Multinational firms and the theory of international trade

how international trade benefits households and firms innovation firm dynamics and international trade journal of political economy and firms in international trade pdf
AbbieBenson Profile Pic
AbbieBenson,United States,Professional
Published Date:13-07-2017
Your Website URL(Optional)
Economics 2535 Lecture Notes Advanced Topics in International Trade: Firms and International Trade Pol Antràs Harvard University Department of Economics Spring 2004Contents 1 Introduction and Basic Facts 4 IFirmsandtheDecisiontoExport 12 2 Intraindustry Heterogeneity with Fixed Costs of Exporting: Melitz (2003) 13 3 Intraindustry Heterogeneity and Bertrand Competition: Bernard, Eaton, Jensen, and Kortum (2003) 27 4 Firms and the Decision to Export: Empirics 38 4.1 ExportingandPlant-LevelPerformance.................. 38 4.2 Evidence on Reallocation Effects:Pavcnik(2002) ............ 40 5 The Relevance of Sunk Costs 47 II Firms and the Decision to Invest Abroad 52 6 Horizontal FDI: Brainard (1997) 53 7 Exports vs. FDI with Asymmetric Countries: Markusen and Ven- ables (2000) 63 8 Exports vs. FDI with Heterogenous Firms: Helpman, Melitz and Yeaple (2003) 71 9 VerticalFDI:TheoryandEvidence 82 1III Intermission: The Boundaries of The Firm 96 10 The Theory of The Firm: Transaction-Cost Approaches 97 11 The Theory of The Firm: The Property-Rights Approach 107 12 The Theory of The Firm: Alternative Approaches 120 12.1 The Firm as an Incentive System: Holmstrom and Milgrom (1994) . . . 120 12.2FormalandRealAuthority:AghionandTirole(1997) ......... 126 12.3AuthorityandHierarchies:Rosen(1982) ................. 129 IV Trade and Organizational Form 137 13 Early Transaction-Cost Approaches 138 13.1Ethier(1986) ................................ 138 13.2EthierandMarkusen(1996) ........................ 144 14 The Transaction-Cost Approach in Industry Equilibrium: McLaren (2000) and Grossman and Helpman (2002) 148 14.1GlobalizationandVerticalStructure:McLaren(2000).......... 148 14.2 Integration vs. Outsourcing in Industry Equilibrium: Grossman and Helpman(2002)............................... 154 15 The Property-Rights Approach in International Trade (I): Antràs (2003a) 165 16 The Property-Rights Approach in International Trade (II): Antràs (2003b) and Antràs and Helpman (2004) 182 16.1 Incomplete Contracts and the Product Cycle: Antràs (2003b) ..... 183 16.2 Global Sourcing with Heterogenous Firms: Antràs and Helpman (2004) 194 2Preface These lecture notes review some of the material that I cover in the advanced graduate courseintheInternationalTradethatIteachatHarvardUniversity. Thecoursefocuses on a firm-level approach to international trade and on selected topics in trade policy. I am teaching this class for the firsttimethisSpring,sothenotesarelikelytocontain several typos and mistakes. Comments, suggestions, and corrections would be most welcome. Pol Antràs Department of Economics Harvard University January 2004 3Chapter 1 Introduction and Basic Facts • In Neoclassical Trade Theory, firms are treated as a black box. The supply sideoftheeconomyischaracterizedbyasetofproductionfunctionsaccordingto whichthefactorsofproduction(capital,labor)aretransformedintoconsumption goods. • Moreover, for the most part, the literature assumes constant returns to scale, under which the size of the firm is indeterminate (the general equilibrium only pins down the size of the sector or industry to which the firm belongs). • NewTradeTheoryintroducedincreasingreturnsandimperfectcompetitionin internationaltrade. Thisresolvedtheindeterminacyofthesizeofthefirm. Asan example, take a Helpman-Krugman type of model with product differentiation. The unique producer of a particular varietyω faces demand given by: −ε , ε 1 q(ω)=Ap(ω) and hence setsq(ω) to maximize: q(ω) (ε−1)/ε 1/ε π(ω)=Aq(ω)−−f, ϕ where 1/ϕ is the marginal cost of production andf is a fixed cost. Profits are strictly concave inq(ω), so there is a well-determined profit-maximizing level of ∗ outputq (ω). 4• Still, as discussed below, New Trade Theory cannot account for important facts in the data. A. Firms and the Decision to Export • The Helpman-Krugman models feature complete specialization: each industry variety is produced by a single firm in just one country, whichexports its out- puteverywhereelseintheworld. Adding transport costs could potentially invalidate this, but not if transport costs are of the iceberg type. In that case, we still get a similar result (the elasticity of demand remains unaffected). The transport cost inflates the marginal cost and reduces profits on foreign sales: à X X 1 1/ε 1/ε (ε−1)/ε (ε−1)/ε π (ω)=Aq (ω) +Aq (ω)−q (ω)+τq (ω)−f, j jj jk jj jk j k ϕ k6=jk6=j ∗ but one can show that the optimalq (ω) satisfies jk µ ¶ ε−1 τ 1 (ε−1)ϕ 1/ε (ε−1)/ε ∗∗ Aq (ω)−q (ω)=A 0 for allk = 6j. k k jk jk ϕεετ Hence, even in the presence of transport costs, a firmcontinuestoexportevery- where else in the world. As we will see, two features of this example are crucial: (i)thattransportcostsaffect only the marginal cost, and (ii) that foreign com- petition does not affect the markup the firm can charge over marginal cost. • In reality, not all domestic producers export to foreign markets. And, more importantly, the literature has “uncovered stylized facts about the behavior and relativeperformanceof exportingfirmsandplantswhichholdconsistentlyacross a number of countries” (Bernard et al., 2003, BEJK hereafter). — Exporters are in the minority. In 1992, only 21% of U.S. plants reported exporting anything. — Exporters sell most of their output domestically: around 2/3 of exporters sell less than 10% of their output abroad. 5— Exporters are bigger than non-exporters: they ship on average 5.6 times more than nonexporters (4.8 times more domestically). — Plants are also heterogeneous in measured productivity; Figures 2Aand 2B in BEJK. — Exporters’ productivity distribution is a shift to the right of the nonex- porter’s distribution. Exporters have, on average, a 33%advantage in labor productivity relative to nonexporters. — Thissuggeststhatthemostproductivefirmsself-selectintoexportmarkets, butitcouldalsoreflect learning by exporting (Clerides et al., 1998) • Furthermore, micro-level studies have also found evidence of substantial reallo- cation effects within an industry following trade liberalization episodes. — Exposure to trade forces the least productive firms to exit or shut-down (Bernard and Jensen, 1999; Aw, Chung and Roberts, 2000; Clerides et al., 1998). — Trade liberalization leads to market share reallocations towards more pro- ductive firms, thereby increasing aggregate productivity (Pavcnik, 2002, Bernard, Jensen and Schott 2003). • These studies suggest that successful theoretical frameworks for studying firms and the decision to export should include two features: 1. Within sectoral heterogeneity in size and productivity. 2. A feature that leads only themostproductive firms to engage in foreign trade: — This could be a sunk cost of exporting as documented by Roberts and Tybout(1997)andBernardandJensen(2004),andformalizedbyMelitz (2003); — Oralimitationonproductdifferentiation(i.e.,afixedmeasureofgoods) that leads to worldwide (price) competition in the production of a par- ticular good, which in turn gives rise to variable markups (BEJK). 6• We will study each of these two approaches and revisit the empirical evidence in light of the theories. B. Firms and the Decision to Invest Abroad • Another important fact that traditional trade theory neglects is that firms have (atleast)twomodesofservicingaforeignmarket. Thefirstmodeistheexporting option, which was discussed above. An alternative mode, however, is to set up multiple productionplants toservice the different foreignmarkets (i.e. engage in foreign direct investment, FDI hereafter). This trade-off was first formalized by Markusen (1984). • Multinational firms may also arise when, in the presence of factor price differ- encesacrosscountries,aproducermayfinditoptimaltofragmenttheproduction processandundertakedifferentpartsof theproductionprocessindifferentcoun- tries. This “vertical” approach to the multinational firm was first developed by Helpman (1984). • Why should we care about multinational firms?Becausetheyplayakeyrolein the global economy: — One-third of the volume of world trade is intrafirm trade. In 1994, 42.7 percent of the total volume of U.S. imports of goods took place within the boundariesofmultinationalfirms,withthesharebeing36.3percentforU.S. exports of goods (Zeile, 1997). — Aboutanotherthirdofthevolumeofworldtradeisaccountedforbytransac- tionsinwhichmultinationalfirmsareinoneofthetwosidesoftheexchange. — Still, is this large? Rugman (1988) estimates that the largest 500 multina- tional firms account for around one-fifth of world GDP. • Furthermore,somestylizedfactsaboutmultinationalfirmsandFDI(Markusen 1995, 2003) provide foundations for theorizing: I. Macro Facts 71. FDI has grown rapidly throughout the world, especially in late 1980s and late 1990s. 2. The bulk of FDI flows between developed countries. In 2000, developed countries were the source of 91 percent of FDI flows and also the recipient of79percent(UNCTAD,2001). Furthermore, 80percentoftheinflows into developing countries went exclusively to Hong Kong, China and Korea. 3. Two-way FDI flows are common between pairs of developed countries. 4. ThereexistslittleevidencethatFDIispositivelyrelatedtodifferences in capital endowments across countries; see, however, Yeaple (2003). 5. Political risk and instability deter inward FDI. II. Firm and Industry Characteristics 1. The relative importance of multinational firms varies by industry. The sig- nificanceishigherinsectorsthat: — have high levels of R&D expenditures over sales — employ large number of nonproduction workers — produce new and/or complex goods — have high levels of product differentiation and advertising — featurehighproductivitydispersion(Helpman,MelitzandYeaple,2003) 2. At the firm level, multinationality is: — negatively associated with plant-level scale economies — positively associated with size, up to a threshold size level — positively associated with trade barriers. • We will study different theoretical approaches to explaining these facts. I will refer to these as technological theories of the multinational firm. • Of particular interest will be the contribution by Helpman, Melitz and Yeaple (2003), which combines insights from this branch of the literature together with 8insights from the literature on within sectoral heterogeneity and the exporting decision discussed above. • We will also briefly discuss another branch of the literature that has focused on studying the effects of FDI. C. Firm Boundaries: Trade and Organizational Form • Inrecentyears,wehavewitnessedaspectacularincreaseinthewayfirmsorganize production on a global scale. Feenstra (1998), citing Tempest (1996), describes Mattel’s global sourcing strategies in the manufacturing of its star product, the Barbie doll: Therawmaterialsforthedoll(plasticandhair)areobtainedfromTaiwanand Japan. Assembly used to be done in those countries, as well as the Philippines, but it has now migrated to lower-cost locations in Indonesia, Malaysia, and China. The molds themselves come from the United States, as do additional paints used in decorating the dolls. Other than labor, China supplies only the cotton cloth used for dresses. Of the 2 export value for the dolls when they leave Hong Kong for the United States, about 35 cents covers Chinese labor, 65 cents covers the cost of materials, and the remainder covers transportation and overheads, including profits earned in Hong Kong. (Feenstra, 1998, p. 35-36). • A variety of terms have been used to refer to this phenomenon: the “slicing of the value chain”, “international outsourcing”, “fragmentation of the production process”, “vertical specialization”, “global production sharing”, and many more. • One-thirdofworldtradeisintrafirmtrade,butnoticethatmultinationalfirms choose not to internalize an equally sizeable volume of their transac- tions. In developing their global sourcing strategies, firms not only decide on where to locate the differentstagesofthevaluechain,butalsoontheextentof control they want to exert over these processes. • The internalization issue is nothing more than the classical “make-or-buy” decision in industrial organization. Firms may decide to keep the production 9of intermediate inputs within firm boundaries, thus engaging in FDI when the integrated supplier is in a foreign country, or they may choose to contract with arm’s length suppliers for the procurement of these components. An example of theformerisIntel Corporation,whichassemblesmostofitsmicrochipsinwholly- owned subsidiaries in China, Costa Rica, Malaysia, and Philippines. Conversely, Nike subcontracts most of the manufacturing of its products to independent producersinThailand, Indonesia, Cambodia, andVietnam, whilekeepingwithin firm boundaries the design and marketing stages of production. • The decision to internalize an international transaction also seems to be system- atically related to certain industry and country characteristics. For instance, Antràs (2003a) reports that the share of intrafirm imports in total U.S. imports is larger in R&D and capital intensive sectors. In a cross-section of exporting countries, this share is also significantly larger in imports from capital-abundant countries. • Antràs (2003b) also reviews some evidence from firm-level studies that suggests thatthechoicebetweenintrafirmandmarkettransactionsissignificantlyaffected by both the degree of standardization of the good being produced abroad and also by the domestic firm’s resources devoted to product development. • The previously discussed approaches to the multinational firm share a common failure to properly model the crucial issue of internalization. These models can explain why a domestic firm might have an incentive to undertake part of its production process abroad, but they fail to explain why this foreign produc- tion will occur within firm boundaries (i.e., within multinationals), rather than through arm’s length subcontracting or licensing. In the same way that a the- ory of the firm based purely on technological considerations does not constitute a satisfactory theory of the firm (cf., Tirole, 1988, Hart, 1995), a theory of the multinational firm based solely on economies of scale and transport costs cannot be satisfactory either. • Inthissectionwewillinsteaddiscusspurelyorganizationalorcontractualtheories 10of the multinational firm.Wewillalsoreviewthetheoriesofthe firm that serve as basis for these new approaches to the multinational firm. • Of particular interest will be the contribution by Antràs and Helpman (2003), which combines insights from this branch of the literature together with insights from the literature on intraindustry heterogeneity. 11Part I Firms and the Decision to Export 12Chapter 2 Intraindustry Heterogeneity with Fixed Costs of Exporting: Melitz (2003) • As argued in the Introduction, the available empirical studies suggest that suc- cessfultheoreticalframeworksforstudyingfirmsandthedecisiontoexportshould incorporate intraindustry heterogeneity in size and productivity. This chapter and the next present two recent theoretical frameworks that elegantly introduce such heterogeneity in otherwise standard models of international trade. • I follow Melitz in discussing first the closed economy model and then moving on to the open economy model. The Closed Economy Model • On the demand side, there is a representative consumer with preferences: ⎡⎤ 1/ρ Z ρ ⎣⎦ U =q(ω)dω , 0 ρ 1, (2.1) ω∈Ω whereΩ denotes the measure of available products andσ=1/(1−ρ) 1 is the constant elasticity of substitution. We will focus on stationary equilibria, so we 13droptimesubscripts. Consumersmaximize(2.1)subjecttothebudgetconstraint Z p(ω)q(ω)dω =R. ω∈Ω It is well-known (prove it yourselves) that this leads to the following demand function for a particular varietyω: µ ¶ −σ Rp(ω) q(ω)= , (2.2) PP where ⎡⎤ 1/(1−σ) Z 1−σ ⎣⎦ P =p(ω)dω . ω∈Ω Because consumers value variety, they are willing to consume positive (although lower) amounts of even relatively expensive varieties. • The supply side is characterized by monopolistic competition. Each variety is produced by a single firm (so we hereafter index varieties byϕ) and there is free entry into the industry. Firms produce varieties under a technology that featuresaconstantmarginalcostandafixedoverheadcostintermsoftheunique composite factor of production (labor), which we take as numeraire. The fixed cost is assumed identical across firms and we denote byf. So far the set up is identical to Krugman (1980). Here are the distinguishing features: 1. Themarginal cost is assumed to vary across firms and is denoted by 1/ϕ, i.e. q(ϕ) TC(ϕ)=f + (2.3) ϕ Firmswithhigherϕarethereforemoreproductive,inthesensethattheyneedto 1 hirefewerworkerstoattainagivenamountofoutput. Higherproductivityfirms also charge lower prices, produce more output, and obtain both higher revenues r(ϕ) and higher profitsπ(ϕ). To see this, notice that with CES preferences, the 1 Ahigherϕ can also be interpreted as higher quality varieties (see Melitz, 2003). 14profit-maximizing price is a constant mark-up over marginal cost: 1 p(ϕ)=, (2.4) ρϕ which by way of (2.2) implies: σ σ−1 q(ϕ)=RP (ρϕ) σ−1 r(ϕ)=p(ϕ)q(ϕ)=R(Pρϕ) (2.5) 1 π(ϕ)=r(ϕ)−f, (2.6) σ where remember thatR andP are common across firms. 2. The other additional assumption is that prior to entry,firms face uncertainty as to how productive they will turn out to be. In particular, to start producing a particular variety a firm needs to bear a fixed cost consisting of f units of labor. Upon paying this sunk cost, the firm draws its productivity e levelϕ from a known distribution with pdfg(ϕ) and associated cdfG(ϕ).After observingthisproductivitylevel,theproducerdecideswhethertoexitthemarket immediatelyorstartproducingaccordingtothetechnologyin(2.3). Inthelatter case, in every period, the firm faces a probabilityδ of exogenous exit, which is common across firms. • Letusnextturntotheequilibrium of the closed economy. Consider firstfirm behavior. Since we focus on steady state equilibria, a firm with productivityϕ earns profitsπ(ϕ) in each period, until it is hit by a shock, at which point it is forced to exit. Hence, a firm that is contemplating starting production expects a (probability) discounted value of profits of ( ) ½ ¾ ∞ X 1 t−s v(ϕ)=max 0, (1−δ)π(ϕ) =max 0,π(ϕ) , (2.7) δ t=s whereweimposethatifa firm anticipates stationary negative operating profits, it will choose to exit the market upon observingϕ.Itisclearfrom(2.6)and ∗ (2.7) that there is a unique threshold productivityϕ such thatv(ϕ) 0 if and 15π (ϕ) π (ϕ) x σ−1 σ−1 (ϕ ) (ϕ ) x σ−1 ϕ -f -f x Figure 2.1: Firm Behavior ∗ only if ϕ ϕ.Thisimpliesthata firm will remain in the market and produce if and only if it is sufficiently productive. Following Helpman, Melitz and Yeaple (2003) and Antràs and Helpman (2003), Figure 2.1 illustrates the equilibrium. σ−1 Notice that profits are proportional toϕ ,andthatπ(0) =−f. • Consider next the industry equilibrium, where we solve for the endogenously determined measureM of firms (and varieties), as well as for the distribution of (active firms’) productivities in the economyµ(ϕ). We follow Melitz (2003) ∗ in expressing all the equilibrium conditions in terms of the cutoffϕ and then obtainingtheremainingvariablesofinterestfromit. Forthatpurpose,itisuseful to start by defining the weighted average productivity measure, ∙ ¸ Z 1/(σ−1) ∞ σ−1 ϕe =ϕµ(ϕ)dϕ , 0 which, as we will see, completely summarizes the relevant information in the 16 0distributionofprobabilities. Noticethattheconditionaldistributionµ(ϕ)equals: ( g(ϕ) ∗ ifϕ≥ϕ ∗ 1−G(ϕ ) µ(ϕ)= , 0 otherwise from which ∙ Z ¸ 1/(σ−1) ∞ 1 ∗σ−1 ϕe(ϕ )=ϕg(ϕ)dϕ , (2.8) ∗ 1−G(ϕ )∗ ϕ ∗ and henceϕe is uniquely pinned down byϕ and the exogenous (unconditional) distributionsg(ϕ) andG(ϕ). Next, we can define average profitsπ =π(ϕe) as à µ ¶ µ ¶ σ−1σ−1 ∗∗∗ r(ϕe)ϕe(ϕ )r(ϕ )ϕe(ϕ ) π =−f =−f =f−1,(ZCP) ∗∗ σϕσϕ (2.9) ∗ 2 where we have used (2.5), (2.6) andπ(ϕ )=0. Finally, free entry ensures that, in the industry equilibrium, the expected dis- 3 counted value of profits for a potential entrant equal the fixed cost of entry, or Z ∞ δf e v(ϕ)g(ϕ)dϕ =f⇔π = . (FE) (2.10) e ∗ 1−G(ϕ ) 0 Notice that (2.9) and (2.10) form a system of two equations in two unknowns ∗ 0∗ π andϕ . BecauseG (ϕ ) 0, it is clear that along the FE scheduleπ is an ∗∗ ∗ increasing function ofϕ and satisfiesπ(0) = δf and limπ(ϕ)=∞.In- eϕ→∞ tuitively, for a given expected value of entryf , the probability of success should e be decreasing in the average profit levelπ. Hence, an increase inπ should be ∗ matchedbyanincreaseinϕ . On the other hand, Melitz (2003) shows that the FE curve is cut by the ZCP curve only once from above, thus ensuring the existence anduniqueness of the equilibrium. Furthermore, under common distri- ∗ butions, the ZCP schedule is downward sloping in the space (π, ϕ ) (see Figure R ∞ 2 Notice that we refer to these as average profits becauseπ =π(ϕ)µ(ϕ)dϕ (go ahead and 0 prove it). R R R ∞∞∞ 3 1∗ 1 Notice thatv(ϕ)g(ϕ)dϕ =π(ϕ)g(ϕ)dϕ =1−G(ϕ )π(ϕ)µ(ϕ)dϕ = ∗∗ 0ϕδδϕ 1∗ 1−G(ϕ )π. δ 17∗ 2.2 below). Intuitively, an increase inϕ will increase the average productivity of the surviving firms Z 2−σ∞ ∗ ¡ ¢ g(ϕ )ϕe 0σ−1 ∗σ−1∗ ϕe(ϕ ) =ϕ−(ϕ )g(ϕ)dϕ 0. 2 ∗ ∗ (σ−1)(1−G(ϕ )) ϕ ∗ Because profits tend to increase with a firm’s productivity, an increase inϕ will have a direct positive effect on profitsπ. But because firm profits are decreasing in the productivity of rivals, there is also an additional effect that goes in the oppositedirection. IfthedistributionG(ϕ)hasafatenoughrighttail, thelatter effect will dominate and the ZCP will be downward sloping. An interesting case ³ ´ k b is that of a Pareto distribution, i.e.,G(ϕ)=1− ,whichyields ϕ ⎡⎤ 1/(σ−1) µ ¶ Z k−1 ∞ 1b ⎢⎥ ∗σ−1 ϕe(ϕ)=ϕ kbdϕ = ⎣⎦ ³ ´ k ∗ϕ b ϕ ∗ ϕ à 1/(σ−1) ∙ ¸ Z 1/(σ−1)σ−1 ∞ ∗ k(ϕ ) k ∗σ−k =k(ϕ )ϕdϕ = , σ−k+1 ∗ ϕ and the ZCP schedule is flat. ∗ • Oncewehavetheequilibriumvaluesofϕ andπ, wecaneasilysolvefortheequi- librium number of firms. Notice that the identical price index in (2.2) becomes simply: ∞ Z Z 1−σσ−1σ−1 1−σ P =p(ω)dω = (ρϕ) Mµ(ϕ)dϕ =M (ρϕe) , ω∈Ω 0 and hence, 1R π =−f. σM 4 Finallynoticethattheequalityofincomeandexpenditure(R =L)impliesthat: L M = , (2.11) σ(π +f) 4 δM In particular,R =Π+L =Mπ +L =f +L =Mf +L =L +L =L. pp∗epeepep 1−G(ϕ ) 18which completes the characterization of the stationary equilibrium of the closed economy. • Notice the following features of the equilibrium: ∗ —ϕ, eϕ,π andµ(ϕ) are independent ofL,whileM is proportional to country size. — Welfareisgivenby ⎛⎞⎛⎞ 1/ρ 1/ρ Z Z µ ¶ ρ L ρσ 1/(σ−1) ⎝⎠⎝⎠ U =q(ω)dω = (ρϕ)µ(ϕ)dϕ = LMρϕe σ−1 M (ρϕe) ω∈Ωω∈Ω — Notice that the aggregate outcome predicted by the model is identical to that generated by a Krugman (1980) model with homogenous firms with productivityϕe. This shows how nicely the model aggregates the sectoral heterogeneity. The Open Economy Model • With this machinery at hand, we can now move to the open economy version ofthemodelandanalyzetheexportingdecisionaswellasthereallocationeffects generatedby trade. If trade openingis just an increase inthe relevant size of the economy, then we know that all firms will export and also, from the equilibrium above, that trade will have no impact on average productivity (see, however, footnote 16 as well as Melitz and Ottaviano, 2003, for the importance of CES preferencesfortheseresults). Melitz(2003)thusintroducestradefrictions. These are of two types: 1. A standard per-unit iceberg costs, so thatτ units need to be shipped for 1 unit to make it to any foreign country; 2. Aninitialfixedcostoff unitsoflabortostartexporting, which is incurred ex once the firm has learnedϕ. 19