- Email: [email protected]
Endogenous market structures and the gains from foreign direct investment
Journal of International Economics 64 (2004) 545 – 565 www.elsevier.com/locate/econbase
Endogenous market structures and the gains from foreign direct investment Roberto A. De Santis a, Frank Sta¨hler b,* b
a European Central Bank, Kaiserstrasse 29, D-60311 Frankfurt am Main, Germany Department of Economics, University of Kiel, Wilhelm-Seelig-Platz 1, D-24098 Kiel, Germany
Received 15 November 1999; received in revised form 27 June 2003; accepted 10 August 2003
Abstract This paper discusses the gains from liberalizing foreign direct investment (FDI) in a two-country setting with endogenous market structure. We investigate two different scenarios. In the first scenario, headquarters costs are large in the foreign country so that the industry is located in the domestic country only. In this case, multinational and national firms may coexist and market concentration may make FDI welfare improving for the foreign country and welfare reducing for the domestic country. In the second scenario, headquarters costs are symmetric and firms will be located in both countries. Here, profitable FDI activities lead to mutual welfare gains, irrespective of market structure effects. D 2003 Elsevier B.V. All rights reserved. Keywords: Foreign direct investment; Multinational firms; Imperfect competition; Welfare JEL classification: F12; F15
1. Introduction This study examines the welfare and the market structure effects of foreign direct investment (FDI) in a two-country model in which the choice of FDI and market entry are both endogenous. The paper is motivated by the fact that these effects of FDI are thus far relatively poorly understood in comparison with the effects of trade. In recent years, however, economic integration has increasingly taken the form of FDI, rather than that of
* Corresponding author. Tel.: +49-431-880-1448; fax: +49-431-880-3150. E-mail address: [email protected] (F. Sta¨hler). 0022-1996/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.jinteco.2003.08.008
546
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
trade. FDI has been growing at a high rate, and trade barriers have fallen extensively at the same time. The World Investment Report of the United Nations estimates sales of foreign affiliates at USD 18.5 trillions in the year 2001, whereas exports of goods and non-factor services amounted to USD 7.4 trillions in the same period.1 In other words, the value of aggregate production of multinational firms in host countries nowadays outweighs aggregate exports. This paper focuses on the case where FDI and exports are perfect substitutes as often found in the empirical literature.2 The model we employ is similar to the setting employed in Horstmann and Markusen (1992), but with free entry/exit.3 Market entry is considered by Markusen and Venables (1998, 2000) but—given the complexity of their models—Markusen and Venables rely on numerical solutions.4 In this paper we develop a way of solving analytically models with multinational firms, international trade and free entry/exit. We employ a model of imperfect competition in which firms may enter the market either as a national firm or as a multinational firm. Both types of firms face the same production costs in serving the domestic market. However, in serving the foreign market, the national firm faces additional trade costs, while the multinational firm faces additional fixed costs in setting up a production plant abroad. It is this trade-off between higher marginal costs and higher fixed costs of production which determines the profitability of FDI. Under the hypothesis that the parameter values are such that FDI is profitable, we study how FDI changes market structure and welfare in both countries by comparing a regime under which FDI is prohibited with a regime under which FDI is allowed. Therefore, FDI liberalization is treated as a policy shift, and the condition that no firm gains unilaterally by switching its type and no firm has an incentive to enter or to leave the market then determines the equilibrium market structure. Following the eclectic paradigm of Dunning (1977), firms are induced to invest abroad if, in addition to location advantages (i.e. lower production costs), they have ownership advantages (i.e. patent, blueprint, or trade secret), and internalization advantages (i.e. strategic reasons to exploit its ownership advantage internally rather than licensing it to a foreign firm). Thus, we assume that firms need specific skills to be able to run their headquarters. We measure this requirement by the necessary labor input. Since the labor input needed to set up the headquarters of an oligopolistic firm may differ substantially from country to country, we explore the economic implications of FDI by using two alternative scenarios as reference points.
1
See UNCTAD (2002), Table I.1, p. 4. See Brainard (1997); Blonigen (2001); Markusen (1998); Markusen and Maskus (2002). Literature distinguishes between horizontal FDI (Markusen, 1984; Horstmann and Markusen, 1992; Brainard, 1993; Markusen and Venables, 1998, 2000), which is undertaken to place production closer to foreign markets; and vertical FDI (Helpman, 1984; Helpman and Krugman, 1985), which is undertaken to exploit lower production costs in order to serve both the domestic and the foreign markets. This paper assumes horizontal FDI which seems to be dominant according to empirical evidence because most of the worldwide FDI occurs between countries that are similar in size, relative endowments and per capita incomes. 3 The number of firms in Horstmann and Markusen (1992) is limited to one or zero in each country. Free entry is not considered (see also Markusen, 2002, Ch. 3). 4 A recent manuscript by Markusen (2002) summarizes the results so far achieved in the literature. 2
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
547
In the first scenario, the specific skills to run the headquarters are asymmetrically distributed across both countries so that input requirements for headquarters services are substantially larger in one country, say the foreign country. In this case, firms cannot be run in the foreign country without suffering losses so that market entry will occur only in the domestic country. In the FDI-allowed regime there are, depending on parameter values, three types of equilibria. If the cost of establishing a foreign plant is sufficiently high, allowing FDI has no effect: all firms remain national and serve the foreign market by exports as in the FDI-prohibited regime. If the cost is sufficiently low, all firms are multinational and serve each market through local production in the FDI-allowed regime. For intermediate values of setting up a foreign plant, national and multinational firms coexist. If FDI is profitable, competition becomes tougher in the foreign country with the emergence of multinational firms, because these firms have a marginal cost advantage relative to national firms. The intermediate case of coexistence of national and multinational firms then warrants that the overall number of (national and multinational) firms is reduced. Due to this change in market structure, domestic welfare declines whereas foreign welfare improves. If FDI costs are low enough, only multinational firms remain profitable. In that case, the foreign country gains from FDI but the domestic welfare change depends (inversely) on the size of FDI costs. In the second scenario, the skills to run headquarters are symmetrically distributed across countries. In this case, firms are established in both countries. If FDI is prohibited, our model coincides with the reciprocal dumping model of Brander and Krugman (1983) in which each country hosts the same number of national firms. Conversely, if FDI is allowed and profitable, we show that FDI does not allow any national firm to survive, and all firms are multinational in equilibrium. Also in this case market concentration may (but does not necessarily) occur because multinational firms have to carry larger fixed costs and their size must be larger than that of national firms. Irrespective of the impact of FDI on market entry, we find that FDI is unambiguously mutually welfare improving. Therefore, this result emphasizes the negative role of trade costs on welfare, as often found in the literature. As intra-industry trade compared to autarky, so intra-industry FDI compared to intra-industry trade is mutually welfare enhancing with endogenous market structures. Table 1 summarizes the main assumptions and results of the paper.
Table 1 Summary of assumptions and results Cost structure Location FDI-prohibited regime FDI-allowed regime Coexistence of both types of firms Impact on domestic welfare Impact on foreign welfare
Headquarters costs are large in the foreign country Headquarters are located in the domestic country only Inter-industry trade FDI and inter-industry trade Possible
Headquarters costs are symmetric
Depends on market structure
Positive
Positive
Positive
Headquarters are located in both countries Intra-industry trade FDI and intra-industry trade Impossible
548
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
The remaining parts of the paper are structured as follows. Section 2 presents the framework of the model. Section 3 explores the effects of FDI liberalization on market structure and welfare, when headquarters are located only in the domestic country. Section 4 discusses the implications of FDI liberalization, when headquarters are located in both countries. Section 5 concludes the paper.
2. The model The model used in this paper builds on the model employed by Horstmann and Markusen (1992). There are two countries, a domestic country d and a foreign country f, and two goods, X and Y. Each country is endowed with a certain amount of a productive factor, such as labor L. In both countries, the homogeneous good Y is produced under perfect competition using only the labor input, so that LY = Y, where the superscript denotes the sector in which labor is used. The price of Y equals the return on L, and Y is the numeraire of the model. Exporting Y is assumed to incur no trade costs. The industry producing X is characterized by imperfect competition and Cournot behavior. Firms produce a homogeneous good, which is sold at home and exported or produced abroad. The technology is characterized by constant marginal cost c, and operating headquarters and a plant in country i involves a fixed cost Fi. Shipping goods between countries costs t per unit. All costs are measured in units of labor required. The consumer behavior in each country is determined by the linear quadratic utility function Ui = U(Xi, Yi) = aXi 0.5bXi2 + Yi. Given the aggregate budget resource constraint Li + Spi = piXi + Yi, ia{d, f}, where Spi denotes the aggregate profits of the oligopolistic industry and pi the price of Xi in terms of the numeraire, maximization of Ui subject to the resource constraint yields the inverse income inelastic demand function pi ¼ a bXi ;
iafd; f g;
a c t > 0:
ð1Þ
Markets are segmented so that each firm is able to regard each country as a separate market. Thus, each firm maximizes its profit with respect to both sales in the foreign market and production for the domestic market, and chooses the profit-maximizing quantity for each country separately. The profits of a national firm operating in d are pn ¼ ðpd cÞxnd þ ½pf ðc þ tÞ xnf Fd ;
ð2Þ
where xdn and xnf represent production for the domestic and the foreign market, respectively, and Fd is the fixed set-up cost for home production. This fixed cost comprises the investment cost for headquarters and for one production plant in the domestic country. A similar expression holds for foreign national firms which have to carry fixed costs of size Ff. The profit of a (domestic) multinational firm is m pm ¼ ðpd cÞxm d þ ðpf cÞxf Fd G;
ð3Þ
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
549
where xdm and xdm represent production for the domestic and the foreign market, respectively, and G is the fixed FDI cost needed to start the production process abroad (i.e. the plant-specific fixed cost). In summary, multinational firms are able to produce for the foreign market without incurring trade costs t, but they have to bear the additional fixed cost of operating a plant abroad. Reasonably, the cost of setting up a production plant abroad should be less than the cost of starting production at home, i.e. G < Fi. Furthermore, we assume that a firm cannot set up a plant abroad without first setting up one at home. The sequence of decisions is as follows: in the first stage, firms decide on market entry as a national firm if FDI is banned. If FDI is allowed, they decide simultaneously on market entry and the type of firm they wish to be, national or multinational. In the second stage, all active firms compete in the Cournot fashion. For convenience, we have relegated all the details concerning the optimal production patterns of individual firms to Appendix A.1.
3. FDI and inter-industry trade This section assumes that the fixed costs to set up the headquarters of a firm are substantially different between the two countries. In other words, we assume that the specific skills needed to set up headquarters are available in the domestic country and they are scarce in the foreign country so that the oligopolistic industry can exist only in the domestic country. More formally, Assumption 1 2t 2 Ff zFd þ b
rffiffiffiffiffiffiffiffiffi! bFd 1 ; s
G < Fd ;
t2 s < Fd < ; b b
specifies this asymmetry where s=(a c)2+(a c t)2 measures the size of the markets for an exporting (national) firm. This implies that headquarters are located in the domestic country only (see Appendix A.1.3). Furthermore, G < Fd guarantees that national firms will always export to the foreign market if they coexist with multinational firms (see Appendix A.3, Fd < s/b guarantees that the market exists and Fd>t2/b guarantees that competition by national firms is not so intense that FDI can never become dominant (see Appendix A.2). In the FDI-prohibited equilibrium, each firm maximizes profit (2) given the equilibrium output of its rivals, and entry occurs until the excess profit of the marginal firm is driven to zero. Since domestic firms are symmetric, they earn zero profit in equilibrium. This condition allows us to determine the equilibrium number of national firms, which under the FDI-prohibited regime (denoted by a star) is equal to rffiffiffiffiffiffiffiffiffi s N* ¼ 1; ð4Þ bFd
550
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
and the equilibrium prices in both countries a þ cN * ¼ c þ ða cÞ pd* ¼ N* þ 1
rffiffiffiffiffiffiffiffiffi bFd ; s
ð5Þ
rffiffiffiffiffiffiffiffiffi a þ ðc þ tÞN * bFd ¼ c þ t þ ða c tÞ p*f ¼ : N* þ 1 s Given Assumption 1, N*>0. Note that the higher Fd, the lower the number of firms and the higher the price levels in equilibrium. What is the impact of FDI liberalization on market structure and welfare? In order to investigate this issue, we solve for the first order conditions of the national and multinational firms under the hypothesis that the parameters of the model make FDI profitable.5 The equilibrium number of national firms (N) and multinational firms (M) can be determined by solving two zero profit conditions f ðN ; M Þ ¼
gðN ; M Þ ¼
ða cÞ2 bðN þ M þ 1Þ
2
þ
2
þ
ða cÞ2 bðN þ M þ 1Þ
½a c ðM þ 1Þt 2
Fd ¼ 0;
ð6Þ
ðFd þ GÞ ¼ 0;
ð7Þ
bðN þ M þ 1Þ2 ða c þ NtÞ2 bðN þ M þ 1Þ2
where f denotes the maximized profit of national firms, and g denotes the maximized profit of multinational firms. In both Eqs. (6) and (7), the first (second) term gives the revenues net of variable costs in the domestic (foreign) market. In order to examine the impact of ¯ denote the FDI on the market structure, the definition of two critical values is useful. Let G lowest fixed cost that makes multinational production unprofitable, and let G denote the largest fixed cost that makes national production unprofitable (see Appendix A.2). By using G G , we can demonstrate Proposition 1. and Proposition 1 . Three different types of a unique equilibrium are possible when headquarters are located in the domestic country only and FDI is allowed. If Ga]G , l[, only national firms are active. If Ga[0, G [, only multinational firms are active. If ¯ Ga[G , G ], national and multinational firms coexist. 5 ¯ Appendix A.2 shows that the nondegeneracy condition G >G is satisfied for some parameter values, so that the intermediate range can emerge as an equilibrium. Additionally, it shows that the zero profit conditions f (N, M) = 0 and g(N, M) = 0 have a negative slope in the (N, M) space, and that the resulting Jacobian determinant is unambiguously positive, which implies that the equilibrium is unique.
Proof . See Appendix A.2.
5
Obviously, multinational firms will be established only in the domestic country.
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
551
In summary, Proposition 1 demonstrates that national and multinational firms can coexist.6 From the viewpoint of a single firm, optimal production for the home market does not depend on its type. But a multinational firm produces more for the foreign market than an exporting national firm to cover its larger fixed costs (see Appendix A.1.2). Hence, prices will decline in the foreign country, and this effect makes subsequent FDI activities less attractive. In other words, once FDI is liberalized, the attractiveness of FDI declines, such that in equilibrium not all active firms may become multinational. More specifically, in the case of coexistence, the equilibrium values of N and M, denoted by e, are: a c bG þ t 2 Ne ¼ ð8Þ pffiffiffi 1 ; t g Me ¼
ac bG t 2 1 pffiffiffi 1; g t
ð9Þ
2tða cÞ pffiffiffi 1; g
ð10Þ
Ne þ Me ¼
where gw 4bFdt2 (bG t2)2>0.7 Appendix A.2 demonstrates that the equilibrium values are positive in case of coexistence. Then, the four first order conditions and the two inverse demand functions Eq. (1) determine the equilibrium prices ped
pffiffiffi g a þ cðN e þ M e Þ ; ¼cþ ¼ Ne þ Me þ 1 2t
pef ¼
a þ ðc þ tÞN e þ cM e bG þ t 2 ¼cþ : e e N þM þ1 2t
ð11Þ
Expression (11) demonstrates a positive relationship between pde and t, and a negative relationship between pde and G. Conversely, pfe and t are negatively related, while pfe and G are positively related. These peculiar relationships can be understood only if one looks at the implication of the FDI cost G on market structure. As the FDI cost declines, more multinational firms having a comparative marginal cost advantage force foreign prices to decline. This effect reduces the total number of firms in equilibrium and, as a result, the domestic price increases. The opposite analysis is valid if t declines. Expression (11) shows also that the domestic price is affected by the fixed cost Fd which is necessary to start the production process (through g), while the foreign price does not change with Fd. An increase in fixed cost warrants firms to produce more to cover the fixed cost, and multinational firms have an advantage in size compared to national firms. The invariance of the foreign price with respect to Fd shows that the negative effect on the 6
Appendix A.3 proves that national firms export and thus compete with multinational firms in the foreign market in case of coexistence if Assumption 1 holds. 7 Appendix A.2 shows that g is unambiguously positive in the case of coexistence.
552
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
total number firms is just compensated by the relative increase in the number of multinational firms. Indeed, Eqs. (8) – (10) show that both the number of national firms and the aggregate number of firms decline, while the number of multinational firms increases if Fd rises. As Proposition 1 has shown, it might happen that only one type of firm, either national or multinational, is active in the market. If f (0, M**) V 0 holds for the profits of national firms, where the double star denotes the equilibrium under the condition Ga[0, G [, market entry is not profitable even for a single national firm. We then derive the following condition for an equilibrium where only multinational firms are active, which is consistent with Proposition 1: f ð0; M**ÞV0ZbGVt
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2bFd t 2 :
ð12Þ
In other words, if the FDI cost G (the trade cost t) is sufficiently low (high) all national firms exit from the market, once FDI is liberalized. It is also interesting to note that the larger is Fd, the greater is the likelihood that multinational firms alone will serve the markets. The reason is that the larger is the fixed cost to start the production process at home, the larger firms have to be in order to avoid losses. If Eq. (12) holds, the equilibrium values for prices and number of firms are:
pd** ¼ p** ¼ f
a þ cM** ¼cþ M ** þ 1
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi bðFd þ GÞ ; 2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 M ** ¼ ða cÞ 1: bðFd þ GÞ
ð13Þ
ð14Þ
If, conversely, the FDI cost G (trade cost t) is sufficiently high (low) such that Ga[G ,l[, then multinational firms are not profitable. This occurs if g(N*, 0) V 0 holds for the profits of multinational firms. We derive the following condition for an equilibrium outcome where only national firms are active in a FDI-allowed regime, which is also consistent with Proposition 1: rffiffiffiffiffiffiffiffiffi bFd gðN *; 0ÞV0ZbGz2tða c tÞ þ t2 : s
ð15Þ
Namely, the greater is the fixed cost to undertake FDI and/or the lower is the trade cost, the greater is the likelihood that only national firms are active in the market. Having identified the conditions under which firms are active in the domestic and foreign markets, investigating the impact of FDI liberalization on market structure is straightforward. Lemmas 1 and 2 summarize the results.
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
553
Lemma 1. Under parameter values such that both types of firms coexist in a regime with FDI, the impact of allowing FDI is to reduce the number of firms. Proof. The total number of active firms is given by Ne + Me. The impact of FDI on market structure can be easily examined by using Eqs. (4) and (10). In fact, Ne + Me < N* if, and only if, sðbG t 2 Þ2 < 4ða c tÞ2 t 2 bFd :
ð16Þ
This expression is fulfilled if, and only if, condition (15) for the equilibrium where only national firms are active is violated. Since this is the case, it follows that allowing FDI leads to market concentration if national firms survive. 5 The intuition for this result can be summarized as follows. National and multinational firms sell the same amount in the domestic market. Given the zero profit condition, the net profit from exporting has to equal the net profit of setting up a foreign subsidiary. If the total number of firms rose after liberalization, a national firm would automatically earn less from exporting, since the foreign market is more crowded, and multinational firms have a marginal cost advantage. This implies that national firms would have to earn more at home, which is inconsistent with entry. Hence, in any coexistence outcome, FDI leads to exit. Lemma 2. Under parameter values such that only multinational firms exist in a regime with FDI, the impact of allowing FDI on the number of firms depends on the fixed costs of a national firm compared with those of a multinational firm: M ** < N *; if
Fd s < ; Fd þ G 2ða cÞ2
M ** > N *; if
Fd s > : Fd þ G 2ða cÞ2
Proof. The condition for M**bN* follows from Eqs. (4) and (14). The condition for the equilibrium with only multinational firms Eq. (12) can be re-arranged as follows: Fd ða cÞ2 þ ða c t tM **Þ2 z : Fd þ G 2ða cÞ2 Expression (17) can be consistent with both M**< N* and M**>N*.
ð17Þ 5
Lemma 2 maintains that there is scope for market concentration even if only multinational firms are active in the market. In particular, the number of active firms under the FDI regime is lower, if national firms are not active at the margin. However, we cannot rule out the possibility that M**>N*. If G is very small, the equilibrium number of multinational firms could be so large that the markets could support more multinational firms.
554
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
Table 2 Scenario 1—FDI and trade with headquarters located in the domestic country only
G Number of national firms Number of multinational firms Domestic price Foreign price Domestic output per firm Exports per national firm Foreign sales per multinational firm
Only national firms*
National and multinational firms**
FDI-prohibited 33
9 12 17
7.760 9.217 2.760
8.152 8.750 3.152
2.717
2.220 3.750
Only multinational firms*** 4.5
29
Only multinational firms*** 0.15
34
8.122 8.122 3.122
7.522 7.522 2.752
3.122
2.752
Parameters of the model: a = 100, b = 1, c = 5, t = 0.3c, Fd = 15, Ff z 19.369. * G > 0.693Fd or FDI is prohibited. ** 0.527Fd V G V 0.693Fd. *** G < 0.527Fd.
Table 2 shows the results of numerical simulations which are based on parameter values as indicated at the bottom of the table.8 In accordance with Proposition 1, the simulations indicate that the markets are served by national firms only if FDI costs are relatively large, i.e. G/Fd>69.3%, or if FDI is prohibited. If FDI is liberalized and 52.7% V G/Fd V 69.3%, coexistence between national and multinational firms occurs. Conversely, if G/Fd < 52.7%, only multinational firms would survive and serve both markets. Note that the overall number of firms declines with a reduction of G in the range in which G is moderately large so that an equilibrium with multinational firms only cannot emerge. Moreover, the number of multinational (national) firms is inversely (positively) related to G. This is the reason why, if G is sufficiently low, multinational firms become dominant, coexistence is impossible and the number of multinational firms may be even larger under FDI liberalization compared to FDI prohibition (see last column of Table 2). Note also that the foreign output of a multinational firm is larger than that of a national firm, because it has to cover larger fixed costs. The impact of FDI on market structure has an effect on the welfare of both the domestic and the foreign country, because the equilibrium prices depend on the equilibrium number of firms and production costs. Based on Lemmas 1 and 2, we prove the following Proposition 2.
8 Note that the fixed cost to start a business in the foreign country ( Ff) has only to be larger than 19.369 in order to prevent the location of headquarters if Fd = 15 (see Assumption 1). Thus, Ff /Fd z 1.29 is sufficient to guarantee that trade and FDI are of the inter-industry type here.
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
555
Proposition 2 . Under parameter values such that both types of firms coexist in a regime with FDI, the welfare of the domestic country declines when FDI is allowed. Under parameter values such that only multinational firms exist in a regime with FDI, the welfare of the domestic country goes the way of the number of firms, while the welfare of the foreign country always increases. Proof. Given the zero profit condition, welfare depends only on prices. Eqs. (5), (11) and (13) show that the domestic consumer is worse off if the total number of active firms under the FDI regime is smaller than the number of national firms under the trade regime. Lemma 1 shows that market concentration occurs if both types of firms coexist, Lemma 2 shows that market concentration may occur if only multinational firms exist. The proof of the foreign country’s welfare improvement can be given by contradiction. Suppose that pf given by Eq. (11) is larger than pf given by Eq. (5). Then, an increase in the foreign price occurs if, and only if, rffiffiffiffiffiffiffiffiffi bFd þ t2 : ð18Þ bG > 2tða c tÞ s However, if this inequality were true, multinational firms would not be active and only national firms would serve the market (see condition (15)). Hence, if coexistence can occur, the foreign price declines when FDI is allowed. This finding also holds for the equilibrium where only multinational firms are active because the foreign price declines unambiguously if G falls (see Eq. (13)). 5 Proposition 2 demonstrates that the domestic country is worse off after the foreign country liberalizes FDI activities, if national and multinational firms coexist. As multinational firms are able to offer goods at lower marginal costs, prices decline in the foreign markets and the overall number of firms is reduced. The reduction in the number of firms which compete also on the domestic market leads to an increase in domestic prices and hence domestic welfare declines in an equilibrium with coexistence of national and multinational firms. As the domestic country is worse off in the case of coexistence, it is obvious that the domestic country can only become better off if multinational firms alone are active in the market. To obtain a welfare gain, it is essential that the foreign market becomes much more profitable after FDI liberalization. If the increase in the profitability of the foreign market is substantial enough so that more multinationals are active than national firms have been active before, domestic welfare will improve. As the profitability of the foreign market depends crucially upon the size of fixed costs, we investigate how small the FDI costs G should be to ensure that the domestic country is not worse off. The critical amount of the plant-specific fixed costs can be computed by setting expression (14) larger than or equal to expression (4). Therefore, the domestic country is not worse off if, and only if, GVwFd where w ¼
ða cÞ2 ða c tÞ2 ða cÞ2 þ ða c tÞ2
:
ð19Þ
Only if G is small enough, does the domestic country gain from FDI, because the number of multinational firms would be larger than the number of firms under the
556
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
FDI-prohibited regime. However, it seems that G should be really small for a welfare enhancing effect to occur with FDI liberalization. In fact, consider a possible case where marginal costs are 5% (30%) of the reservation price and trade costs are 30% of the marginal costs. In this case, the critical parameter w would be equal to 0.024 (0.137). In other words, for the domestic consumer to be better off, the fixed costs of setting up a plant abroad should be lower than 2.4% (13.7%) of the fixed costs of starting the production process at home. These results are also confirmed by the numerical simulations presented in Table 2, which assumes that marginal costs are 5% of the reservation price. Under the parameter values reported in Table 2, the domestic price increases from 7.76 under the FDIprohibited regime to 8.152 if coexistence occurs with FDI liberalization. Conversely, if G is sufficiently low, only multinational firms would survive and the domestic price would either increase to 8.122 if G is 30% of Fd or decline to 7.522 if G is 1% of Fd. On the other hand, the foreign consumer is always better of with FDI liberalization, regardless of the impact on market structure.
4. FDI and intra-industry trade The previous section has shown the effects of FDI liberalization on market structure and welfare under the assumption that the headquarters of the oligopolistic industry are located only in the domestic country. This section assumes that the fixed costs to start a business are symmetric and hence low enough also in the foreign country: Assumption 2 Fd ¼ Ff ¼ F;
t2 s
Assumption 2 implies that headquarters are located in both countries. In particular, F < s/b guarantees that the markets exist and F >t2/b guarantees that exports occur (see Appendix A.1.3). If fixed costs are too low compared to trade costs, intra-industry trade would not occur because the presence of too many local firms would make exports not worthwhile. If FDI is banned, the FDI-prohibited regime is characterized by intra-industry trade, and the model coincides with the reciprocal dumping model of Brander (1981); Brander and Krugman (1983). They show that, with respect to autarky, trade is mutually welfare improving if the market structure is endogenous, as it induces firms to move down their average cost curves. If FDI is banned, we need to consider only the domestic country because the number of domestic and foreign national firms coincides. If more domestic national firms than foreign national firms were active, competition in the domestic country would be tougher than in the foreign country due to trade costs, and hence domestic firms would make less profits than foreign firms, so that the zero profit condition would be violated. In the intra-industry FDI-prohibited
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
557
regime, denoted by a tilde, the equilibrium number of national firms and the equilibrium price in both countries is, respectively, equal to ða cÞ þ ða c tÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi N˜ ¼ 1: 2bF t 2 and p˜ ¼ c þ
tþ
ð20Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2bF t 2 : 2
ð21Þ
We will now discuss the impact on market structure and welfare of allowing FDI under the hypothesis that the parameters of the model are such that FDI occurs in equilibrium. In this setting, multinational firms may run their headquarters in either country. But since countries are symmetric, it does not make any difference whether a certain firm locates its headquarters and a plant in the domestic country and an additional plant in the foreign country or vice versa. Therefore, we are able to determine the aggregate equilibrium number of multinational firms but we cannot say where their headquarters are located. The location of national firms, however, is sensitive, because their exports suffer from the trade cost disadvantage. We know that the number of national firms is identical across countries if they coexist with multinational firms, because multinational firms produce the same amount of output for each country, and the residual demand faced by national firms is identical across countries. Therefore, the number of national firms will coincide also with FDI for the same reason as in the reciprocal dumping model. Since all types of firms are possibly present in the market, we define N as the number of all national firms located in both countries and M as the number of all multinational firms. In other words, N/2 national firms are located in each country, but the headquarters location of M multinational firms cannot be determined. The solution of the standard profit maximization problems faced by the firms yields the equilibrium price pðN ; M Þ ¼
a þ cN þ 0:5tN þ cM ; N þM þ1
ð22Þ
and two zero profit conditions, /ðN ; M Þ ¼
cðN ; M Þ ¼
ða c þ 0:5tN Þ2 bðN þ M þ 1Þ
2
2ða c þ 0:5tN Þ2 bðN þ M þ 1Þ2
þ
ða c t 0:5tN tM Þ2 bðN þ M þ 1Þ2
F G ¼ 0;
F ¼ 0;
ð23Þ
ð24Þ
where / denotes the maximized profit of national firms, and c denotes the maximized profit of multinational firms. In Eq. (23), the first (second) term gives the revenues of national firm net of variable costs in the domestic (foreign) country. The first term in Eq. (24) collects the symmetric revenues of multinational firms net of variable costs in both markets. / and c would allow us to determine the equilibrium number of both national and multinational firms, respectively. However, the Jacobian determinant of / and c with
558
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
respect to N and M is zero, which implies that / and c are linearly dependent and no equilibrium solution with positive N and M exists. Consequently, the equilibrium in a symmetric Cournot oligopoly with entry can only be a corner solution, which leads to the conclusion that either only multinational firms or only national firms are in the market. Proposition 3. If the oligopolistic industry is located in both countries, firms are either all national or all multinational. Proof . See Appendix A.4.
5
In this setting, coexistence of both types is not possible. Proposition 3 demonstrates that national firms cannot survive if multinational production becomes profitable because multinational production affects competition in both countries and is a threat to national firms in both countries. Since the pre-liberalization equilibrium is represented by Eq. (20) with national firms making zero profits, FDI is profitable if, and only if, c(N˜, 0) z 0. The latter condition implies that multinationals may enter without making losses and would— as Proposition 3 has shown—dominate the market. Once FDI is liberalized, the condition for the equilibrium where only multinational firms are active is cðN˜ ; 0Þ ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2ða c þ 0:5t N˜ Þ2 F Gz0ZbGVt 2bF t 2 : 2 bðN˜ þ 1Þ
ð25Þ
Condition (25) coincides with Eq. (12) in the first scenario of asymmetric fixed costs Fi described in Section 3. Under condition (25), national firms do p notffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi survive andffi the equilibrium number of multinational firms, M**, is given by ða cÞ 2=ðbðF þ GÞÞ 1 (see Eq. (14)).9 With regard to the impact on the market structure, we find that M**< N˜ does not violate condition (25), so that industry concentration may also occur in the case of intra-industry FDI. But the fixed costs to set up a plant abroad could also be so small that the equilibrium number of firms rises. More importantly, the welfare results are unambiguous. Proposition 4. Although the impact on the market structure of the intra-industry FDIallowed regime in comparison with the intra-industry FDI-prohibited regime depends upon parameter values, FDI liberalization generates mutual welfare gains. Proof . Under FDI, the equilibrium price is rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi bðF þ GÞ : ð26Þ p** ¼ c þ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p** is less than p˜ if bGVt 2bF t 2 which coincides with condition (25) that multinational firms are active and dominant. 5 This is a remarkable result because it stresses the role of trade costs. Consider a firm switching from being a national to being a multinational firm after FDI has been liberalized. Then, a decline in the domestic price implies a decrease in the individual domestic output. Furthermore, the multinational firm has to bear the additional plant-specific fixed cost. From 9
Conversely, if /(0, M**) z 0, the market would be served by national firms only.
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
559
Table 3 Scenario 2—FDI and trade with headquarters located in both countries Only national firms* G Number of national firms Number of multinational firms Domestic price Foreign price Domestic output per firm Exports per national firm Foreign sales per multinational firm
FDI-prohibited 35 8.384 8.384 3.384 1.884
Only multinational firms** 6 28 8.240 8.240 3.240 3.240
Parameters of the model: a = 100, b = 1, c = 5, t = 0.3c, F = 15. * G > 0.527F or FDI is prohibited. ** G V 0.527F.
these profit-reducing effects, we would expect that prices rise. However, the foreign market has become so profitable that the increase in individual production for the foreign market is able to overcompensate these effects. Proposition 4 demonstrates that the profit-reducing effects are offset by vanishing trade costs. The numerical simulations reported in Table 3 confirm the analytical results. The markets are served by national firms only, if FDI is prohibited or if G/Fd>52.7%. Conversely, if G/Fd V 52.7%, only multinational firms survive with FDI liberalization. They serve both markets at lower prices because average costs have declined due to the lack of trade costs. Note also that the size of a multinational firm (6.48) is larger than that of a national firm (5.27), because it has to cover larger fixed costs.
5. Concluding remarks This paper has examined the impact of FDI on market structure and welfare. We have distinguished between two different scenarios concerning headquarters costs:
Headquarters costs are relatively large in the foreign country so that they are located in the domestic country only. In this case, trade and FDI are of the inter-industry type. Headquarters costs are symmetric in both countries. In that case, trade and FDI are of the intra-industry type. Given these two different cases, we have studied the impact of liberalizing FDI on market structure and welfare by comparing the FDI-allowed regime with the FDIprohibited regime. We have shown that, if the headquarters are located in the domestic country only, the FDI-allowed regime as compared with the FDI-prohibited regime leads to welfare losses for the domestic country, but to welfare gains for the foreign country when national and multinational firms are both active. National and multinational firms may coexist for moderately large FDI costs because FDI affects competition only in the foreign market. If only multinational firms emerge, the welfare of the foreign country improves, whereas the impact on the welfare of the domestic country depends on the size of FDI costs.
560
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
FDI liberalization also has an impact on market structure: if both firms coexist, the number of active firms decreases; if only multinational firms are active, their number may be smaller or larger than the number of national firms in the case of no FDI. In this latter case, we have shown that domestic welfare improves only if FDI increases the number of active firms. It must be noted, however, that an increase in the number of firms and—due to entry of multinational firms—an increase in aggregate fixed costs alone cannot be expected to improve welfare, but would imply an increase in prices. In order to improve domestic welfare, it is essential that multinational firms save large trade costs. In other words, the increase in the profitability of the foreign market must be sufficiently large so that the number of active firms increases. The results are different if the headquarters are located in both countries. As FDI alters competition in both markets, coexistence of both types of firms is not possible so that either only national or only multinational firms are active. Under parameter values that make FDI profitable, the FDI-allowed regime implies that national firms are completely replaced by multinational firms. In this case, we have shown that the FDI-allowed regime leads to mutual welfare gains in comparison with the FDI-prohibited regime. The reason is that consumers have to bear trade costs in the FDI-prohibited regime which are saved by multinational firms. Furthermore, multinational firms dominate national firms if and only if the additional fixed costs are sufficiently low and trade costs are sufficiently substantial so that average costs are lower with FDI. This case demonstrates that FDI, as compared with trade, leads to mutual welfare gains just as trade, as compared with autarky, leads to mutual welfare gains if the market structure is endogenous and firms’ headquarters may locate in either of the countries involved.
Acknowledgements We would like to thank Jonathan Eaton, Horst Raff, Gerald Willmann and two anonymous referees for valuable comments, and Wilfred Ethier, Jim Markusen and Anthony Venables for helpful suggestions. The opinions expressed herein are those of the authors and do not represent those of the European Central Bank. All errors are authors’ responsibility.
Appendix A A.1 . Output levels A.1.1 . Output levels under inter-industry trade without FDI sffiffiffiffiffiffi ac Fd xd* ¼ ¼ ða cÞ ; bðN * þ 1Þ sb sffiffiffiffiffiffi x*f ¼
act ¼ ða c tÞ bðN * þ 1Þ
Fd : sb
xd* denotes production for the domestic market and xf* denotes exports.
ðA:1Þ
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
561
A.1.2 . Output levels under inter-industry trade and FDI pffiffiffi g ac me ; xne ¼ ¼ x ¼ d d bðN e þ M e þ 1Þ 2bt xne f ¼
a c tðM e þ 1Þ bG t 2 ¼ ; bðN e þ M e þ 1Þ 2bt
xme f ¼
a c þ tN e bG þ t 2 ¼ : bðN e þ M e þ 1Þ 2bt
ðA:2Þ
me xdne(xne f ) denotes domestic production (exports) of national firms, xf denotes production of multinational firms for each market. Note that the size of the multinational firm is larger ne than the size of the national firm (xme f xf = t/b), because the former has to bear larger ne aggregate fixed costs. Note also that xf >0 requires bG>t2 which is guaranteed by Assumption 1 (see Appendix A.2).
A.1.3 . Output levels under intra-industry trade without FDI The equilibrium output levels per firm are xdd ¼
a c þ tnf ; bðnd þ nf þ 1Þ
xdf ¼
a c tðnf þ 1Þ ; bðnd þ nf þ 1Þ
xff ¼
a c þ tnd ; bðnd þ nf þ 1Þ
xfd ¼
a c tðnd þ 1Þ ; bðnd þ nf þ 1Þ
ðA:3Þ
where subscripts (superscripts) denote the destination of production (location of the firm). nd(nf) denotes the number of domestic (foreign) national firms. With asymmetric fixed costs, these output levels lead to maximized profits of zd ¼
zf ¼
ða c þ tnd Þ2 þ ða c tðnd þ 1ÞÞ2 bðnd þ nf þ 1Þ2 ða c þ tnf Þ2 þ ða c tðnf þ 1ÞÞ2 bðnd þ nf þ 1Þ2
Fd ;
Ff :
ðA:4Þ
qffiffiffiffiffiffi 2 bFd Under the condition that Ff zFd þ 2tb 1 , zd = zf = 0 implies that nf V 0 so s that no foreign firm will enter the market. If fixed cost are symmetric, i.e. Fd = Ff = F, output levels are pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2bF t 2 þ t a c þ 0:5N˜ t ; x˜ d ¼ ¼ 2b bðN˜ þ 1Þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2bF t 2 t a c ð0:5N˜ þ 1Þt x˜ f ¼ : ðA:5Þ ¼ ˜ 2b bðN þ 1Þ Note that F >t2/b (see Assumption 2) guarantees that exports x˜f are positive.
562
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
A.1.4 . Output levels under intra-industry trade and FDI m xnd ¼ xm d ¼ xf ¼
a c þ 0:5tN ; bðN þ M þ 1Þ
xnf ¼
a c t 0:5tN tM : bðN þ M þ 1Þ
ðA:6Þ
xdn(xnf) denotes production for the home market (exports) of national firms, and xdm(xmf ) denotes production of a multinational firm for the home market (host market of FDI). A.1.5 . Output levels under FDI without trade ac ¼ xd** ¼ xf** ¼ bðM ** þ 1Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Fd þ G : 2b
ðA:7Þ
x**d (x** f ) denotes production of a multinational firm for the home market (host market of FDI). A.2 . Proof of Proposition 1
G (and N ) is determined by setting f (N, 0) = g(N, 0) = 0, whereas G (and M) is determined by setting f (0, M) = g(0, M) = 0: rffiffiffiffiffiffiffiffiffi! t bFd t þ 2ða c tÞ G ¼ ; ðA:8Þ b s t pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi 2bFd t : G¼ b
ðA:9Þ
2 Note that Fd>t2/b (see Assumption 1) guarantees that G>G >t /b so that national firms export to the foreign market (see Eq. (A.2)). If Fd were too small, a large number of national firms would enter the markets and FDI would never be able to become dominant, irrespective of G, a case we are not interested in. The minimum Fd guarantees that the profitability of multinational production depends on the FDI cost G. Only national firms will emerge if G>G, because G is determined from Eqs. (6) and (7) by setting M equal to zero. Any higher G would only deepen the cost disadvantage of multinational firms. By contrast, only multinational firms will emerge if G < G , because G is determined from Eqs. (6) and (7) by setting N equal to zero. Any lower G would only raise the profitability of multinational firms. If, however, G > G , a range of G exists in which a mixed equilibrium occurs, i.e. an equilibrium in which national and multinational firms coexist. For the mixed equilibrium proof, define Dw(b/t)(G G ). If D > 0, a mixed equilibrium may exist. First, let us explore the behavior of the D-function in the relevant range. We observe that D = 0 if Fd = s/b. However, this level of fixed costs implies that the market does not exist, because even a single firm is not able to recoup its fixed costs (see Assumption 2 and Eq. (4)). Hence, the relevant fixed costs must be below this level. If D decreases with Fd in the relevant range, we have shown that D is positive, since it reaches zero at the border where
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
563
the market does not exist under the FDI-prohibited regime. Differentiation yields ! BD act 1 ¼ b pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ðA:10Þ BFd sbFd 2bFd t 2 This term is negative if t > bFd[2(a c) t]/[a c t]2, which is always fulfilled. Hence, a mixed equilibrium may exist. Uniqueness is proven if the Jacobian determinant has an unambiguous sign for the equilibrium values of N and M. Differentiation of the zero profit conditions yields Bf 2ðða cÞ2 þ ða c ðM þ 1ÞtÞ2 Þ ¼ < 0; BN bðN þ M þ 1Þ3
ðA:11Þ
Bf Bg 2½ða cÞ2 þ ða c þ NtÞða c ðM þ 1ÞtÞ ¼ ¼ < 0; BM BN bðN þ M þ 1Þ3
ðA:12Þ
Bg 2ðða cÞ2 þ ða c þ NtÞ2 Þ ¼ < 0: BM bðN þ M þ 1Þ3
ðA:13Þ
For the negative signs of all derivatives see also Eq. (A.2). The Jacobian is then AJ A ¼
Bf Bg Bf Bg 4ða cÞ2 t 2 g2 ¼ 2 ¼ >0 BN BM BM BN b ðN þ M þ 1Þ4 ½2bða cÞt 2
ðA:14Þ
where gw 4bFdt2 (bG t2)2>0 because G < G implies that g must be larger than 4bFdt2 (a c)2/s>0. Note also that Ne>0 if, and only if, G < G, while Me>0 if, and only if, G>G satisfying, therefore, the conditions for coexistence shown in Proposition 1. Expression (A.14) shows that the sign of the Jacobian is unambiguously positive. Note that this property holds also for the corner solutions characterized by f (0, M) = g(0, M) = 0 or f (N, 0) = g(N, 0) = 0, and hence proves that the equilibrium is unique. The proof can be given by contradiction. Suppose that multiple mixed equilibria exist. In this case both the f- and the g-function should intersect at least twice. However, both the f- and the g-function are twice continuously differentiable and have a negative slope. If they intersect twice, the second intersection should imply a Jacobian with an opposite sign. However, the Jacobian is always positive. Hence, any mixed equilibrium must be unique. 5 A.3 . Positive exports of national firms The proof for positive exports will be given by contradiction. Suppose that national firms serve only the domestic market. Output levels per firm are then me xne d ¼ xd ¼
bðN e
ac ; þ M e þ 1Þ
xne f ¼ 0;
xme f ¼
ac ; bðM e þ 1Þ
ðA:15Þ
564
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
leading to the following zero profit conditions: zne ¼
ða cÞ2 bðN e þ M e þ 1Þ2
Fd ¼ 0
ðA:16Þ
and zme ¼
ða cÞ2 bðN e þ M e þ 1Þ2
þ
ða cÞ2 bðM e þ 1Þ2
Fd G ¼ 0:
ðA:17Þ
Eqs. (A.16) and (A.17) imply that ac M e ¼ pffiffiffiffiffiffiffi 1 bG
ðA:18Þ
ac 1 1 N e ¼ pffiffiffi pffiffiffiffiffi pffiffiffiffi : Fd b G
ðA:19Þ
and
Clearly, N e>0 if, and only if, G >Fd, which contradicts the reasonable Assumption 1 that the costs of setting up headquarters and a plant are less than the costs of setting up an additional plant. 5 A.4 . Proof of Proposition 3 Proposition 3 claims that the Jacobian of Eqs. (23) and (24) is zero. Differentiation of Eqs. (23) and (24) yields B/ ½ða cÞ þ ða c tÞ tM 2 ¼ < 0; BN bðN þ M þ 1Þ3
ðA:20Þ
B/ Bc 2ða c þ 0:5tN Þ½ða cÞ þ ða c tÞ tM ¼ ¼ < 0; BM BM bðN þ M þ 1Þ3
ðA:21Þ
Bc 4ða c þ 0:5tN Þ2 ¼ < 0: BM bðN þ M þ 1Þ3
ðA:22Þ
By using Eqs. (A.20) –(A.22), we find that
AJ A ¼
B/ Bc B/ Bc ¼ 0: BN BM BM BN
ðA:23Þ 5
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
565
References Blonigen, B.A., 2001. In search of substitution between foreign production and exports. Journal of International Economics 53, 81 – 104. Brainard, S.L., 1993. A simple theory of multinational corporations and trade with a trade-off between proximity and concentration. NBER Working Paper 4269. Brainard, S.L., 1997. An empirical assessment of the proximity-concentration trade-off between multinational sales and trade. American Economic Review 87, 520 – 540. Brander, J.A., 1981. Intra-industry trade in identical commodities. Journal of International Economics 11, 1 – 14. Brander, J.A., Krugman, P.R., 1983. A ‘‘reciprocal dumping’’ model of international trade. Journal of International Economics 15, 313 – 323. Dunning, J.H., 1977. Trade, location of economic activity and MNE: a search for an eclectic approach. In: Ohlin, B., Hesselborn, P.O., Wijkman, P.M. (Eds.), The International Allocation of Economic Activity. Macmillan, London, pp. 395 – 418. Helpman, E., 1984. A simple theory of trade with multinational corporations. Journal of Political Economy 92, 451 – 571. Helpman, E., Krugman, P.R., 1985. Market Structure and Foreign Trade. MIT Press, Cambridge. Horstmann, I.J., Markusen, J.R., 1992. Endogenous market structures in international trade (natura facit saltum). Journal of International Economics 32, 109 – 129. Markusen, J.R., 1984. Multinationals, multi-plant economies, and the gains from trade. Journal of International Economics 16, 205 – 226. Markusen, J.R., 1998. Multinational firms, location and trade. World Economy 21, 733 – 756. Markusen, J.R., 2002. Multinational Firms and the Theory of International Trade. The MIT Press, Cambridge. Markusen, J.R., Maskus, K.E., 2002. Discriminating among alternative theories of the multinational enterprise. Review of International Economics 10, 694 – 707. Markusen, J.R., Venables, A.J., 1998. Multinational firms and the new trade theory. Journal of International Economics 46, 183 – 203. Markusen, J.R., Venables, A.J., 2000. The theory of endowment, intra-industry, and multinational trade. Journal of International Economics 52, 209 – 234. UNCTAD, 2002. World Investment Report 2002. Transnational Corporations and Export Competitiveness, New York.
Endogenous market structures and the gains from foreign direct investment Roberto A. De Santis a, Frank Sta¨hler b,* b
a European Central Bank, Kaiserstrasse 29, D-60311 Frankfurt am Main, Germany Department of Economics, University of Kiel, Wilhelm-Seelig-Platz 1, D-24098 Kiel, Germany
Received 15 November 1999; received in revised form 27 June 2003; accepted 10 August 2003
Abstract This paper discusses the gains from liberalizing foreign direct investment (FDI) in a two-country setting with endogenous market structure. We investigate two different scenarios. In the first scenario, headquarters costs are large in the foreign country so that the industry is located in the domestic country only. In this case, multinational and national firms may coexist and market concentration may make FDI welfare improving for the foreign country and welfare reducing for the domestic country. In the second scenario, headquarters costs are symmetric and firms will be located in both countries. Here, profitable FDI activities lead to mutual welfare gains, irrespective of market structure effects. D 2003 Elsevier B.V. All rights reserved. Keywords: Foreign direct investment; Multinational firms; Imperfect competition; Welfare JEL classification: F12; F15
1. Introduction This study examines the welfare and the market structure effects of foreign direct investment (FDI) in a two-country model in which the choice of FDI and market entry are both endogenous. The paper is motivated by the fact that these effects of FDI are thus far relatively poorly understood in comparison with the effects of trade. In recent years, however, economic integration has increasingly taken the form of FDI, rather than that of
* Corresponding author. Tel.: +49-431-880-1448; fax: +49-431-880-3150. E-mail address: [email protected] (F. Sta¨hler). 0022-1996/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.jinteco.2003.08.008
546
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
trade. FDI has been growing at a high rate, and trade barriers have fallen extensively at the same time. The World Investment Report of the United Nations estimates sales of foreign affiliates at USD 18.5 trillions in the year 2001, whereas exports of goods and non-factor services amounted to USD 7.4 trillions in the same period.1 In other words, the value of aggregate production of multinational firms in host countries nowadays outweighs aggregate exports. This paper focuses on the case where FDI and exports are perfect substitutes as often found in the empirical literature.2 The model we employ is similar to the setting employed in Horstmann and Markusen (1992), but with free entry/exit.3 Market entry is considered by Markusen and Venables (1998, 2000) but—given the complexity of their models—Markusen and Venables rely on numerical solutions.4 In this paper we develop a way of solving analytically models with multinational firms, international trade and free entry/exit. We employ a model of imperfect competition in which firms may enter the market either as a national firm or as a multinational firm. Both types of firms face the same production costs in serving the domestic market. However, in serving the foreign market, the national firm faces additional trade costs, while the multinational firm faces additional fixed costs in setting up a production plant abroad. It is this trade-off between higher marginal costs and higher fixed costs of production which determines the profitability of FDI. Under the hypothesis that the parameter values are such that FDI is profitable, we study how FDI changes market structure and welfare in both countries by comparing a regime under which FDI is prohibited with a regime under which FDI is allowed. Therefore, FDI liberalization is treated as a policy shift, and the condition that no firm gains unilaterally by switching its type and no firm has an incentive to enter or to leave the market then determines the equilibrium market structure. Following the eclectic paradigm of Dunning (1977), firms are induced to invest abroad if, in addition to location advantages (i.e. lower production costs), they have ownership advantages (i.e. patent, blueprint, or trade secret), and internalization advantages (i.e. strategic reasons to exploit its ownership advantage internally rather than licensing it to a foreign firm). Thus, we assume that firms need specific skills to be able to run their headquarters. We measure this requirement by the necessary labor input. Since the labor input needed to set up the headquarters of an oligopolistic firm may differ substantially from country to country, we explore the economic implications of FDI by using two alternative scenarios as reference points.
1
See UNCTAD (2002), Table I.1, p. 4. See Brainard (1997); Blonigen (2001); Markusen (1998); Markusen and Maskus (2002). Literature distinguishes between horizontal FDI (Markusen, 1984; Horstmann and Markusen, 1992; Brainard, 1993; Markusen and Venables, 1998, 2000), which is undertaken to place production closer to foreign markets; and vertical FDI (Helpman, 1984; Helpman and Krugman, 1985), which is undertaken to exploit lower production costs in order to serve both the domestic and the foreign markets. This paper assumes horizontal FDI which seems to be dominant according to empirical evidence because most of the worldwide FDI occurs between countries that are similar in size, relative endowments and per capita incomes. 3 The number of firms in Horstmann and Markusen (1992) is limited to one or zero in each country. Free entry is not considered (see also Markusen, 2002, Ch. 3). 4 A recent manuscript by Markusen (2002) summarizes the results so far achieved in the literature. 2
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
547
In the first scenario, the specific skills to run the headquarters are asymmetrically distributed across both countries so that input requirements for headquarters services are substantially larger in one country, say the foreign country. In this case, firms cannot be run in the foreign country without suffering losses so that market entry will occur only in the domestic country. In the FDI-allowed regime there are, depending on parameter values, three types of equilibria. If the cost of establishing a foreign plant is sufficiently high, allowing FDI has no effect: all firms remain national and serve the foreign market by exports as in the FDI-prohibited regime. If the cost is sufficiently low, all firms are multinational and serve each market through local production in the FDI-allowed regime. For intermediate values of setting up a foreign plant, national and multinational firms coexist. If FDI is profitable, competition becomes tougher in the foreign country with the emergence of multinational firms, because these firms have a marginal cost advantage relative to national firms. The intermediate case of coexistence of national and multinational firms then warrants that the overall number of (national and multinational) firms is reduced. Due to this change in market structure, domestic welfare declines whereas foreign welfare improves. If FDI costs are low enough, only multinational firms remain profitable. In that case, the foreign country gains from FDI but the domestic welfare change depends (inversely) on the size of FDI costs. In the second scenario, the skills to run headquarters are symmetrically distributed across countries. In this case, firms are established in both countries. If FDI is prohibited, our model coincides with the reciprocal dumping model of Brander and Krugman (1983) in which each country hosts the same number of national firms. Conversely, if FDI is allowed and profitable, we show that FDI does not allow any national firm to survive, and all firms are multinational in equilibrium. Also in this case market concentration may (but does not necessarily) occur because multinational firms have to carry larger fixed costs and their size must be larger than that of national firms. Irrespective of the impact of FDI on market entry, we find that FDI is unambiguously mutually welfare improving. Therefore, this result emphasizes the negative role of trade costs on welfare, as often found in the literature. As intra-industry trade compared to autarky, so intra-industry FDI compared to intra-industry trade is mutually welfare enhancing with endogenous market structures. Table 1 summarizes the main assumptions and results of the paper.
Table 1 Summary of assumptions and results Cost structure Location FDI-prohibited regime FDI-allowed regime Coexistence of both types of firms Impact on domestic welfare Impact on foreign welfare
Headquarters costs are large in the foreign country Headquarters are located in the domestic country only Inter-industry trade FDI and inter-industry trade Possible
Headquarters costs are symmetric
Depends on market structure
Positive
Positive
Positive
Headquarters are located in both countries Intra-industry trade FDI and intra-industry trade Impossible
548
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
The remaining parts of the paper are structured as follows. Section 2 presents the framework of the model. Section 3 explores the effects of FDI liberalization on market structure and welfare, when headquarters are located only in the domestic country. Section 4 discusses the implications of FDI liberalization, when headquarters are located in both countries. Section 5 concludes the paper.
2. The model The model used in this paper builds on the model employed by Horstmann and Markusen (1992). There are two countries, a domestic country d and a foreign country f, and two goods, X and Y. Each country is endowed with a certain amount of a productive factor, such as labor L. In both countries, the homogeneous good Y is produced under perfect competition using only the labor input, so that LY = Y, where the superscript denotes the sector in which labor is used. The price of Y equals the return on L, and Y is the numeraire of the model. Exporting Y is assumed to incur no trade costs. The industry producing X is characterized by imperfect competition and Cournot behavior. Firms produce a homogeneous good, which is sold at home and exported or produced abroad. The technology is characterized by constant marginal cost c, and operating headquarters and a plant in country i involves a fixed cost Fi. Shipping goods between countries costs t per unit. All costs are measured in units of labor required. The consumer behavior in each country is determined by the linear quadratic utility function Ui = U(Xi, Yi) = aXi 0.5bXi2 + Yi. Given the aggregate budget resource constraint Li + Spi = piXi + Yi, ia{d, f}, where Spi denotes the aggregate profits of the oligopolistic industry and pi the price of Xi in terms of the numeraire, maximization of Ui subject to the resource constraint yields the inverse income inelastic demand function pi ¼ a bXi ;
iafd; f g;
a c t > 0:
ð1Þ
Markets are segmented so that each firm is able to regard each country as a separate market. Thus, each firm maximizes its profit with respect to both sales in the foreign market and production for the domestic market, and chooses the profit-maximizing quantity for each country separately. The profits of a national firm operating in d are pn ¼ ðpd cÞxnd þ ½pf ðc þ tÞ xnf Fd ;
ð2Þ
where xdn and xnf represent production for the domestic and the foreign market, respectively, and Fd is the fixed set-up cost for home production. This fixed cost comprises the investment cost for headquarters and for one production plant in the domestic country. A similar expression holds for foreign national firms which have to carry fixed costs of size Ff. The profit of a (domestic) multinational firm is m pm ¼ ðpd cÞxm d þ ðpf cÞxf Fd G;
ð3Þ
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
549
where xdm and xdm represent production for the domestic and the foreign market, respectively, and G is the fixed FDI cost needed to start the production process abroad (i.e. the plant-specific fixed cost). In summary, multinational firms are able to produce for the foreign market without incurring trade costs t, but they have to bear the additional fixed cost of operating a plant abroad. Reasonably, the cost of setting up a production plant abroad should be less than the cost of starting production at home, i.e. G < Fi. Furthermore, we assume that a firm cannot set up a plant abroad without first setting up one at home. The sequence of decisions is as follows: in the first stage, firms decide on market entry as a national firm if FDI is banned. If FDI is allowed, they decide simultaneously on market entry and the type of firm they wish to be, national or multinational. In the second stage, all active firms compete in the Cournot fashion. For convenience, we have relegated all the details concerning the optimal production patterns of individual firms to Appendix A.1.
3. FDI and inter-industry trade This section assumes that the fixed costs to set up the headquarters of a firm are substantially different between the two countries. In other words, we assume that the specific skills needed to set up headquarters are available in the domestic country and they are scarce in the foreign country so that the oligopolistic industry can exist only in the domestic country. More formally, Assumption 1 2t 2 Ff zFd þ b
rffiffiffiffiffiffiffiffiffi! bFd 1 ; s
G < Fd ;
t2 s < Fd < ; b b
specifies this asymmetry where s=(a c)2+(a c t)2 measures the size of the markets for an exporting (national) firm. This implies that headquarters are located in the domestic country only (see Appendix A.1.3). Furthermore, G < Fd guarantees that national firms will always export to the foreign market if they coexist with multinational firms (see Appendix A.3, Fd < s/b guarantees that the market exists and Fd>t2/b guarantees that competition by national firms is not so intense that FDI can never become dominant (see Appendix A.2). In the FDI-prohibited equilibrium, each firm maximizes profit (2) given the equilibrium output of its rivals, and entry occurs until the excess profit of the marginal firm is driven to zero. Since domestic firms are symmetric, they earn zero profit in equilibrium. This condition allows us to determine the equilibrium number of national firms, which under the FDI-prohibited regime (denoted by a star) is equal to rffiffiffiffiffiffiffiffiffi s N* ¼ 1; ð4Þ bFd
550
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
and the equilibrium prices in both countries a þ cN * ¼ c þ ða cÞ pd* ¼ N* þ 1
rffiffiffiffiffiffiffiffiffi bFd ; s
ð5Þ
rffiffiffiffiffiffiffiffiffi a þ ðc þ tÞN * bFd ¼ c þ t þ ða c tÞ p*f ¼ : N* þ 1 s Given Assumption 1, N*>0. Note that the higher Fd, the lower the number of firms and the higher the price levels in equilibrium. What is the impact of FDI liberalization on market structure and welfare? In order to investigate this issue, we solve for the first order conditions of the national and multinational firms under the hypothesis that the parameters of the model make FDI profitable.5 The equilibrium number of national firms (N) and multinational firms (M) can be determined by solving two zero profit conditions f ðN ; M Þ ¼
gðN ; M Þ ¼
ða cÞ2 bðN þ M þ 1Þ
2
þ
2
þ
ða cÞ2 bðN þ M þ 1Þ
½a c ðM þ 1Þt 2
Fd ¼ 0;
ð6Þ
ðFd þ GÞ ¼ 0;
ð7Þ
bðN þ M þ 1Þ2 ða c þ NtÞ2 bðN þ M þ 1Þ2
where f denotes the maximized profit of national firms, and g denotes the maximized profit of multinational firms. In both Eqs. (6) and (7), the first (second) term gives the revenues net of variable costs in the domestic (foreign) market. In order to examine the impact of ¯ denote the FDI on the market structure, the definition of two critical values is useful. Let G lowest fixed cost that makes multinational production unprofitable, and let G denote the largest fixed cost that makes national production unprofitable (see Appendix A.2). By using G G , we can demonstrate Proposition 1. and Proposition 1 . Three different types of a unique equilibrium are possible when headquarters are located in the domestic country only and FDI is allowed. If Ga]G , l[, only national firms are active. If Ga[0, G [, only multinational firms are active. If ¯ Ga[G , G ], national and multinational firms coexist. 5 ¯ Appendix A.2 shows that the nondegeneracy condition G >G is satisfied for some parameter values, so that the intermediate range can emerge as an equilibrium. Additionally, it shows that the zero profit conditions f (N, M) = 0 and g(N, M) = 0 have a negative slope in the (N, M) space, and that the resulting Jacobian determinant is unambiguously positive, which implies that the equilibrium is unique.
Proof . See Appendix A.2.
5
Obviously, multinational firms will be established only in the domestic country.
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
551
In summary, Proposition 1 demonstrates that national and multinational firms can coexist.6 From the viewpoint of a single firm, optimal production for the home market does not depend on its type. But a multinational firm produces more for the foreign market than an exporting national firm to cover its larger fixed costs (see Appendix A.1.2). Hence, prices will decline in the foreign country, and this effect makes subsequent FDI activities less attractive. In other words, once FDI is liberalized, the attractiveness of FDI declines, such that in equilibrium not all active firms may become multinational. More specifically, in the case of coexistence, the equilibrium values of N and M, denoted by e, are: a c bG þ t 2 Ne ¼ ð8Þ pffiffiffi 1 ; t g Me ¼
ac bG t 2 1 pffiffiffi 1; g t
ð9Þ
2tða cÞ pffiffiffi 1; g
ð10Þ
Ne þ Me ¼
where gw 4bFdt2 (bG t2)2>0.7 Appendix A.2 demonstrates that the equilibrium values are positive in case of coexistence. Then, the four first order conditions and the two inverse demand functions Eq. (1) determine the equilibrium prices ped
pffiffiffi g a þ cðN e þ M e Þ ; ¼cþ ¼ Ne þ Me þ 1 2t
pef ¼
a þ ðc þ tÞN e þ cM e bG þ t 2 ¼cþ : e e N þM þ1 2t
ð11Þ
Expression (11) demonstrates a positive relationship between pde and t, and a negative relationship between pde and G. Conversely, pfe and t are negatively related, while pfe and G are positively related. These peculiar relationships can be understood only if one looks at the implication of the FDI cost G on market structure. As the FDI cost declines, more multinational firms having a comparative marginal cost advantage force foreign prices to decline. This effect reduces the total number of firms in equilibrium and, as a result, the domestic price increases. The opposite analysis is valid if t declines. Expression (11) shows also that the domestic price is affected by the fixed cost Fd which is necessary to start the production process (through g), while the foreign price does not change with Fd. An increase in fixed cost warrants firms to produce more to cover the fixed cost, and multinational firms have an advantage in size compared to national firms. The invariance of the foreign price with respect to Fd shows that the negative effect on the 6
Appendix A.3 proves that national firms export and thus compete with multinational firms in the foreign market in case of coexistence if Assumption 1 holds. 7 Appendix A.2 shows that g is unambiguously positive in the case of coexistence.
552
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
total number firms is just compensated by the relative increase in the number of multinational firms. Indeed, Eqs. (8) – (10) show that both the number of national firms and the aggregate number of firms decline, while the number of multinational firms increases if Fd rises. As Proposition 1 has shown, it might happen that only one type of firm, either national or multinational, is active in the market. If f (0, M**) V 0 holds for the profits of national firms, where the double star denotes the equilibrium under the condition Ga[0, G [, market entry is not profitable even for a single national firm. We then derive the following condition for an equilibrium where only multinational firms are active, which is consistent with Proposition 1: f ð0; M**ÞV0ZbGVt
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2bFd t 2 :
ð12Þ
In other words, if the FDI cost G (the trade cost t) is sufficiently low (high) all national firms exit from the market, once FDI is liberalized. It is also interesting to note that the larger is Fd, the greater is the likelihood that multinational firms alone will serve the markets. The reason is that the larger is the fixed cost to start the production process at home, the larger firms have to be in order to avoid losses. If Eq. (12) holds, the equilibrium values for prices and number of firms are:
pd** ¼ p** ¼ f
a þ cM** ¼cþ M ** þ 1
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi bðFd þ GÞ ; 2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 M ** ¼ ða cÞ 1: bðFd þ GÞ
ð13Þ
ð14Þ
If, conversely, the FDI cost G (trade cost t) is sufficiently high (low) such that Ga[G ,l[, then multinational firms are not profitable. This occurs if g(N*, 0) V 0 holds for the profits of multinational firms. We derive the following condition for an equilibrium outcome where only national firms are active in a FDI-allowed regime, which is also consistent with Proposition 1: rffiffiffiffiffiffiffiffiffi bFd gðN *; 0ÞV0ZbGz2tða c tÞ þ t2 : s
ð15Þ
Namely, the greater is the fixed cost to undertake FDI and/or the lower is the trade cost, the greater is the likelihood that only national firms are active in the market. Having identified the conditions under which firms are active in the domestic and foreign markets, investigating the impact of FDI liberalization on market structure is straightforward. Lemmas 1 and 2 summarize the results.
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
553
Lemma 1. Under parameter values such that both types of firms coexist in a regime with FDI, the impact of allowing FDI is to reduce the number of firms. Proof. The total number of active firms is given by Ne + Me. The impact of FDI on market structure can be easily examined by using Eqs. (4) and (10). In fact, Ne + Me < N* if, and only if, sðbG t 2 Þ2 < 4ða c tÞ2 t 2 bFd :
ð16Þ
This expression is fulfilled if, and only if, condition (15) for the equilibrium where only national firms are active is violated. Since this is the case, it follows that allowing FDI leads to market concentration if national firms survive. 5 The intuition for this result can be summarized as follows. National and multinational firms sell the same amount in the domestic market. Given the zero profit condition, the net profit from exporting has to equal the net profit of setting up a foreign subsidiary. If the total number of firms rose after liberalization, a national firm would automatically earn less from exporting, since the foreign market is more crowded, and multinational firms have a marginal cost advantage. This implies that national firms would have to earn more at home, which is inconsistent with entry. Hence, in any coexistence outcome, FDI leads to exit. Lemma 2. Under parameter values such that only multinational firms exist in a regime with FDI, the impact of allowing FDI on the number of firms depends on the fixed costs of a national firm compared with those of a multinational firm: M ** < N *; if
Fd s < ; Fd þ G 2ða cÞ2
M ** > N *; if
Fd s > : Fd þ G 2ða cÞ2
Proof. The condition for M**bN* follows from Eqs. (4) and (14). The condition for the equilibrium with only multinational firms Eq. (12) can be re-arranged as follows: Fd ða cÞ2 þ ða c t tM **Þ2 z : Fd þ G 2ða cÞ2 Expression (17) can be consistent with both M**< N* and M**>N*.
ð17Þ 5
Lemma 2 maintains that there is scope for market concentration even if only multinational firms are active in the market. In particular, the number of active firms under the FDI regime is lower, if national firms are not active at the margin. However, we cannot rule out the possibility that M**>N*. If G is very small, the equilibrium number of multinational firms could be so large that the markets could support more multinational firms.
554
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
Table 2 Scenario 1—FDI and trade with headquarters located in the domestic country only
G Number of national firms Number of multinational firms Domestic price Foreign price Domestic output per firm Exports per national firm Foreign sales per multinational firm
Only national firms*
National and multinational firms**
FDI-prohibited 33
9 12 17
7.760 9.217 2.760
8.152 8.750 3.152
2.717
2.220 3.750
Only multinational firms*** 4.5
29
Only multinational firms*** 0.15
34
8.122 8.122 3.122
7.522 7.522 2.752
3.122
2.752
Parameters of the model: a = 100, b = 1, c = 5, t = 0.3c, Fd = 15, Ff z 19.369. * G > 0.693Fd or FDI is prohibited. ** 0.527Fd V G V 0.693Fd. *** G < 0.527Fd.
Table 2 shows the results of numerical simulations which are based on parameter values as indicated at the bottom of the table.8 In accordance with Proposition 1, the simulations indicate that the markets are served by national firms only if FDI costs are relatively large, i.e. G/Fd>69.3%, or if FDI is prohibited. If FDI is liberalized and 52.7% V G/Fd V 69.3%, coexistence between national and multinational firms occurs. Conversely, if G/Fd < 52.7%, only multinational firms would survive and serve both markets. Note that the overall number of firms declines with a reduction of G in the range in which G is moderately large so that an equilibrium with multinational firms only cannot emerge. Moreover, the number of multinational (national) firms is inversely (positively) related to G. This is the reason why, if G is sufficiently low, multinational firms become dominant, coexistence is impossible and the number of multinational firms may be even larger under FDI liberalization compared to FDI prohibition (see last column of Table 2). Note also that the foreign output of a multinational firm is larger than that of a national firm, because it has to cover larger fixed costs. The impact of FDI on market structure has an effect on the welfare of both the domestic and the foreign country, because the equilibrium prices depend on the equilibrium number of firms and production costs. Based on Lemmas 1 and 2, we prove the following Proposition 2.
8 Note that the fixed cost to start a business in the foreign country ( Ff) has only to be larger than 19.369 in order to prevent the location of headquarters if Fd = 15 (see Assumption 1). Thus, Ff /Fd z 1.29 is sufficient to guarantee that trade and FDI are of the inter-industry type here.
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
555
Proposition 2 . Under parameter values such that both types of firms coexist in a regime with FDI, the welfare of the domestic country declines when FDI is allowed. Under parameter values such that only multinational firms exist in a regime with FDI, the welfare of the domestic country goes the way of the number of firms, while the welfare of the foreign country always increases. Proof. Given the zero profit condition, welfare depends only on prices. Eqs. (5), (11) and (13) show that the domestic consumer is worse off if the total number of active firms under the FDI regime is smaller than the number of national firms under the trade regime. Lemma 1 shows that market concentration occurs if both types of firms coexist, Lemma 2 shows that market concentration may occur if only multinational firms exist. The proof of the foreign country’s welfare improvement can be given by contradiction. Suppose that pf given by Eq. (11) is larger than pf given by Eq. (5). Then, an increase in the foreign price occurs if, and only if, rffiffiffiffiffiffiffiffiffi bFd þ t2 : ð18Þ bG > 2tða c tÞ s However, if this inequality were true, multinational firms would not be active and only national firms would serve the market (see condition (15)). Hence, if coexistence can occur, the foreign price declines when FDI is allowed. This finding also holds for the equilibrium where only multinational firms are active because the foreign price declines unambiguously if G falls (see Eq. (13)). 5 Proposition 2 demonstrates that the domestic country is worse off after the foreign country liberalizes FDI activities, if national and multinational firms coexist. As multinational firms are able to offer goods at lower marginal costs, prices decline in the foreign markets and the overall number of firms is reduced. The reduction in the number of firms which compete also on the domestic market leads to an increase in domestic prices and hence domestic welfare declines in an equilibrium with coexistence of national and multinational firms. As the domestic country is worse off in the case of coexistence, it is obvious that the domestic country can only become better off if multinational firms alone are active in the market. To obtain a welfare gain, it is essential that the foreign market becomes much more profitable after FDI liberalization. If the increase in the profitability of the foreign market is substantial enough so that more multinationals are active than national firms have been active before, domestic welfare will improve. As the profitability of the foreign market depends crucially upon the size of fixed costs, we investigate how small the FDI costs G should be to ensure that the domestic country is not worse off. The critical amount of the plant-specific fixed costs can be computed by setting expression (14) larger than or equal to expression (4). Therefore, the domestic country is not worse off if, and only if, GVwFd where w ¼
ða cÞ2 ða c tÞ2 ða cÞ2 þ ða c tÞ2
:
ð19Þ
Only if G is small enough, does the domestic country gain from FDI, because the number of multinational firms would be larger than the number of firms under the
556
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
FDI-prohibited regime. However, it seems that G should be really small for a welfare enhancing effect to occur with FDI liberalization. In fact, consider a possible case where marginal costs are 5% (30%) of the reservation price and trade costs are 30% of the marginal costs. In this case, the critical parameter w would be equal to 0.024 (0.137). In other words, for the domestic consumer to be better off, the fixed costs of setting up a plant abroad should be lower than 2.4% (13.7%) of the fixed costs of starting the production process at home. These results are also confirmed by the numerical simulations presented in Table 2, which assumes that marginal costs are 5% of the reservation price. Under the parameter values reported in Table 2, the domestic price increases from 7.76 under the FDIprohibited regime to 8.152 if coexistence occurs with FDI liberalization. Conversely, if G is sufficiently low, only multinational firms would survive and the domestic price would either increase to 8.122 if G is 30% of Fd or decline to 7.522 if G is 1% of Fd. On the other hand, the foreign consumer is always better of with FDI liberalization, regardless of the impact on market structure.
4. FDI and intra-industry trade The previous section has shown the effects of FDI liberalization on market structure and welfare under the assumption that the headquarters of the oligopolistic industry are located only in the domestic country. This section assumes that the fixed costs to start a business are symmetric and hence low enough also in the foreign country: Assumption 2 Fd ¼ Ff ¼ F;
t2 s
Assumption 2 implies that headquarters are located in both countries. In particular, F < s/b guarantees that the markets exist and F >t2/b guarantees that exports occur (see Appendix A.1.3). If fixed costs are too low compared to trade costs, intra-industry trade would not occur because the presence of too many local firms would make exports not worthwhile. If FDI is banned, the FDI-prohibited regime is characterized by intra-industry trade, and the model coincides with the reciprocal dumping model of Brander (1981); Brander and Krugman (1983). They show that, with respect to autarky, trade is mutually welfare improving if the market structure is endogenous, as it induces firms to move down their average cost curves. If FDI is banned, we need to consider only the domestic country because the number of domestic and foreign national firms coincides. If more domestic national firms than foreign national firms were active, competition in the domestic country would be tougher than in the foreign country due to trade costs, and hence domestic firms would make less profits than foreign firms, so that the zero profit condition would be violated. In the intra-industry FDI-prohibited
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
557
regime, denoted by a tilde, the equilibrium number of national firms and the equilibrium price in both countries is, respectively, equal to ða cÞ þ ða c tÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi N˜ ¼ 1: 2bF t 2 and p˜ ¼ c þ
tþ
ð20Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2bF t 2 : 2
ð21Þ
We will now discuss the impact on market structure and welfare of allowing FDI under the hypothesis that the parameters of the model are such that FDI occurs in equilibrium. In this setting, multinational firms may run their headquarters in either country. But since countries are symmetric, it does not make any difference whether a certain firm locates its headquarters and a plant in the domestic country and an additional plant in the foreign country or vice versa. Therefore, we are able to determine the aggregate equilibrium number of multinational firms but we cannot say where their headquarters are located. The location of national firms, however, is sensitive, because their exports suffer from the trade cost disadvantage. We know that the number of national firms is identical across countries if they coexist with multinational firms, because multinational firms produce the same amount of output for each country, and the residual demand faced by national firms is identical across countries. Therefore, the number of national firms will coincide also with FDI for the same reason as in the reciprocal dumping model. Since all types of firms are possibly present in the market, we define N as the number of all national firms located in both countries and M as the number of all multinational firms. In other words, N/2 national firms are located in each country, but the headquarters location of M multinational firms cannot be determined. The solution of the standard profit maximization problems faced by the firms yields the equilibrium price pðN ; M Þ ¼
a þ cN þ 0:5tN þ cM ; N þM þ1
ð22Þ
and two zero profit conditions, /ðN ; M Þ ¼
cðN ; M Þ ¼
ða c þ 0:5tN Þ2 bðN þ M þ 1Þ
2
2ða c þ 0:5tN Þ2 bðN þ M þ 1Þ2
þ
ða c t 0:5tN tM Þ2 bðN þ M þ 1Þ2
F G ¼ 0;
F ¼ 0;
ð23Þ
ð24Þ
where / denotes the maximized profit of national firms, and c denotes the maximized profit of multinational firms. In Eq. (23), the first (second) term gives the revenues of national firm net of variable costs in the domestic (foreign) country. The first term in Eq. (24) collects the symmetric revenues of multinational firms net of variable costs in both markets. / and c would allow us to determine the equilibrium number of both national and multinational firms, respectively. However, the Jacobian determinant of / and c with
558
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
respect to N and M is zero, which implies that / and c are linearly dependent and no equilibrium solution with positive N and M exists. Consequently, the equilibrium in a symmetric Cournot oligopoly with entry can only be a corner solution, which leads to the conclusion that either only multinational firms or only national firms are in the market. Proposition 3. If the oligopolistic industry is located in both countries, firms are either all national or all multinational. Proof . See Appendix A.4.
5
In this setting, coexistence of both types is not possible. Proposition 3 demonstrates that national firms cannot survive if multinational production becomes profitable because multinational production affects competition in both countries and is a threat to national firms in both countries. Since the pre-liberalization equilibrium is represented by Eq. (20) with national firms making zero profits, FDI is profitable if, and only if, c(N˜, 0) z 0. The latter condition implies that multinationals may enter without making losses and would— as Proposition 3 has shown—dominate the market. Once FDI is liberalized, the condition for the equilibrium where only multinational firms are active is cðN˜ ; 0Þ ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2ða c þ 0:5t N˜ Þ2 F Gz0ZbGVt 2bF t 2 : 2 bðN˜ þ 1Þ
ð25Þ
Condition (25) coincides with Eq. (12) in the first scenario of asymmetric fixed costs Fi described in Section 3. Under condition (25), national firms do p notffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi survive andffi the equilibrium number of multinational firms, M**, is given by ða cÞ 2=ðbðF þ GÞÞ 1 (see Eq. (14)).9 With regard to the impact on the market structure, we find that M**< N˜ does not violate condition (25), so that industry concentration may also occur in the case of intra-industry FDI. But the fixed costs to set up a plant abroad could also be so small that the equilibrium number of firms rises. More importantly, the welfare results are unambiguous. Proposition 4. Although the impact on the market structure of the intra-industry FDIallowed regime in comparison with the intra-industry FDI-prohibited regime depends upon parameter values, FDI liberalization generates mutual welfare gains. Proof . Under FDI, the equilibrium price is rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi bðF þ GÞ : ð26Þ p** ¼ c þ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p** is less than p˜ if bGVt 2bF t 2 which coincides with condition (25) that multinational firms are active and dominant. 5 This is a remarkable result because it stresses the role of trade costs. Consider a firm switching from being a national to being a multinational firm after FDI has been liberalized. Then, a decline in the domestic price implies a decrease in the individual domestic output. Furthermore, the multinational firm has to bear the additional plant-specific fixed cost. From 9
Conversely, if /(0, M**) z 0, the market would be served by national firms only.
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
559
Table 3 Scenario 2—FDI and trade with headquarters located in both countries Only national firms* G Number of national firms Number of multinational firms Domestic price Foreign price Domestic output per firm Exports per national firm Foreign sales per multinational firm
FDI-prohibited 35 8.384 8.384 3.384 1.884
Only multinational firms** 6 28 8.240 8.240 3.240 3.240
Parameters of the model: a = 100, b = 1, c = 5, t = 0.3c, F = 15. * G > 0.527F or FDI is prohibited. ** G V 0.527F.
these profit-reducing effects, we would expect that prices rise. However, the foreign market has become so profitable that the increase in individual production for the foreign market is able to overcompensate these effects. Proposition 4 demonstrates that the profit-reducing effects are offset by vanishing trade costs. The numerical simulations reported in Table 3 confirm the analytical results. The markets are served by national firms only, if FDI is prohibited or if G/Fd>52.7%. Conversely, if G/Fd V 52.7%, only multinational firms survive with FDI liberalization. They serve both markets at lower prices because average costs have declined due to the lack of trade costs. Note also that the size of a multinational firm (6.48) is larger than that of a national firm (5.27), because it has to cover larger fixed costs.
5. Concluding remarks This paper has examined the impact of FDI on market structure and welfare. We have distinguished between two different scenarios concerning headquarters costs:
Headquarters costs are relatively large in the foreign country so that they are located in the domestic country only. In this case, trade and FDI are of the inter-industry type. Headquarters costs are symmetric in both countries. In that case, trade and FDI are of the intra-industry type. Given these two different cases, we have studied the impact of liberalizing FDI on market structure and welfare by comparing the FDI-allowed regime with the FDIprohibited regime. We have shown that, if the headquarters are located in the domestic country only, the FDI-allowed regime as compared with the FDI-prohibited regime leads to welfare losses for the domestic country, but to welfare gains for the foreign country when national and multinational firms are both active. National and multinational firms may coexist for moderately large FDI costs because FDI affects competition only in the foreign market. If only multinational firms emerge, the welfare of the foreign country improves, whereas the impact on the welfare of the domestic country depends on the size of FDI costs.
560
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
FDI liberalization also has an impact on market structure: if both firms coexist, the number of active firms decreases; if only multinational firms are active, their number may be smaller or larger than the number of national firms in the case of no FDI. In this latter case, we have shown that domestic welfare improves only if FDI increases the number of active firms. It must be noted, however, that an increase in the number of firms and—due to entry of multinational firms—an increase in aggregate fixed costs alone cannot be expected to improve welfare, but would imply an increase in prices. In order to improve domestic welfare, it is essential that multinational firms save large trade costs. In other words, the increase in the profitability of the foreign market must be sufficiently large so that the number of active firms increases. The results are different if the headquarters are located in both countries. As FDI alters competition in both markets, coexistence of both types of firms is not possible so that either only national or only multinational firms are active. Under parameter values that make FDI profitable, the FDI-allowed regime implies that national firms are completely replaced by multinational firms. In this case, we have shown that the FDI-allowed regime leads to mutual welfare gains in comparison with the FDI-prohibited regime. The reason is that consumers have to bear trade costs in the FDI-prohibited regime which are saved by multinational firms. Furthermore, multinational firms dominate national firms if and only if the additional fixed costs are sufficiently low and trade costs are sufficiently substantial so that average costs are lower with FDI. This case demonstrates that FDI, as compared with trade, leads to mutual welfare gains just as trade, as compared with autarky, leads to mutual welfare gains if the market structure is endogenous and firms’ headquarters may locate in either of the countries involved.
Acknowledgements We would like to thank Jonathan Eaton, Horst Raff, Gerald Willmann and two anonymous referees for valuable comments, and Wilfred Ethier, Jim Markusen and Anthony Venables for helpful suggestions. The opinions expressed herein are those of the authors and do not represent those of the European Central Bank. All errors are authors’ responsibility.
Appendix A A.1 . Output levels A.1.1 . Output levels under inter-industry trade without FDI sffiffiffiffiffiffi ac Fd xd* ¼ ¼ ða cÞ ; bðN * þ 1Þ sb sffiffiffiffiffiffi x*f ¼
act ¼ ða c tÞ bðN * þ 1Þ
Fd : sb
xd* denotes production for the domestic market and xf* denotes exports.
ðA:1Þ
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
561
A.1.2 . Output levels under inter-industry trade and FDI pffiffiffi g ac me ; xne ¼ ¼ x ¼ d d bðN e þ M e þ 1Þ 2bt xne f ¼
a c tðM e þ 1Þ bG t 2 ¼ ; bðN e þ M e þ 1Þ 2bt
xme f ¼
a c þ tN e bG þ t 2 ¼ : bðN e þ M e þ 1Þ 2bt
ðA:2Þ
me xdne(xne f ) denotes domestic production (exports) of national firms, xf denotes production of multinational firms for each market. Note that the size of the multinational firm is larger ne than the size of the national firm (xme f xf = t/b), because the former has to bear larger ne aggregate fixed costs. Note also that xf >0 requires bG>t2 which is guaranteed by Assumption 1 (see Appendix A.2).
A.1.3 . Output levels under intra-industry trade without FDI The equilibrium output levels per firm are xdd ¼
a c þ tnf ; bðnd þ nf þ 1Þ
xdf ¼
a c tðnf þ 1Þ ; bðnd þ nf þ 1Þ
xff ¼
a c þ tnd ; bðnd þ nf þ 1Þ
xfd ¼
a c tðnd þ 1Þ ; bðnd þ nf þ 1Þ
ðA:3Þ
where subscripts (superscripts) denote the destination of production (location of the firm). nd(nf) denotes the number of domestic (foreign) national firms. With asymmetric fixed costs, these output levels lead to maximized profits of zd ¼
zf ¼
ða c þ tnd Þ2 þ ða c tðnd þ 1ÞÞ2 bðnd þ nf þ 1Þ2 ða c þ tnf Þ2 þ ða c tðnf þ 1ÞÞ2 bðnd þ nf þ 1Þ2
Fd ;
Ff :
ðA:4Þ
qffiffiffiffiffiffi 2 bFd Under the condition that Ff zFd þ 2tb 1 , zd = zf = 0 implies that nf V 0 so s that no foreign firm will enter the market. If fixed cost are symmetric, i.e. Fd = Ff = F, output levels are pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2bF t 2 þ t a c þ 0:5N˜ t ; x˜ d ¼ ¼ 2b bðN˜ þ 1Þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2bF t 2 t a c ð0:5N˜ þ 1Þt x˜ f ¼ : ðA:5Þ ¼ ˜ 2b bðN þ 1Þ Note that F >t2/b (see Assumption 2) guarantees that exports x˜f are positive.
562
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
A.1.4 . Output levels under intra-industry trade and FDI m xnd ¼ xm d ¼ xf ¼
a c þ 0:5tN ; bðN þ M þ 1Þ
xnf ¼
a c t 0:5tN tM : bðN þ M þ 1Þ
ðA:6Þ
xdn(xnf) denotes production for the home market (exports) of national firms, and xdm(xmf ) denotes production of a multinational firm for the home market (host market of FDI). A.1.5 . Output levels under FDI without trade ac ¼ xd** ¼ xf** ¼ bðM ** þ 1Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Fd þ G : 2b
ðA:7Þ
x**d (x** f ) denotes production of a multinational firm for the home market (host market of FDI). A.2 . Proof of Proposition 1
G (and N ) is determined by setting f (N, 0) = g(N, 0) = 0, whereas G (and M) is determined by setting f (0, M) = g(0, M) = 0: rffiffiffiffiffiffiffiffiffi! t bFd t þ 2ða c tÞ G ¼ ; ðA:8Þ b s t pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi 2bFd t : G¼ b
ðA:9Þ
2 Note that Fd>t2/b (see Assumption 1) guarantees that G>G >t /b so that national firms export to the foreign market (see Eq. (A.2)). If Fd were too small, a large number of national firms would enter the markets and FDI would never be able to become dominant, irrespective of G, a case we are not interested in. The minimum Fd guarantees that the profitability of multinational production depends on the FDI cost G. Only national firms will emerge if G>G, because G is determined from Eqs. (6) and (7) by setting M equal to zero. Any higher G would only deepen the cost disadvantage of multinational firms. By contrast, only multinational firms will emerge if G < G , because G is determined from Eqs. (6) and (7) by setting N equal to zero. Any lower G would only raise the profitability of multinational firms. If, however, G > G , a range of G exists in which a mixed equilibrium occurs, i.e. an equilibrium in which national and multinational firms coexist. For the mixed equilibrium proof, define Dw(b/t)(G G ). If D > 0, a mixed equilibrium may exist. First, let us explore the behavior of the D-function in the relevant range. We observe that D = 0 if Fd = s/b. However, this level of fixed costs implies that the market does not exist, because even a single firm is not able to recoup its fixed costs (see Assumption 2 and Eq. (4)). Hence, the relevant fixed costs must be below this level. If D decreases with Fd in the relevant range, we have shown that D is positive, since it reaches zero at the border where
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
563
the market does not exist under the FDI-prohibited regime. Differentiation yields ! BD act 1 ¼ b pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ðA:10Þ BFd sbFd 2bFd t 2 This term is negative if t > bFd[2(a c) t]/[a c t]2, which is always fulfilled. Hence, a mixed equilibrium may exist. Uniqueness is proven if the Jacobian determinant has an unambiguous sign for the equilibrium values of N and M. Differentiation of the zero profit conditions yields Bf 2ðða cÞ2 þ ða c ðM þ 1ÞtÞ2 Þ ¼ < 0; BN bðN þ M þ 1Þ3
ðA:11Þ
Bf Bg 2½ða cÞ2 þ ða c þ NtÞða c ðM þ 1ÞtÞ ¼ ¼ < 0; BM BN bðN þ M þ 1Þ3
ðA:12Þ
Bg 2ðða cÞ2 þ ða c þ NtÞ2 Þ ¼ < 0: BM bðN þ M þ 1Þ3
ðA:13Þ
For the negative signs of all derivatives see also Eq. (A.2). The Jacobian is then AJ A ¼
Bf Bg Bf Bg 4ða cÞ2 t 2 g2 ¼ 2 ¼ >0 BN BM BM BN b ðN þ M þ 1Þ4 ½2bða cÞt 2
ðA:14Þ
where gw 4bFdt2 (bG t2)2>0 because G < G implies that g must be larger than 4bFdt2 (a c)2/s>0. Note also that Ne>0 if, and only if, G < G, while Me>0 if, and only if, G>G satisfying, therefore, the conditions for coexistence shown in Proposition 1. Expression (A.14) shows that the sign of the Jacobian is unambiguously positive. Note that this property holds also for the corner solutions characterized by f (0, M) = g(0, M) = 0 or f (N, 0) = g(N, 0) = 0, and hence proves that the equilibrium is unique. The proof can be given by contradiction. Suppose that multiple mixed equilibria exist. In this case both the f- and the g-function should intersect at least twice. However, both the f- and the g-function are twice continuously differentiable and have a negative slope. If they intersect twice, the second intersection should imply a Jacobian with an opposite sign. However, the Jacobian is always positive. Hence, any mixed equilibrium must be unique. 5 A.3 . Positive exports of national firms The proof for positive exports will be given by contradiction. Suppose that national firms serve only the domestic market. Output levels per firm are then me xne d ¼ xd ¼
bðN e
ac ; þ M e þ 1Þ
xne f ¼ 0;
xme f ¼
ac ; bðM e þ 1Þ
ðA:15Þ
564
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
leading to the following zero profit conditions: zne ¼
ða cÞ2 bðN e þ M e þ 1Þ2
Fd ¼ 0
ðA:16Þ
and zme ¼
ða cÞ2 bðN e þ M e þ 1Þ2
þ
ða cÞ2 bðM e þ 1Þ2
Fd G ¼ 0:
ðA:17Þ
Eqs. (A.16) and (A.17) imply that ac M e ¼ pffiffiffiffiffiffiffi 1 bG
ðA:18Þ
ac 1 1 N e ¼ pffiffiffi pffiffiffiffiffi pffiffiffiffi : Fd b G
ðA:19Þ
and
Clearly, N e>0 if, and only if, G >Fd, which contradicts the reasonable Assumption 1 that the costs of setting up headquarters and a plant are less than the costs of setting up an additional plant. 5 A.4 . Proof of Proposition 3 Proposition 3 claims that the Jacobian of Eqs. (23) and (24) is zero. Differentiation of Eqs. (23) and (24) yields B/ ½ða cÞ þ ða c tÞ tM 2 ¼ < 0; BN bðN þ M þ 1Þ3
ðA:20Þ
B/ Bc 2ða c þ 0:5tN Þ½ða cÞ þ ða c tÞ tM ¼ ¼ < 0; BM BM bðN þ M þ 1Þ3
ðA:21Þ
Bc 4ða c þ 0:5tN Þ2 ¼ < 0: BM bðN þ M þ 1Þ3
ðA:22Þ
By using Eqs. (A.20) –(A.22), we find that
AJ A ¼
B/ Bc B/ Bc ¼ 0: BN BM BM BN
ðA:23Þ 5
R.A. De Santis, F. Sta¨hler / Journal of International Economics 64 (2004) 545–565
565
References Blonigen, B.A., 2001. In search of substitution between foreign production and exports. Journal of International Economics 53, 81 – 104. Brainard, S.L., 1993. A simple theory of multinational corporations and trade with a trade-off between proximity and concentration. NBER Working Paper 4269. Brainard, S.L., 1997. An empirical assessment of the proximity-concentration trade-off between multinational sales and trade. American Economic Review 87, 520 – 540. Brander, J.A., 1981. Intra-industry trade in identical commodities. Journal of International Economics 11, 1 – 14. Brander, J.A., Krugman, P.R., 1983. A ‘‘reciprocal dumping’’ model of international trade. Journal of International Economics 15, 313 – 323. Dunning, J.H., 1977. Trade, location of economic activity and MNE: a search for an eclectic approach. In: Ohlin, B., Hesselborn, P.O., Wijkman, P.M. (Eds.), The International Allocation of Economic Activity. Macmillan, London, pp. 395 – 418. Helpman, E., 1984. A simple theory of trade with multinational corporations. Journal of Political Economy 92, 451 – 571. Helpman, E., Krugman, P.R., 1985. Market Structure and Foreign Trade. MIT Press, Cambridge. Horstmann, I.J., Markusen, J.R., 1992. Endogenous market structures in international trade (natura facit saltum). Journal of International Economics 32, 109 – 129. Markusen, J.R., 1984. Multinationals, multi-plant economies, and the gains from trade. Journal of International Economics 16, 205 – 226. Markusen, J.R., 1998. Multinational firms, location and trade. World Economy 21, 733 – 756. Markusen, J.R., 2002. Multinational Firms and the Theory of International Trade. The MIT Press, Cambridge. Markusen, J.R., Maskus, K.E., 2002. Discriminating among alternative theories of the multinational enterprise. Review of International Economics 10, 694 – 707. Markusen, J.R., Venables, A.J., 1998. Multinational firms and the new trade theory. Journal of International Economics 46, 183 – 203. Markusen, J.R., Venables, A.J., 2000. The theory of endowment, intra-industry, and multinational trade. Journal of International Economics 52, 209 – 234. UNCTAD, 2002. World Investment Report 2002. Transnational Corporations and Export Competitiveness, New York.