James R. Markusen
Department of Economics
University of Colorado
and NBER

This study examines foreign direct investment into developing countries, and the stake those countries have in liberalizing or restricting these long-term investments. We begin with a review of the characteristics of multinationals and the characteristics of countries which are sources and hosts to such investments. Then we propose and analyze a general theoretical framework which is consistent with a wide range of empirical evidence. Subsequent sections review further theoretical and empirical evidence on the consequences of investment liberalization, and public policies which might impact on the levels and consequences of inward investment. These sections set the stage for a discussion and analysis of participation by developing countries in Multi-Lateral Agreements on Investment. We examine the consequences and advisability of adopting various rules and commitments that may be proposed in international negotiations.

This draft: October 7, 1998

This study has been prepared by the author for the World Bank. It reflects the views of the authors and not necessarily the view of the Bank.

Over the past two decades, direct investment by multinational firms has grown significantly faster than trade flows, particularly among the world's most developed economies. International economic activity increasingly takes the form of non-arm's-length trade, and instead takes the form of foreign production by multinational firms and intra-firm trade by those same companies. It is now estimated that about 30% of world trade is intra-firm.¹ Yet we have a poor understanding of the ways in which direct foreign investment is just a simple substitute for trade, and ways in which it is something quite different. In the 1970s, many host countries and some economists viewed multinational investment as detrimental to host-economies' welfare and development, creating monopoly situations that exploited those economies and stifled local competition. The view in the 1990's is considerably different and more optimistic, suggesting that multinationals are important vehicles for transferring technical knowledge, labor and management skills, and even improving the competitive environment in host economies.

This study is intended to serve as a background document for analyzing and evaluating the developing countries' stake in a new round of multilateral negotiations on investment. The study will be largely conceptual, examining the determinants of foreign direct investment into developing countries and the consequences of such investments. However, I hope that it will be well grounded in extensive empirical evidence. Direct investment is inherently a second-best phenomenon, associated with a variety of distortions, and hence the answer to many questions all too often takes the form "it depends". The existing empirical evidence is very helpful in greatly delimiting the range of possible outcomes and allowing us to identify "best guess" parameterizations which permit a much sharper set of predictions. The results from positive analysis will then be used to discuss the developing countries' interests in committing to international rules on investment, and to suggest the advantages of alternative forms of these rules.

It is helpful to note early on what I will cover, but also what will be omitted. First, I will consider exclusively direct foreign investment, and not examine portfolio capital flows at all. Nor will I consider the financial flows that may accompany direct investment, leaving aside the question of whether firms raise funds at home or in local host-country capital markets. There is no suggestion here that financial flows are uninteresting and unimportant, rather their omission is a reflection of time constraints and my own lack of expertise in macro economics.

Second and closely related, I will be examining long-run structural issues for host economies. I will not consider issues related to short-term fluctuations and macroeconomic stability issues that may be connected to direct investment. Such issues are obviously of great concern to developing countries, but they fall into a somewhat different category than rules dealing with a multilateral code on direct investment.

Before reviewing the theory and evidence, a short taxonomy is needed. First of all, a multinational is a firm that undertakes direct investments in foreign countries, defined as acquiring a controlling interest in foreign subsidiary, either through a new ("greenfield") investment or through the acquisition of an existing foreign firm. Direct investments seem to be motivated largely by returns to specific business opportunities rather than by more general returns to capital which motivate portfolio investments.

Second, multinationals can be characterized by according to the relationship of their activities at home and in the host economy. Horizontal investments refer to ones in which the multinational carries on basically the same activity in the host country as at home. For example, a manufacturing firm that makes the same product (or components) in the home and host countries or a service firm that runs hotels in both countries is a horizontal multinational. Vertical investments are those in which the production process is geographically fragmented by stages. For example, capital-intensive intermediates may be produced at home and labor-intensive assembly might be done in a low-wage host country. The most superficial type of direct investment is in wholesaling, distribution, and servicing facilities where virtually no transformation of product is done. We will generally ignore direct investments of this last type in which there is very little value added.

Horizontal investment has been much more important quantitatively than vertical investment and tends to take place between the high-income countries. Vertical direct investment tends to take place between a high income and a low-wage partner or a resource-rich host country. In other words, vertical investment are related to difference in relative factor endowments between countries, while horizontal investments are associated with similarities in endowments and high per-capita incomes.

Another important taxonomy relates to what is referred to as the "mode" of entry by firms into foreign markets. One extreme is direct investment in a wholly owned subsidiary, and the other is arm's-length exports sold to foreign wholesalers. As noted above, one intermediate option is a direct investment in a sales branch in which the multinational does little or no production, but gains control of the distribution, sales, and servicing functions in the foreign market. Licensing, on the other hand, is an option which may involve substantial foreign production, but that production is carried on by independent foreign firms. Items licensed range from product and process technology, to patents and trademarks. Joint-ventures might be thought of as an intermediate option between a wholly owned subsidiary and a licensing arrangement.

Modes of entry thus differ with respect to the amount of actual foreign production that takes place in the host country and the degree of proprietary control over the production and distribution process. Modes of entry are endogenous, and influenced by a wide variety of market and public-policy characteristics, ranging from trade and investment barriers to legal infrastructure and the protection of intellectual property. Insofar as different modes of entry have different implications for employment levels and technology transfer, influencing entry mode becomes a policy issue.

In the next section of the study, I will turn to some empirical evidence on multinational firms and the theories that have been developed to explain these data.

Theoretical models should be empirically relevant models: models that are closely consistent with stylized facts in their major assumptions and which generate predictions that are similarly consistent with the facts and provide testable predictions. This is especially important in analyses with multiple distortions, insofar as theoretical models based on different assumptions can generate radically different policy recommendations. The empirical evidence can be used to discard a number of apparently irrelevant alternatives.

In this section, we review some of these stylized facts which will provide a basis for models to follow. First, a number of studies have documented what types of firms and industries are dominated by multinationals relative to firms and industries which are primarily national in scope. These are as follows.²

1. Multinationals are associated with high ratios of R&D relative to sales.
2. Multinationals employ large numbers of scientific, technical, and other "white collar" workers as a percentages of their work forces.
3. Multinationals tend to have a high value of "intangible assets"; roughly, market value minus the value of tangible assets such as plant and equipment.
4. Multinationals are associated with new and/or technically complex products.
5. Evidence suggests that multinationality is negatively associated with plant-level scale economies.
6. Multinationals are associated with product-differentiation variables, such as advertising to sales ratios.
7. A minimum or "threshold" level of firm size seems to be important for a firm to be a multinational, but above that level firm size is of minimal importance.
8. Multinationals tend to be older, more established firms.

This evidence suggests that multinational firms are found in industries in which knowledge capital and knowledge-intensive production are important. Next, consider some findings about what types of countries are home and hosts to foreign direct investment.

1. The high-income developed countries are not only the major source of direct investment, they are also the major recipients. Most direct investment seems to be horizontal.
2. There has been a major boom of direct investment into the developing countries in the 1990s, but most of it has gone to the more advanced LCDs and to China. Little goes to the least developed countries.³
3. Direct investment stocks have grown significantly faster than trade flows over the last two decades, even though trade barriers have fallen dramatically.
4. High volumes of direct investment are associated with similarities among countries in terms of relative factor endowments and per capita incomes, not differences.
5. Point 4 notwithstanding, that portion of affiliate output which is exported back to the parent country seems to depend on differences in factor endowments between the home and host country.
6. A high volume of outward direct investment is positively related to a country's endowment of skilled labor and insignificantly or negatively related to its physical capital endowment.
7. There is weak evidence that direct investment is primarily motivated by tariff avoidance or measurable transport costs.
8. There is mixed evidence that tax avoidance and/or risk diversification are important motives for direct investment. Some evidence does suggest that political risk discourages inward investment.
9. Infrastructure, skill levels, and a minimum threshold level of per capita income seem to be very important determinants of direct investment.
10. There is evidence that agglomeration effects are important in direct investment. But it is admittedly difficult to distinguish agglomeration effects from firms being drawn to the same (unobserved) site-specific resources.

The pattern that emerges from this evidence is that multinationals are active primarily between the similar, high-income countries and that outward direct investment in particular is associated with skilled-labor abundance. The data do not permit a clear separation between horizontal and vertical investments. But limited data suggest that firms which export a large portion of their output back to the home country are probably vertical investments, and the importance of this type of firm is positively related to differences in relative endowments between home and host countries.

Most attempts to conceptually organize these data begin with a relatively simple and plausible logical argument. If foreign multinational enterprises are identical to domestic firms, they will not find it profitable to enter the domestic market. After all, there are added costs of doing business in another country, including communications and transport costs, higher costs of stationing personnel abroad, barriers due to language, customs, and being outside the local business and government networks. The multinational enterprise must, therefore, arise due to the fact that it possesses some special advantage. This point is found in Hymer's 1960 dissertation (published in Hymer, 1976) and the logic of the argument remains persuasive. It implies that a multinational enterprise brings inherent advantages, such as technology, which potentially constitute an important gain for the host country. There may, however, be offsetting costs such as increased monopoly power resulting in the transfer of rents away from indigenous host-country firms (Hymer, 1976; Kindleberger, 1969, 1984).⁴

A limited but very useful organizing framework for inquiring into the nature of firm-specific advantages was proposed by John Dunning (1977, 1981). Dunning proposed that there are three conditions needed for firms to have a strong incentive to undertake direct foreign investments.

Ownership Advantage: the firm must have a product or a production process such that the firm enjoys some market power advantage in foreign markets.

Location Advantage: the firm must have a reason to want to locate production abroad rather than concentrate it in the home country, especially if there are scale economies at the plant level.

Internalization Advantage: the firm must have a reason to want to exploit its ownership advantage internally, rather than license or sell its product/process to a foreign firm.

An important task of theory is to connect these ideas with the firm and country characteristics in a consistent way. This is something that was undertaken in a number of papers including Markusen (1984) Ethier (1986), Helpman (1984), Horstmann and Markusen (1987a,b, 1992), Brainard (1993a), Ethier and Markusen (1996), and Markusen and Venables (1998). I will refer to this as the "knowledge-capital" model, although I note that this is not a widely used term.

Consider first ownership advantages. We noted that this evidence indicate that multinationals are related to R&D, marketing, scientific ad technical workers, product newness and complexity, and product differentiation. This suggests that multinationals are firms which are intensive in the use of knowledge capital. This is a broad term which includes the human capital of the employees; patents, blueprints, procedures, and other proprietary knowledge, and finally marketing assets such as trademarks, reputations, and brand names.

The crucial question is then why should knowledge capital be associated with multinationals while physical capital is not? I have suggested that the answer lies in two features of knowledge capital. First, the services of knowledge capital can be easily transported to foreign production facilities, at least relative to the services of physical capital. Engineers and managers can visit multiple production facilities with some ease (although stationing them abroad is costly) and communicate with them in a low-cost fashion via telephone, fax, and electronic mail. This property of knowledge capital is important to firms making either horizontal or vertical investments.

The second property of knowledge capital that leads to the association of multinationals with knowledge capital is the fact that knowledge capital often has a joint-input or "public-good" property within the firm. Blueprints, chemical formula, or even reputation capital may be very costly to produce, but once they are they can be supplied at relatively low cost to foreign production facilities without reducing the value or productivity of those assets in existing facilities. The blueprint, for example, can yield a flow of services in multiple locations simultaneous. This property of knowledge capital, which does not characterize physical capital, is particularly important to horizontal multinationals. But it may be quite important to vertical multinationals as well insofar as the "blueprint" indicates exactly how the geographically fragmented activities, components, and products must fit and work together. In the knowledge-capital framework, multinationals are then exports of the services of knowledge-based assets: managerial and engineering services, financial services, reputations and trademarks.

The sources of location advantages are somewhat more ambiguous, primarily because they can differ between horizontal and vertical firms. Consider horizontal firms that produce the same goods and services in each of several locations. Given the existence of plant-level scale economies, there are two principal sources of location advantages in a particular market. The first is the existence of trade costs between that market and the MNEs home country, in the form of transport costs, tariffs and quotas, and more intangible "proximity" advantages. Indeed, if trade costs were truly zero, production would be concentrated in a single location (again, assuming plant-level scale economies) with the other location served by exports. That is, some sort of trade costs seem to be a necessary condition for horizontal multinationals to exist. The second source of location advantage, again following from the existence of plant-level scale economies, is a large market in the potential host country. If that market is very small, it will not pay a firm to establish a local production facility but the firm will instead service that market by exports.

The sources of location advantage for vertical multinationals are somewhat different. Suppose for example (as we will do in the next section) that a MNE exports the services of its knowledge capital and perhaps other intermediate inputs to a foreign production facility for final assembly and shipment back to the MNEs home country. This type of investment is likely to be encouraged by low trade costs rather than by high trade costs. Secondly, the most logical situation in which this type of fragmentation arises is when the stages of production have different factor intensities and the countries have different relative factor endowments. Then for example, skilled-labor-intensive R&D and intermediate goods should be produced in the skilled-labor abundant country and less-skilled-labor-intensive final assembly should be done in a country with low-wage unskilled labor. Fragmentation arises to exploit factor-price differences across countries.

Internalization advantages are the most abstract of the three. The topic quickly gets into fundamental issues such as what is a firm, and why and how agency problems might be better solved within a firm than through an arm's-length arrangement with a licensee or contractor. Basically, it is my view that internalization advantages arise from the same joint-input, public-goods property of knowledge that create ownership advantages. The property of knowledge that makes it easily transferred to foreign locations makes it easily dissipated. Firms transfer knowledge internally in order to maintain the value of assets and prevent asset dissipation. Licensees can easily absorb the knowledge capital and then defect from the firm or ruin the firm's reputation for short-run profit.

The formal models mentioned at the beginning of this sub-section incorporate these ideas into general-equilibrium models that in some cases have clear testable implications. Suppose that we are interested in industry X. The principal assumptions in a formal model begin with the notion that firms must make initial investments in firm-specific, knowledge-based assets (e.g., R&D). More specific assumptions then include the following.

1. Transportability: the services of knowledge-based assets are easily supplied to geographically separate facilities.
2. Factor Intensity: Knowledge capital is skilled-labor intensive relative to final production.
3. Jointness: the services of knowledge-based assets are (at least partially) joint ("public") inputs into geographically separate production facilities.

Properties (1) and (2) creates a motive for the vertical fragmentation of production. Knowledge-based assets and activities, such as R&D, will be located when skilled labor is abundant/cheap, while routine production activities may be located when less skilled labor is abundant/cheap. Property (3) implies firm-level scale economies and creates a motive for horizontal investments which replicate the same products or services in different locations. Multiple plants will be located in potentially many countries to exploit the joint input nature of knowledge capital. Once created, the services of blueprints, formulae, procedures, or trademarks can be costlessly supplied to additional production facilities.

What then is being traded when we observe multinational production? Basically, multinational enterprises in this framework are exporters of the services of firm-specific assets. If these exports were arm's length rather than intra-firm, they would generally be classified as "producer services". They include management, engineering, marketing, and financial services, many of which are based on human capital. They also include the "services" of patents and trademarks which are other knowledge-based assets. Subsidiaries import these services in exchange for repatriated profits, royalties, fees, or output.

The theoretical model just outlined fits well with simple observations. First, it predicts a large volume of direct investment, among large, high-income countries and this is a striking empirical regularity. Second, it suggests that the source of direct investment should be the skilled-labor abundant countries, equally true empirically. Third, it suggests that for investments into developing countries, we should observe a significant portion of the output shipped back to the source country (e.g., the investments are in assemble plants, using cheap unskilled and semi-skilled labor). This is also the case.

More formal econometric treatments are found in Brainard (1993b, 1997), Ekholm (1995, 1997a, 1997b), and Carr, Markusen, and Maskus (1998). All of the results support the general theoretical model outlined above, although the conformance of the theoretical predictions on vertical investments with the data are weaker than those for horizontal investments. The last of these papers (Carr, Markusen, and Maskus) has very strong results in estimating equations derived from a formal general-equilibrium model. Thus I will use this model in the remainder of the study as our basic conceptual framework.

Another conceptual approach that has attracted much attention in the literature could be referred to as the "internalization" or "transactions-cost" approach to the multinational. Some authors might view these as distinct, but frankly I don't see much fundamental difference. Important authors in this tradition would include Casson (1987) and Rugman (1981, 1985). Basically, these approaches revolve around the existence of market failures, which lead firms to exploit and transfer intangible assets internally within the firm rather than through arm's length markets. Specific types of transactions costs often mentioned include asymmetric information, moral hazard, and asset-specific investments with the latter in particular leading to hold-up risk.

It is my view that some of the ideas advanced in this literature are quite compatible with the knowledge-capital approach outlined above. As I suggested, the quasi-public-good aspect of knowledge makes it difficult and risky to transfer in arm's length markets. The value of knowledge capital can be dissipated, as licensees or other contractors learn trade secrets and can "defect" to start rival firms. This type of transactions costs leads the firm to transfer assets internally to owned subsidiaries, even if this is in principle a costly mode of serving a market.

While there is clearly considerable validity to internalization theory, I have a couple of problems with it that I have never been able to resolve in discussions with leading authors. First, there are corresponding difficulties internal to firms just as with independent contractors. Employees can learn the technology and then quit to start rival firms. Most importantly and somewhat related, I do not see that this approach is operational in the sense of generating any testable predictions. I have seen virtually no empirical work trying to fit the theory to data, largely because the theory gives us few predictions to work with. Thus I am going to largely ignore this approach, but at the same time I wish to emphasize again that (a) I am not denying its inherent validity, and (b) I believe that many of its principal views are compatible with the fact that MNEs are trading the knowledge-based assets within the firm, a type of trade associated with considerable problems of moral hazard, asymmetric information, asset specific investments, and hold-up risk (see Ethier 1986, Ethier and Markusen, 1996, and Horstmann and Markusen 1987b, 1996 for formal models).

Finally, there is the monopoly-power view of Hymer (1976) and Kindleberger (1969, 1984), in which firm-specific assets seem to confer monopoly power on an MNE with no offsetting advantages to a host country. I tend to dismiss this view as frankly absurd. MNEs gain market power advantages by providing useful and/or inexpensive products and services. While one can surely find examples of abusive practices right up to overthrowing governments, I have never seen any systematic evidence of their importance. Far stronger evidence probably exists for abusive practices by government-owned corporations in developing countries.

The previous paragraph not withstanding, there is clearly a nexus among multinationals, firm-level scale economies, oligopoly, externalities, and spillovers. The expense of developing firm-specific assets clearly implies scale economies, which are then associated empirically with concentration and oligopolistic market structures. Because of the quasi-public-goods nature of knowledge capital, we also have a "smoking gun" for the existence of various positive spillovers and externalities for host countries.

Thus any discussion of the effects of liberalizing investment is immensely complicated by the possible existence of both positive and negative externalities. It is unquestionably true that any shrewd researcher could create a model in which inward direct investment is unambiguously welfare improving or worsening for the host country.

The theory that I have outlined, along with strongly-supporting empirical evidence, does however put limits on the types of outcomes that we might observe. I will use this model to guide the discussions of policy to follow. But first, I think that it is useful to present a bit more positive theory, outlining the predictions of this approach for the liberalization of investment restrictions on production, trade, and factor markets.

In this section, I will analyze the implications of trade and investment liberalization, and the relationship between the two. The knowledge capital approach outlined above will serve as the basic analytical tool. In particular, we would like some insight into whether or not trade and investment liberalization are substitutes or (in some sense) complements. There is an old and rather deeply held notion, due to Mundell (1957) that trade in goods and factors are substitutes. This was later shown to be a very special property of the Heckscher-Ohlin model which Mundell used (Markusen, 1983). In addition, it is not clear that direct investment relates closely to trading physical factors of production. As noted above, multinationals in our view are exports of knowledge-intensive producer services, not in general exporters of physical capital.

Suppose that we have two countries, source and host (h and f), producing two homogeneous goods, X and Y, from two factors of production, skilled and unskilled labor (S and L). Good Y is produced by a competitive industry with constant returns to scale and is unskilled labor intensive.

Good X is produced with increasing returns to scale at both the firm and the plant level. Firm-level scale economies are due to the existence of knowledge-based assets that have a jointness property across geographically separated production facilities. Good X is skilled labor intensive overall. There is a firm-level fixed cost to entering X production (e.g., R&D), plant-level fixed costs, and a constant marginal cost of final production. There are transport costs (using unskilled labor) for shipping X between markets.

There are three "types" of X-sector firms. National firms are single-plant firms with their headquarters and plant in the same country, and are referred to as type-n firms. Two-plant (horizontal) multinationals are firms which have plants in both countries and headquarters in either h or f. These are referred to as type-m firms. Type-v firms will refer to (vertical) multinational firms which have their headquarters in one country and their plant in the other country. Types n, m and v firms may be headquartered in country h or f, so in a sense there are six firm types: n_h, n_f, m_h, m_f, v_h, and v_f.

Factor-intensity assumptions are crucial to the results that will be derived below. These are guided by what I believe are some empirically relevant assumptions. (1) headquarters activities are more skilled-labor intensive than production plants (including both plant-specific fixed costs and marginal costs). This obviously implies that an "integrated" type-n firm with a headquarters and plant in the same location is more skilled-labor intensive than a plant alone. (2) We assume that a plant alone (no headquarters) is more skilled labor intensive than the composite Y sector. This is much less obvious, but some evidence suggests that this is probably true for developing countries: branch plants of foreign multinationals are more skilled-labor intensive than the economy as a whole.⁵ Assumptions on the skilled-labor intensity of activities are therefore:

[headquarters only] > [integrated X] > [plant only] > [Y]

(3) We assume that two-plant type-m firms are more skilled labor intensive than single-plant type-n or type-v firms. Two-plant firms are assumed to need additional skilled labor in the source country in order to manage the overseas facility, as well as needing some skilled labor in the host country branch-plant as well. Single-plant type-n or type-v firms are assumed to use unskilled labor in shipping costs if they supply the market in which the plant is not located. Assumptions on the skilled-labor intensity of firm types are therefore:

[type-m firms] > [type-v and type-n firms]

Markusen (1997) provides a full formal analysis of this type of model and the types of firms which are supported in equilibrium as a function of country characteristics (relative sizes, relative factor endowments), trade costs, and investment costs. Suppose that country h is large and skilled-labor abundant. Country f is small and skilled-labor scarce. Table 1 outlines the results from Markusen (1997) for country f in a 4x4 tableau of possible regimes, which depend on combinations of trade and investment restrictions.

Let country f be a small, skilled-labor scarce host country. Panel (A) of Table 1 shows the type of X firms active and the trade pattern for country f as a function of the investment and trade regimes. The bottom panel gives production levels in X and factor market effects (a discussion of the latter is temporarily postponed). When both trade and investment are restricted, local national firms may produce X in country f (type-n_f). Production is likely to be low and relatively inefficient in the small, skilled-labor scarce market.

In the top right of the matrix, we have restricted investment and liberal trade. A likely outcome in this situation is that country f firms are force to exit, and all X is supplied by production from country h national firms, n_h. Country f is specialized in good Y.

In the bottom left of the matrix, investment is liberalized and trade is restricted. This is likely to result in production in country f being dominated by branch plants of horizontal multinationals from country h, type-m_h firms. Production may well be higher than when both trade and investment are restricted (top left box), but headquarters services are produced in country h and imported into country f. This may correspond to situations observed in Latin America, where inward investments were permitted but inward barriers to trade were high. The auto industry may provided a good example, with plants of US and European multinationals supplying the local Latin American markets.

The bottom right box of the matrix in Table 1 considers the outcome when both trade and investment are liberalized. In this case, production in country f may be dominated by plants of vertical firms from country h. Headquarters services are produced in country h and exported to country f. Some or indeed most of the output of the plants in country f may be exported back to country h. In my examples found in Markusen (1997), X production is highest in the regime with both trade and investment liberalized.

What we see in these results is that trade and investment liberalization are clearly not substitutes, nor is the combination of trade and investment liberalization in any sense a combination of the two individual liberalizations. X production is the lowest under trade liberalization and restricted investment, higher under trade restricted and investment liberalized, and highest when both are liberalized.

Finally, consider the pattern of trade in the four regimes. Under restricted investment X is imported although imports could be small or zero if trade is also restricted (autarky in X). Imports may also be low or zero under liberal investment but restricted trade, when branch plants of country h firms produce X in country f. In the case of both regimes liberal, the pattern of X sector trade is reversed, with country f exporting X. Liberalization in investment also introduces service trade, the-real-if-unobserved trade in headquarter services (HS). The value of these services is implicitly defined in this simple model as the merchandise account imbalance of trade in X and Y. The results in Markusen (1997) are as follows.

While these results are derived from a very specific model, we noted above that empirical tests of the predictions of this model have stood up very well in empirical testing. At very least therefore, they should lead us to abandon the Mundellian notion that trade and direct investment are substitutes. I gather that the phrase "modal neutrality" has been advanced as a policy recommendation that countries treat trade and direct investment, different modes of acquiring goods, as symmetric. While this may or may not be a valid policy prescription, it is clearly not true that trade liberalization, investment liberalization, or the two together have similar or redundant effects.

Liberalization of direct investment restrictions allow the host country to directly import foreign producer services, which are produced with skilled labor which is locally scarce in the host country. This immediately suggests that inward direct investment may substitute for local skilled labor, lowering its demand, price, and in the long run its quantity. But there are some important subtleties implied by the knowledge-capital approach to the MNE outlined above. In this section, we will examine the implications of the model for the effects of inward investment on factor markets, particularly the demand for skilled and unskilled labor. The implications of the model seem to fit well with some empirical evidence, although there have been only a few studies on this issue to date. A discussion of the evidence itself is postponed until a later section.

Consider again the lower panel of Table 1 along with our earlier assumptions about the skilled-labor intensity of headquarters and plant production in the X sector. X sector demand for skilled labor, and hence its wage in general equilibrium, is related to the output level of X in the local economy, but also to the demand for skilled labor in headquarters services. Entering multinationals might produce more X than local firms, but they may also use "imported" headquarters services that substitute for local equivalents.

Figures 1 presents a competitive "parable" for the model drawn from Markusen (1997). Figure 1 gives the Edgeworth box for host-country f, with skilled labor S on the vertical axis and unskilled labor L on the horizontal axis. Factor allocations to the X sector are graphed from the SW corner and for Y from the NE corner. We will first use Figure 1 to discuss the effects of investment liberalization. Suppose that point A is an initial equilibrium in country f with all X production by type-n_f national firms. This implicitly defines an equilibrium factor-price ratio (w/z)₀, where w is the price of unskilled labor and z is the price of skilled labor.

Now liberalize investment, replacing some or even all of the type-n_f firms with branch plants of type-m_h or type-v_h firms. First of all, there is a "substitution effect", holding the output of Y constant. This is shown as the movement from A to B in Figure 1. The branch-plants are less skilled-labor intensive than the national firms that they displace because the former do not involve factor demands for skilled-labor-intensive headquarters services. The integrated-firm (type-n) isoquant X₀ is more skilled-labor intensive than the branch-plant isoquant X*₀. This substitution effect toward a less skilled-labor-intensive technology would unambiguously increase w/z in country f: skilled labor experiences at least a relative wage decline.

But there will generally be a "scale effect" as well. The new technology must in some sense be more cost efficient and hence production of X should expand relative to the initial equilibrium at A. Production could go to point C in Figure 1 for example, an interesting outcome because relative factor prices are just restored to their initial level at A. Or production might increase even more, to a point like D where there is a fall in w/z relative to A, an increase in the relative wage of skilled labor. If the entering branch plants draw more resources from the Y sector than from the exiting type-n_f firms, then the relative price of skilled labor may rise. Finally, we can complete the parable by noting that we are getting in some sense a "better" technology at points B, C, and D than at point A. Comparing A and C, note that the relative factor prices are the same, but the two isoquants are drawn from different technologies in effect. Thus the real prices of both L and S should be higher at C than at A, and it is possible that the real wage of unskilled labor a D is higher than at A even though the relative wage of unskilled labor has fallen.

Figure 2 shows the approximate actual outcomes for the numerical experiments in Markusen (1997) under the assumptions on country size and relative endowments mentioned earlier. In the initial equilibrium HP (for "high protection": both trade and investment restricted), there is some production of X in country f, and a moderately low w/z. Trade liberalization eliminates the type-n_f firms and all X production, resulting in an equilibrium at TL, with w/z rising (w/z is monotonic in the slope of the ray from the Y origin at the NE corner, since Y is homogeneous of degree one). The investment liberalization regime gives an equilibrium at point IL in Figure 2, with a moderate level of production and the lowest w/z (highest z/w). The FF equilibrium (for "free-free": liberal trade and investment) involves the highest level of X production in country f, since the type-v_h firms active in this equilibrium are shipping output back to country h. w/r lies between its values at HP and IL.

These results are from a very specific model as I continue to emphasize. It indicates a skilled-labor bias to direct investment, but that is consistent with the limited evidence we have as I will also try to show later. By expanding the scale of production, entering multinationals may well expand the local demand for skilled labor more than any demand that is lost by the closer of some local firms. Even though multinationals are exporting skilled-labor intensive producer services to the host country, these services may well be a general-equilibrium complement to local skilled labor.

One other comment is in order before concluding this section. I don't have too much to say about entrepreneurship and ownership. It may well be that branch plants of multinationals do displace some local firms, which may clearly be viewed as a bad thing, other things equal. But entrepreneurship is a type of skilled labor, and it is plausible that more liberal investment rules will indeed expand demand for these types of individuals. Recall that the brain drain refers precisely to the loss of the most talented local skilled labor due to the lack of opportunity at home. The scenarios outline above, while special, suggest that inward investment may increase the demand for local skilled labor.

The issue of ownership per se is not one that neoclassical economics gives us much of a handle on. In the types of models that I and others have developed, free entry gives little meaning to the concept and foreign firms behave no differently to those headquartered and owned domestically. If there are serious implications regarding who owns what, I am afraid that they are beyond my areas of expertise and competence. But clearly these issues have been of great concern, if only in a somewhat emotional sense, in many developing (and a few developed) economies.

The view that multinationals are transferring the services of knowledge-based assets to developing host economies, and that these investments may have a skilled-labor bias immediately suggests the possibility of important dynamic effects and various spillovers and externalities. Let me look at these in turn, considering the theoretical arguments in this section. Extensive empirical evidence has been recently summarized in Blomstrom and Kokko (1998), and will be discussed later in the study.

Extending the basic model to allow skilled labor to be endogenous suggests an interesting and potentially important two-way causality. First, investment liberalization may lead to a rise in the real wage of skilled labor as just noted. But if skilled labor is endogenous, this is turn should lead to skill upgrading in the workforce as workers have the incentive to acquire more skilled and training. But this can in turn make the host country more attractive for further investments, leading to a second-round increase in inward investment. This effect is examined in a numerical general-equilibrium model by Markusen and Venables (1999a). Combined with results alluded to in Markusen (1997) noted above, they find that we most expect to find this effect when a small, skilled-labor-scarce country liberalizes with a larger, skilled-labor abundant country. Inward investment (possibly displacing a few local firms) raises the aggregate demand for skilled labor and leads to a higher ratio of skilled to unskilled labor in a new steady-state equilibrium.

In a slightly different context, Markusen and Venables (1997) note another implication of these results (really a corollary, its the same result): by raising the skilled wage in small, skilled-labor-scarce countries, investment liberalization reduces the likelihood of brain drain, the phenomenon of skilled labor migrating out of the skilled-labor-scarce country.

The concept of linkages is a an old one. The idea is that a new firm (as in a new investment) may have "backward linkages" to local supplier firms, or "forward linkages" in which it sells output to local firms (e.g., the MNE is producing intermediate goods). There are a couple of obvious difficulties. First, almost any conceivable pattern of forward and/or backward linkages might be observed. Theory really doesn't have too much to say about this. Second, the whole concept doesn't have much meaning in traditional general-equilibrium theory. In particular, the input-output structure of an economy which summarizes these linkages is of no particular welfare significance.

Very little has been done on this problem in the new theoretical literature with endogenous multinationals where, due to scale economies and imperfect competition, the concept of linkages may have important welfare implications. Exception are papers by Rodrigues-Clare (1996) and Markusen and Venables (1999b) The latter is motivated by several empirical studies including Hobday (1995). These papers generally do not come to any strong conclusions one way or another. Evidence indicates that multinationals trade more, both imports and exports, than do local firms. One bad scenario is that multinationals could displace local firms and import more intermediates from their home country, creating a negative linkage effect on supplying firms in the host country.

Markusen and Venables (1999b) do note one interesting, positive scenario (again, motivated by some empirical evidence which will be presented later). In this scenario, the multinationals entry into local "downstream" (final) production helps the economy jump over what might be termed a coordination failure. By entering the economy instead of exporting to it, the MNE creates demand for local intermediate inputs that didn't exist before, so none were produced. Local intermediate ("upstream") firms enter, and this in turn provides output of services to potential local downstream firms in the same or in other industries as the entering multinational. This generates a second-round effect of favorable development. Markusen and Venables even provide a cute example where the entry of downstream multinationals generates the seeds of their own destruction, kicking the economy over (via creating the upstream industry) to a new equilibrium where local firms dominate in both the upstream and downstream industries.

Empirical evidence presented later suggests that these linkage effects may be important and positive, but the theory is these effects is not well developed (it is much better developed in new trade theory models with no multinationals, see Fujita, Krugman, and Venables (1998)).

Several sources of externalities or spillovers have been identified in the theoretical and empirical literature. We need to be clear what we mean by these items. Generally, we mean situations in which MNEs cannot capture all of the productivity and efficiency benefits they bring to a host country. Even this is a bit vague. For example, many theoretical articles identify situations in which the entry of MNEs into a host country generates gains to the latter through lower product prices or higher real wages. This is analogous to the classic "gains from trade" proposition in which both parties to voluntary exchanges are better off. The fact that the host country's real income rises is not usually termed a "spillover" or "externality". These terms are usually reserved for situations in which an economic good, service, information, et. cetera is involuntarily transferred from one party to the other: the transferor would like to receive payment but there is no mechanism by which to extract payment from the transferee. This must be kept conceptually distinct from a voluntary transaction in which both parties reap a surplus: gains from trade. Keeping in mind the latter, we must also emphasize that the existence of spillovers and externalities are not necessary conditions for an investment to be beneficial!

(1) Training. It is frequently suggested that multinationals benefit host countries by training employees, who may continue to work for those firms or may lead to join or start local firms. The latter are not necessarily competitors, they often become suppliers to their former employers. When thinking about spillovers and externalities, the question is whether or not the employees are able to capture a surplus from that training, or whether that surplus is extracted through low wages, for example, during a training period.

Several examples have been produced in the theoretical literature in which the MNE cannot extract all rents, and some are shared with the employees (Fosfuri and Motta (1997), Ethier and Markusen, 1996, Markusen, 1998b). The situation is that a multinational introduces a product into host-country production that last two time periods. In the first time period, the manager learns the technology and can defect in the second period to start a rival firm. The MNE faces a "participation constraint" on the part of the employee: over two periods the employee must earn at least his/her opportunity cost. The MNE also faces an "incentive compatibility" constraint: in the second period the employee must be guaranteed a return that prevents defection. While the various papers just mentioned differ in details, several show that there exist plausible situations where the participation constraint is slack in equilibrium: the MNE must share rents on current and future production in order to satisfy the incentive compatibility constraint. But the papers also note that if the necessary rent share becomes too large, the MNE will not make the investment in the first place, preferring to dissipate a part of the potential rents in exports or producing in an alternative location rather than share a large portion of rents. This delicate tradeoff will be important when we come to discuss host-country participation in binding rules on international investment.

Once again, theory does not give us a very complete guide to the situation. However, there are clearly plausible situations as just noted in which the investing MNEs must indeed share rents in order to prevent their newly training employees from defecting to start rival firms.

It should be emphasized that these effects are not always entirely harmful to the MNE. Many cases have been documented in the literature where former employees become suppliers to the MNE, so that the value of the transferred knowledge is partially returned to the MNE in the form of better quality and/or cheaper inputs (Fosfuri and Motta, 1997, provide the theory behind this).

(2) Demonstration. The simplest ideas in the literature are demonstration effects. Local firms and workers can simply watch (or spy on) what the multinational is doing, in terms of its technology, managerial practices, marketing and so forth. Watching is free or rather it is prohibitively expensive for MNEs to prevent anyone from watching (e.g., construction). There is a real economic transaction here, the transfer of knowledge, but the MNE cannot charge for it.

Following up the comment made at the end of the last sub-section, this practice is not always a loss for the MNE. Potentially, the "watchers" could become suppliers to the MNE, and putting up a shroud around the construction site might prove counter productive. That is, one can imagine situations where giving away information, while worse than selling it, is better than not transferring it at all.

(3) Exporting. A common suggestion is that the entry of MNEs into a developing country improves the ability of local firms to export. Note that this is a positive question and should not be confused with a normative reaction that such an ability is surely a good thing.

This suggestion is clearly plausible (and some evidence supports it) if there are some "public good" aspects to a new foreign firm entering a lot of importing and exporting. A simple example would be the construction or improvements to physical facilities, such as seaports or airport, the benefits of which cannot be entirely captured by the investment MNEs. Telecommunications improvements may have similar properties. There may be a public good (jointness and non-exclusion) aspect per se, or these infrastructure facilities may be subject to sharply falling average cost, so that additional users (MNEs) can greatly lower costs to existing or potential local users.

Access to information networks provided by new investing multinationals might be very important, if more intangible. Multinationals can surely provide information and contacts to local firms at much lower cost that the local firms would have to incur in order to develop such information themselves. However, such a statement certainly does not imply that MNEs would necessarily be willing to do so without attempting to extract the potential surplus from the local firms.

(4) Pro-Competitive, and X-Efficiency Effects. Much has been written about the effect of MNE entry on local competitive conditions, including both price competition and "effort" competition or X-efficiency. Once again, theory is not a very helpful guide, since is it surely possible to construct a theoretical model that produces almost any result. The question is, what theoretical model and assumptions are most plausible and consistent with evidence?

As I noted above, early writings of Hymer and Kindleberger seemed to emphasize a market power dimension that cast MNEs in a decidedly unfavorable light. MNEs exercise market power that has unfavorable effects on local competition, may raise local prices in equilibrium and transfer rents back to the rich countries.

Neither Hymer or Kindleberger ever provided much in the way of formal analysis, and modern theory demands that researchers back up these statements with a formal model that generates the asserted outcome as an equilibrium of some game. What Hymer and Kindleberger consistently neglected was the reasons why firms had market power in the first place. In general it must come through offering new products or jobs at favorable prices and wages, or through offering existing products or services at lower prices. While market power can arise through artificial barriers to entry such as licensing or regulation with no offsetting advantages, these phenomenon are more generally associated with domestic (especially government owned ) firms that with foreign MNEs.

Looking at the existing theoretical literature, it is quite easy to construct cases in which multinational entry unambiguously benefits host economies. In several papers, Markusen and Venables show that the smaller, skilled-labor scarce country must gain and that it is possible that the large, skilled-labor abundant country may be the loser if there is indeed a loser. These are free-entry models, and local firms in the host country can only be driven from the market by more efficient firms and lower prices (higher real wages) in equilibrium.

The formal theoretical models identify several channels of competition effects (the informal literature identifies others). First, consumers in the host country gain through lower prices or conversely increased real wages if MNE entry results in more firms. But even if an entering MNE just displaces a local firm, the country gains through a lower price if the MNE produces with lower marginal cost. For this not to occur, it has to be the case that the MNE has the same marginal cost as the local firm, but has an advantage in lower fixed costs.

Secondly, there is the issue of the existence and distribution of possible monopoly rents. In the Markusen-Venables approach, these rents are reduced to zero by entry, so they play no role in welfare. But when there exist barriers to entry, it is conceivable that the local firm earns rents, which are counted as part of the local income stream, and that these rents are transferred to an entering MNE. It is possible to construct situations where the entry of an MNE reduces host country welfare. The MNE might for example, drive out a local rent-earning firm, yet not lower price because its advantage is in fixed costs, not marginal costs. But such an example relies very much on the existence of initial local monopoly rents (some barrier to entry, possibly just limited market size which can support only a single firm) and equal marginal costs.

Table 2 presents a couple of specific examples of an entry game followed by, for example, Cournot competition (Table 2 presents the normal form, with the second-stage payoffs entered into the payoff matrix). In panel (A), the equilibrium is that the multinational and the local firm both enter, and earn positive profits. Standard duopoly assumptions (such as Cournot) assure us that the local price will be lower with two firms in the market than if the local firm was a monopolist (lower left-hand entry). The local firm might lose some rents to the MNE as shown, but in general consumer surplus gains outweigh such a transfer.

Panel (B) shows an outcome in which the MNE is sufficiently advanced, or the local market is sufficiently small, such that the local firm is driven out in equilibrium ((5, 0) is the equilibrium). Theory does not dismiss this possibility. However, this outcome is by no means sufficient to imply that the host-country is worse off. If the MNEs advantage lies in marginal cost, then the new equilibrium monopoly price will lie below the old monopoly price of the local firm, and the gain in consumer surplus may well outweigh any loss of rents to the local firm.

How likely is a bad outcome of the form taken in panel (B) of Table 2? As suggested above, the argument is destroyed by free entry of firms, since there is no rent to transfer and the consumer price must generally be lower (consumer surplus higher) if the MNE(s) drive out the local(s). Note that the pro-gains argument does not rely on the MNE not earning rent, it only relies on the exiting local firm not earning rent. Theory does not allow us to say much more, but I suppose that most economists would suggest that the free-entry assumption is a reasonable one for many industries once economies reach some modest size. More will be said later in this study about situations where there are positive rents associated with entry, and the distribution of these rents is an important policy issue.

The theory of the multinational is both fascinating and frustrating. The frustration, as I have noted several times, comes from the inherent second-best nature of the problem, which in turn implies that all sorts of bad outcomes are possible, and all sorts of weird policies (including autarky) may be second-best optimal in some model. Theory by itself has no way of distinguishing among these possibilities. Therefore, what I have tried to do in the preceding sections, is to move back and forth between the theory and the data, selecting among many possible theoretical frameworks on the basis of what characteristics of a theory are most consistent with those aspects of the data that we can reliably observe. When we have this consistency, we can have some confidence about policy prescriptions based upon the theory.

I have suggested we work with what I call the knowledge-capital approach to the multinational enterprise. This theory states that

1. MNEs are firms which develop skilled-labor-intensive, knowledge-based firm-specific assets
2. The services of these assets can be provided at relatively low cost to geographically separate production facilities
3. These assets have a joint-input characteristic such that they can be supplied to additional production facilities with (appreciably) affecting their productivity in existing facilities
4. This same joint-input characteristic subjects arm's length trade in these assets to problems of moral hazard, asymmetric information, hold-up risk, and asset dissipations.

Although this may seem somewhat vague and general, in fact such a theory has very specific, testable empirical predictions about how direct investment should relate to country characteristics such as their relative sizes, relative endowments, trade costs and so forth. Empirical evidence by Brainard (1997, 1993b), Ekholm (1995, 1997a, 1997b), and Carr, Markusen, and Maskus (1998) strongly supports the model.

I would characterize the implications of this model as "optimistic" for host countries. Multinationals provide knowledge-intensive services which are costly for developing host economies to develop by themselves, and yet in many circumstances may be complements to local skilled labor. The model suggests that entering multinationals should general lower prices and/or raise real wages. Sets of assumptions can be constructed such that an entering multinational might make a host country worse off (capturing rents that had been earned by an inefficient local firm without lowering prices), but one has to work a bit to do so. Furthermore, the theoretical and empirical association of multinationals with knowledge-intensive service transfers to host countries, combined with point (d) above is a "smoking gun" suggesting the possibility of significant spillovers and externalities to host countries above and beyond the normal gains-from-trade argument.

In what follows, I will therefore proceed under the assumption that inward direct investment is generally a good thing for host economies, and disregard most of the perverse possible cases as either empirically irrelevant or at least the exception to the rule.

In the next section, we will look at some more evidence connecting inward direct investment to policy variables in developing countries. In section 8, we will turn explicitly to the question of the developing countries stake in a new multilateral agreement on investment, using the theory and evidence we have developed as a background.

In this section we will review some empirical evidence on inward FDI in order to further aid us in settling on the most relevant approach to the policy discussions to follow.

Consider first the effects of inward investment on the profitability and competitiveness of indigenous domestic firms in the same sector. Note that there may be a conflict of course between the interests of the indigenous firms' equity owners and domestic consumers with respect to these effects as we noted in the theory section above. Domestic firms may be made worse off from multinational entry, while domestic consumers may be made better off by increased competition. Older writings, such as those of Hymer and Kindleberger tended to emphasize the negative effects of multinational entry on domestic firms and the host-country competitive environment, while more recent thinking emphasizes potential increases in competitiveness. (Blomstrom, 1991, Blomstrom and Kokko, 1998, Blomstrom, Kokko, and Zejan, 1994, Chung, Mitchell, and Yeung, 1994). Borensztein, De Gregorio and Lee (1995) conclude in a recent paper that the "crowding out" effect of inward direct investment on local firms is negative. That is, inward direct investments actually raise the investment levels of local firms.

The simplest conceptual model suggests that the effects of inward investment on domestic firms should depend on the horizontal or vertical relationship between the multinational and the domestic firms, and whether or not the investment is largely just replacing goods that were previously imported. The worst-case scenario is that the MNE enters to produce goods in direct competition to domestic suppliers, goods which were not previously exported to the country by the subsidiary's parent. Effects are much less damaging if the new subsidiary's production largely displaces previous imports. Finally, domestic firms can of course benefit if the relationship is vertical with the multinational's subsidiary becoming a customer or supplier to domestic firms in the sector.

The question of the effects of inward investment is thus not enlightened very much by theory, and it becomes largely an empirical matter. Estimation of these effects is a very difficult problem, even if data are available. The few studies that we are aware of come to relatively optimistic conclusions as just noted or find no effects of direct investment on local firms. We know of no study that documents destructive effects of multinational entry on local industry, at least in manufacturing. Aitken and Harrison (1991) and Haddad and Harrison (1993) find essentially no effects of direct investment on local firms.

On the other hand, a number of other recent studies conclude that multinational entry improves the productivity of local firms, and increases the competitiveness of the local business environment. Blomstrom and Kokko (1998, henceforth BK) review evidence of the effects on multinationals on competitiveness and industry structure in developing host countries. Substantial evidence seems to exist, especially from Blomstrom's own work, that multinationals put competitive pressure on local firms, forcing them into more efficient work practices ("X-efficiency") and more modern technology. Findings by Blomstrom, Lipsey, and Zejan (1994), Chung, Mitchell, and Yeung (1994), Kokko (1997), and Kokko, Tansini and Zejan (1996a,b) combined with the less optimistic findings of Aitken, Harrison and Lipsey (1996) and Haddad and Harrison (1993) suggest that the pro-competitive effect is strong the higher the level of development of the host country.⁶ It is possibly the case that in quite backward countries, local firms are rarely competitors for multinationals and hence pro-competitive effects are not operative.

The effects on industry structure, including the number of competing local firms, are difficult to measure. BK note that a central problem is the "the question whether multinational entry and presence explain industry structure, or whether industry structure determines if multinationals will enter or not". An empirical finding of many studies is that there is a positive correlation between foreign entry and concentration in host countries. Multinationals seem to exist in equilibrium in industries with substantial firm-level scale economies, implying concentration in equilibrium especially in small economies. BK's summary of the evidence is that the entry of multinationals probably does tend to increase the long-run concentration in a host-country's industry. However, because of the characteristics of these industries, modest increases in concentration may be consistent with increased productivity and efficiency.

Considerable caution is advised on this question since the studies on this problem are few and sometimes apparently contradictory in their findings. As noted earlier for example, Borensztein, De Gregorio and Lee (1995) find a "crowding-in" phenomenon in a study of 69 countries. Overall levels of increased investment were about 1.5-2.3 times the increase in foreign direct investment inflows. One possible reconciliation of this finding with some of those reported in BK is that inward direct investment may lead to the net exit of some local competitors, but this disinvestment may be outweighed by investments in vertically-linked activities by suppliers and customers.

In any case, the weight of the (meager) evidence is that multinationals improve the productivity (although not necessarily profits) of local firms and increase the competitiveness of local firms, or at least do no harm.

A second important and difficult question relates to the effects of inward direct investment on local skilled labor. Our earlier theory discussion and supporting empirical evidence suggest two conflicting mechanisms. On the one hand, multinationals are exporters of skilled-labor-intensive producer services. Imports of these producer services may substitute for local skilled labor, depress its wage, and reduce long-run incentives to accumulate skills. This scenario seems most likely when the entering multinationals displace local firms. On the other hand, multinationals will generally require some amount of local skilled labor in a broad range of occupational classifications, ranging from managers, to technicians, to accountants and lawyers. If multinationals are not displacing local firms, but rather introducing new products, displacing imports, or producing for export, then their entry may well lead to an increased demand for local skilled labor. The imported "blueprints" by multinational subsidiaries then create a complementary demand for host-country skilled labor.⁷

A somewhat different approach has recently been adopted by Feenstra and Hanson (1996a,b, 1997) who concentrate on vertical direct investments and the sourcing of components and other intermediate inputs. In their model, direct investment fragments the production process, moving the production of less-skilled-labor-intensive intermediates to low wage countries. However, these components are at the same time skilled-labor-intensive relative to local industry as a whole in the host country, thus raising the demand for skilled labor. Thus direct investment can raise the skilled-labor-wage in both countries. The authors find strong empirical support for this for the US and Mexico, and in particular find that US direct investment in Mexico tends to raise the share of skilled labor in wage payments in Mexican industry.

Next, there is the issue of how inward investment affects the physical capital stocks in developing countries. We are not aware of much work in this area, but basic trade theory suggests some conceptual ground work is needed. For example, if world capital markets were highly integrated then inward direct investment really has little effect on local capital markets and factor prices and thus on welfare from this perspective. In a world of factor-price equalization, inflows of capital only change the composition of output toward the capital-intensive sector without affecting factor rewards.

In an imperfect world with barriers to trade in goods and capital, the effects of the entering multinational's investment depends on whether it brings financial capital from abroad or raises it in local markets. In the former case, a simple Heckscher-Ohlin approach suggests that the investment will raise local wages and depress the local return to capital. If the multinational brings only "blueprints" and raises financial capital locally, then Heckscher-Ohlin theory suggests that local capital might benefit and local labor might be harmed. However, to the extent that the multinational is bring new technology, this simple Heckscher-Ohlin story must be modified and it is quite possible that all local factors gain with multinational entry. One experiment in the context of a traditional Heckscher-Ohlin model would be a technical improvement in one sector, possibly accompanied by a capital inflow. In any case, the best empirical evidence we have on this issue is the finding of Borensztein, De Gregorio, and Lee (1995) noted above, that direct investment significantly stimulates domestic investment rather than displaces it.

The theoretical approach of Feenstra and Hanson (1996a,b, 1997), supported by strong empirical results, is appealing for understanding the effects of physical capital accumulation. To recap, their model has a range of intermediate inputs ranked by skill intensity. Capital accumulation in the developing country shifts the range of inputs produced toward that country, so that the high-income developed country loses its most unskilled-labor-intensive production while these same inputs become the most skilled-labor-intensive products in the developing country. Capital accumulation via direct investment or other means thus leads to an increase in the relative wage of skilled labor in the developing country. Physical capital and skilled labor are general-equilibrium complements.

The analysis of spillovers, externalities, and related effects is a murky area, in part because it is difficult to define exactly what is and is not a spillover and secondly because they are notorious difficult to measure. In this section, we will review some of the relevant literature and attempt to draw a few inferences. We will rely extensively on the excellent new survey by Blomstrom and Kokko (BK) (1998).

BK note early in their paper the difficulty with defining spillovers. One approach is to make the term synonymous with externality and thus restricted to situations in which benefits or costs imposed by one economic actor on another are not priced by the market. Benefits that the MNE transmits to local firms and workers but for which the MNE is not able to internalize the full value might thus be called positive spillovers. But of course, there is a "surplus value" (gains from trade) in many situations and the fact that one party does not capture the entire surplus on a transaction does not imply the existence of externalities as the term is normally used. Is it a spillover if multinationals cause local firms to price more competitively and invest in new technology? Obviously, we could get bogged down in a discussion with little payoff. We will therefore follow BK in using the term spillover fairly broadly, and include under this heading a number of effects that not obviously externalities in the traditional sense yet important in various ways to host countries.

BK begin by dividing spillovers into those that affect productivity and those that affect market access. They in turn divide productivity spillovers into those that are allocative in nature and those that are technical. Allocative effects include the entry of multinationals into domestic sectors that are monopolistic and inefficient, thereby improving the performance of domestic firms by creating competitive pressures. Technical effects include making new technology available to domestic firms, and training workers who may later move to or startup domestic firms.⁸

Market access spillovers have to do with the improved ability of local firms to export. It is argued (theoretically) that MNEs have better international networks and tend to be much more active in trade than many domestic firms in developing countries. There are various fixed costs in establishing export sales that deter domestic firms from exporting, and these costs are reduced by the entry of multinationals. The simplest form these might take occurs when local firms become suppliers and subcontractors to the multinational. Indirect effects occur when the local firms learn from contacts with the multinational or hire former employees of the multinational who have gained a better understanding of international business.

BK begin their review of the empirical evidence with a discussion of linkages between multinationals and their local suppliers and customers. The evidence on backward linkages to suppliers is somewhat mixed. Several studies document favorable spillovers in this regard, as multinationals lead to an increase in productivity and quality in local suppliers. Purchases from local suppliers tend to increase over time as the multinational substitutes local inputs for imported ones. But some evidence, notably Aitken and Harrison (1991) points in the opposite direction. They attribute their finding of negative backward linkages in Venezuelan manufacturing to the fact that foreign firms divert demand for domestic inputs into imported inputs, meaning that local suppliers cannot achieve economies of scale. No account is taken of possible increased interaction over time. Similar results are found in Aitken, Harrison, and Lipsey (1996), but more optimistic results (at least for some sectors) are found in Kokko (1997), and Kokko, Tansini and Zejan (1996a,b).

Evidence on forward linkages to customers through local distributors and sales organizations is scarce. Studies that BK do document (including Aitken and Harrison, 1991) suggest that forward linkages may be significantly more important than previously thought. There does appear to be a tendency for the local market to grow in importance over time, and the presence of exporting multinationals has a positive influence on the exports of nearby local firms (Aitken, Hanson, and Harrison, 1997).

Next, BK review the evidence on training of local employees. Evidence strongly supports the hypothesis that multinationals transfer both technical and managerial skills to local employees. In a study of Hong Kong, Chen (1983) found that training of workers was more important than the introduction of new techniques and products. Some studies emphasize managerial training more than production skills, and note that many managers of local firms received training in multinational corporations Katz (1987), Hobday (1995). They suggest that managerial skills are less firm specific than technical skills, so that spillovers from management training may be quite significant. It also seems to be the case that local affiliates perform more R&D than is generally presumed. Finally, Feenstra and Hanson (1996a,b, 1997) find that investments by US multinational firms raise the demand for skilled labor in Mexico, and other studies find that multinationals generally pay higher wages than local firms. While these are not spillovers in any externality sense, they do have long-run implications for the incentives to accumulate skills and education.

Although it is somewhat out of order, we might emphasize again that there is substantial evidence of two-way causality between direct investment and human capital levels. There seem to be definite "threshold" effects with respect to human capital levels in a potential host country, below which there is little inward investment. Higher education and skill levels attract direct investment, which then provides a further stimulus to education and skills accumulation (Borensztein, De Gregorio, and Lee (1995), Blomstrom, Lipsey, and Zejan (1994), Kokko and Blomstrom (1998), Eaton and Tamura (1994), Schneider and Fry (1985), Ekholm (1995)).

Empirical evidence does give rather clear support to the existence of market-access spillovers. Multinationals do much more importing and exporting than local firms which makes one suspect linkages in the first place. Spillovers from the multinational's networks and expertise affect both their affiliates and local firms. Evidence is provided by Aitken, Hanson, and Harrison (1997), Eaton and Tamura (1994), and Lipsey (1995).

Finally, we want to raise a final issue that has not been dealt with to the best of our knowledge. This is the possible existence of fiscal externalities. Strong evidence exists concerning agglomeration economies and the importance of infrastructure in direct investment as we noted earlier. But agglomeration or simply the attraction of foreign firms in general may lower the cost of providing infrastructure and other public intermediate inputs. To the extent that there is partial jointness in public inputs, then additional firms significantly lower the average cost per firm of such inputs. Modest tax rates or initial infrastructure investments that generate inward direct investment may thus later generate positive fiscal impacts in terms of better public inputs and/or lower taxes for local firms. This will be one question discussed in the next section.

Country infrastructure has been identified in empirical studies as a very important determinant of inward direct investment (Hackett and Srinivasan 1996, Wheeler and Mody 1992). Public intermediate goods in ranging from road and telecommunications to public education are surprisingly important determinants of the ability of countries to attract investment. Of course, one could point out that infrastructure is highly correlated with per capita income and other measures of development. But most of the studies we have examined take this into account and perform multiple regression estimation in order to separate out the different contributions of correlated measures of development. Morrison and Schwartz (1996) arrive at similar conclusions looking across US state.

It is probably frustrating for countries to be told that higher levels of investment in infrastructure will help attract investment. That might be rather obvious and, in addition, investments in lots of things might improve growth. And, of course, investment funds are limited. We simply wish to point out the empirical research suggests that the quantitative importance of infrastructure seems to be much greater than we would have guessed, and thus it merits careful attention in development plans.

These items are much more intangible than physical infrastructure and accordingly they are much more difficult to quantify. Most researchers think that well defined property rights, enforceable contracts, and related legal infrastructure is important to attracting direct investment, but no one can make very definite quantitative statements. A small amount of theoretical work suggests that well developed legal institutions are particularly important in encouraging licensing agreements in which (arguably) more local learning will occur than in wholly-owned subsidiaries of foreign multinationals. As with the previous point, it is easy to suggest improvements in this regard and their role is probably fairly obvious. But the role of legal institutions in determining the mode in which a firm enters a host country is perhaps not so obvious and worth considering. A stylized theoretical model might suggest that complete lack of enforceable property rights and contracts might lead a foreign multinational to chose exporting over any form of local production, modest legal infrastructure leads to a choice of a wholly owned subsidiary, and strong legal systems lead to more licensing and other modes with higher levels of local participation. If more local participation leads to higher levels of spillovers and skill accumulation, then investments in legal infrastructure have a higher payoff. Existing evidence that increasing patent protection has a positive impact on trade and inward direct investment includes Maskus and Penubarti (1995) and Seyoum (1996). Gould and Gruben (1996) and Park and Ginarte (1997) provide evidence that intellectual property rights stimulates growth.

Agglomeration effects have been cited in a number of studies as being quite important (Head, Ries, and Swenson, 1995, Wheeler and Mody, 1992). Foreign firms tend to cluster and an existing group of foreign investors significantly raises the probability of more entry. Of course, this phenomenon could be confused with the fact that a group of firms is simply attracted to the same site-specific resources or good infrastructure. But these authors have been careful to control for these effects. One consequence of agglomeration economies is the possibility of multiple equilibria. One low-level outcome in which little or no foreign investment is attracted, and another one in which there is substantial amounts attracted (this is examined in an interesting theoretical article by Bose and Purkayastha, 1994). Public policy then has the difficult role of trying to push the equilibria over to the high-investment outcome. Initial investments in improving infrastructure and attracting a few firms may have very high payoffs as the process becomes self sustaining. This is perhaps a point that should be noted by international agencies in formulation their aid strategies.

A number of studies have emphasized that direct investment is attracted to countries with educated and skilled labor forces (Blomstrom, Lipsey, and Zejan, 1994, Borensztein, De Gregorio, and Lee, 1995, Eaton and Tamura, 1994, Kokko and Blomstrom, 1998). There seems to be a definite threshold, below which inward direct investment does not occur.⁹ There is likely a two-way causality here, with good education levels attracting direct investment and direct investment increasing worker skill levels. As in the case with other infrastructure, educational investments may thus have very large payoffs in sending the country onto a self-sustaining growth path.

Evidence on the effects of local taxation on inward direct investment is mixed. There is of course a reason to suspect that these effects are limited, which is that taxes paid to host-country governments by multinational firms are in large part deductible from taxes due in their home countries. Similarly, tax breaks and other fiscal inducements are in part mitigated by increased tax liabilities at home on the increased profits. Insignificant effects of host-country taxes on inward direct investment have been found by Brainard (1993a,b), Morck and Yeung (1991), and Wheeler and Mody (1992). Negative effects are on the other hand found by Grubert and Mutti (1991, 1996), and in a number of articles in Feldstein, Hines, and Hubbard (1995). Several partial resolutions to the apparent contradictory evidence are contained in these papers by looking at a more disaggregated level. For example, taxes matter more in footloose vertical investments (e.g., electronics assembly for export) than in horizontal investments to serve the local market (e.g., food processing), and matter more on the margin such as choosing which location within Europe than they do for deciding whether or not to invest in Europe at all (Devereux and Griffin, 1998).

While the contradictory empirical evidence does not permit very clear policy recommendations, my sense of the results is that a government facing fiscal constraints might be better advised to invest in good physical, educational, and legal infrastructure than try to minimize taxes.

Governments have often enacted policies to pressure multinational firms to undertake more local production. Measures include minimum local content requirements, export performance requirements, and so forth. Several studies have suggested that these are counter-productive to attracting foreign direct investment, including Kokko and Blomstrom (1998), Hackett and Srinivasan (1996), and Lopez-de-Silanes, Markusen, and Rutherford (1996). Kokko and Blomstrom find that performance requirements seem to discourage transfers of knowledge capital, but have no clear effect on imports of machinery and equipment by multinationals. Lopez-de-Silanes, Markusen, and Rutherford suggest that domestic content provisions in NAFTA will discourage foreign auto firms from producing inside North America and induce some switch to exporting to North America. The beneficiaries will be the North American auto firms.

Moran (1992) argues the opposite. In a study of Mexico, Columbia, and Saudi Arabia, he found that firms often resist moving production to a country even if the cost structure makes it advantageous to do so. He found that content and performance requirements possibly combined with fiscal or infrastructure benefits to foreign firms, can induce entry and higher local output at little cost or need for long-term subsidies. This can make sense if there are significant fixed costs to entering a new market. In the presence of such costs, firms will resist entering even if the marginal cost of producing in the host country is significantly lower. If they can be induced by either carrot or stick measures to bear these entry costs, local production will prove sustainable and beneficial to the host country. This seems to have happened with the Mexican auto industry, for example. The industry was created due to severe import restrictions, but it is now profitable, self-sustaining, and exporting a significant proportion of its output. Perhaps the implication is that, used with considerable caution, there may be a role for performance and content requirements, but only when fixed costs are clearly the barrier to entry and production will be profitable and self-sustaining once these costs are met.

Analyses of regional integration have always been rather inconclusive become of the fact that multiple outcomes are possible. An analysis of direct investment in regional integration is no exception. Norman and Motta (1993) note in a theoretical paper that several outcomes are possible for an integrating region. First, if there was little direct investment into the separate countries before integration, then the larger market provided by integration will attract investment into the region. On the other hand, if many firms had plants in the individual protected markets prior to integration, then integration may lead these firms to consolidate their operations in one country, thus reducing investments in some countries and possibly cutting their overall level of investment. The likelihood that inward investment direct investment would rise following integration is strengthened by the fact that integration should raise aggregate incomes and incomes per capita, and both of these variables have been proved to be strong attractors of inward direct investment.

The fact that all countries are unlikely to attract equal shares of investment combined with the facts that many investment projects are large and the firms may shop around for a location, raises concerns about fiscal (tax) competition for direct investment by developing countries. Multinationals may allow countries to complete for investment projects in terms of low taxes, promises of improved infrastructure, etc. This process will transfer more of available rents to the multinationals. We have seen this happen within the United States in competition between individual states for large investment projects, and certainly the same things happen in Europe (e.g., some countries are unhappy with Irish investment location incentives). International agencies possibly have some role to pay in preventing such tax competition. Little systematic evidence exists as to the effects of fiscal inducements, but limited evidence on inter-state location decisions suggests that they range from second-order importance to firms making location choices (Freidman, Gerlowski and Silberman, 1992) to high importance (Hines, 1996).

Many authors and surveys identify political conditions as being of significant importance for trade and investment flows. Corruption and instability are mentioned, but it should be noted that these two things are quite different. Corruption is a transactions cost, which might be large, but it may be a stable and "reliable" form of business costs. Instability on the other hand, involves substantial, possibly extreme, uncertainly. Some survey evidence that I have seen suggests that this is a far larger deterrent to inward direct investment, because the firm runs the risk of very large losses. Empirical evidence on these issues is weak. Some evidence to support these ideas is found in Schneider and Fry (1985). More recently, Anderson and Marcouiller (1998) and Shang-Jin Wei (1998) have done interesting analyses of trade as related to the insecurity and uncertainty that foreigners feel about a country. They show that insecurity and uncertainly are highly detrimental to trade, which fits well with the "gut" feelings of many researchers, policy makers, and businessmen. It seems likely that if this is true for trade, it surely must be true for direct investments where investing foreigners have a lot more at risk.

A few others references, which are not focussed on direct investment might be noted. One is Bardhan's review article (Bardhan, 1997), and another is a general op ed piece by Tanzi (1995). Bardhan cites extensive empirical evidence regarding the detrimental effects of corruption on growth and development. Mauro (1998) finds that corruption reduces spending on education. Dollar and Svensson (1997), and Dollar and Pritchett (1997) present evidence that domestic political-economic factors strongly influence the success or failure of aid and structural adjustment assistance, and that aid work best when the recipient countries have strong domestic institutions. This evidence relates to the discussion of the ability of governments to commit and/or whether or not they should commit to international rules later in this study.

At this point, we have a good understanding of a theory of the multinational enterprise, and an understanding of a large body of empirical evidence consistent with that theory. The previous section reviewed a great deal of related empirical evidence on factors that are often under the influence of public policy. While I fully understand that some might disagree with aspects of the theory or with my interpretation of evidence, I am going to proceed as if the theory and evidence presented above are "true".

The conclusions to this point are that inward direct investment generally has favorable effects on host countries, both direct effects on prices and real wages, and quite likely indirect effects through the transfer of skills, technical knowledge, spillovers to local firms and so forth. MNEs in turn are attracted to countries with good skilled labor, strong and reliable physical and institutional infrastructure, and so forth.

We will therefore turn to the implications of all these finding for the stake that developing countries have in participating in and signing onto a new international agreement on investment.

It might be useful to remind ourselves at the outset that the objectives of multilateral rules and indeed the whole push toward the liberalization of trade and investment may have limited relationship to the theorist's notions of efficiency and welfare. While all of us surely have different views as to the actual driving forces behind liberalization, I think that perhaps I should reveal my views, which condition what follows.

I don't view the US and other OECD countries as either attempting to maximize their own individual welfare or global welfare in setting international rules. Indeed, I don't think that their objectives have much to do with the economists concept of welfare at all. Primarily, I see the liberalization process as an exchange of market access. Country A will lower its barriers to country B's goods only if country B reciprocates. Often, theory might suggest that country A should lower its barriers regardless of what B does (unilateral liberalization), but we do not see this very often. It is primarily this absence of unilateral self-interested moves that leads me to this view.

Closely related is the concept of "fairness". Rules are designed to promote someone's idea of fairness in the international market place, which is often (fortunately) consistent with efficiency, though not necessarily so. National treatment is a cherished cornerstone of fairness and I suspect that it is indeed generally consistent with economic efficiency. But we should not view the two as always the same. In particular, what is fair for one country might be (and often is) viewed as inherently unfair from another country's perspective.

Finally, we should note that countries are often acting as agents on behalf of domestic firms in international negotiations. The objective is to increase profits for domestic firms, which may or may not be consistent with either the home or host country's welfare. I do not know if the story is true, but it was rumored that the US almost had a trade war with China, because the US government was preoccupied with acting as an agent of Disney Studios in trying to stop the copying of films and videos in China. True or not, the general thrust of such stories is entirely plausible, and developing countries must not be blamed for being suspect of OECD countries' objectives in negotiations.

The question of "what is" often gets mixed up with the question of "what ought to be". Obviously, economists are going to recommend that both source and host countries be more concerned about the efficiency of the allocation process. In addition, I would suggest that on the basic of extensive theory and empirical evidence, that developing host countries consider the following.

(1) Develop a Consensus on the Big Issues. My reading of the evidence is that inward investment is generally a good thing, and I will not summarize the evidence yet again. This may seem trivial, but the first thing countries might want to do is to form a consensus about this most basic of all questions. This consensus can then steer discussions on the details.

(2) Minimize transactions costs, rent seeking, and uncertainty. If inward investment is a good thing, then we want to minimize unnecessary costs imposed on investing firms, unnecessary in the sense that they are almost entirely wasteful: costs to the firm without any benefits to the country. Dealing with investment proposals on and individual, ad hoc basis may have occasional advantages, but such a policy generates high transactions costs, corruption, and uncertainty.

(3) Offset other distortions. There may be distortions connected with investing firms such as monopoly power, but often there are other distortions in the economy which interfere with investments or indeed stimulate investments into the wrong sector (small scale import substitution in industries with significant scale economies). As suggested below, weak institutions can even be a distortion in the sense of a country being unable to commit to stable and reliable policies. First best policies generally remove distortions at the source, but often offsetting policies are needed as second-best options.

(4) Maximize the transfer of public-goods aspects of knowledge capital. If we accept the theoretical view I have laid out, then there will often exist a positive rent that can be bargained over. The MNE may be able to transfer knowledge capital at essentially zero cost, but would like to extract a high price, up to the host-country's reservation price. There may exist a huge gap between zero and the latter price, and it is perfectly legitimate for the host country to want a share of this surplus value. The question is how to do this without creating high transactions costs of the type just alluded to above.

This is the crux of the policy problem that developing countries have to deal with in formulating their position as to new multilateral rules on investment. What I want to do in this section is present some of the basic tradeoffs of this dilemma.

In order to set the stage, consider first Figure 3, which is based on the general "findings" of previous sections. On the vertical axis we have host-country welfare, and on the horizontal axis, total rents generated by an investment project. Let's assume this is a horizontal project, replacing imports from the MNE by local production. The point labeled "export" gives the consumer surplus gain to the host and the profits to the MNE from serving the market by exports. If the MNE invests in the country, the price of the good(s) falls, generating a consumer-surplus gain plus rents. A crucial question is, "who gets the rents"? The possible outcomes in Figure 3 are drawn as a negatively sloped locus of points ("Invest Locus"), with "All Rents to the MNE" at the lower right, and "All Rents to Host" at the upper left. Any point on this locus is Pareto improving for the MNE and the host over the exporting option. It is natural that there is going a tendency of the host economy to want to bargain over the distribution of the rents.

One of the critical issues to investing firms is what is know as "holdup risk" in game theory. The investment may involve a lot of sunk, immobile fixed capital investment. Once the investment is made, the host country can renege on tax or other agreements, and directly or indirectly expropriate rents from the MNE. After investing, the MNE worries that it could end up at a point like H in Figure 3, in which it is actually worse off than with the exporting option. In such a case, the MNE will not make the investment, potential rents are dissipated, and both parties are worse off relative to other but perhaps infeasible outcomes.

The usual way out of this mess in game theory is to try to find a commitment mechanism, a way in which the host country can bind itself not to expropriate ex post. This idea is developed in Figure 4, where I will just refer to the commitment mechanism as posting a bond. H, M, and W are the actors: H for host country, M for multinational, and W for the world authority that holds the bond. Initially, H can post a bond or not with W. If it does not post a bond (top part of tree), M then decides to invest or not invest (export). If M invests, H then plays again, deciding to not change the rules (honor) or to expropriate (defect). I have arbitrarily assigned payoffs, where the first number is the payoff to the host and the second number is the payoff to the MNE. The equilibrium of the stage game, if no bond is posted is that the MNE does not invest, and serves the market by exports, earning a return of 2. If the MNE does invest, it knows it will be expropriated, and its payoff will be 1.

In the bottom section of the game tree in Figure 4, the host country posts a bond with W in the amount 3. The game tree is identical, except that W moves last, returning the bond to the host country if the MNE chooses export, or if it chooses invest and the host plays honor. But if the host plays defect, W transfers the bond to the MNE. In this case, the equilibrium of the game is invest-honor, with the host earning 3 and the MNE 6. This is a Pareto improvement over the outcome in which no bond is posted.

This is a well-known type of result in game theory. A player may wish to bind him or herself to not having certain options. A commitment, which appears to make the agent worse off by sacrificing flexibility, in fact is welfare improving for the agent. If the agent has the option, the agent never gets to exercise that option because the other player makes sure that the game never arrives at that decision node.

Figure 5 puts this game tree in normal form. In panel (A), there is no bond, which we could think of as no international treaty and/no commitment to certain rules, or intellectual property protection (IPP), et. cetera. The equilibrium is for the MNE to choose exporting. In panel (B), the host government posts a bond, signs a treaty, or whatever. This commitment changes the payoffs in the defect-invest outcome. Even though this outcome was not played before, it nevertheless is of crucial importance to the solution to the game. Now invest-honor is the equilibrium, and both players are better off. This illustrated the fundamental value of commitment: if you don't commit to giving up something tempting, you are going to get something even worse that if you do commit. This example suggests that committing to international rules on investment, intellectual property protection, etc. may be strongly in the interests of a developing country.

This discussion lays out the general case for the advantages to a country of committing to rules. There are few additional, more specific points that can be made.

(1) Reputation Effects. A reputation means that other individuals or firms can predict and rely on your future behavior from past actions. Reputations are assets of individuals, firms and countries alike. And like other assets, they require that firms or countries give up something today to build capital for tomorrow. It may well be in the interest of an individual firm or country to make the investment if it is far sighted, and does not necessarily require some special form of commitment. On the other hand, decisions might be made by individual firms which have long-term consequences for the country and not to the firm making the decision.

The situation is illustrated in Figure 6. We assume that a MNE, denoted M1, makes an investment in the country. The local manager can then choose to honor the agreement or defect, or the government may honor the investment or expropriate (defect). The manager/government is denoted H. The payoffs to the manager/government and the MNE respectively are (3,6) and (5,1) as in our previous example. Now a second firm M2 decides to invest or export to the country. If the previous investment has been honored, M2 invests, if not, M2 exports to the country assuming that the manager/government will continue as before. Suppose that this process is repeated for infinitely many periods with discount rate r. If the manager/government always honors, the present values of future payoffs are (3/r, 6/r) and if exporting occurs forever the presents values are (0/r, 2/r).

If the same agent H is making the honor/defect decision each period, that agent will honor each agreement in the sequence unless r is very large (high discounting). The manager/government builds a reputation in their own self interest and there is no need for any special enforcement mechanism.

On the other hand, each successive investment may involve a different local manager. In that case, the present values of future investment projects are a return to the country as a whole, but accrue to different agents. Each agent in succession has an incentive to defect, ignoring the present value of future rents forgone, and imposing an "externality" on the country as a whole through defection. In this case, there is an role for the government to commit to laws or international agreements that prevent individual agents from defecting (whatever defecting might mean in any specific circumstance). Enforcement on individuals builds a reputation for the country as a whole. This constitutes another reason for a government to enter into commitments to binding rules on such issues of intellectual property rights, contract enforcement, and other forms of property rights.

One idea that was mentioned by a number of analysts in the NAFTA debate was that a treaty between the US, Canada, and Mexico had a considerable benefit for Mexico in terms of binding future political leaders in all countries, or at least making it far more difficult for leaders in any country to deviate from the agreement. It was suspected that the Mexican government wanted to bind its own leaders, preventing them from backsliding on liberalization according to the political whims of the day, and similarly they wanted to bind future leaders in the US from altering US policy as the political climate might change in the US. The Mexican leadership (so it was reported) viewed a treaty that bound tariffs at zero as far different from an agreement that (temporarily) set them at 1%.

To make the point somewhat more generally, we might say that a treaty, whether on trade or investment, has some "insurance value". The gains from a treaty for free trade versus an agreement for 1% tariffs has a trivial welfare effect if nothing changes. However, the treaty gives economic agents much more confidence that the trade or investment policy will not be changed in the face of future economic and political fluctuations. This insurance value can lead firms to make large, sunk-cost investments when they would not do so otherwise. A treaty is seen as a much more credible commitment than a renegotiable bargain to set tariffs near or at zero. Again, when agents face the prospect of investments with large sunk costs, they are subject to hold-up risk, and a treaty gives them more confidence that they will not be held up in the future even if they trust the current government.

I suppose it should be mentioned that there is of course the possibility that the current government may be foolish or evil in some way. Thus commitment may backfire, with the current government committing future wise or benevolent governments to continue bad policies. I will comment on this again later in discussing the "option value", more or less the opposite of insurance value, in a later section.

There is a large theoretical literature in public finance about tax competition between jurisdictions for mobile capital. Tax revenue is need to provide pubic goods but high tax rates may drive mobile capital elsewhere. I don't know much in the way of empirical literature on this topic, but I seem to remember reading that location incentives offered to firms within the US are considered of second-order importance by those firms.

For present purposes, the idea is that two developing countries might compete with one another to offer location incentives, which is another type of prisoner's dilemma problem that transfers rents to multinational firms in equilibrium. This is shown in Figure 7, where I present an extremely simple example of two countries deciding on whether or not to offer incentives to a multinational to locate in their country or not. Suppose that if both offer incentives or both refrain from doing so that the probability that the MNE will choose each country is one half. The payoff numbers in Figure 7 are expected values when both firms choose the same strategy (e.g., (3,3) means each country has a payoff of 6 with probability 1/2 and a payoff of 0 with probability of 1/2). If each country maximizes its expected return, the outcome of the game is that both offer incentives to the MNE and the MNE chooses one country. The first number is the (expected) payoff to country 1 and the second number the payoff to country 2. The result of this is that potential rents are transferred to the MNE in equilibrium.

This example emphasizes the value of commitment, in this case a commitment not to use selective investment incentives, as jointly efficient among the developing countries rather than focussing on one country individually. In a sense, the developing countries are entering into a binding international agreement on investment incentives not with the "North", but really among themselves, colluding together to maximize the joint rent transfer from investment. On any given investment project, of course, only one country benefits, but one hopes that in the long run things even out. Yet no economist (much less a mere theorist) can make such a promise.

(4) Rent Seeking and Corruption. Another motive for commitment to international rules relates to strictly internal considerations within a particular country (and relates a bit to the binding-future-governments problem). If all negotiations with foreign firms are on an ad hoc, case-by-case basis, the stage is set for considerable rent seeking and corruption by local officials. In a sense, there are surely many cases where a official could, for example, offer a tax concession to an entering firm in exchange for a "gift" when no such concession is needed to draw the firm to the country. Tax revenue is converted into a direct transfer to a corrupt individual.

I am not sure how to model this although I am sure that it can be done. Basically, when there are not fixed, transparent procedures and policies, the situation is open to negotiations between investing companies and domestic officials. Committing to fixed and open policies and procedures amounts to a restriction that certain strategies may not be played. Eliminating discretion may in some cases be harmful (discussed in the next section), but in many others may lead to investments that would not have occurred otherwise (i.e., bribery does not occur in equilibrium, the firm goes elsewhere), and/or makes it more likely that rents go to the central treasury.

(1) Bargaining over the distribution of rents. It must be pointed out that there are cases where one would clearly not like to commit, since in doing so the agent sacrifices flexibility and possible bargaining power. Let's return to our discussion of rent (surplus value) in Figure 3. In Figures 4 and 5 we are just looking at investment versus export, presumably under the assumption that the MNE captures all of the rent surplus noted in Figure 3 (the host is still better of through the capture of consumer surplus through lower prices, higher wages). Here the situation can get more tricky, and has been analyzed in a couple of papers including Ethier and Markusen (1996), Fosfuri and Motta (1998) and Markusen (1998b). It is possible that the there is a sustainable, self-enforcing arrangement where the MNE invests, but the host country (firm manager or whoever) captures some of the rents. If there is no contract or intellectual property protection enforcement, then the host or local manager may be able to credibly threaten to defect on an agreement and so extract a positive rent share. But of course the worst threat that can be made by the local manager or government must be less severe than expropriation, and must leave the MNE with ex post profits (if the threat is carried out) greater than from exporting.

Suppose for example that there is no contract enforceability or intellectual property protection. Suppose that an investment in a technology lasts two period, perhaps to be follow by a series of newer technologies later. The local manager learns the technology in the first period and can credibly threaten to leave and start a rival firm in the next period. There may exist a credible, self-enforcing arrangement between the MNE and the local manager to share rents over future technologies and products so that the contract will be (voluntarily) honored and the investment occurs. In this situation, a commitment to contract enforceability or intellectual property protection can make the local manager worse off, by removing from him the threat which allows him to extract rents without such enforceability.

This idea is illustrated in Figure 8, where we assume that the MNE has three strategies, exporting, an investment with no rent sharing, or an investment with rent sharing. The payoffs are as in Figures 4 and 5, except now we have added a payoff of (5,4) if the MNE offers to share rents and the local manager honors the agreement. The equilibrium of this game is that the MNE offers a contract to share rents and the manager accepts and honors that agreement.

The lower panel of Figure 8 adds the "bond", a restriction on the host government which is then imposed on the manager to penalize the manager for defecting. As in panel (B) of Figure 5, this lowers the payoff to the manager from defecting. However, in panel (B) of Figure 8, the effect of this is to remove the ability of the manager to effectively threaten the MNE into rent sharing. The MNE can now offer a contract with no rent sharing, and this contract is accepted and honored. The payoff to the local manager is reduced from 5 (panel (A)) to 3 (panel (B)). Note again that, while "defection" does not actually occur in either case, the threat value does determine what outcome does occur in equilibrium.

So Figures 4, 5 and 8 illustrate the dilemma. With reference back to Figure 3, the host government would like to commit sufficiently to attract the MNE investment, but not so much as to surrender all local bargaining power. In Figure 3, the host wants entry so it wants to commit the country not to be able to go to point H, but it would like to stay as far up and to the left on the invest locus as it can subject to the MNE entering.

(2) Inability to discriminate among investment projects. If there was just one investment project under consideration, then no problem says the theorist. The developing host country should just tinker with its commitment level and other policies to induce the MNE to enter, but leave the host with the maximum possible surplus. But there are many investments opportunities over time. If the host country does not commit to certain policies, then we have the various messes alluded to in the previous section. If it does commit to policies such as contract enforcement and intellectual property protection, then it will surely loose out on a larger possible rent share on the more lucrative projects.

Figure 9 makes an analogy to price discrimination. Suppose that there are two investment projects, with project 1 yielding a larger surplus (in the sense of Figure 3) than project 2. If the local government commits to a "low" level of contract enforcement and IPP, then all rents can be extracted from the first project, but the second firm says "forget it". If policies are set at a tougher standard, then the second project occurs, but infra-marginal rents are lost on the first project.

Figure 9 is much like the idea behind price discrimination: the firm (in our case the host country) can do better be selling to different customers at different prices. However, in trying to do so, we open the Pandora's box discussed in the previous section. How to maintain flexibility with the substantial advantages of commitment is an elusive ideal.

(3) Inability to adjust for externalities. We can think of a number of situations in which governments would like to have some flexibility. In a sense, all are examples of what we were just discussing: projects might differ widely in their potential net benefits to a society. I will just mention a couple quite quickly, because the general idea is probably clear. First, projects carry (real or perceived) positive and negative externalities, so a country would like to "personalize" its policies according to the individual investment proposed. From our discussions of spillovers in the form of training and knowledge transfer, linkages and so forth, many projects are much more appealing than others. Inward investments in which the knowledge transfer is chiefly in the form of brand identification and advertising, such as American fast food restaurants, are obvious much less attractive than manufacturing investments requiring more substantial skills and providing outputs than governments might judge (rightly or wrongly) to be of greater value.

(4) Offsetting other distortions. Many incentives that a government may offer, or costs that it may impose (performance requirements) may be justified as a response to other distortions that cannot be removed directly. For example, suppose a MNE has to make a substantial investment in sunk capital costs for a project, and the local government cannot commit to a stable future tax rate. Then an upfront "incentive payment" may be a logical response by the government to offset the distortion caused by its inability to commit on something else.

In the other direction, suppose that a government is going to buy aircraft. Frequently, the government asks for "offsets", in which some parts or assembly for the aircraft are done locally. This is an attempt at rent sharing by the host government. In addition, it makes sense if there are substantial externalities to the host country in the form of technology transfer and training, so that the value to the host is much greater than the cost to the aircraft firm from sourcing locally.

(5) Option value. The opposite of insurance value discussed above is option value. Retaining discretion by not committing retains flexibility to respond to unknown future contingencies, opportunities, or crises.

It is clear that theory can take us only so far. To put it another way, we can probably dream up a theory model to produce any result that we want, justify any policy that we desire. There are pros and cons to surrendering discretion in favor of rules.

In the end, we have a "gut feeling" and make recommendations on the basis of an intuitive process that weighs all of these pros and cons in some informal way. My bottom line is that a commitment to a general set of principles is probably strongly in the interests of developing countries. The elimination of the worst forms of holdup risk and the creation of a sense of confidence and reliability for investors seems likely to far outweigh the various forms of disadvantages we have discussed. Still, some forms of flexibility, if and only if they do not lead to holdup risk and uncertainly, would be desirable for developing countries so that they can bargain for a share of rents and maximize desirable transfers of technology and knowledge capital. In the next section, we will try to discuss these ideas in terms of specifics.

In this section, we will try to draw together the theory and empirical evidence presented in earlier sections in order to make a few comments about some specific areas in which international agreements may be proposed and in which there may be pressure on developing countries to sign on. These are all very broad and complex problems, and I am an expert on none of them. My comments will therefore be strictly limited to lessons suggested by the theory and evidence.

These three items are closely related when it comes to inward foreign direct investment, and they derive from the cornerstone principles of the old GATT. They are basically commitments to treat foreign goods and firms in the same way as domestic goods and firms are treated. Adoption of these principles is a fundamental commitment to give up certain strategies, to bind oneself to not engaging in pre or post-investment discrimination against foreign investors. I see a strong case for developing countries signing on to these principles, and do not see substantial advantages from failing to do so. If a country does not sign on, foreign investors must surely view that country as unreliable and the firm will view itself in a position of substantial hold-up risk, if not from the current political leaders, then from future ones.

This has been an issue of considerable policy importance and controversy. Intellectual property is easily copied (e.g., software, videos) and MNE are concerned about the loss of rents when such copying takes place. The difference of opinion can be appreciated in terms of the theory we have reviewed here. If something is easily copied by local firms, then it clearly must be the case that it can be provided to the host country at essentially zero marginal cost by the MNE (the marginal cost of software is the cost of the CD or disk). The reservation price of the host country may be very high, and thus there is a huge gap between the willingness of the host country to pay and the marginal cost of serving the market by the MNE. It is perfectly understandable that developing countries want to bargain over this surplus, or more to the point fail to protect intellectual property so as to transfer the surplus to domestic consumers by default. Developed country firms and their governments have consistently tried to deny that there is a large surplus which is a legitimate target of bargaining.

My view is that there needs to be a distinction between intellectual property protection and legitimate negotiations over the surplus associated with the joint-input (public-good) property of knowledge capital. I don't know exactly how it would work, but an ideal principle might consist of two parts. First, an agreement by developing countries to respect intellectual property of foreign corporations. Second, a recognition that the marginal cost of providing knowledge-based assets is often almost zero, and hence there is a legitimate bargaining problem of the distribution of the surplus value. Personally, I think that it is perfectly appropriate for a developing country to designate a single buyer to negotiate with Microsoft or pharmaceutical companies in order to extract benefits for the developing country. No doubt the MNE would claim that such restrictions contradict national treatment or right of establishment. But a bilateral monopoly situation is preferable to either the extreme of no IPP (effective monopoly power to the host country, with the possible outcome of no investment at all) or all market power to the foreign MNE.

National treatment of course, could mean that everyone is in a free-for-all, with little or no legal infrastructure such as contract enforcement available to anyone. I think that it is clearly in developing countries' interests to try as hard as possible to develop strong, reliable legal and institutional infrastructure. The limited evidence we have suggests that these features are strong determinants of growth and inward investment. Developing countries might want to bargain hard, but once the bargain is signed, it should be credibly and reliably honored.

The example of incentives competition in Figure 7 illustrates the advantage of countries making a commitment not to compete with one another. But of course the flip side of this is that a country also surrenders bargaining power with respect to specific investments in doing so. The thing that we must keep in mind is that the developed countries, and especially cities, provinces, or states in them regularly use incentives of various kinds in order to attract industry. It is entirely normal business procedures for localities to write special contracts with prospective investors. I don't particularly believe that developing countries should give up this form of flexibility, and in any case it is almost surely a non-starter politically given to common use of selective incentives in the developed country. I cannot see a tough code on locations subsidies, tax holidays, and related incentives as being a viable part of an international agreement.

For years, many developing countries wanted to exempt certain sectors from foreign investment and indeed often these sectors were reserved for government-owned corporations. These sectors were generally services, ranging from utilities and telecommunications to banking and insurance. This is much less of an issue today, with many of these activities being privatized; however, a couple of comments might be in order.

I detect two arguments in favor of preserving these sectors for government monopolies or at least total domestic control (a positive, not a normative statement). First, there is the possibility of substantial hold-up risk in these activities, since the economy can grind to a halt without electricity and telecommunications. However, regulation is not the same as ownership, and regulatory mechanisms can be applied regardless of who owns the phone company. Also, it is not clear to me that hold-up risk is less with domestic or government owners. I lived for a number of years in Canada, where the general public is subjected to horrible abuses by public-sector unions on a regular basis. Most would prefer almost any alternative to Canada Post.

Second, it may be argued that government ownership is needed to pursue non-economic or non-profit maximizing objectives. That is probably wrong because again, it confuses regulation with ownership. The US airline, telecommunications, and broadcasting industries have always had regulations requiring certain non-profit-maximizing behavior, but the US government has never viewed it as necessary to own them. Perhaps more worrying, it that this type of argument is generally an opening for corruption and shear laziness on the part of such government-owned corporations and their workers. Developing countries might be doing themselves a favor by "giving up" the right to exempt some sectors from foreign investment. This is the greatest of all bargaining successes: get a concession from the foreigners by giving up what you would like to unilaterally give up anyway.

Regulatory takings refers to situations in which a change in government policy reduces the value of a firm's or individual's asset. A common example is a change in zoning or land-use law that restricts the uses of land, thereby imposing a capital loss on the land's owner. I gather that there has been considerable discussion as to whether or not a multilateral investment code should contain statements requiring host countries to compensate MNEs for regulatory takings (as defined by the MNEs!). This is clearly a difficult area, in which it is hard to discriminate between a "legitimate" taking and one that could be viewed as subtle expropriation. I think that host countries should probably resist a general clause on regulatory takings, but agree to a principle of national treatment: if national firms are subject to the same laws and changes therein, then takings are non-discriminatory and permitted. I recall hearing about a case in which a multinational was claiming regulatory takings by a host government that was imposing tougher environmental standards. Surely all governments are going to retain the right to make such changes without paying off everyone who is disadvantaged, and as long as such changes are non-discriminatory I cannot see an international codes prohibiting them or requiring compensation. A question of interpretation could of course arise if a multinational, for example a chemical company, is the only firm in the industry and thus the only one affected by the regulatory change.

This is another case where a strong multilateral code may be a non-starter. The developed countries clearly do not compensate domestic firms for a wide variety of regulatory takings, and so how on earth can they demand that developing host countries do so? My suggestion is that the principle of national treatment be extended to cover regulatory takings as both a feasible and fair step.

Competition policy has been a lively area of discussion and debate among the developed countries. To be perfect honest, I have lost track of the status of these discussions and the principal point of disagreement. But recent David Richardson (Richardson, 1998) wrote an interesting general article about competition policy in a general context. He argues, and I quite agree, that competition policy is rarely about competition or even efficiency, it is often about "fairness", much like my earlier discussion about the general objectives of a multilateral code on investment. Richardson makes a convincing case that it is precisely this problem which explains a great many controversies. What is fair to an incumbent is hardly fair to a potential entrant for example. He also makes some persuasive points about the extent to which the notion of fairness is culturally rooted. It can mean equality of opportunity, fairness of process, equality of outcomes, or loyalty in long-term relationships. In an international context, one country might see fairness in terms of one definition which conflicts with the views of its neighbors. Also, he notes that fairness and efficiency often conflict.

All of this of course begs the question of what position developing countries might adopt if called upon by the developed countries to adopt a competition code as part of a multilateral agreement on investment. As I said earlier, I just don't know very much about the current debate. My view, to the extent that I have one, is that a modest competition code might be just a natural extension of national treatment. Foreign investors should be subject to the same codes of conduct as domestic firms. But I also suspect that a comprehensive investment code may be premature and will remain so until a strong consensus develops among the OECD countries themselves. If the OECD countries can ever sort out the mess they have made of the international air transport industry, I will take that as a signal that it might be time to look for a more comprehensive international competition code.

We have reviewed a great deal of theory and empirical evidence in this study, and attempted to use the results to offer some opinions on the stake developing countries have in participating in and adopting a new multilateral code on direct investment. Several times, I have noted that theory is limited in its ability to chart us a clear path forward. Basically, it is possible to cook up a theory model to explain almost anything and justify almost any policy option. By appealing to empirical evidence however, we can greatly narrow the range of sensible theories as candidates for use in policy analysis. I have dubbed a general approach the "knowledge-capital" model of the multinational enterprise, and this approach is both theoretically appealing and consistent with a broad range of informal evidence and formal econometric results.

The range of possible policy recommendations remains broad, even when we stick to a central model. The reason for this lies in the fact that direct investment is inherently a second-best problem, in which substantial potential benefits to host economies are inherently mixed in with scale economies and imperfect competition. It is in the very nature of knowledge-intensive production that MNEs will be found in concentrated industries and may involve a variety of externalities and spillovers.

It is my view of the theoretical and empirical evidence that developing countries should generally view inward direct investment as a beneficial phenomenon. The transfer of knowledge-based assets and producer services inherent in direct investment can bring not only lower prices and higher real wages, but the transfer of technical knowledge, managerial techniques, improved information and access to world markets for domestic firms, linkage effects and possibly pro-competitive gains.

The adoption of such a view is however just a first step in formulating a negotiating position. Many specifics have to be filled in. The most fundamental of these is willingness to sacrifice discretion and adopt multilateral rules and commitments. It is easy to make a strong case for the benefits of commitment. It eliminates hold-up risks and uncertainty for multinational firms, and makes them much more willing to make investments with expensive sunk capital (including training locals). Further, it may eliminate opportunities for domestic corruption and competition with other developing countries. But it has to be noted that there do exist counter arguments, generally revolving around the fact that inward investments may generate substantial potential rents and that policy will influence the distribution of these rents (given that the investment indeed occurs). The joint-input nature of knowledge capital means that the MNEs cost of supplying it to a new market may be orders of magnitude less than the host-country's willingness to pay.

It is my view that developing countries should adopt strong commitments to national treatment, right of establishment, contract enforcement, and the protection of intellectual property. These are essential to attracting good-quality inward investment. Beyond that however, things are much less clear in my mind. Given the substantial surpluses that may (at least occasionally) exist, developing countries may want to retain the flexibility to bargain hard with MNEs over a wide range of issues. These could include both offering benefits (tax holidays) on projects viewed as having substantial spillovers (training), or extracting various concessions (again, training). This is not easy and it invites possible corruption, but I believe that developing countries may be able to retain some bargaining power while building and maintaining reputations for strong institutions, contract enforcement, and honoring international commitments.

Aitken, Brian and Ann Harrison (1991), "Are there Spillovers from Foreign Direct Investment? Evidence from Panel Data for Venezuela", MIT and World Bank Working Paper.

Aitken, Brian, Ann Harrison and Robert E. Lipsey (1996), "Wages and Foreign Ownership: A Comparative Study of Mexico, Venezuela and the United States", Journal of International Economics 40, 345-71.

Aitken, Brian, Gordon Hanson, and Ann Harrison (1997), "Spillovers, Foreign Investment, and Export Behavior", Journal of International Economics 43, 103-32.

Anderson, James E. and Douglas Marcouiller (1998), "Trade, Location, and Security", Boston College and NBER working paper.

Bardhan, Pranab (1997), "Corruption and Development: A Review of the Issues", Journal of Economic Literature 35, 1320-1346.

Blomstrom, Magnus (1991), "Host Country Benefits of Foreign Direct Investment", in D.G. McFetridge, editor, Foreign Investment, Technology and Economic Growth. Toronto: University of Toronto Press.

Blomstrom, Magnus, and Ari Kokko (1998), "Foreign Investment as a Vehicle for International Technology Transfer", in G. Barba Navaretti, P. Dasgupta, K.-G. Maeler, and D. Siniscalco (editors), Creation and Transfer of Knowledge: Institutions and Incentives, Berlin: Springer-Verlag.

Blomstrom, Magnus, Ari Kokko and Mario Zejan (1994), "Host Country Competition and Technology Transfer by Multinationals", Weltwirtschaftliches Archiv 130, 521-533.

Blomstrom, Magnus, Robert E. Lipsey, and Mario Zejan (1994), "What Explains the Growth of Developing Countries", in W. Baumol, R. Nelson and E. Wolff (editors), Convergence of Productivity : Cross-National Studies and Historical Evidence. Oxford: Oxford University Press, 243-259.

Blomstrom, Magnus and Edward Wolff (1994), "Multinational Corporations and Productivity Convergence in Mexico", in W. Baumol, R. Nelson and E. Wolff (editors), Convergence of Productivity : Cross-National Studies and Historical Evidence. Oxford: Oxford University Press, 263-284.

Borensztein, Eduardo, Jose De Gregorio, and Jong-Wha Lee (1995), "How Does Foreign Direct Investment Affect Economic Growth?", working paper.

Bose, Pinaki and Dipankar Purkayastha (1994), "Infrastructural Development, Spillover Effects, and the Strategic Adoption of New Technologies in the LDCs", International Economic Journal 8, 1-14.

Brainard, S. Lael, "A Simple Theory of Multinational Corporations and Trade with a Trade-off between Proximity and Concentration", NBER Working Paper No. 4269, February 1993a.

Brainard, S. Lael, "An Empirical Assessment of the Factor Proportions Explanation of Multinationals Sales", NBER Working Paper No. 4580, December 1993b.

Brainard, S. Lael, "An Empirical Assessment of the Proximity-Concentration Tradeoff between Multinational Sales and Trade", American Economic Review, 1997, 87, 520-544.

Casson, Mark (1987), The Firm and the Market: Studies in Multinational Enterprise and the Scope of the Firm. Oxford: Blackwell and Cambridge: MIT Press.

Carr, David, James R. Markusen and Keith Maskus (1998), "Estimating the Knowledge-Capital Model of the Multinational Enterprise", University of Colorado working paper.

Caves, Richard E. (1982), Multinational Enterprise and Economic Analysis. London: Cambridge University Press.

Chen, E.K.Y. (1983), Multinational Corporations, Technology and Employment. London: Macmillan.

Chung, Wilbur, Will Mitchell and Bernard Yeung (1994), "Foreign Direct Investment and Host country Productivity: The Case of the american Automotive Components Industry", University of Michigan Working Paper.

Devereux, Michael and Rachel Griffith (1998), "Taxes and the Location of Production: Evidence from a Panel of US Multinationals", Journal of Public Economics 68, 335-67.

Dollar, David and Jakob Svensson (1997), "What Explains the Success or Failure of Structural Adjustment Programs?", World Bank working paper.

Dollar, David and Lant Pritchett (1977), "Rethinking Aid", World Bank working paper.

Dunning, John H. (1981), International Production and the Multinational Enterprise. London: George Allen and Unwin.

Dunning, John H. (1977), "Trade, Location of Economic Activity and MNE: A Search for an Eclectic Approach", In Ohlin, B., P. O. Hesselborn, and P.M. Wijkman (eds.), The International Allocation of Economic Activity. London: Macmillan.

Dunning, John H. (1993), The Globalization of Business, London: Routledge.

Eaton, Jonathan and Akiko Tamura (1994), "Bilateralism and Regionalism in Japanese and US Trade and Direct Foreign Investment Patterns", Journal of the Japanese and International Economies 8, 478-510.

Ekhlom, Karolina, Multinational Production and Trade in Technological Knowledge, Lund Economic Studies, number 58, 1995.

Ekholm, Karolina (1997), "Headquarter Services and Revealed Factor Abundance", Review of International Economics.

Ekholm, Karolina (1997b), "Factor Endowments and the Pattern of Affiliate Production by Multinational Enterprises", CREDIT working paper no. 97/19, University of Nottingham.

Ethier, Wilfred J. (1986), "The Multinational Firm", Quarterly Journal of Economics.

Ethier, Wilfred J. and James R. Markusen (1996), "Multinational Firms, Technology Diffusion and Trade", Journal of International Economics 41, 1-28.

Feenstra, Robert C. and Gordon H. Hanson (1997), "Foreign Direct Investment and Relative Wages: Evidence from Mexico's Maquiladoras", Journal of International Economics 42, 371-393.

Feenstra, Robert C. and Gordon H. Hanson (1996a), "Foreign Investment, Outsourcing, and Relative Wages", in R.C. Feenstra, G.M. Grossman, and D.A. Irwin (editors), The Political Economy of Trade Policy: Papers in Honor of Jagdish Bhagwati, MIT Press.

Feenstra, Robert C. and Gordon H. Hanson (1996b), "Globalization, Outsourcing, and Wage Inequality", American Economic Review 86, 240-245.

Feldstein, Martin, James R. Hines Jr. and R. Glenn Hubbard, editors (1995), The Effects of Taxation on Multinational Corporations, Chicago: University of Chicago Press.

Fosfuri, Andrea, Massimo Motta and Thomas Ronde (1997), "Foreign Direct Investments and Spillovers through Workers' Mobility", Universitat Pompeu Fabra working paper.

Freidman, Joseph, Daniel A. Gerlowski and Jonathan Silberman (1992), "What Attracts Foreign Multinational Corporations? Evidence from Branch Plant Location in the United States", Journal of Regional Science 32, 403-418.

Froot, Kenneth A. (1993), Foreign Direct Investment, Chicago: University of Chicago Press.

Fujita, Masahisa, Paul Krugman and Anthony J. Venables (1997), The Spatial Economy: Cities, Regions, and International Trade, manuscript.

Gould, David and William Gruben (1996), "The role of Intellectual Property Rights in Economic Growth", Journal of Development Economics 48, 323-50.

Grubert, Harry and John Mutti (1991), "Taxes, Tariffs, and Transfer Pricing in Multinational Corporate Decision Making", Review of Economics and Statistics, 285-293.

Grubert, Harry and John Mutti (1996), "Do Taxes Influence where U.S. Corporations Invest?", Working Paper.

Hackett, Steven C. and Krisna Srinivasan, " A Transaction Cost Analysis of the Multinational Firm: Evidence from Japan and the United States".

Haddad, M. and Ann Harrison (1993), "Are There Positive Spillovers from Direct Foreign Investment?", Journal of Development Economics 42, 51-74.

Head, Keith, John Ries, and Deborah Swenson (1995), "Agglomeration Benefits and Location Choice: Evidence from Japanese Manufacturing Investments in the United States", Journal of International Economics 38, 223-247.

Helpman, Elhanan, "A Simple Theory of Trade with Multinational Corporations", Journal of Political Economy, 1984, 92, 451-471.

Hines, James R. Jr. (1996), "Altered States: Taxes and the Location of Foreign Direct Investment in America", American Economic Review 86, 1076-1094.

Hobday, Michael (1995), Innovation in East Asia: The Challenge to Japan, Aldershot: London.

Horstmann, Ignatius J. and James R. Markusen (1987a), "Strategic Investments and the Development of Multinationals," International Economic Review 28, 109-121.

Horstmann, Ignatius J. and James R. Markusen (1987b), "Licensing Versus Direct Investment: A Model of Internalization by the Multinational Enterprise," Canadian Journal of Economics 20, 464-481.

Horstmann, Ignatius J. and James R. Markusen (1992), "Endogenous Market Structures in International Trade," Journal of International Economics 32, 109-129.

Horstmann, Ignatius J. and James R. Markusen (1996), "Exploring New Markets: The Multinational Firm's Choice of Direct Investment versus Licensing", International Economic Review, forthcoming.

Hymer, Stephen H. (1976), The International Operations of National Firms: A Study of Direct Foreign Investment. Cambridge: MIT Press.

Katz, J.M. (1987), Technology Creation in Latin American Manufacturing Industries. New York: St. Martin's Press.

Kindleberger, Charles P. (1969), American Business Abroad, New Haven: Yale University Press.

Kindleberger, Charles P. (1984), Multinational Excursions, Cambridge: MIT Press.

Kokko, Ari and Magnus Blomstrom (1995) "Policies to Encourage Inflows of Technology Through Foreign Multinationals", World Development 23, 459-468.

Kokko, Ari (1997), "Technology, Market Characteristics, and Spillovers", Journal of Development Economics.

Kokko, Ari, R. Tansini and M. Zejan (1996a), " Local Technologic Capability and Spillovers from FDI in the Uruguayan Manufacturing Sector", Journal of Development Studies 34, 602-611.

Kokko, Ari, R. Tansini and M. Zejan (1996b), " Trade Regimes and Effects of FDI: Evidence from Uruguay", Stockholm School of Economics working paper.

Lipsey, Robert E., "Trade and Production Networks of US MNC's and Exports by their Asian Affiliates", NBER Working Paper 5255.

Lopez-de-Silanes, Florencio, James R. Markusen and Thomas F. Rutherford (1994), "Compementarity and Increasing Returns in Imported Intermediate Inputs", Journal of Development Economics 45, 101-119.

Low, Patrick and Arvind Subramanian (1996), "Beyond TRIMs: a Case for Multilateral Action on Investment Rules and Competition Policy?, in Will Martin and L. Alan Winters (editors), The Uruguay Round and the Developing Countries, Cambridge: Cambridge University Press.

Markusen, James R. (1983), "Factor Movements and Commodity Trade as Complements," Journal of International Economics 13, (1983), pp. 341-356.

Markusen, James R. (1984), "Multinationals, Multi-Plant Economies, and the Gains from Trade", Journal of International Economics 16, 205-226.

Markusen, James R. (1995), "The Boundaries of Multinational Firms and the Theory of International Trade", Journal of Economic Perspectives 9, 169-189.

Markusen, James R. (1997), "Trade versus Investment Liberalization", NBER working paper 6231.

Markusen, James R. (1998), "Multinational Firms, Location and Trade", The World Economy forthcoming.

Markusen, James R. (1998), "Contracts, Intellectual Property Rights, and Multinational Investment in Developing Countries", NBER working paper 6448.

Markusen, James R., Thomas F. Rutherford, and Linda Hunter (1995), "Trade Liberalization in a Multinational-Dominated Industry", Journal of International Economics 38, 95-117.

Markusen, James R. and Anthony J. Venables (1997), "The Role of Multinational Firms in the Wage-Gap Debate", Review of International Economics 5, 435-451.

Markusen, James R. and Anthony J. Venables (1998), "Multinational Firms and the New Trade Theory", Journal of International Economics forthcoming, NBER Working Paper 5036.

Markusen, James R. and Anthony J. Venables (1999a), "Multinational Firms, Skilled Labor and Real Wages", in Richard Baldwin (editor), Dynamic Issues in Applied Commercial Policy Analysis, Cambridge University Press, forthcoming. NBER working paper 5483, 1996.

Markusen, James R. and Anthony J. Venables (1999b), "Foreign Direct Investment as a Catalyst for Industrial Development", European Economic Review, forthcoming, NBER working paper 6241.

Maskus, Keith and Mohan Penubarti (1995), "How Trade-Related are Intellectual Property Rights?", Journal of International Economics 39, 227-48.

Mauro, Paolo (1998), "Corruption and the Composition of Government Expenditure", Journal of Public Economics 69, 263-279.

Morrison, Catherine J. and amy Ellen Schwartz (1996), "State Infrastructure and Productive Performance", American Economic Review 86, 1095-1111.

Moran, Theodore H. (1992), "Strategic Trade Theory and the use of Performance Requirements to Negotiate with Multinational Corporations in the Third World: Exploring a "New" Political Economy of North-South Relations in Trade and Foreign Investment", The International Trade Journal 7, 45-83.

Morck, Randall and Bernard Yeung, "Why Investors Value Multinationality", Journal of Business, 1991, 64:2, 165-187.

Mundell, Robert (1957), "International Trade and Factor Mobility", American Economic Review 47, 321-335.

Norman, George and Massimo Motta (1994), "Eastern European Economic Integration and Foreign Direct Investment", Journal of Economics and Management Strategy 2, 483-507.

Park, Walter and Juan Carlos Ginarte (1997), "Intellectual Property Rights and Economic Growth", Contemporary Economic Policy 15, 51-61.

Richardson, J. David (1998), ""Competition Policies and Trade in an Asia-Pacific Context", paper prepared for the US International Trade Commission.

Rodriguez-Clare, Andres (1996), "Multinationals, Linkages, and Economic Development", American Economic Review 86, 852-873.

Rugman, Alan M., Inside the Multinationals: The Economics of Internal Markets, London: Croom Helm and New York: Columbia University Press. 1981.

Rugman, Alan M., "Internalization is Still a General Theory of Foreign Direct Investment", Weltwirtschaftliches Archiv, 1985.

Schneider, Friedrich and Bruno S. Fry (1985), "Economic and Political Determinants of Foreign Direct Investment", World Development 13.

Seyoum, B. (1996), "The Impact of Intellectual Property Rights on Foreign Direct Investment", Columbia Journal of World Business 31, 50-59.

Slaughter, Matthew J. (1999), "Multinational Corporations, Outsourcing, and American Relative Wage Divergence", Journal of International Economics, NBER working paper 5253.

Tanzi, Vito (1995), "Corruption, Governmental Activities and Markets", Finance and Development 32, 24-27.

UNCTAD (1993), World Investment Report 1993: Transnational Corporations and Integrated International Production, New York: United Nations.

UNCTAD (1996), World Investment Report 1996: Investment, Trade and International Policy Arragnements, New York: United Nations.

Wei, Shang-Jin (1998), "Why is Corruption so Much More Taxing than Tax? Arbitrariness Kills", Harvard University and NBER working paper.

Wheeler, David and Ashoka Mody (1992), "International Investment Location Decisions", Journal of International Economics 33, 57-76.

1 | For a fairly comprehensive review of the literature on trade and direct investment, see Markusen's 1995 review article. A recent collection of papers with a more macro outlook are found in Froot (editor) (1993). Recent comprehensive treatments include Caves (1996) and Dunning (1993).

2 | Again, I will not provide a detailed survey of where these results may be found, since that would be redundant with Markusen (1995), revised and updated in Markusen (1998a). Many of the key references can be found in those two articles.

3 | Good data is found in the 1996 UNCTAD World Investment Report. The 48 least developed countries receive a much smaller share of investment than their share of world income. China is receiving about 40% of all investment going to developing countries and it alone accounts for most of the "boom" in direct investment to developing countries.

4 | In fact, Hymer (1976) and Kindleberger (1969, 1984) concentrate rather heavily on the market power dimensions of multinationals, without inquiring very deeply as to the efficiency advantages that might go hand-in-hand with market power.

5 | Evidence supporting the assumption that multinational branch plants are more skilled-labor intensive than the overall economy (at least for developing economies) is inferred from Feenstra and Hanson (1996a,b, 1997), and Aiken, Harrison, and Lipsey (1995). Slaughter (1999) gives data on the labor-force composition of US multinationals' home operations versus their affiliates abroad, but no comparable data is available for the overall economy.

6 | Indeed, this point is explicitly made by Aitken, Harrison and Lipsey, who find spillovers within the US but not from foreign multinationals to local firms in Mexico and Venezuela.

7 | Empirical evidence is meager. Aitken, Harrison, and Lipsey (1996) find that multinationals pay higher wages than local firms, but find no effect of direct investment on the wages paid by local firms.

8 | With reference to the previous paragraph, the training of local workers by multinationals may not create any externality in the strict sense of the word. In a world of good information and perfect foresight, the value of the training that the worker receives would be captured or capitalized into the salary paid by the firm. Workers who will receive valuable training will be paid less in equilibrium than identical workers who will not. We have no evidence as to whether or not firms or workers captures these benefits in practice.

9 | To put this another way, wages are an indicator of unit labor cost, which is the wage divided by a productivity index. A recent issue of the Economist (November 4-10, 1995 page 86) presents a few figures. The Philippines, India, Malaysia, and Thailand all have substantially lower manufacturing wages than Mexico and South Korea but, due to productivity differences, the latter two countries have substantially lower unit labor costs.