In our model the distinction between horizontal and vertical FDI is somehow
blurred, as it is often in reality.2 The production process consists of two activi-
ties: an ‘upstream’ activity that we interpret as the production of intermediate
inputs, and a ‘downstream’ activity that we interpret as assembly (or commer-
cialization). To keep the model as simple as possible, we focus on a local market
where the final product is supplied only by foreign firms. These firms choose
their supply mode between exports and FDI. This latter option, however, is
available only in the downstream stage. This assumption is supported by em-
pirical evidence. For instance, the share of value added on sales for US foreign
affiliates is lower compared with domestically owned establishments based in
the US, which means that only a subset of production stages are performed in
foreign subsidiaries (see Hanson, Mataloni and Slaughter, 2001). So, in spite of
the fact that FDI in our model is carried out with the aim of serving the local
market (as it is typical of horizontal FDIs), FDI activity is associated with the
geographical separation of production stages (as it is typical for vertical FDIs).
As implied by the proximity-vs-concentration framework, the choice of sup-
ply mode entails a trade-off. Exportation faces trade costs but saves on the
additional costs of distant operations, while the opposite is true for FDI. In
particular, we assume that carrying out FDI requires a local investment in as-
sembly lines. The larger the investment, the lower the variable costs but the
higher the plant-specific fixed costs. A firm that undertakes such investment
is a multinational enterprise (henceforth, MNE) and faces a choice in terms of
intermediate supply. The MNE can produce its intermediates in its country of
origin and then ship them to the assembly lines. Alternatively, it can outsource
their porduction to local suppliers.3 Therefore, FDI always concerns a subset
of activities, but these may or may not entail intra-firm trade depending on
whether self-production or outsourcing are chosen.
The choice between self-production and outsourcing of intermediates intro-
duces a second trade-off. Self-production incurs in trade costs because interme-
diates have to be shipped to the distant assembly line. Outsourcing saves on
trade costs but faces additional costs on its own. We focus on the transaction
costs associated with the outsourcing agreement between the MNEs and their
local suppliers.4 Specifically, we follow Grossman and Helpman (2002a, 2002b)
2 Recent evidence shows that the benchmark distinction between horizontal and vertical
FDI does not capture the growing importance of MNEs’ expansion strategies which involve
FDIs having both horizontal and vertical features (see Hanson, Mataloni and Slaughter, 2001).
3 See, e.g., Hummels, Rapoport and Yi (2001) for empirical evidence on the growing impor-
tance of within-sector trade in intermediate inputs. Hanson, Mataloni and Slaughter (2001)
report that the share of input purchases over sales for US foreign affiliates in 1994 is 12 per
cent for manufacturing as a whole, with shares over 20 per cent in electronics and transport
equipment.
4 Transaction costs are at the centre of the theory of the firm since the pioneering work
of Coase (1937). Williamson (1985) revived the insights of Coase relating the efficient de-
termination of firms’ boundaries to investment incentives in the presence of asset specificity
and incomplete contracting. The first formalization of the hold-up problem in the presence of
asset specificity and incomplete contracting is found in Grout (1984). Hart and Moore (1990)
study the implications of alternative ownership structures on investment incentives within the
firm.