmann and Markusen, 1987; Brainard,
1993; Markusen and Venables, 1998). When FDI is vertical (Helpman, 1984),
aimed at saving on costs in a particular production stage, it is instead more
likely directed towards relatively close markets. The reason is that in this case
FDI is not a substitute for trade, but rather a complement, since trade flows will
occur intra-firm, and economic distance adds costs to the MNE. As illustrated
5 A negative sign for distance (and a positive one for trade barriers in host countries) is also
found in Carr, Markusen and Maskus (2000) that analyse the determinants of US affiliates’
sales.