Multinational firms, technology diffusion and trade

Wilfred J. Ethier, James R. Markusenb

Attempts to add multinational firms to the new trade theories generate predictions about direct
investment dramatically at odds with the facts. A failure to treat internalization adequately seems
responsible at least in part. Empirical evidence also indicates a close association between
multinational firms and knowledge capital; we exploit this to examine the role of internalization
and its relation to locational factors. We model a firm which must choose between costly exporting
and the possible dissipation of its knowledge capital by producing abroad. The paper examines the
choice between exporting, licensing, and acquiring a subsidiary in this environment. We analyze
the cost and technology parameters that support alternative modes of serving the foreign market,
and we describe the international equilibrium that jointly determines the pattern of specialization
and the market mode. We find that similarities in relative factor endowments may promote direct
investment.
First, the new trade theory emphasizes just those features that appear central to multinationals:
economies of scale, imperfect competition,strategic considerations, and technological change and
diffusion.
The second reason the new trade theory is important for the theory of the multinational firm, a
reason which is still much less appreciated, is that the theory of direct investment has its own
strong tension between existing theory and apparent fact
With patterns of direct investment increasingly mimicking patterns of trade, it is perhaps not
surprising that a large share of trade is now intra-$rm as well as intra-industry. Thus the new
trade theory itself supplies a picture of multinational firms that contrasts sharply with the stylized
facts5 of contemporary direct investment
This paper addresses the nexus of increasing returns to scale, international trade,growth and
technological competition and dissernination. We instead endogenize market structure and link it
to the intemational dissemination of knowledge. Multinational enterprises (MNEs) are closely
linked to knowledge-based capital. It is not surprising then to find an association between MNEs
and the production of high-tech, R&D intensive goods.
Also central is the choice of how the firm services a foreign market. We consider three choices:
(1) exporting, (2) licensing a foreign firm, and (3) establishing a subsidiary (i.e. becoming a true
MNE).
Mansfield (1994) presents survey evidence that the perceived degree of intellectual property
protection provided by developing countries influences the willingness of multinational firms to
establish subsidiaries or undertake joint ventures and, even more strongly, influences the extent
of technology transfer to such operations industry-specific characteristics significantly affect the
ability of host-country firms to exploit lax protection
We model source-country firms that continually compete to introduce new products. A successful
firm has a temporary proprietary advantage in its new product (the ownership advantage) and
must decide how to exploit this advantage abroad The MNE must then choose between costly
exporting and the possible dissipation of its proprietary asset.

We show that there are five distinct equilibrium outcomes possible: (1) the source-country firm
exports in both periods, (2) a subsidiary arrangement in which the source-country firm extracts all
potential rents, (3) a subsidiary arrangement in which the source-country firm shares the rents
with its subsidiary employees, (4) exporting in the first period followed by a one-period license in
the second, or (5) two successive one-period licenses with distinct host-country firms.
If (l), (2) or (3) is chosen, all rents are extracted from the host-country market; (3) involves the
sharing of rents between the source-country firm and the subsidiary while the source-country firm
captures all rents in (1) and (2). With (4) and (5) some rents are dissipated but the licensee(s)
earn no rents: the source-country firm prefers dissipating some rent to sharing the maximum rent
with a subsidiary.
Results suggest how the type of equilibrium relates to the importance of pure public-good
knowledge capital (intellectual property)

Concluding remarks
This paper investigates the interplay of locational and internalization considerations when direct
investment emerges. We draw on the empirical association between MNEs, knowledge-based
assets, and advanced technology to analyze the choice of mode for serving a foreign market. The
joint input property of knowledge-based capital that supports MNE market structures in the first
place also implies a risk of asset dissipation when the knowledge capital is transferred to a hostcountry firm. We assume that foreigners eventually learn to produce a good on their own, and
learn faster if the good is actually produced in their country than if it is imported. We also assume
that no contracts can enforce protection of intellectual property (knowledge capital), which may
be a preferred assumption, in the international context, to complete enforceability.
A source-country firm with a new product must choose between costly exporting and the early
loss of the value of its knowledge as a result of producing abroad. When producing within the host
country, the source-country firm mustchoose between a product-specific license and a subsidiary
(becoming an MNE).
The MNE structure is chosen when (1) knowledge capital is of medium or high importance relative
to physical capital; (2) discounting is low so that commitments to future products are valuable; (3)
exporting costs are medium, so that exporting is not chosen yet some disincentive is provided to
subsidiary employees not to defect and face competition from exports; (4) the wage in H is low
relative to S to encourage production in Z-Z, but not so low as to furnish no disincentive to
subsidiary employees to defect; and (5) fewer S firms compete in product development, so that
the subsidiary attaches a higher probability to successful new products. The importance of
knowledge capital and the results that higher international wage disparities and lower transfer
costs do not necessarily encour
The firm may also resort to successive one-period licenses or to exporting followed by a license
when there exists no second-period share that would not lead one partner to defect. The firm thus
chooses among five modes of serving the host-country market. We related each choice to the five
basic parameters. When exporting, RCM, or RSM is chosen, all possible rents are extracted from
the host-country market. The source-country firm retains all rents in the first two cases but shares
them in the RSM mode. With successive one-period licenses or exporting followed by a one-period
license, some rent is dissipated. The source country firm finds the moral hazard problem severe
enough to dissipate some rent rather than give up the share necessary to induce a
subsidiary/licensee not to defect.

Our model differs from others, such as Ethier (1986), in allowing locational and internalization
considerations to work simultaneously. We conclude with a description of general equilibrium in
terms of basic parameters that highlights this interplay in determining whether MNEs emerge.
Intuitively, the two countries must be sufficiently dissimilar for the S firm to find it advantageous
to allow production in H (locational), but they must be sufficiently similar for a multiperiod
agreement to be sustainable (internalization). Depending on parameters, either influence may
dominate. We examine the effects of relative factor endowments and of relative country size on
the pattern of specialization, the choice of market mode, equilibrium relative wages and the
intensity of research. A key result is that similarities in relative factor endowments may promote
direct investment when account is taken of the desire to protect knowledge-based capital.
We do not claim either that an association of direct investment with relative endowment
similarities requires a model like ours - Ethier (1994) and Markusen (1995) survey alternative
possibilities - or that our model necessarily generates such an association: the parameters matter.