HD28 /^' ^':K/ HD28 .M414 ALFRED P. WORKING PAPER SLOAN SCHOOL OF MANAGEMENT A Simple Theory of Multinational Corporations and Trade with a Trade-off between Proximity and Concentration S. Lael Brainard Massachusetts Institute of Technology WP#3492-92-EFA December 1992 Revised MASSACHUSETTS INSTITUTE OF TECHNOLOGY 50 MEMORIAL DRIVE CAMBRIDGE, MASSACHUSETTS 02139 A Simple Theory of Multinational Corporations and Trade with a Trade-off between Proximity and Concentration S. Lael Brainard Massachusetts Institute of Technology WP#3492-92-EFA December 1992 Revised JAU 1 3 1993 A Simple Theory of Multinational Corporations and Trade with a Trade-off between Proximity and Concentration Abstract This pap>er develops a two-sector, two-country model, where firms in a differentiated products sector choose between exporting and multinational expansion as alternative modes of foreign market penetration, based on a tradeoff between proximity and concentration advantages. production, with increasing returns at The differentiated sector is characterised the corporate level associated with some by multi-stage activity such as R&D, scale economies at the plant level, and a variable transport cost that rises with distance. A pure multinational equilibrium, where two-way horizontal expansion across borders completely supplants two-way trade in differentiated products, is possible even in the absence of factor proportion differences. It is more likely the greater are transport costs relative to fixed plant costs, The model and the greater are increasing returns also establishes conditions for a * First draft ** I am mixed equilibrium, in at the corporate level relative to the plant level. which national and multinational firms coexist. August 1992. grateful to Gene Grossman, Paul Krugman and Julio National Science Foundation for support under grant SES-9111649. Rotemberg for helpful discussions, and to the Introduction I. (MNEs) Multinational enterprises Alan international trade. Rugman (1988) estimates that the largest 500 multinationals control over one-half of global trade flows, and one-fifth of global network of overseas in are a substantial and imperfectly understood part of the landscape of GDP. For affiliates, affiliate sales tend to countries such as the swamp exjxjrt flows. markets such as Brazil and the EC, local sales by U.S. -owned US, which have a well-developed DeAnne affiliates are five Julius (1990) estimates that times the level of imports from the U.S. It is also striking that a large and growing share of foreign direct investment involves industrialized Between 1961 and nations as both the source and destination markets, rather than flowing from North to South. 1988, over half of all direct investment outflows generated countries; this share had risen to nearly 70 percent by 1988 by G-5 countries were absorbed by other G-5 absorbed by developed market nations more broadly was 81 percent by the (1992)). These trends suggest a large and growing considerations of market access. phenomenon. Indeed, that The share of (Julius (1990)). 1980s (Krugman and Graham late of two-way activity across borders motivated by level But multinational expansion motivated by market access in her study direct investment is not just a recent of American multinationals from the 1700s to 1914, Mira Wilkins concludes market access was the primary motivation for multinational expansion: TTiese early international businesses established their foreign stakes foremost in the areas around the world where there were customers as - we have in Eurojje seen - first and and Canada primarily. where there was demand for their offerings - in the wealthiest The businesses described ... went abroad to sell and to make profits through sales in the countries in which they made investments; if they went into manufacturing, refining, or real estate, it was with marketing considerations paramount. They did not go abroad for supplies or to They made their principal investments nations in the world. produce abroad for sale The implications of in the United States. (Wilkins, 1970, p. these trends for trade and for policy have not been fully explored, partly because research on trade and on multinationals proceeds along separate tracks. recent years in explaining the Meanwhile, the managerial 1966) phenomenon of literature Trade theory has made great intraindustry trade, and in modelling imperfect competition'. on multinationals has evolved to incorporate recent of the boundaries of the firm.^ However, there have been only a few attempts characteristics that have been adduced strides in to explain intraindustry trade 1 - advances in the theory at synthesis, even though the imperfect competition, scale economies. and differentiated products - are precisely the characteristics that are associated with the internalization and ownership advantages believed to motivate the formation of multinationals. This f>aper develops a two-sector, two-country model, where firms in a differentiated products sector choose between exporting and cross-border expansion as alternative modes of foreign market penetration. differentiated sector is characterized that can be spread by increasing returns among any number of production at the facilities firm level due to some input, such as The R&D, with undiminished value, scale economies at the plant level, such that concentrating production lowers imit costs, and a variable transport cost that rises with distance. The transport cost stands in for a variety of advantages from locating close according to Wilkins, have been a chief motivation for expansion of 'investments abroad were made to save customer service and to avoid damage US to the customer, multinationals historically: on transportation costs and warehousing expenses, in shipping a perishable product which, (p. 67)." The to obtain superior decision whether to expand abroad via trade or via investment hinges on a trade-off between these proximity advantages and scale advantages from concentrating production in a single location. In this model, a pure multinational equilibrium, supplants merchandise trade in the differentiated sector, differences, when proximity advantages where two-way intraindustry investment completely is possible even in the absence of factor proportion provides a rationale for two-way horizontal expansion across borders, which rationale for vertical expansion based of two-way investment in on factor endowment automobile assembly for example US investments in semiconductor assembly and - facilities in contrast to factor-based theories This equilibrium is more crowds out trade differentials. between the of multinational test facilities in With a simple two-stage production process, trade in invisibles completely Thus, are strong relative to concentration advantages. in there is US is distinct this from the model traditional This model formalizes an explanation and the activities, EC (eg, which are Volkswagen and GM), better suited to explaining Southeast Asia. a pure multinational equilibrium in which two-way goods, even in the presence of symmetric factor proportions. likely the greater are transport costs relative to plant-level scale greater are increasing returns at the corporate level relative to those at the plant level. economies, and the Under circumstances, a symmetric equilibrium with purely national firms obtains, in which there is the opposite balanced two-way trade in differentiated products. economies and of returns at the For intermediate ranges of transport costs corporate level relative to the plant level, there multinational and national firms coexist in the differentiated sector. two-way is a mixed equilibrium in which In a symmetric mixed equilibrium, there is trade in both differentiated products and invisibles. The incorporation of a third stage of production, such as sales, two-way, intrafirm, intraindustry trade when relative to plant-level scale makes possible an equilibrium in intermediates replaces trade in final goods. in which This equilibrium holds concentration advantages dominate in upstream activities, and proximity advantages dominate downstream. The presence of multinationals has an ambiguous effect reduction in transport costs, but there the absence of multinationals, they imports, so that there equal quantities of When is a is home and a concomitant reduction in the consume home-market on consumer welfare; consumers benefit from a bias, number of a larger quantity of differentiated varieties, all else equal. In goods produced locally than of whereas with two-way multinational production, they consume foreign varieties. there are both factor proportions and proximity advantages to overseas expansion, the location of each stage of the production process dej>ends on both considerations. single-plant multinationals may emerge, while When factor considerations dominate, dual-plant multinationals prevail in the presence of sufficiently strong proximity advantages. I. Related Research i. The model combines elements from both literature on multinationals. In recent the imperfect competition trade literature and the strategy years, the strategy literature on multinationals has converged on a conceptual framework based on transactions cost analysis of the boundaries of the firm. expansion is Roughly, multinational explained as a firm's optimal response to the conjuncture of three advantages. ownership advantage in some unique asset that gives the firm market power and is The first is an associated with increasing returns across the firm (network capital, proprietary process technology or product designs, or an established reputational capital); the motivation for expansion mtemalization advantage, such that the firm is is to maximize the returns unable to realize the full to this asset. Second, there is an value of the asset through the market. due to transactions costs or other Third, there licensing. makes is first failures. This determines the choice of direct investment over an advantage to locating production near consumers or factors across borders, which international expansion While the market more profitable than exporting. two types of advantages are closely linked transactions costs, the concept of location advantages is to literature loosely defined, and This disjuncture between location advantages and trade theory theory.' is on the theory of the firm and on makes no reference to trade surprising in light of the important advances made by trade theorists in recent years in incorporating maricet power and increasing returns to scale into trade models. The economics literature on multinationals has gone part way in recent years to incorporate these Broadly, two strands of research have considered multinational activity. insights. choice between licensing and investing across borders. market failure that prevents firms from It The first focuses on the models internalization as an endogenous response fully exploiting their market power through the market. to a Ethier (1986) incorporates the internalization decision into a general equilibrium trade model based on specific factor endowments with a differentiated manufacturing sector. imperfections in contracting under uncertainty. downstream distribution must be located The reputation for quality. equilibrium framework. in the destination internalization decision of the firm is a response to location choice is captured by an ad hoc assumption that in the destination Horstmaim and Markusen (1987) focus on framework, where production The consumption market. the internalization decision in a partial equilibrium market may be chosen over licensing in order to maintain a Ethier and Markusen (1991) similarly focus on the internalization decision in a partial The decision between exporting and overseas production via licensing or investment dep>ends in part on a trade-off between variable transport costs and a fixed production cost, but the general equilibrium implications are not explored. A second strand of research focuses on the choice between exporting and investing across borders, which hinges on location advantages. Krugman Krugman (1982), Helpman (1984), Markusen (1984), Helpman and (1985), and Horstmann and Markusen (1992) incorporate models, which start from the premise that firms benefit MNEs into general equilibrium trade from internalization due to increasing returns at the level The model of the firm, and focus attention on the location decision. research in focusing on the location decision. vertical I develop below follows this line of But while the Helpman and Helpman, Knigman models explain expmnsion across borders in terms of factor proportions differences, the model presented in Section II formulates the motivation for overseas expansion in terms of a trade-off between proximity and concentration The model advantages.^ models in a in Section II integrates the central features more general model; the Krugman model' develops of the Horstmann, Markusen and a differentiated products model Knigman that focuses on proximity advantages alone, while the Horstmann, Markusen model' incorporates a tradeoff between proximity and concentration advantages I briefly in the context of Coumot summarize the main findings of the duopoly. factor proportion comparison. The Helpman and Helpman/Krugman model trade model, where one sector is incorpwrates multinationals into a factor proportions characterized by product differentiation and multiple-stage production. are multiplant economies of scale associated with a firm-specific input production of the input this is assumed to be relatively model, multinationals arise only factor endowments are no incentive this more There which has a public goods character, and capital-intensive than production of the final good. presence of sufficiently great factor endowment differences. In When sufficiently similar that factor price equalization obtains in the trade equilibrium, there is in the differentiated sector locate Thus, in the for cross-border investment. good production models here, for purposes of When factor prices are not equalized under trade, some of production of the input in the relatively capital-abundant economy and final in the relatively labor-abundant economy, and export back model explains cross-border investment to the capital-abundant as a response to factor price differentials, and it multinational activities will only arise in a single direction within an industry, in single-plant firms. effectively explains one-way differences, but has little to the firms direct investment flows economy. predicts that It between economies with strong factor proportions say about two-way intraindustry investment flows between economies with similar factor proportions. This contrasts sharply with the model I develop below, in which multinational activity can take place in the absence of factor prop>ortion differences and in multiplant firms. To make two directions the findings directly comparable, I in the same industry, and is undertaken by develop a general equilibrium model that closely parallels the Helpman and Helpman, Knigman framework, which also has the virtue of making the findings comparable with a monopolistic competition trade model. Section II describes the structure of preferences and production in a framework with symmetric factor endowments and a two-stage production process, and derives Section III the equilibrium pattern of trade and investment. elaborates the production structure in the differentiated sector to incorporate an additional stage. Section IV elaborates the model to incorporate factor proportions differences as a motivation for overseas expansion, in addition to the proximity/concentration trade-off. Section V concludes. Two-Stage Production with Symmetric Factor Endowments n. There are two neighboring countries, A and B, between which factors and consumers are immobile. endowments and in the distribution at a distance The D apart. analysis will initially Countries are defined by areas assume symmetry in factor of consumers across countries, in order to isolate the role of proximity advantages in stimulating multinational activity. n. Consumption i. There are two classes of goods: agricultural goods, which are homogeneous, and manufactured goods, which are differentiated. Consumers have identical, homothetic preferences over classes of goods. assumption permits separation of the consumption decision into two stages: the allocation of expenditure between the two aggregate goods, and the suballocation of manufacturing exptenditure among It is further assumed that consumers have identical preferences This among varieties. varieties of the differentiated good, which take the form of a constant-elasticity-of-substitutionsubutility function of the Dixit-Stiglitz-Spence type. Denoting the differentiated product Q and the share of expenditure allocated to differentiated goods Eg, the consumer's second tier utility maximization problem can be represented as: ^^^ where market Max q^, is a, the quantity of variety and n, is the number of UQ{q^^^,...A^gA^...) i originating in market a varieties from market a. 6 s.t. ( EE = A,B), The ''i.'^fa Pj, is ^ ^o the price of variety i produced in subutility function is defined over all varieties of good q available at the consumer's location: B where <t is the elasticity of substitution consumption plan variety is among I '. varieties. Solving for the first order conditions yields an optimal where the share of manufacturing expenditure allocated that includes all varieties, to each inversely related to the relative price of that variety: p-" 9b, = (3) "^^-^ ^^-T~^Q k a Differentiating with respect to the price yields an expression for the elasticity of .^1^1^ (4) I a. a-A^ (1-q) drop the second term k in the elasticity expression. analysis without substantially changing the results, for a sufficiently large Returning to the first-tier utility for each variety: ^^^ = a Following Helpman, Krugman, demand maximization problem, This vastly simplifies the number of varieties.' total utility for a representative consumer is defined: U (5) where utility from consumption of the agricultural good, Y. = UIu/IuqO] agricultural good, u^(.), Assuming a Cobb-Douglas simply assumed to equal consumption of the is utility function, the consumer's first-tier utility maximization decision can be represented as: Max (6) where Py is Q'Y'-' s.t. the price of the agricultural good, and I is Pyy*JiY.P^q^ = I Solving for the first income. ex{)enditure shares for each of the aggregate goods that are independent of income: B (7) ". EE f'kJlka a=A k=\ & ^E^^Ib I (8) — ^ = 1-5 -'fy = '(!-*) order conditions yields II. Production ii. There are two factors of production: which used in both sectors. is The supply of both preferences specified above, the The price of the agricultural good good is constant, and its The manufacturing is sector is L,=L and = facilities C\w) is i=Afi characterized by product-specific increasing returns. treasury, in the manufacturing sector production all R&D in These Adopting a familiar from increasing costlessly separable into two activities that have a and can be spread among any number of activities are Such unique to each variety. activities product design or process technology, advertising, or services such as persoimel, and planning; for simplicity, total costs is is The corporation performs some facilities equally, without diminishing their value to any one. could include Thus, the marginal cost of the transportable at negligible cost between countries. The business system public good character; they benefit in the just equal to the marginal cost: is i corporate activities and production activities. activities: Given the The production technology Chamberlinian approach with free entry will yield monoptolistic competition, as returns trade models. Gi = G. and labor, L, used in the agricultural sector. and perfect competition prevails. price in country 1 good specific to the agricultural sector, will be taken as the numeraire. (9) In addition, the agricultural is factors is symmetric: supply of land full agricultural sector has constant returns to scale, agricultural G, which land, per firm located in market (10) I i will refer to them as R&D. Denoting the level of R&D activity as r, are: -^^^ C'iw,r) ^ i=/t^ dr which are simply assumed to be nondecreasing. Production operations are characterized by increasing returns a variable cost associated with each manufacturing plant. "" CV„r^,) = at the plant level; there is a Production costs for a plant located in market F(^)*V{w^^) ^ < or To nuke (12) the analysis more tractable, it is useful to assume V{w^^) fixed cost and = 8 that marginal costs are constant: V{w^)q^ i are: This assumption is not critical; the results require only that the cost function exhibit bounded returns to scale. Notice that variable costs indef>endent of the number of fall as the level of corporate activity increases, and that this relationship is Tlie firm chooses r to minimize costs over both markets such plants in operation. that: dC'(w^) ^3^ V'-''' dV^^^) dViw^r) = dr dr Finally, there is It is from destination markets, such as shipping The fraction of output that is lost in the distance travelled. and the transport „ ^ j=B when . a, J , . t=A, ^ and ^ the reverse intended to stand in for a variety of disadvantages of distance costs, trade barriers, linguistic or cultural differences, transport cost assumed is in transit (similar to Thus, for some amount shipment to a foreign market — dr ' a transport cost. responsiveness to consumers. '- q,+ Krugman q, to increase with distance: (1979)), which produced in I it is and slow modelled as a specify as exponentially increasing market a = A,B, the amount that survives smaller by an amount that increases in the distance between the two markets, D, is cost, T: ''"" (14) daa=^' <^ak=D ^a b=B when a-A, and a=Afi, The competition among firms equilibrium in each stage. in the differentiated sector takes place in First, firms in full stages, and there is a = A,B) Solving the yields a price to location] as a first order conditions for profit markup over the marginal cost of production and transport: P« (15) . The markup hinges on II. iii. = a, the elasticity l-I F(.)(' " for a=A^ and j=A^ o of substitution between varieties. Market Equilibrium With constant Nash knowledge of competitors' configuration decisions, firms simultaneously choose prices, yielding a Bertrand equilibrium. at a( two both markets simultaneously decide whether to enter the industry and choose their plant configurations. Then, with maximization for a firm located the reverse returns in the agricultural sector, the equilibrium product of labor: 9 wage is the value of the marginal w, = -^^^^^^ (16) With symmetric factor supplies, and i=A^ identical preferences, the wage is equal across countries. Together, the markup pricing policy in equation (IS) and the elasticity expression in equation (4) imply that all varieties of the differentiated product produced similarly for varieties all of varieties produced B at location B. market B by an amount that in by a consumer in market a( = A,B) qJP^) - (17) = will be priced equally is varieties — -^ o produced from both markets. a at P^P.„...P, * In^^n^e^^-') The quantity demanded of imported • IJP'") (18) r varieties A;.-^. ' . "^''^.^a number of varieties available is: ^-^AJ' Tt)~r(n^.„^,'«i-<')) P'.J'bj' Pn.' "-''a^b the Te\>erse The optimal allocation produced a single location, and smaller amounts of imported varieties, yielding a varieties implies that each consumer consumes equal amounts of 1 facilities, all firms operate a single production facility, or there will characterize the conditions for each type of equilibrium in turn. plant configurations, variable profits where B t on sales of each variety denotes the single-plant equilibrium. produced is all varieties all home market Depending on the parameter values, there are three possible types of equilibrium: production demanded is: a=B when b=A, and at exceeds that ^-TDo — _] - among A at in each market, the quantity increasing in the distance between the markets, and decreasing in the is produced proportional to the transport cost differential, and of a representative variety produced K(.) a given location, and at In maiicet B, the price of varieties Given equilibrium prices and numbers of the reverse. which at produced A at bias. firms operate two a mixture of both configurations. Starting with the equilibrium with single- at A and sold in A can be expressed Variable profits on sales of each variety produced at A in market are: (20) Profits are higher the closer n'^iPrn) is = -^ the production location to the destination market, 10 as: and the fewer competing varieties there are. Thus, an equilibrium where in markets for a firm located Assuming ft^e entry, the Further, given the at A all firms have a single production facility, total profits over both are: number of firms producing symmetry assumptions made above, symmetric across firms and markets, and the wage number of firms that the at located at number of firms A and B are equal in the optimal choice of the firm-specific input, both markets in equilibrium, located in each market is equal, which r, is in turn implies that the nA=nB=n,. Together, these two conditions imply is: *^ n, - (22) each location adjusts endogenously to yield zero profits. <3[F{w)*C\w/)-\ The number of the elasticity of substitution between varieties, To check whether configuration. A this is and in the level of per-firm and per-plant fixed costs. an equilibrium, consider the incentive for a firm to change firm will open a second production facility only if the increase in its a plant located at (23) its A opens an additional production facility at B. Prices and profits in its production fixed costs compyensated by the increase in variable profits from locating closer to foreign consumers. firm lowers and decreasing in varieties is increasing in the expenditure allocated to differentiated products, A is more than Suppose a firm with are unchanged. price in market B, to reflect the reduction in the variable costs of serving that market: p'^ = The equilibrium: F(w) (26) Comparative plant is more statics ^ (i-^nxi-.)) on the equilibrium conditions establish that the equilibrium in likely to prevail the higher is the fixed production cost, the distance between the markets, and the lower to 0, a single-plant equilibrium is The always prevails. effect defection on the market price index, the elasticity of substitution. is it number of of substitution do not take is is ambiguous. into account the effect of their more likely to hold the smaller is firms, a single-plant equilibrium is more likely the foreign market. In this case, a firm serves market that firms lower are the transport cost and the elasticity can be shown that the equilibrium In addition, for a fixed Next consider an equilibrium production of the which each firm has a single In the limit, as the corporate cost goes the corporate fixed cost. However, by simplifying the no-defection condition, so the smaller n,(Ug"^'-''>)-g"^'-') facility at B. consumers in in which all market A firms have production facilities in both markets, denoted m. from its production facility at A, and consumers in B from Solving for equilibrium prices and quantities yields sales for each variety in each of: qZiP^^J (27) - —^-^ «= ^3 associated with variable per-firm profits in each market of: 5/ (28) «"(^.^J Total per-firm profits over both markets (29) is: n"(/',,yij = 27i-(/'_,nJ-2/^H')-C'(>v^) Again, given symmetry in factor proportions and demand, the number of firms in each market Da— ig=n„. Then assuming equilibnum free entry, and solving for the number of firms under zero-profits is equal, in a dual-plant yields: 8/ n_ = (30) o[2Fiw)*C'iw^y\ Again the number of firms is decreasing in the level of total fixed costs, 12 but here the per-plant fixed costs are its weighted more heavily. Notice that under plausible conditions, there are fewer varieties in total in the dual- plant equilibrium, regardless of the level of transport costs.' A down firm is tempted to defect from the dual-plant equilibrium a production facility foreign market. more than Consider a firm however, the firm's price if the offset the loss in variable profits that shuts down its savings in fixed costs from shutting due to higher transport costs to the Pricing and profits in plant at B. A are unchanged. In B, rises to reflect the increase in transptort costs: P' = -2-V{.)e^ (31) o-l yielding lower variable profits of: «T ^AB^P'J'.^J (32) Thus, firms have no incentive to defect — = -7D(I-o) from a dual-plant configuration only if: gTD(l-o) (33) Fiw) i -^ a Combining the zero-profit '^"m (n.-l)+< ^ 2n,(l-."X'-°))-(l-."^'-'>) In^-a-e^^-'^ 2F(>v)+C'(H'^) A fully multinational equilibrium, smaller is never sustainable. ambiguous. elasticity where all firms have production facilities in both markets, is more likely the the fixed production cost, the higher are transport costs and the distance between markets, and the higher are economies is 7I)(l-o) and no-defection conditions yields equilibrium conditions: 2F(w) (34) J By at the corjKDrate. Here, as the corporate fixed cost goes to 0, a dual-location equilibrium Again, the effect of an increase in the elasticity of substitution between varieties simplifying the no-defection condition as above, it can be shown that an increase in the of substitution makes the multinational equilibrium more likely to hold. the dual-plant equilibrium is more likely to hold the larger is the foreign market. single-plant equilibria are characterized is For a fixed number of firms, Thus, the dual-plant and by precisely reversed conditions. In the intermediate range of parameter values, there multinational firms coexist with national firms. In the is a third possible equilibrium, in mixed equilibrium, some which fraction, a, of firms have a single production facility and export, and the remaining fraction have production facilities in both markets. 13 The condition for the mixed equilibrium can be derived from combining no-defection conditions for both single plant and dual-plant firms: (35) n^(2-a(l-e"^'-°')) > l+«"^'-> where n^ is the number of firms mixed equilibrium. in the When equalized, the expression for the proportion of single plant firms where r; proportion with production additional With R&D two configurations are is: 2 W (l-^rocJ-)) a„^ (F(k')+C'(h'^j)-C'(h-^,)) the profit-maximizing level of is returns to the facilities in R&D input for a firm with i plants. For a given number of firms, the both markets declines with the size of the fixed production cost and the investment, and rises with the size of the transport cost and the distance between markets. free entry, this equilibrium holds only when the following conditions on the fixed costs and the transport costs are satisfied: The fixed production costs must be greater the greater relative to a dual-plant firm, If this condition II. iv. It is is satisfied, is the size of the and the smaller are the transport R&D cost, distance, investment for a single-plant firm and the elasticity of substitution. any mixture of firm types can be supported. Factor Market Clearing now possible to go back and solve for equilibrium conditions in the labor market. returns in the agricultural sector imply that the per unit labor input function with respect to local wages. Thus, (36) in is Constant simply the derivative of the unit cost = A,B: market i ,,(,^ __ ^^^(^ dw In the differentiated sector, the demand demand for labor can be separated into corporate and production activities. for labor for corporate activities is the derivative of the cost function with respect to the local the equilibrium level of corporate services: 14 wage The at (37) ,.(.^, . i£^ dw Production activities generate a demand for labor associated with the fixed production cost, which is simply the derivative with respect to the local wage: LVJ (38) And they generate a demand output produced at location i for labor for production and sold in j I'^iwj-) = e^'^ (39) ^ and transport of the differentiated goods; for a unit of demand this -—-^ = is: d.=0, d^=D; iMA *=« ^hen iM, and the reverse ^w^ Labor demand varies according wages across countries, in the single-plant equilibrium, factor clearing conditions in each L ('^0) where q' is to the type of equilibrium in the differentiated sector. = With symmetric market are: P{w)Y'*n\L\w/)*L''(.w)*l^(w/)q'\ per firm output in the single-plant equilibrium. In the pure dual-plant equilibrium, factor clearing implies: 1 (41) where q™ is = /''(>v)y"+njL''(>v^)+LV)+2/<?(H',r)4"] per firm output in the dual-plant equilibrium. In the mixed equilibrium, factor demands are straightforward to calculate, and depend on the mix of firms. II. V. Commodity Market Clearing The income-expenditure equality can then be expressed in market ('*2) /- while in the single-plant equilibrium Then commodity market A = y+2«.[/»7<7;(/>-)] it is: clearing requires a single market clearing condition. this is: (44) in the dual-plant equilibrium as: (l-») = L n=m^c I" 15 For agriculture in equilibrium n n. Trade Balance vi. Given synimetry agricultural good and in population size for differentiated products. with purely national firms. and the Given symmetric intraindustry trade elasticity The single-plant equilibrium is in both markets for the essentially a trade equilibrium own demand factor proportions, each country satisfies its for the and there are balanced flows of two-way trade in the differentiated manufacturing sector: agricultural good, The volume of and income, there are symmetric demands is a decreasing function of the transport cost, the distance between markets, of substitution, and an increasing function of income and the expenditure share allocated to differentiated goods. In sharp contrast, there is trade in goods each direction It rises no trade in goods in the pure multinational equilibrium. in the differentiated sector, there is balanced two-way trade Instead of in corporate services. two-way The flow in is: with the amount of labor employed in corporate activities and with the number of firms, which in turn is an increasing function of the expenditure allocated to differentiated products, and a decreasing function of the of substitution and fixed costs. elasticity Thus, this model yields an equilibrium in which multinational production arises in the absence of factor proportion differences, and two-way intraindustry trade in invisibles completely crowds out intraindustry trade in These goods. differences, factor results contrast sharply with the predictions where the only motivation for cross-border expansion is factor price differentials caused endowments. In a factor proportions in the capital-intensive by different framework, the presence of multinationals reduces the share of trade accounted for by intraindustry trade, but need have headquartered from a model based on factor proportions no effect on the total volume of total trade, since varieties country and produced in the labor-intensive country are exported back to the capital-intensive country. Lastly, in the differentiated sector. mixed equilibrium, there is balanced two-way trade in both goods and invisibles in the Multinational sales are a relatively larger share of total foreign sales for each country, the 16 higher the proximity advantage relative to the concentration advantage. Three-Stage Production with Symmetric Factor Endowments ni. So far, the model suggests the stark result that, given large benefits to proximity relative to benefits from concentration, multinational enterprises sits uneasily alongside two stylized facts. First, there is a large share of intraindustry trade in total trade (Loertaches, Walter (1981)), and second, multinationals account for a significant share of trade Indeed, there (1988)). is evidence that suggests that multinational activity on balance rather than a substitute for trade The first fact (Swedenborg (1979), Blomstrom et. al. is (Rugman complementary to (1984)). could be explained in the model developed above by a combination of low benefits to proximity and large scale economies. The second fact is not explained by the model developed above with a simple two-stage production function in the differentiated sector. would explain This will either in part or totally substitute for intraindustry trade. the second fact in two ways. First, multinationals, finished goods are exported back A factor proportions account of multinationals given a two-stage production process and single-plant from the country of production to the headquarter country. Second, with a three-stage production process, multinationals produce intermediate inputs in the economy that relatively abundant in the factor downstream production is used relatively intensively in intermediate production, and exf)ort them to facilities in the intensively in final goods processing. economy A that is relatively factor proportions well-endowed with the factor used relatively model thus predicts that either final goods or intermediates will flow across borders in a single direction, contributing to interindustry rather than intraindustry trade. Alternatively, the with both stylized facts. model developed above can be elaborated By in a way that is sensible and accords well elaborating the production structure in the differentiated products industry to incorporate additional stages characterized by different proximity/concentration trade-offs, the model predicts that multinationals will generate intraindustry, intrafirm trade under plausible conditions. This prediction differs from a factor proportions account, which would predict one-way flows of intrafirm Again, the contrast between commodity semiconductors and cars serves to 17 trade."* illustrate the differences in the two models. facilities in In the 1970s, US commodity made chip firms substantial investments in Southeast Asia, which were motivated entirely by factor price differentials." assembly and Tlie test test and assembly stages were intensive in low-skilled labor, in contrast to the design and manufacturing stages, which were intensive in engineering skills and capital. manufacturing complexes to assembly and destinations in the US Semiconductors were shipped from the and elsewhere. '* expansion into destination markets. been substantial two-way investment and multinational A typical pattern of expansion distribution and sales facilities in the destination market, followed was would in two-way multinational economies and proximity advantages provides more with the establishment of facilities when the share in the Factor price differentials provide sufficient to warrant the fixed investment." relative factor prices in periods of substantial start by assembly EC into these investment patterns, since the markets in question, such as the scale design and Southeast Asia, and then shipped back to sales test facilities in In contrast, in the automobile industry, there has foreign market US and the US, have had similar activity. scale insight: insight little Instead, the relationship between economies are substantial but bounded auto assembly, and smaller in sales and distribution, trade barriers into several countries in the EC are high, and the advantages of a local presence and customization to the local market are high. To formalize an explanation for the latter pattern, additional stage, which has I modify the production structure a distinct proximity /concentration trade-off. manufacturing (q) stages detailed above, there are sales activities (s), '* to incorporate an Thus, in addition to the whose production R&D (r) and requires a fixed cost and a constant marginal cost: (45) C'iw^^^) = F'iwJ^V'iw^^ys a=Afi Similar to the upstream relationship between corporate activities and manufacturing, the variable cost in sales a decreasing function of the input from the manufacturing stage, R&D input, which is proprietary to the firm. however, the manufactured input lowers the cost only of the sales In addition, there is a transport cost, simplify notation, I facility that both for the manufactured input, and for the replace the exponential transport cost with a both the distance between markets and the shipping cost. 18 The more general parameter, uses Unlike the it. final output. t, To which subsumes cost of transporting the output of each stage, is ti> 1 (i = q,s), is the reciprocal of the fraction lost in shipment; only the portion of output that borders incurs the transport cost. The is shipped across transport cost between manufacturing and sales can be interpreted in terms of a co-location advantage. Each firm chooses its production configuration to trade off additional fixed costs against additional variable costs of transtmrt at each stage of production. commonly activities dispersed near consumers, while manufacturing R&D such as are the most concentrated. equilibrium, in prices. w^ = Wg, and Given symmetry the (or confine attention to firms headquartered in market it can invest in sales facilities in R&D A and costless is equal in markets output can A and B, so by in we sales in a single both markets, and concentrate manufacturing in a single market; The it can There are four possible without loss of generality. can invest in manufacturing it facilities in can both fourth configuration can be ruled out, since the firm by shutting down a manufacturing activities are concentrated in a single location, facility. In any due to the combination of scale economies by deriving conditions for each pure type of equilibrium - where market have identical configurations.'* The equilibrium conditions are derived market. R&D transport.'^ I start in Section that and demand, factor prices are equal investments) fixed production costs and transport costs R&D assume II, I Each firm can concentrate both manufacturing and markets, and concentrate sales in a single market. configuration, be concentrated, and corporate in production configurations is followed spread both sales and manufacturing activities across borders; or would save both likely to 1). in factor proportions number of firms configurations in the differentiated sector. more Nash equilibrium Again, firms compete in two stages: Nash equilibrium is Accordingly, as in Section be transported costlessly to manufacturing plants (t,= market; Casual empiricism suggests that sales activities are most II, all firms headquartered in each in precisely the by combining zero-profit conditions with no-defection conditions for all same manner firms located in each However, with three possible configurations, there are two no-defection conditions for each equilibrium. Firms choose configurations and prices taking as given the prices and configurations of Variable profits for each firm across both markets in equilibrium 19 j are: their rivals. as *' -^ ./p/-^ _ i^iPiJ^i^n) = (46) where Dj is the number of firms The number of firms free entry. in located in each market under market structure each equilibrium is n, = (47) II, j. determined endogenously by the zero-profit condition under Thus, in an equilibrium where firms concentrate Just as in Section H,m^ all activities in a single location (denoted t), it is: —[CXw^)*F*(w)^F'iw)] in each equilibrium the no-defection conditions are obtained by comparing the change in variable profits associated with each alternative production configuration with the corresponding change in fixed costs. The equilibrium is then obtained by solving the no-defection and zero-profit equations simultaneously, yielding conditions on the relative size of the transport costs to the fixed costs, and on the activity-specific scale economies Combining the condition ruling out defection relative to the firm-wide scale economies.'^ to a fully diversified configuration with the zero-profit condition yields an equilibrium condition: C'(w^) where the cost of (U2l,'"') trans[)orting intermediate inputs raises variable costs 7-,= by a proportion, T , defined: (^;CV^^ (K,(.)+K,(.)) And the equilibrium condition ruling out defection to a In the equilibrium where only downstream equilibrium number of firms (50) is: (Tl-''-tl")[n,(l*tyW,-'] F'(w) (49) downstream diversified configuration activities are dispersed across borders (denoted d), the is: « 'd = ^[C'(,w^)*F^(w)*2F'(w)]-^ There are two no-defection conditions, corresponding to the two alternative configurations. Combining the zero-profit condition with the condition ruling out defection to a single-location configuration yields: 20 (r;"'-rj-')[>./u7:;-')-r;-'] ^ (51) n/UTl-')iU2tl"'-Tl-'')-iTi-'-tl-'')(l-Tl-') And combining f.(^) C'iw,r)*F*(w) the zero-profit condition with the condition that firms have no incentive to extend their multinational activities by opening an additional manufacturing facility yields: F^(w) (52) (i-r;->^u7:;-)-r;-°] ^ C\w^)*2F'(w) Lastly, the number of firms across borders (denoted m) nJ,l*Tl")2f^"Hl-'li'')il*2Tl") in the equilibrium where firms diversify both manufacturing and sales is: (53) = n *^[C'(H'^)+2F«(>v)+2F'(H')]-' a The combination of production the zero-profit condition with the condition that firms have facility yields the down one to close equilibrium condition: (2n„-l)(l-7^-'') f,(H,) 4«_r'-° m'l C'(w^)*2F'{w) (54) And no incentive the equilibrium condition ruling out defection to a concentrated, single-location configuration (2n,-l)(l-r,'"°) (55) is: ^ F'iw)+F''(w) Notice that together, equations (47), (50), and (53) imply that under plausible conditions, the number of varieties is greatest in the single-location equilibrium, and smallest in the fully diversified dual-location equilibrium.'* Comparative statics on the equilibrium conditions establish firms establish sales and manufacturing facilities in both markets, manufacturing, the greater are fixed costs in intermediate input (T^" and the final good concentrated, single-location equilibrium fixed cost of the smaller R&D, is the lower is is R&D, (t,), A likely the smaller are fixed costs in is the elasticity of substitution.^ A fully likely the greater is the fixed cost in sales, the smaller is the the transport cost of the final the elasticity of substitution. more where the greater are the costs of transporting both the and the greater more is that a fully diversified equilibrium, good relative to that downstream-diversified equilibrium 21 of the intermediate input, and is more likely to obtain the smaller is the fixed cost in sales, the greater is the fixed cost in manufacturing, transporting the intermediate input relative to that of the final good. equilibrium when only possible is is is the cost of In addition, the downstream-diversified the transport cost of the intermediate Similar to the two-stage case, a diversified equilibrium and the smaller is smaller than that of the final good. increasingly likely the larger are the returns to scale at the corporate level relative to those in manufacturing and sales. The pattern of trade and production associated with each equilibrium can be derived following the analysis in Section commodity II, so that I clearing conditions. will simply discuss the The main conclusions here, without detailing the factor and addition of a stage of production generates demand for labor, analogous to Conditions for factor market clearing and commodity market clearing can then be equations (38) and (39). derived analogously to equations (40) through (44), with a similar set of conditions for the downstream diversified equilibrium. patterns will emerge Just as in Section in the II, with equal relative factor endowments, identical production two markets; given symmetric demand, trade in the differentiated sector will be balanced, and the agricultural sector will be autarkic. The equilibrium which in all firms concentrate trade patterns identical to those in a pure trade proportions: there are balanced flows of intrafirm trade. all model with two-way trade three production stages in a single market results in differentiated products and in differentiated products, symmetric factor and no interindustry or Just as in a trade model, the flows will be greater, the larger is the share of expenditure allocated to differentiated products, the higher The equilibrium results in balanced which in all is income, and the lower are transport costs. firms invest in both manufacturing and sales facilities in both markets two-way flows of intrafirm trade in R&D completely crowding out trade in goods. Here multinational production completely substitutes for trade. The addition of a production stage introduces the possibility of a third equilibrium, in which firms diversify their sales activities across borders but concentrate their production activity in a single market. In this equilibrium, there are balanced two-way flows of intraindustry, intrafirm trade in intermediates, which substitute completely for trade in final stylized facts cited above: goods and in invisibles. This equilibrium is thus consistent with both of the the existence of multinationals generates intraindustry trade. 22 Asymmetric Factor Proportions IV. In reality, firm configuratioos are likely to reflect considerations of both factor and proximity advantages; in this section I a pure factor proportions account of relatively abundant in the factor in MNEs predicts which R&D which in factors are completely mobile - is relatively intensive, the integrated equilibrium hybrid model of multinationals, since there in analyzing a equilibrium when transport costs are introduced. Instead, benchmark and analyze how firm configurations and I - It is is in the country and a single manufacturing plant Such accounts use as intensive. is configurations and trade patterns of dividing up the supply of factors. approach R&D configurations in which a firm locates country abundant in the factor in which manufacturing world As described above, develop a hybrid model to accommodate both considerations. their in the benchmark a and analyze the implications for firm somewhat complicated no obvious counterpart to use this to the integrated adopt the symmetric equilibrium as an alternative become trade patterns change as factor proportions increasingly unbalanced. This section returns to the simplified two-stage production structure of the are assumed to be used are divided between the wages of skilled in the production of both goods: two countries and immobile. Type workers to that for skilled labor in activity that the factor intensity relative wage l.V^/KXW) > with of unskilled workers as its first W= R&D paid wage Wj'/Wj'. W, the most skill-intensive activity section. and unskilled labor Wj' in country i. Two (L^). factors The Define the factors ratio of Next, define the equilibrium unit input and similarly for unskilled labor. of production in the three productive activities is and agriculture the Further, ranked consistently least: VO^/^z'O^ at any -* 1/(W)/1,^(W). In addition, assume that there locate j is k as 1,'tW), given relative wages assume rate, skilled labor (L,) first is manufacturing plant with some small advantage its R&D to co-location, facilities, all else equal. ^' factor proportions, the existence of a co-location advantage, such that a firm chooses to co- In the equilibrium with equal however small, eliminates single-plant multmationals. Starting in the benchmark case with equal fully multinational equilibrium where all factor endowments, there are three possible equilibria: a firms in the differentiated sector have manufacturing facilities in both 23 markeUi, a pure trade equilibrium where firms have a single manufacturing plant, and a mixed equilibrium. all two markets, and there In all three cases, given equal factor proportions, factor prices are equalized across the IS symmetry in the number of will analyze the effect B in the Suppose firms. of moving skilled labor from market same proportions B of skilled to unskilled labor A to market B is the monopoly supplier of unskilled the center along the diagonal toward the northwest comer labor. pressure on the relative imbalance by moving equalization. R&D wage of activities to skilled labor in to B, there is is the Helpman diagram. in the A, and upward pressure A, while agriculture expands A corresponds to Tliis A In the pure trade equilibrium, as the ratio of skilled to unskilled labor rises in downward A economy We is Lj/Lj. and unskilled labor from market as the aggregate skilled to unskilled ratio, so that in the limit monof)oly supplier of skilled labor, and movement from that the aggregate ratio and in B. falls in Firms respond to the B, thereby maintaining factor price in In this case, firms adopt multinational configurations in response to factor availabilities; these configurations are distinct from those found in the symmetric multinational equilibrium in that each firm has a Here there are two-way flows of single plant. differentiated products of the same symmetric equilibrium, but these flows may be unbalanced, reflecting income additional trade flows of agriculture As from B to A R&D in return for diverge, and different production techniques are adopted in the concentrated before agriculture. Take the point at additional increment of skilled labor from plant will move to A B to A which the last two-way in from B, and agriculture will continue to move B to one-way flows of facility has A that moved to B. R&D from A A in to With unequal becomes A, and transfer an a manufacturing sufficiently strong A, and manufacturing B and Here, there is B to A. means of factor prices, the payoff to concentrating manufacturing activities 24 is split unbalanced, agriculture from interesting feature of this equilibrium is that firms continue to use trade as the sole foreign penetration. intensifies, R&D to Now B from A. With and/or R&D skills in Assume countries. R&D to B. that juncture, factor prices unequal proportions and utilizing different production techniques. trade in differentiated products and The At and of unskilled labor from factor imbalances, either agriculture concentrates completely in between the two two A abundance of R&D. and there are differentials, from services reallocation of factor supplies continues, so that the relative eventually one country becomes the sole producer of agriculture or overall magnitude as in the must be at least as great as it is with equal factor prices. Thus, the introduction of factor imbalances gives rise only to single-plant multinationals, if any. Performing the same experiment starting from a pure multinational equilibrium, the progressive unbalancing of factor proportions B, and factor price equalization that there are flows is is similar. R&D moves to B to response to a market A, agriculture moves to market maintained, along with plant configurations. of agriculture from initial A, balanced by reverse flows of The R&D pattern of trade adjusts, so services. Eventually, the imbalance in factor endowments becomes sufficiently extreme that one industry concentrates in a single market. Now a factor price differential between the two markets economies motivation for concentrating manufacturing activities in a single location. is added TTius, a to the scale few firms will concentrate production in the market with favorable factor conditions, and exptort, while others will maintain production facilities in although the both markets. total extent In equilibrium, the returns to both configurations will be equal. So here, of multinational activity may remain unchanged, the number of firms with multiplant configurations will diminish, and there will be unbalanced two-way trade in differentiated products along with flows of R&D services and agriculture. Thus, the introduction of different factor proportions across the two markets permits a broader range of possible trade and production patterns. With unequal production process depends on two considerations: factor intensities. factor proportions, the location of each stage of the the relative concentration/proximity trade-off and relative Unequal factor proportions introduce the possibility of single-plant multinationals, and make the likelihood of single-plant firms higher outside the range of factor price equalization. V. Conclusion The introduction of an explicit trade-off between scale economies and proximity advantages in a differentiated products investment. model of trade permits a much richer set of predictions about the pattern of trade and In particular, consideration of this trade-off introduces the possibility of two-way, horizontal multinational activity, even in the absence of factor price differentials, and of intrafirm trade in intangibles completely supplanting trade in goods. It also introduces the possibility of multi-plant configurations across 25 These borders. results, which apf)ear to be of increasing importance empirically, are inexplicable in an account of multinational expansion driven purely by factor proportions differentials. An equilibrium characterized by multinational production and multiplant configurations is more likely the higher are returns to scale at the corporate level relative to the plant level, the higher are transport costs across markets, the greater is expenditure on differentiated products in the foreign market, and the higher of substitution between elasticity is the varieties. This framework also can explain equilibria in which national firms coexist with multinational firms in the same industry. returns to the A mixed equilibrium two configurations differ is by Consumption patterns vary with sustainable for intermediate ranges of the parameter values, less than the differential the type of equilibrium in the differentiated sector. multinational equilibrium, equal quantities of pure trade equilibrium, there Consumers there is benefit is a between the associated fixed all home market varieties are consumed, regardless of costs. In a pure their origin, while in a bias in the consumption of the differentiated product. from the presence of multinationals, due to the reduction a concomitant loss in consumer surplus from a reduced overall The when of the price of imported goods, but number of varieties. prediction that intraindustry flows of intangibles substitute for trade depends on the structure of production in the differentiated sector and the relative proximity/concentration tradeoff of each stage of production. activities, With multiple production stages, if concentration advantages dominate in upstream production while proximity advantages dominate downstream, then intraindustry, intrafirm flows of intermediates supplant intraindustry flows of final goods. In this case, multinational activity is complementary to trade in goods, and multinationals generate intraindustry trade. When both factor proportions differences and proximity advantages are incorporated within the same framework, the pattern of trade and investment depends on their relative strength. When concentration advantages are high relative to proximity advantages, the introduction of moderate factor proportions differences results in the formation agriculture and in R&D of single-plant multinationals, and there are unbalanced one-way flows of trade services in the reverse direction. factor price differentials arise, and flows of trade With more extreme in differentiated 26 in factor proportion differences, products become increasingly unbalanced, as manufacturing facilities are distributed increasingly unevenly across the two markets. When proximity advantages dominate, moderate factor proportions differences result in unbalanced one-way flows of trade in agriculture in return for R&D services, but firms maintain multiplant, multiiuitional configurations. As factor proportions become increasingly unbalanced, the resulting factor price differmtials create an incentive to concentrate production in a single plant. This model has strong implications for policy. transport costs, and Most obviously, makes penetration by multinational production more of intangibles (or profits) has the reverse the imposition of a tariff raises attractive, while a tax on the earnings Further, the relative structure of tariffs on final goods and effect. intermediate inputs influences firms' decisions about which activities to concentrate and which to disperse across borders. The model further implies that restrictions on foreign direct investment may be more damaging to welfare than restrictions on trade in industries where natural proximity advantages are high and scale economies at the corptorate level are high relative to those at the plant level, trade blocs are not deleterious to global welfare. natural barriers are If and suggests conditions under which regional regions are geographically coherent areas within which low and between which they are high, and if different regions have similar region-wide factor proportions (while the factor proportions of countries within regions could differ significantly), then the welfare gains from free trade may be exhausted within regions, while free interregional investment flows raise welfare by permitting maximal returns to investment at the corporate level, and a greater variety of products. Future research will analyze these conditions. Further, this model lends itself to testable hypotheses. In particular, it suggests that firms' observed choices between exports and multinational expansion as alternative modes of foreign market {jenetration should reflect trade-offs between measurable proximity and concentration advantages. employs a gravity equation approach to test this hypothesis. 27 Research currently underway . Seminal contributions to 1. and growing this large literature include Knigman (1979), Lancaster (1980), Helpman (1981), and Ethier (1982). See Caves (1982) and Dunning (1988). 2. This 3. may be attributable in part to the failure of earlier attempts to explain multinational activity in terms capital flows in a factor proportions trade In the 4. Markusen model, factor of framework with perfect conq>etition. endowments are symmetric, but sector-specific capital plays a key role in explaining multinational production. The Krugman model 5. differs in its In the model presented below, assumptions about the production function. firms incur an additional fixed cost to open production facilities abroad, and save on variable transport costs. the Krugman model, multiplicative factor. there no is fixed production cost, and variable costs differ between markets by In some The predictions differ in two respects. First, in the Krugman model, only uniform equilibria is a mixed equilibrium. Second, in the model below, the decision whether to produce are possible, while here there abroad depends on the potential scale of production, and thus on the market size, while in the Krugman model, firms do not discriminate between markets based on size or share. The Horstmann, Markusen model assumes that there are 2 potential entrants, one headquartered in each country, is Coumot-Nash equilibrium if both enter the market. Below, the number of firms is determined endogenously in a monopolistic competition framework. This framework lends itself more readily to predictions about the pattern of trade, and introduces an additional welfare effect from the number of varieties produced. 6. and there The 7. analysis would be messier in the absence of this simplification, but the qualitative results would not change significantly. The no-defection conditions 8. fixed the level of a fixed R&D that the configuration by 10. input. reflect the additional costs that are required to open a production holding facility, This can be interpreted either in terms of a sequential decision-making process, or requirement. Assuming 9. R&D R&D more than investment for a single-plant configuration does not exceed that for a dual -plant the size of the fixed production cost. Unfortimately, the empirical work in this area does not distinguish between the two types of trade when referring to multinationals. See Yoffie (1992) for a complete account of investment and trade patterns in the global semiconductor industry. 1 1 12. The current pattern of trade and investment in the semiconductor industry is probably best explained by a hybrid between the two models (see Section IV). 13. In the interests of tractability, I ignore dynamic considerations in this model. relationship between exports and multinational expansion in a 14. in The model can be generalized country a is assumed to dynamic any number of production stages, to entail a fixed cost, a variable used as an mput, q,.,, at and each stage all is assumed to where each stage of production i located production cost, and a variable transjxjrt cost: I where the variable cost Future research will model the setting. a'*! '(a. -a')] be a decreasing function of the output from the previous stage mtemally produced inputs are assumed proprietary 28 to the firm. In addition, there is . a multiplicative transport cost for each unit of output that above is is transported to the foreign market. The model developed just a sp>ecial case. I assume it is always optimal for a firm to locate its first manufacturing plant in the same market as R&D, due some arbitrarily small co-location advantage between R&D and manufacturing. This assumption is useful because makes the flows of invisibles determinate, and sensible because it accords well with casual empiricism. It could 15. to it be modelled formally as 16. t, > 1 confine attention to uniform equilibria for simplicity; mixed equilibria can be derived similarly to Section I Similar to Section 17. II, the no-defection condition excludes changes in the R&D investment. interpreted either in terms of a sequential decision-making process, or in terms of a fixed 18. if it Assuming that the R&D Tq is This can be input requirement. investment does not decrease with the number of production and sales locations, or, does, that the increase in the fixed costs of production or sales 19. R&D II. more accurately described is at least as great as the R&D savings. as the ratio of the variable cost gross of the cost of transporting the intermediate input to the variable cost net of transport costs. 20. This result is derived by differentiating the no-defection conditions, simplified so that firms do not take into account the effect of their configuration decisions on the market price index. This of a large number of firms, used in simplifying the 21. elasticity is consistent with the assumption expression This assumption can be formalized in a variety of ways, such as assuming a variable transport cost between R&D and manufacturing, or a lower fixed cost for manufacturing when it is co-located with R&D. simply follow Helpman in making an ad hoc assumption, since the results are essentially the same. 29 Instead, I " References Blomstrom, Magnus, Robert Lipsey, and Ksenia Kulchycky (1985): "U.S. and Swedish Direct Investment and Exports," in Robert Baldwin (ed.), Trade Policy and Empirical Issues (Chicago: University of Chicago Press). Multinational Enterprise and Economic Analysis (Cambridge: Caves, Richard (1982): Cambridge University Press). Dunning, John (1988): Erdilek, Asim (1985): Explaining International Production (London: Multinationals as Mutual Invaders Ethier, Wilfred (1982): Harper Collins). (New York: "National and International Returns American Economic Review 72, pp. 389-405. St. to Scale in the Martin's Press). Modem Theory of International Trade, . "The Multinational Firm," Quarterly Journal of Economics November, pp. 806-33. Ethier, Wilfred (1986): . Ethier, Wilfred, and Henrik Horn (1990): "Managerial Control of International Firms and Patterns of Direct Investment," Journal of International Economics and James Markusen (1991): Pap)er No. 3825, August. Ethier, Wilfred, . 28, pp. 25-45. NBER "Multinational Firms, Technology Diffusion, and Trade," Working Helpman, Elhanan (1981): "International Trade in the Presence of Product Differentiation, Economies of and Monopolistic Competition," Journal of International Economics 11, pp. 305-40. Scale, . "A Simple Theory of International Trade with Multinational Corporations," Journal of Helpman, Elhanan (1984): Political Economy , p. 451. Helpman, Elhanan, and Paul Krugman (1985): Market Structure and Foreign Trade (Cambridge: M. I. T. Horstmann, Ignatius, and James Markusen (1987): "Licensing vs. Direct Investment: by the MNE," Canadian Journal of Economics August, p. 464. A Press). Model of Intemalisation . Horstmann, Ignatius, and James Markusen (1992): "Endogenous Market Structures of International Economics 32, pp. 109-129. in International Trade," Journal . Julius, DeAnne (1990): Global Companies and Public Policy (London: Royal Krugman, Paul (1979): "Increasing Returns, Monopolistic Competition, and International Economics 9, pp. 469-79. Institute of International Affairs). International Trade," Journal of . Krugman, Paul (1983): "The "New Theories" of International Trade and the Multinational Enterprise," in D.B. (eds.). The Multinational Corporation in the 1980s (Cambridge: MIT Audretsch and Charles Kindleberger . Press). Krugman, Paul, and Edward Graham (1992): "The Surge in Foreign Direct Investment in the Late 1980s: Conceptual Issues," forthcoming in Ken Froot (ed.). Foreign Direct Investment Today (NBER). Lancaster, Kelvin (1980): Economics , "Intra-industry Trade under Perfect Monopolistic Competition," Journal of International 9, pp. 151-75. "Multinationals, Multi-plant Economies, and the Gains from Trade," Journal of Markusen, James (1984): International Economics 16, p. 205-26. . Rugman, Alan (1988): "The Multinational Enterprise," International Management (New York: Wiley). Swedenborg, Birgitta (1979): "The Multinational in Ingo Walter and Tracy Murray (eds.). Handbook of Operations of Swedish Firms," (Stockholm: Industheos Utredningsinstitut). Wilkins, Mira (1970): The Emergence of Multinational Enterprise: Era to 1914 (Cambridge: Yoffie, American Business Abroad from the Colonial Harvard University Press). David (1992): "Foreign Direct Investment in the Semiconductor Industry," forthcoming Foreign Direct Investment Today (NBER). .J / b G / U in Ken Froot (ed.), Date Due m. V Ji-. ;-^ 1 1994 4 \ii\i4 . 9 1S84 Lib-26-67 MIT L III IRRARIF*: ill 1 1 ii[ 'M,,m:^m^M^^^m^^