Foreign Direct Investment to Developing Countries: Technological Externalities and Welfare Gains

Contents

List of Figures

List of Variables

1 Introduction

2 Review of Literature
2.1 Theory about Foreign Direct Investment
2.2 Spillovers from Foreign Direct Investment and Absorptive Capacity
2.2.1 Theoretical Literature
2.2.2 Empirical Evidence

3 The Model
3.1 Model Description
3.1.1 Technology and Market Structure
3.1.2 Fixed Cost Assumptions
3.1.3 Specification for Technology Transfer
3.1.4 Consumption
3.2 Equilibrium Conditions

4 Impact Effects and Partial Equilibrium Analysis
4.1 Factor Price Effects
4.2 A Simple Supply-Side Version of the Model
4.2.1 International Wage Differentials for Unskilled Labor
4.2.2 International Wage Differentials for Skilled Labor
4.2.3 Wage Differentials in a North-South Context
4.3 Trade Costs Effects and the Degree of Accessability of the Host Country

5 Numerical General Equilibrium Analysis
5.1 Calibration and Replication Check
5.2 Model without Spillover Effects (δ = 0)
5.2.1 The Equilibrium Regime
5.2.2 Welfare Gains from FDI Liberalization
5.3 Model with Spillover Effects (δ = 1)
5.3.1 The Role of Skilled Labor in the Absorption of FDI Spillovers
5.3.2 Technological Capacity of Domestic Firms

6 Conclusion

A Appendix A

A.1 Optimal Consumer Behavior
A.2 Optimal Firm Behavior

B Appendix B
B.1 Micro-Consistent Input Data and Calibration Issues
B.2 GAMS (MPS/GE) Code for the Model
B.2.1 Program for Calibration Check
B.2.2 Program used for Analysis in Section 5.2 onward

References

Figures

List of Figures

1 Equilibrium regimes (benchmark case I )

2 International wage differentials for unskilled labor in equilibrium (benchmark case I )

3 International wage differentials for skilled labor in equilibrium (benchmark case I)

4 Number of type-v firms along row 0.5 of Fig. 1

5 Number of type-v firms along column 0.5 of Fig. 1 (different values for Fi)

6 Percentage welfare change in country j due to FDI liberalization (scenario I )

7 Number of type-v firms in equilibrium after FDI liberalization (scenario I )

8 Percentage of price for country j’s unskilled labor due to FDI liberalization (scenario I )

9 Percentage of price for country j’s skilled labor due to FDI liberalization (sce- nario I )

10 Allocations of country j’s factors before and after FDI liberalization (scenario I )

11 Allocations of country i’s factors before and after FDI liberalization (scenario I )

12 Percentage change of country j’s income due to FDI liberalization (scenario I )

13 Percentage change of price for X (in terms of Y ) in country j due to FDI liberalization (scenario I )

14 Change in price for factor technology in country j due to FDI liberalization (scenario I )(in %)

15 Crowding out of type-d firms in country j due to FDI liberalization (scenario I) along row 0.75

16 Number of type-v firms along column 0.5 (different levels for tz )

17 Welfare gains in country j due to FDI liberalization (scenario II : tz = 0.1)

18 Welfare gains in country j due to FDI liberalization (scenario II : tz = 0.2)

19 Welfare gains in country j due to FDI liberalization (scenario II : tz = 0.3)

20 Number of type-v firms after FDI liberalization (tz = 0) (benchmark case II )

21 Technology level in country j after FDI liberalization (tz = 0)(benchmark case II )

22 ΔW for FDI liberalization (tz = 0)

23 ΔW (80%) − ΔW (100%)

24 ΔW (60%) − ΔW (80%)

25 ΔW (40%) − ΔW (60%)

List of Variables

illustration not visible in this excerpt

1 Introduction

The spectacular growth of foreign direct investment (FDI) by multinational enterprises (MNEs) represents a distinctive feature of the present phase of globalization of the world economy. FDI has become the single most important component of private capital flows to developing countries and has emerged as the most important source of international resource transfers to developing countries¹. Many countries have liberalized their FDI regimes and are pursuing domestic policies to maximize the benefits of foreign presence in their economies.

The aim of the work presented here is to understand under what conditions FDI flows to developing countries occur and what host country characteristics determine the extent of benefits associated with FDI. Using a numerical two-country general equilibrium model in which (vertical) FDI arise endogenously I compare welfare situations for the developing country before and after FDI liberalization. I specifically focus on the welfare potential of externalities arising from the activity of MNEs that carry out FDI². These spillover effects stemming from the transfer of foreign technology and skills to local industries are shown to be welfare-improving if the host country, in particular local firms, possess the ability to absorb benefits from FDI. It is thus argued that policies of developing countries should be geared towards enhancing this absorptive capacity.

The importance of absorptive abilities for developing countries has long been an issue of vital research for economists. Empirical studies have consistently identified various deter- minants that seem to play a crucial role. At the same time, theoretical work provided an underpinning for these findings. However, one common drawback that applies to a good deal of this theoretical literature is that the existence of MNEs is taken as granted when analyzing technological externalities from FDI (e.g. Glass and Saggi (2002), Das (1987)). Employ- ing partial equilibrium analyses (e.g. Markusen and Venables (1999)) and/or game-theoretic frameworks (e.g. Fosfuri and Ronde (2001), Wang and Blomstroem (1992)) these approaches tend to neglect what drives MNEs to choose a particular location for foreign production. This suppresses general equilibrium effects that occur once spillovers have materialized which is likely to influence the scale of multinational firms’ operations in the host country. Thus, one motivation for the work presented here is to analyze whether determinants of the absorptive capacity that have been identified by previous work carry through in a general equilibrium framework that explicitly takes into account endogeneity of multinational firms.

The organization of the text is as follows. Section 2 briefly reviews the theory on FDI and provides an overview of theoretical literature and empirical findings on spillovers arising from FDI in developing countries. In section 3 the model is presented and equilibrium conditions are stated. In order to develop some intuition for general equilibrium analysis, section 4 conducts a number of partial equilibrium experiments varying factor prices and trade costs. Numerically solving for the full general equilibrium, I investigate in section 5.2 welfare implications of FDI liberalization for the developing country when there are no spillover effects. Section 5.3 goes on to analyze how welfare gains from FDI liberalization are affected when technological externalities from MNEs are present and under what circumstances spillovers from FDI are absorbed by the host country. Section 6 concludes.

2 Review of Literature

2.1 Theory about Foreign Direct Investment

This section very briefly reviews the theory on FDI and multinational enterprises. I want to give an idea what distinguishes a MNE from a domestic firm and what is being traded when we observe multinational production. Moreover, I very shortly describe the origins and evolution of this particular body of international trade literature explaining which theoretical frameworks have been used to explain the investment decision of a multinational firm.

When a firm operates in several countries it is a multinational firm (MNE), and the investment made is referred to as foreign direct investment (FDI)³. Two types of MNEs are distinguished: the vertical multinational firm that chooses to operate its headquarter in one country and has production facilities in another country and the horizontal case where the MNE operates its headquarter in one country but has production facilities in multiple countries. In contrast, national firms (henceforth also: domestic or local firms) are only located in one country and serve foreign markets by exports.

In comparison to national firms, MNEs are believed to possess some special advantage such as superior technology or lower costs due to scale economies that makes it profitable for them to set up foreign production plants rather than to serve local markets by exports since "after all, there are added costs of doing business in another country, including communication and transport costs, higher costs of stationing personnel abroad, barriers due to language, customs, and being outside the local business and government networks" (Markusen, 1995, p.173). A non-formal but very useful framework for inquiring into the nature of these advantages has been put forward by Dunning (1977) who proposed that three conditions are needed for firms to have a strong incentive to undertake FDI:

1. Ownership advantage: The firm must have a product or production process/technology such that the firm enjoys some market power advantage in foreign markets.
2. Location advantage: The firm must have a reason to want to locate production abroad rather than to concentrate it in the home country.
3. 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.

These fundamental insights imply that developing countries that are exposed to multinational activities can potentially benefit from the advanced technology and knowledge that MNEs bring into the country.

FDI flows over the last decades have been rising more or less steadily⁴ and have shown to partly displace trade in goods. When a MNE chooses to serve a foreign market by setting up a foreign production plant, exports to this country diminish. Trade in goods is thus replaced but what is it substituted by? When we observe multinational activity it is mainly services of firm-specific assets that are being traded across countries. These services "include management, engineering, marketing and financial services, many which are based on human capital. They also include the "services" of patents and trademarks [. . .] " (Markusen, 1995, p.175).

It took some time until these ideas were integrated into international trade models. The field of international trade developed largely as a study of trade in goods. Mundell (1957) wrote an important article in which he noted that trade in goods and factors were substitutes: the same equilibrium in terms of commodity prices, factor prices and welfare can be achieved by trading either goods or factors. Basically, trade flows between countries occurred due to differences in factor endowments (Heckscher-Ohlin model) or differences in technology (Ricardian model) or some combination of both (see e.g. Feenstra (2004)). However, this traditional understanding of trade did not leave room for trade in services and intra-firm trade to have any impact on trade patterns.

The first approaches to foreign direct investment viewed the activities of multinationals as essentially a part of the macroeconomic theory of capital flows (i.e. Kemp (1962), MacDougall (1960), Jones (1971)). In particular, there was no attempt to theoretically differentiate direct investment from portfolio investment. Using workhorse models of international trade theory which assumed constant returns to scale in production and perfectly competitive markets there was still little scope for firms and trade in services to have any significant influence on trade flows.

These early approaches to direct investment began to break down in the 1980’s, as the growing activity of MNEs in an increasingly globalized world economy rendered these models incapable of explaining observed trade patterns. Researchers noted that the directions of investment did not seem to bear any particular relationship to interest rates or other measures of returns to capital (Markusen and Maskus, 2001). A new stream of research emerged: the so-called "new trade theory" in which "trade and gains from trade can arise independently of any pattern of comparative advantage (as traditionally understood) as firms exploit economies of scale and pursue strategies or product differentiation in an imperfectly competitive environment" (Markusen, 1995).

Theoretical studies of the activities of multinational firms draw from quite different sub- areas of economic theory using different methodological tools. In general, the theoretical body of literature on MNE/FDI can be divided into two categories. The first group is mainly con- cerned to motivate and understand the multinational firm’s decision whether to internalize its superior technology within the boundaries of the firm or to license or sell its product/process externally to a foreign firm Markusen (1995). This and related questions of internalization have been addressed in industrial organization models using tools of game and information theory (e.g. Ethier (1986), Horstman and Markusen (1996), Markusen (2001)). Taking the internalization as given, the second line of research incorporates the multinational firm into the microeconomic, general-equilibrium theory of international trade. To achieve this, it was necessary to extend the traditional Arrow-Debreu framework with constant returns to scale and perfect competition to allow for economies of scale in production and imperfectively com- petitive markets. In this class of models MNEs arise endogenously as determined by country characteristics, including relative size and relative endowment differences, trade costs, and investment costs (e.g. Helpman (1984), Markusen (1984)).

The general equilibrium model presented below builds on the model of Markusen and Zhang (1998) which can be subsumed within the second category.

2.2 Spillovers from Foreign Direct Investment and Absorptive Ca- pacity

This section gives a short review of theoretical literature and empirical findings about externalities arising from FDI in developing countries and identifies determinants of the host country’s (firms’) ability to absorb such spillovers.

2.2.1 Theoretical Literature

There are a number of models showing through which channels MNEs can generate technological externalities for domestic firms, most of them also suggesting under what circumstances these externalities are more likely to be absorbed by firms in the host country.

One transmission channel of spillovers from FDI are linkages between different sectors that arise from the input-output structure of the economy. For instance, backward linkages refer to the interaction of firm that manufacture an intermediate good and supply it to final good producers. In a partial equilibrium framework Markusen and Venables (1999) address the question how the entry of a multinational firm affects the domestic industry of a developing country. They show that there is a pro-competitive effect and a linkage effect from multinational activity. The linkage effect back to local suppliers of intermediate inputs creates complementarities which reduces cost for domestic final-good producers. In their model, the expansion of local production due to backward linkages cause the economy to develop and thus MNEs may work as a catalyst for industrial development. In a static general equilibrium setup Rodriguez-Clares (1996) also focusses on backward linkages as a channel of transmission of externalities from FDI. He shows that the linkage effect of multinational on the underdeveloped host country is more likely to be positive when the final good that multinationals produce, uses locally produced intermediate goods intensively. However, if the interaction between MNEs and the local intermediate industry is too small the MNEs could even hurt the developing economy.

Other important transmission channels work through imitation and learning by local firms in the host country. These type of channels thus argue that available human capital resources in the host country plays a crucial role in determining the absorptive capacity of developing countries that are exposed to FDI. In a dynamic oligopolistic price-leadership model Das (1987) examines the optimal behavior of an affiliate of a MNE in a host country when learning about production techniques occurs by local firms that operate in the same sector. He shows that the rate of increase in the efficiency of local firms is positively related to the level of activity of the MNE’s affiliate and that this learning process can constitute a major gain for a technologically backward country. However, he argues that local firms in the host country do not necessarily benefit from the better technology used by the MNE’s affiliate, while the host country as a whole always does. Fosfuri and Ronde (2001) build a dynamic game theoretical model where a MNE trains a local worker to run its subsidiary in the host country. Later, the MNE and a local firm compete for the services of the trained worker. The overall effect on the host country’s welfare is positive: the MNE manages to keep the worker only if it offers higher wages than the local firm giving rise to pecuniary spillovers from MNE. Alternatively, if the worker hires at the local firm technological spillovers occur due to the fact that the worker brings knowledge to his employer. The authors also show that a low level of absorptive capacity by the local firm, which might be due to technological backwardness, reduces the potential for spillovers generated by FDI. Also in a game theoretic context, Wang and Blomstroem (1992) develop a model in which international technology transfer through FDI emerges in equilibrium as a result of strategic interaction between foreign subsidies of MNEs and host country firms. The analysis points to the importance of learning efforts of local firms in increasing the rate at which MNEs transfer technology. At a macro level, Keller (1996) analyzes the circumstances under which a developing country can reap technological spillovers from more advanced countries triggering sustained gains in a country’s growth rate. He argues, however, in order for the developing country to adopt technologies that have been invented abroad more skilled labor must be built up to enhance the country’s absorptive capacity.

The theoretical literature identifies another determinant of the ability of local firms in the host country to absorb spillovers from MNEs. Two opposing positions can be found. In a dynamic general equilibrium setting Findlay (1978) emphasizes the importance of relative backwardness in determining rates of technology transfer between MNEs and domestic firms in the steady state. Relative backwardness refers to the distance between two economies in terms of technological advancement. He argues that the greater this distance, the greater is the number of available opportunities in the less advanced economy that have not been exploited yet, the greater the pressure for change and therefore the more rapidly new tech- nology is adopted from the MNE. In his model, the technology transfer will be more rapid if the MNE establishes (downstream and upstream) networks to local firms. Glass and Saggi (2002) also see a key role for technological distance between the host and home country but a quite different one to Findlay. Focusing on labor mobility as a channel of technology transfer they construct an oligopoly model in which a multinational firm has a superior knowledge compared to local firms. Workers employed by the multinational acquire this superior knowl- edge. The authors that the technological distance, or technology gap, signals something to the MNE about the absorptive capacity of the host country. The bigger it is, the less likely the host country is to have the human capital, physical infrastructure and distribution networks to support FDI and therefore the lower is the potential for productivity spillovers.

2.2.2 Empirical Evidence

Empirical studies on FDI spillovers to developing countries can be divided into two groups with respect to the type of data and/or the estimation techniques they employ. The first group comprises earlier studies which generally use static cross-sectional data. Blomstroem and Persson (1983) use industry level cross sectional data from the Mexican 1970 Census of Manufactures. Focussing on intra-industry spillovers using OLS regression anal- ysis they find that, on average, there are productivity spillovers from foreign-owned plants to domestically-owned ones. Using the same data set as Blomstroem and Persson (1983), Kokko (1994) investigates how spillovers observed in static cross-section analyses are related to various proxies for the complexity of MNEs’ technology and the technology gap between locally-owned host country firms and MNE affiliates. He finds that industries with large technology gaps and relatively low skilled labor experience lower spillovers than other in- dustries. Kokko argues that these industries are characterized by little interaction between multinationals and domestic firms, hence, there is little scope for spillovers. Similarly, Kokko and Zejan (1996) hypothesize, based on plant-level data from the Uruguayan manufacturing sector in 1988, that domestic firms can only benefit from spillovers if the technology gap between the multinational and the domestic firm is not too wide so that firms can absorb the knowledge available from the MNE. Thus domestic firms having a very low technological capacity and low employment of skilled workers may be unable to learn from multination- als and therefore only small or no spillovers effects occur. Based on the data set used by Blomstroem and Persson (1983), Kokko (1996) also argues along these lines, finding that the absorption of spillovers depend on the complexity of technology transferred by multinationals, and the technology gap between domestic firms and MNEs and finds no evidence for spillovers in industries where multinationals use complex technologies. By contrast, Sjoeholm (1999) finds for Indonesian manufacturing data of 1991 that productivity spillovers from foreign to domestic firms are larger the larger the technology gap/the lower the technological capacity of local firms. Using the same data set, Blomstroem and Sjoeholm (1999) find that labor productivity is higher in Indonesian manufacturing industries where foreign presence is high. Balasubramanyam (1998) concludes that FDI can be a powerful instrument of development, but only in the presence of a threshold of human capital and well developed infrastructure. He also finds that the amount of technologies imported by MNEs varies systematically with host country characteristics. These imports seem to be larger and more technology intensive in countries and industries where the educational level of the local labor force is higher and where local competition is tougher. Blomstroem and Zejan (1994) find that the amount of spillovers to local firms seems to be influenced by MNE affiliates’ levels of technology or tech- nology imports to the host country. At a macro level, Borensztein and Lee (1998) analyze the effects of FDI flows from industrial countries to 69 developing countries from 1970 to 1989. They find that FDI is an important channel through which advanced technology is transferred to developing countries. There is a strong positive interaction between FDI and the level of educational attainment (their proxy for human capital). However, this holds only if the host country has a minimum threshold stock of skilled labor.

The second group of empirical work on FDI spillover comprises more recent studies that use panel data methods. In general, they tend to be more pessimistic with respect to the presence of spillovers from multinational activity to domestic firms. Aitken and Harrison (1999) use plant-level data for Venezuelan manufacturing between 1976 and 1989 to test the impact of foreign presence on total factor productivity growth. They conclude that domestic firms exhibited higher productivity in sectors with larger foreign share, but argue that it may be wrong to conclude that spillovers have taken place if MNE affiliates systematically locate in more productive sectors. Using panel data on Moroccan firms in the manufacturing sector from 1985-1989, Haddad and Harrison (1993) test for the presence of any externalities from foreign-owned firms to local firm and find mixed evidence on FDI spillovers. Sectors with high levels of foreign investment have a lower dispersion of productivity levels across firms, however, there is no significant relationship in the sample between higher productivity levels of domestic firms and greater foreign presence in the sector. In addition, they hypothesize that if the productivity gap is too big, there are no significant transfers of technology. Focusing on the backward linkages between multinational firms and local intermediate goods suppliers Alfaro and Rodriguez-Clares (2003) explore the channels through which positive externalities may be materializing. Using panel data from 1997 to 2000 for several Latin American countries they compare multinationals’ linkages potential to that of domestic firms. For Brazil, Chile and Venezuela they find that multinationals’ linkage potential is significantly higher whereas for Mexico they cannot reject the hypothesis that foreign and local firms have similar potential. In summary, there seems to be strong evidence pointing to the potential for significant spillover benefits from FDI, but also mixed evidence indicating that spillovers do not mate-rialize automatically. Whether these potential spillovers will be realized or not depends on the ability of domestic firms to absorb foreign knowledge and skills. The above review of theoretical literature and empirical evidence suggests that

1. The extent of spillovers from FDI positively depends on the technology level of MNEs’ affiliates and technology imports to the host country.
2. Skilled labor supply in the host country tends to positively affect the ability of domestic firms to absorb knowledge from multinational firms.
3. There is mixed evidence on how productivity gaps influence the absorptive capacity of local firms. It has been hypothesized that the lower the technological capacity of domestic firms/the bigger the productivity gap, the poorer is the ability to benefit from spillovers. On the other hand, increasing productivity gaps were found to exert a positive influence on the absorption of spillovers.

3 The Model

My model builds on the general equilibrium model of Markusen and Zhang (1998) but differs in that it allows for spillover effects that arise from multinational activity⁵. The original model has been modified with respect to the following elements of which I want to give a first flavor here:

1. Externalities from multinational activity need to be absorbed somehow by local firms in the developing country. To achieve this, I introduce an additional production sector in the developing country which manufactures the same homogenous final good as multinational firms.
2. I introduce a spillover mechanism which increases the available stock of technology in the developing country depending on the level of multinational activity in equilibrium. To this end, I need to introduce an additional production factor which represents this technological change.

Two further modifications are rather technical: First, to simplify exposition, I set transporta- tion costs for the final good to zero. Second, I introduce efficiency parameters for production in the competitive sectors of both countries which will prove to be helpful for partial equilib- rium analysis.

3.1 Model Description

The model has two countries indexed by k = i, j. The two countries are assumed to be identical with respect to production technologies and utility functions of the representative consumer. There are three factors of production: a technology factor, denoted by T , unskilled labor, denoted by L⁶, and skilled labor, denoted by S. Factors are mobile between production sectors but are internationally immobile. Let Sk and Lk denote country k’s endowment of skilled and unskilled labor, respectively. Country k’s endowment of technology is denoted by Tk.

In the numerical analysis, I will vary factor endowments of countries tracing out cases where either country i or j is relatively poorer endowed with respect to some factors. However, in general, I will refer to country i as the developed "North", which is assumed to be relatively skilled-labor abundant and unskilled-labor scarce whereas country j, the underdeveloped "South", is characterized by relative skilled-labor scarcity and unskilled-labor abundance.

3.1.1 Technology and Market Structure

In both countries a homogenous good Y is produced under perfect competition with a constant returns to scale technology using L and S

illustration not visible in this excerpt

where θk is a country-specific efficiency parameter and Sky and Lky denote skilled and unskilled labor used in the Y -sector in country k, respectively. Markets for Y in country i and j are not segmented and there are no transportation costs for Y . Hence, free trade implies that prices for Y in both countries are identical. Y will be used as the numeraire good throughout this text.

In both countries a homogenous good X is produced with increasing returns to scale under imperfect competition. Markusen and Zhang assume an oligopolistic market structure for X in both countries assuming that firms get engaged in Cournot competition⁷. There is free entry/exit to the X markets in both countries. Markets for X are assumed to be segmented, that is firms operating in each market face a different demand function, so X in country i and j is priced independently. X is produced in two stages. In the first stage an intermediate Z is produced with S alone. To produce one unit of Z, cz units of skilled labor from country k are required. cz is assumed to be constant. In the second stage, X is assembled using the intermediate input Z, unskilled labor L, technology T and S. Each unit of X requires one unit of Z. Exploiting the dual relationship between production and cost functions (see e.g. (Varian, 1992)), the technology of the X sector is represented using cost functions.

There are two types of firms that manufacture X: multinational firms and domestic firms. Domestic firms (henceforth: type-d firms) exist in both countries and assemble Xk in country k using the intermediate product Zk, and factor T and L from country k. Type-d firms in country i and j are assumed to be identical with respect to the production technology they use. Output is divided between supply to the local market and exports. The skilled labor requirement for a type-d firm in country k is given by

illustration not visible in this excerpt

where F is the fixed skilled-labor input requirement of type-d firms in both countries and Zd k type-d firm’s demand for the intermediate input, all measured in units of skilled labor S. Let Xdkl, l = i,j, denote sales in country l by a type-d firm producing in country k. The unskilled labor requirement for a type-d firm in country k is

illustration not visible in this excerpt

where G is the unskilled labor fixed input requirement for a X production plant, cx unskilledlabor variable input requirement⁸, all measured in units of unskilled labor. Let ct denote costs per unit of X incurred in terms of T , so type-d firms in country k require

illustration not visible in this excerpt

units of technology for production. Let wk and zk denote the prices of skilled labor and unskilled labor in country k, respectively, and ptk the price for technology in country k.

Then from (2), (3) and (4) and exploiting the one-to-one relationship between X_k and Z_k,

[illustration not visible in this excerpt] it follows that the cost function for a type-d firm in country k is given

by

illustration not visible in this excerpt

Multinationals firms (henceforth: type-v firms) are vertical⁹ and geographically fragment their production process. A skilled-labor intensive intermediate input Zi is produced in the developed "North" and exported to a foreign production plant in the less developed "South". There the final good X is assembled using country j’s unskilled labor together with the imported intermediate input Zi. Transporting one unit of the intermediate input from i to j incurs trade costs of tz units of unskilled labor of country i. The output of type-v firms is locally sold in country j and/or can be shipped back to country i. As opposed to type-d firms, type-v firms draw skilled labor from both countries

illustration not visible in this excerpt

where Fi is the fixed skilled-labor input requirement in country i, Fj the fixed skilled-labor input requirement in country j and Zv type-v firm’s demand for the intermediate input, all measured in units of skilled labor S, and cz defined as before. Let Xv i andXvj denotesales of a type-v firm in country i and j, respectively. The input requirement for unskilled labor in country i and j, respectively, is given by

illustration not visible in this excerpt

where G and cx are as defined before for type-d firms. This means that transportation of the intermediate input Z from country i to j requires unskilled labor from country i. Type-v draw the technology only from country i, so the input requirement for this factor is given by

illustration not visible in this excerpt

Putting together (6), (7), (8), (9) and (10) and using again the one-to-one relationship between X and Z, [illustration not visible in this excerpt], gives the cost function for a type-v firm

illustration not visible in this excerpt

3.1.2 Fixed Cost Assumptions

All fixed input requirements are assumed to be strictly positive, so[illustration not visible in this excerpt] [illustration not visible in this excerpt] and [illustration not visible in this excerpt] Additionally, Markusen and Zhang assume the following with respect to the fixed skilled-labor input requirements:

illustration not visible in this excerpt

From (12) it follows that Fj > 0. Type-v firms have fixed skilled-labor requirements in country j, Fj , but the total fixed skilled-labor requirement, Fi + Fj , is higher than for type-d firms. This is motivated by the fact that doing business abroad involves higher costs. Due to the fragmentation of the production process by vertical multinationals, the skilled-labor requirement for a type-v firm in country i, Fi, is lower than that for domestic type-d firms, F.

3.1.3 Specification for Technology Transfer

In section 2.2 it has been argued that the existence of multinational firms implies potential transfers of superior technology and skills via different channels to local firms¹⁰. However, when incorporating spillovers into the model, I do not explicitly model how spillovers are transmitted. Instead, I simply assume that —regardless of the channel of transmission— there is a positive relationship between multinational activity and the technology endowment of country j.

Multinational activity can be measured by either output of type-v firms in country j or by the amounts of inputs that are used in production. Section 2.2 presented empirical evidence indicating that the extent of spillovers from FDI positively depends on the quantity of superior knowledge and skills that are imported by multinational firms’ affiliates in the host country. In the model, there are two candidates that can be thought of as potentially embodying such superior production techniques of a multinational firm: the technology input factor, Ti, and the intermediate input used by multinationals, Zv . I argue that it is more plausible to conceive Zv — which is produced using skilled labor of the developed country only — as a vehicle for transfer of superior knowledge. Knowledge of advanced production techniques is to a good deal embodied in highly qualified employees, e.g. engineers and managers. Availability of pure technologies in a MNE affiliate’s production plant in the developing country j, for instance blueprints for certain products or production processes — represented by T —, are of limited use if complementary human capital needed to adopt these technologies is not available. It is exactly these human-capital based technologies that are embodied in the intermediate product Z which are transferred to the developing country by making the intermediate product available at a multinational firm’s foreign production site in country j. I therefore assume that the technology endowment of country j is positively related to Zv , the amount of Z used by type-v firms for final good production in country j and choose the following specification for the spillover mechanism

illustration not visible in this excerpt

The first term on the right hand side of the equation above is not a "spillover term". The parameter [illustration not visible in this excerpt] simply (partly) determines the level of technology endowment in country j in relation to the technology stock in the developed country i¹¹. The right hand term of (13) is the actual spillover specification. To assess the full general equilibrium impact of spillovers arising from FDI it will be important to refer to a benchmark case where this mechanism is deactivated, δ = 0, whereas δ = 1 indicates the case where spillovers are active¹².

3.1.4 Consumption

Tastes across countries are assumed to be homogenous. Let Ndk denotethenumberofdomestic firms in country k and let Nv indicate the number of type-v firms. Then

illustration not visible in this excerpt

denotes aggregated sales of X in country k. The utility of the representative consumer in country k is represented by a Cobb Douglas function

illustration not visible in this excerpt

where [illustration not visible in this excerpt] 1.

3.2 Equilibrium Conditions

We want to determine prices and quantities which maximize firms’ profits and consumers’ utility given our exogenous parameters including preferences, technologies and factor endowments as described above. To derive the system of equilibrium conditions, we solve the optimization problem for each agent (representative consumers in country i and country j, type-d firms based in country i and j and type-v firms).

Utility maximizing behavior for the representative consumer in country k; k = i, j yields

the demand functions for X and Y

illustration not visible in this excerpt

illustration not visible in this excerpt

(see Appendix A.1), respectively, where pk denotes the price of X in country/market k and Mk is the income of country k¹³. Due to free entry and exit in the X sector in each country, there are no profits made in equilibrium. Mk is thus given by

illustration not visible in this excerpt

Optimal factor demand in the competitive Y sector in country k requires that factors are paid their marginal product, so skilled and unskilled labor demand are given by

illustration not visible in this excerpt

illustration not visible in this excerpt

respectively (see Appendix A.2).

The X sector in each country is characterized by Cournot competition and each firm (type-d firms in both countries and type-v firms) maximize revenues minus total costs which is

pk (Xk) X −C˜ (21)

illustration not visible in this excerpt

where [illustration not visible in this excerpt]standsfor kl whensolvingthemaximizationproblem of a domestic firm that is based in country k and selling in country l and for Xvk inthecase of a multinational firm that is selling in country k. Similarly,[illustration not visible in this excerpt]isthe associated cost function of the respective firm as given in (5) and (11)¹⁴. Optimal price setting then requires to equate marginal revenue and marginal costs which is ∂Rf

illustration not visible in this excerpt

where [illustration not visible in this excerpt] denotesrevenuesforatype-ffirm,f = d, v, selling in country/market k. For each firm "type", namely type-d firms based in country i and j and type-v firms, the right hand side of ([22] ) is easily obtained by differentiating the right hand side of ([5] ) and ([11] ) with

respect to X. Marginal revenue is given by

illustration not visible in this excerpt

Let [illustration not visible in this excerpt] denote the Marshallian price elasticity of demand for X. In the case of pk ∂Xk Cobb-Douglas utility homogenous of degree one as in (15), η is equal to −1, which is easily seen by taking the derivative of the demand function for X given in (16) with respect to p_k

illustration not visible in this excerpt

From the definition of [illustration not visible in this excerpt], it follows that a one-unit increase in the market supply of X by a single firm increases the total market supply by one-unit, that is [illustration not visible in this excerpt] Let m denote the proportional markup over marginal costs

illustration not visible in this excerpt

When maximizing profits of domestic firms m = {mdkl;mk} is to be read as md kl whichdenotes the proportional markup for type-d firms based in country k selling in market/country l whereas in the case of a multinational firm it stands for mvk whichdenotestheproportional markup for a type-v firm in market/country k. In our model, the optimal markup formulae for a firm is hence given by its market share

illustration not visible in this excerpt

Writing out the left and right hand side of (22) for each "type" of firm in market k, we get six pricing equations: four for type-d firms (two for each type in both countries) and two for type-v firms (one for each country). I write them out in complementary-slackness formulation with the associated variable in brackets. For each weak inequality (except for auxiliary equations) of this model there exists a complementary variable, i.e. a Kuhn-Tucker multiplier, which is zero if the associated weak inequality holds as a strict inequality, i.e. the constraint is not binding/slack, and which is positive if weak equality holds with equality, i.e. an interior solution. The six pricing equations below are complementary to six output levels for X. Output is strictly positive if marginal revenues equal marginal cost and is zero if marginal revenues are less than marginal costs. The six pricing equations are

illustration not visible in this excerpt

where p_i and p_j denote the price of X in country i and j, respectively.

(16) and (17) the markups for type-d and type-v firms are given by

illustration not visible in this excerpt

Substituting the markup equations into the pricing equations (27)-(32) gives expressions for demand or output

illustration not visible in this excerpt

The number of firms active in equilibrium is determined by the zero-profit conditions. There are three free-entry/zero-profit conditions: one for type-d firms in country i and j and one for type-v firms. Zero-profits require that markup revenues (the left hand side of equations below) are less than or equal to fixed costs (the right hand side of equations below), that is

illustration not visible in this excerpt

Again, variables in brackets are complementary variables. Hence, the number of firms is positive if zero-profit are earned, i.e. left hand side equals right hand side. A firm type is inactive if negative profits are earned, that is fixed costs are strictly greater than markup revenues. Substituting the terms for demand/output (39)-(44) into these equations, (45)-(47) can be written as

illustration not visible in this excerpt

Finally, factor market clearing requires that supply equals demand

illustration not visible in this excerpt

The general equilibrium of this model is then described by a system of equations that em- bodies the optimizing behavior of firms and consumers. To summarize the X-sector, the six inequalities (39)-(44) are associated with six output levels. The number of firms are comple- mentary to the three inequalities (48)-(50). Additionally, goods prices are given by (16) and (17), the two income levels from (18) and factor prices are determined by (51)-(56) together with labor demand (19) and (20) from Y .

4 Impact Effects and Partial Equilibrium Analysis

Before numerically analyzing the full general equilibrium model, I want to develop some intuition about what drives general equilibrium results. To this end, I first conduct a number of thought experiments where I change factor prices to see how the profitability of each firm type is affected. In a simple partial equilibrium version of the model I then go on to investigate how the welfare level in country j responds to changes in factor prices.

To simplify analysis, assume for this entire section that [illustration not visible in this excerpt]

4.1 Factor Price Effects

In order to make zero-profit equations more handy, rewrite (48)-(50) as

illustration not visible in this excerpt

where [illustration not visible in this excerpt] a and b terms denote the the corresponding expressions in brackets (including the exponent) and d terms denote respective fixed costs on the right hand side of (48)-(50). Then, for example, ai [illustration not visible in this excerpt] represent markup revenues of type-d firms based in country i selling in country i and j; and similar for other firm types.

Now, consider a situation in which countries are initially identical, so that commodity prices, factor prices and incomes are the same in both countries — which I refer to as the symmetric case. Then from the definitions of a, b and d terms it follows: [illustration not visible in this excerpt] dv (due to assumption (12)), [illustration not visible in this excerpt] and [illustration not visible in this excerpt] Given these assumptions, inspecting (57)-(59) reveals that variable costs (the right hand sides of (43) and (44)) and fixed costs of type-v firms are always higher than those of type-d firms in both countries implying that profits of type-d firms are always higher than those of a vertical multinational firm [illustration not visible in this excerpt]. Therefore, if countries are symmetric, the technology of type-v firms is "dominated" by that of type-d firms implying that it is always more profitable to become a type-d rather than a type-v firm. Crucial for this result is of course the assumption about the level of fixed costs of type-v firms in comparison to type-d firms. If fixed costs of multinationals are sufficiently low to outweigh the disadvantage in variable costs relative to domestic firms, higher profits could be made by operating as a type-v firm and the result would not hold. However, as mentioned above, due to costs of doing business abroad it is plausible to assume that MNEs display higher fixed cost than national firms¹⁵ which is reflected by fixed costs assumptions in section 3.1.2.

Change in w : [illustration not visible in this excerpt] 0. Now, consider the effect of a fall in the price for skilled labor in country j and an equal rise in the price for skilled labor in country i while holding all other prices constant. Given the assumption that countries are symmetric initially, markup revenues for domestic firms in country i decrease [illustration not visible in this excerpt]¹⁶ and fixed costs rise [illustration not visible in this excerpt] whereas markup revenues for domestic firms in country j rise [illustration not visible in this excerpt] and fixed costs fall [illustration not visible in this excerpt] Therefore profits of type-d firms based in country i (j) fall (rise)[illustration not visible in this excerpt] (due to the assumption that [illustration not visible in this excerpt] and fixed costs decrease [illustration not visible in this excerpt]

[...]

---

¹ UNCTAD World Investment Report (2004).

² The terms "FDI" and "MNEs" are used synonymously throughout the text.

³ FDI is defined as acquiring sufficient assets in a foreign firm to exercise some managerial control (Feenstra, 2004).

⁴ After a steady annual increase until the end of the 1990s, world-wide real inflows of FDI were declining from 1998-2003, but as of today, growth rates seem to have recovered to pre-1998 levels (UNCTAD, 2004).

⁵ See Markusen and Zhang (1998) and Markusen (2002) for a general motivation why this model seems to be appropriate for analysis of FDI in a "North-South" context.

⁶ Markusen (2002) notes that in order to avoid a too strict interpretation of L as unskilled labor, L should rather be seen as a "composite" factor including other factors like physical capital and land.

⁷ Rather than having Cournot competition, an alternative approach would be to assume product differentiation in the imperfectively competitive sector of the economy, as i.e. in so-called "Dixit-Stiglitz large-group monopolistic competition model". However, Markusen (2002) finds that the results of Markusen and Zhang (1998) are not sensitive with respect to the form of competition.

⁸ Instead of variable input requirement one could also label it marginal input requirement. I use the former term which simply means that in order to produce one unit of X, cx units of unskilled labor are needed. Fixed input requirements refer to units of a factor that are need to produce one unit of fixed costs.

⁹ Horizontal MNEs that manufacture the final good in multiple countries are not considered in this model. Markusen (2002) finds that FDI between developed and less developed countries is mainly vertical in nature.

¹⁰ Or alternatively, MNEs can be thought of as being incapable to prevent some of their firm-specific knowledge from spilling over to local firms.

¹¹ I could have specified the relationship between Ti and Tj elsewhere. To do it here, is merely due to syntax requirement in the software that I use for numerical solution of this model. In the program code, which is given in Appendix B.2.2, there will be an auxiliary constraint corresponding to equation (13).

¹² Instead of assuming (13), I also experimented with [illustration not visible in this excerpt] I find that this changes quantitative simulation results very little and — more importantly — qualitative results are unaffected. In what follows, I therefore assume [illustration not visible in this excerpt] as in (13).

¹³ Recall that the price of the numeraire good Y is normalized to one.

¹⁴ This additional notation seems to be cumbersome but it circumvents stating the optimizing problem for each X firm individually.

¹⁵ This is a standard assumption also used by e.g. Markusen and Zhang (1998), Ethier (1986) and Helpman (1984).

¹⁶ Given the initial symmetry of countries, changes in incomes cancel out: ai [illustration not visible in this excerpt]