Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Lecture 5: Multinational Enterprises and Foreign Direct Investment Gregory Corcos Eco572: International Economics
15 October 2014
Introduction
HFDI Proximity/Concentration
Outline
1
Definitions and Stylized Facts
2
Horizontal FDI: The Proximity-Concentration Trade-Off
3
Vertical FDI: Differences in factor prices
VFDI Input Costs
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Definitions
Foreign Direct Investment (FDI) is the purchase of a lasting management interest in a non-resident company. In practice, FDI includes all investments of at least 10% of a non-resident firm’s equity capital. Greenfield versus Mergers & Acquisitions investment: - Greenfield investment: creation of a new firm - M&A: purchase of an existing firm - M&As represented roughly 1/3 of all FDI projects in 2013 (UNCTAD).
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Definitions (2)
Multinational entreprises (MNE) are firms that own and operate production facilities in more than one country. The parent firm is the legal entity that owns the foreign firm. The affiliate is the foreign firm. International trade flows between a parent and its foreign affiliate are called intrafirm trade.
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Definitions (3)
Horizontal versus vertical FDI: - Horizontal : MNE and affiliate sell the same products (alternative to exports). Mostly North-North. Ex: Toyota Motor Sales in the US. - Vertical: the MNE sells an input to the affiliate or vice-versa. The MNC splits the production process across borders, usually to exploit factor price differences. Mostly from an industrial country to a developing country. Ex: Intel’s assembly and test plant in Costa Rica.
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Stylized facts about MNEs
There are about 77,000 MNEs in the world and 770,000 foreign affiliates (UNCTAD). In 2013, foreign affiliates were responsible for around 10% of the world GDP ($7.5 tn), 1/3 of trade in goods and services ($ 7.7 tn) and they were employing 70.7 million people. In developed countries, foreign-owned companies represent roughly 20-25 percent of manufacturing employment. A large share of privately funded R&D is conducted by MNEs. Most statistics on FDI and MNEs come from UNCTAD’s TNC database and World Investment Report.
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Stylized facts about FDI flows 2.5E+12
2E+12
1.5E+12
World High Income
1E+12
Asia Low income
5E+11
0 1970
1975
1980
1985
1990
1995
2000
2005
2010
-5E+11
FDI Net Inflows in current US dollars. Source: World Bank.
Over the period USA, UK, France, Germany, Netherlands, Canada, have the largest inflows, but China was the second FDI recipient in 2013.
Introduction
HFDI Proximity/Concentration
Stylized facts about FDI flows (2) Figure: Map of Foreign Direct Investment in 2000
Source: Feenstra & Taylor from OECD and UN data
VFDI Input Costs
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Stylized facts about FDI flows (3) 2. Two-way FDI flows are common between pairs of developed Figure: German FDI, by partner country (2001) countries
Source: Buch et al. (2005).
Interpretation: Grubel-Lloyd index measures the intensity of two-way activities (≈ horizontal FDI) based on data on outward and inward sales from/to German MNEs Activities with developed countries take place mainly on a 2-way basis
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
The OLI framework Hymer (1965), Dunning (1978) explain why MNEs exist instead of other arrangements: Ownership: MNEs need to have some competitive advantage to operate in an unfamiliar environment (Technology, Economies of scale, Specific factors, ...) Location: There needs to be a motive to operate in more than one location (Input price differences, Quality of inputs, Transport costs, Market access...) Internalization: Some market failure must explain why the MNEs owns foreign plants instead of importing from independent suppliers (Transaction costs, Moral hazard on quality and quantity, Uncertainty, ...)
Introduction
HFDI Proximity/Concentration
Horizontal FDI The proximity-concentration trade-off
VFDI Input Costs
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Evidence on horizontal FDI
Table: 2004 Sales of Affiliates of US MNEs, $ billion (Source: BEA)
Total Local To US To other foreign
All Affiliates 3,312 100% 2,051 62% 355 11% 907 27%
Manuf. Affiliates 1.560 100% 895 57% 203 13% 462 30%
By comparison, total US exports $806 billion, by US parents $373 billion, intrafirm $145 billion ⇒ Most sales of foreign affiliates are realized in the foreign market
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
The Proximity-Concentration Framework
Markusen (1984), Horstmann and Markusen (1989), Brainard (1993) Firms invest abroad to avoid trade costs (proximity benefit). Firms prefer exporting to avoid duplicate fixed costs (concentration benefit). When entry barriers are high or destination markets are large, each firm can sell more and marginal costs become more important relative to fixed costs.
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Assumptions
Two identical countries with labor endowments L. Wages normalized to unity. Free entry in the differentiated good sector. CES preferences with elasticity of substitution σ. Firm-level fixed cost fE (headquarter services). Plant-level fixed cost fI . Homogenous productivity across entrants. Iceberg trade cost τ > 1.
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Preliminaries Demand for variety ω in country i: pi (ω) −σ Ei xi (ω) = Pi Pi with Ei = wi Li the nominal expenditure in country i and Z 1 1−σ 1−σ pi (ω) dω Pi = Optimal prices for a firm facing marginal cost ci (ω) of serving country i: σ ci (ω) pi (ω) = σ−1 Revenues on i’s sales: 1−σ σ pi (ω)xi (ω) = ci (ω) wi Li Piσ−1 σ−1
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
The proximity-concentration trade-off
Proximity benefit: The marginal cost of serving the foreign market is higher than for the domestic market: τ if exports ci (ω) = 1 if domestic sales The benefit is greater, the higher the sales. Concentration benefit: The fixed cost is higher for multinational firms than for exporters: fE + fI if exports f = fE + 2fI if domestic sales The benefit is greater, the higher plant-level fixed costs
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Profit comparison FDI is prefered over export if πiM > πiE ie if: 1 σ
σ σ−1
1−σ
Pi 0 σ−1 Ei 0 (1 − τ 1−σ ) − fI > 0
More likely when i) MPi 0 = Pi 0 σ−1 Ei 0 , the foreign market potential, is large, ii) τ , the marginal transportation cost, is high, iii) fI , the plant-level fixed cost, is low. Market potentials are pinned down by the (firm-level) entry costs and the size of the population. Note that the only mixed equilibrium in which FDI and export coexist corresponds to the knife-edge case where πiM = πiE
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Equilibrium 1. Assume all firms export: Free-entry implies πiE = 0 and πiE0 = 0 ie: 1−σ σ τ 1−σ MPi 0 σ−1 1−σ 1−σ σ 1 σ 1 MPi 0 + τ 1−σ MPi σ σ−1 σ σ−1 1 σ
σ σ−1
1−σ
MPi +
1 σ
=
fE + fI
=
fE + fI
which implies: MPi = MPi 0 = σ
σ σ−1
σ−1
fE + fI 1 + τ 1−σ
This is an equilibrium if no firm has an incentive to produce abroad (πiM < 0 given MPi ): 2(fE + fI ) − fE − 2fI < 0 1 + τ 1−σ fE > fE + 2fI
πiM = ⇒
τ 1−σ
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Equilibrium (2) 2. Assume all firms invest abroad: Free-entry implies πiM = 0 and πiM0 = 0 ie: 1−σ σ MPi 0 σ−1 1−σ 1−σ σ 1 σ 1 MPi 0 + MPi σ σ−1 σ σ−1 1 σ
σ σ−1
1−σ
MPi +
1 σ
= fE + 2fI = fE + 2fI
which implies: MPi = MPi 0 = σ
σ σ−1
σ−1
fE + 2fI 2
This is an equilibrium if no firm has an incentive to export (πiE < 0 given MPi ): πiE = ⇒ τ 1−σ
(fE + 2fI )(1 + τ 1−σ ) − fE − fI < 0 2 fE < fE + 2fI
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Equilibrium (3)
FDI is an equilibrium if: Variable transportation costs are high Plant-level fixed costs are low Firm-level fixed costs are high (through a positive effect on the country’s market potential)
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Testing the model
Brainard (AER, 1997) Confidential US FDI data from the Bureau of Economic Analysis. Estimated equation: ij ShX ij ShX
+
ij ShM
= βτ ln τhij + βfc ln FCh + βE ln Endowj + εijh
where the dependent variable measures the ratio of exports over (Exports+local affiliate sales) by country and industry
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Testing the model (2) Table: Estimation results, Brainard 1997
FREIGHT=freight costs() (charges toTwo: import value), TARIFF=ad valorem tariffs10at / 64the industry Lecture Trade Costs and Horizontal FDI Stephen Ross Yeaple level, PWGDP=Per capita GDP, TAX=Average effective tax rate, TRADE=Measure of trade barriers (include NTBs) country-specific, FDI=Barriers to FDI country-specific, PSCALE=Nber of employees in the median plant as a proxy for plant-level economies fo scale, CSCALE=Advertising and R&D intensity as a proxy for corporate scale economies, ADJ=Dummy for common border, LANG=Dummy for same language, EC=Dummy for membership to the European Community
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Testing the model (3)
High freight and tariff rates are associated with a shift in multinational production relative to exports High plant-level costs are associated with less MP High corporate scale economies are associated with more MP Large differences in endowments are associated with less MP (need to extend the model to asymmetric countries) Extension (Helpman, Melitz & Yeaple 2004): heterogeneous firms invest, export or sell only on the domestic market according to productivity.
Introduction
HFDI Proximity/Concentration
Vertical FDI: Differences in factor prices
VFDI Input Costs
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Evidence on vertical FDI 2. Pourquoi s’internationaliser
2. Pourquoi s’internatio
The previous model assumes FDI and exports are substitutable ways to 2.4. IDE Verticaux 2.4. IDE Verticaux 2.4. IDE Ve 2.4. IDE Ve reach foreign markets. But simple correlations between FDI and trade flows suggest some complementarity.
•
Mais les flux d’IDE semblent plutôt complémentaires du • Mais les flux d’IDE semblent plutôt complément commerce international commerce international commerce international commerce international
Source: Toubal based on Brainard (1997)
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Helpman (1984): Assumptions
Two countries, Home and Foreign Two factors: Labor (L and L∗ ) and a general purpose input (H and H ∗ ). Mobile across sectors but immobile internationally Two goods: - Homogenous, CRS good Y produced under perfect competition: cY (wL , wH ) = 1 (numeraire). No fragmentation - Differentiated, IRS good X produced under monopolistic competition. Fragmentation - Costlessly traded
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Assumptions (2) Each country is inhabited by a representative consumer with homothetic preferences (identical in both countries) u(Y , X ) = Y α X 1−α hR
σ−1 σ
i
σ σ−1
is the CES subutility level where X = x(ω) dω attained in the consumption of differentiated varieties. With CES preferences, the demand addressed to a differentiated good producer is: p(ω) −σ (1 − α)E x(ω) = PX PX where PX is the CES price index and E is nominal revenue
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Assumptions (2)
Good X is produced with labor and the general purpose input H. Each firm needs headquarters, produced by adapting h units of H at cost CH (wL , wH , h) = g (wL , wH , h) + wH h. Headquarters are firm-specific, but common to all of the firm’s plants. Each firm produces the differentiated good under increasing returns to scale, at cost CP (wL , wH , h, x) = fp (wL , wH ) + g1 (wL , wH , x, h), where fP is a plant-specific fixed cost and g1 () is linearly homogenous in (h, x). ⇒ Single-plant total cost function: CX (wL , wH , x) ≡ minh [CP (wL , wH , h, x) + CH (wL , wH , h, x)]
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Industry equilibrium In sector Y , producers engage in marginal cost pricing (PY = 1) In sector X , firms equate marginal revenue to marginal costs and free entry brings zero profits: ∂CX (wL , wK , x) ∂x px − CX (wL , wH , x) = 0 MR(p, n) =
where p is the optimal price, n is the total number of producers, MR() is marginal revenue ⇒ R(p, n) = θ(wL , wH , x) p where R(p, n) = MR(p,n) is the ratio of the average and marginal revenues and θ(wL , wH , x) = the marginal cost
CX /x ∂CX /∂x
the ratio of the average and
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Market clearing conditions Factor markets clear: aLY (wL , wH )Y + aLX (wL , wH , x)nx = L aHY (wL , wH )Y + aHX (wL , wH , x)nx = H
By Shepard’s lemma, the cost-minimizing input level of factor ∂c () i per unit of good j is given by aij = ∂wj i Together with the firms’ optimal behavior, market-clearing conditions determine the equilibrium values of wL , wH , h, p, x, Y and n, in a closed economy or in an integrated world in which factors are mobile.
Introduction
HFDI Proximity/Concentration
Integrated economy Assume that the homogeneous good is the most labor-intensive: ∂CX /∂wH ∂CY /∂wH > ∂CX /∂wL ∂CY /∂wL Further assume that, within sector X , the production of headquarter services is more intensive in the H factor: ∂CP /∂wH ∂CH /∂wH > ∂CH /∂wL ∂CP /∂wL Finally, assume home is relatively rich in factor H H H∗ > ∗ L L
VFDI Input Costs
Introduction
HFDI Proximity/Concentration
Integrated economy (2) Figure: Patterns of Production in the set OXO ∗
Source: Antras (2005): http://www.courses.fas.harvard.edu/~ec2535/Lectures_Notes/Lecture_9.pdf
Source: Antras, 2005 OY OX OH HX
is is is is
employment employment employment employment
of of of of
factors factors factors factors
in in in in
sector sector sector sector
Y X X for headquarter services X for plant production
VFDI Input Costs
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Integrated economy (3)
Without any fragmentation, the model is essentially a 2 × 2 × 2 Helpman-Krugman model: Home specializes in X For endowment points in the OXO ∗ set, trade causes (relative) factor price equalization. Firms do not benefit from fragmenting their production process. ⇒ As long as relative factor endowments are similar enough, multinational firms do not emerge.
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Fragmented economy Figure: Patterns of Production in the set OHX Figure 2: Pattern of Production in the set OHX L* O* B
K
X
E
Em
H EH C
Y K* wL/wK
B’
O L
Source: Antras, 2005
• Figure 2 depicts such an equilibrium. The vectors OEH and EH E correspond
to factor employment at Home in the production of headquarter services and in plant production, respectively. Home does not produce homogenous goods. On
OEH is employment ofhand, factors at home in the production headquarter services the other Foreign produces the whole world demand for good Y of (vector O∗ X), and also employs a positive amount of factors in the production of both EH E is employment of factors at home in plant production h and x. EEm measures factor usage in the production of x in firms with headquarters at • Notice that the vector EEm measure factor usage in the production of x correHome but producing x in Foreign sponding to firms for which headquarter services are produced at Home, whereas
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Fragmented economy (2)
For endowment points in the OHX set, if headquarter services and plant production are located in the same country, factor prices fail to equalize. Given H/L > H ∗ /L∗ , the wL /wH ratio is higher at home than abroad. For high enough factor price differences, firms have an incentive to fragment their production process and locate the production of headquarter services in Home while undertaking plant production in Foreign. Fragmentation will tend to push the demand of labor up in Foreign and reduce it at home, thus creating an additional force towards convergence in factor prices.
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Fragmented economy (3)
FDI emerges when factor prices are different across countries ⇒ Fragmentation improves the allocation of ressources. The measure of multinational firms is increasing in relative factor endowment differences. The model implicitly features intrafirm trade (in headquarter services). The more different the countries, the greater the share of intrafirm trade.
Introduction
HFDI Proximity/Concentration
VFDI Input Costs
Testing the Helpman model: Yeaple (2003) Tests Helpman’s main prediction that FDI should occur between countries with large factor endowment differences in the host country’s comparative advantage industries. This prediction is applied to skilled labor: FDI should flow in skill-abundant countries’ skill-intensive industries. Estimated equation: FDIij
= βτ τij + βfc FCi + βMS MSizej + βE HCj +βFI SKi + βI HCj × SKi + εij
FC : fixed costs. HCj : relative human capital abundance of country j. SKi skilled-labor intensity of industry i. We expect βI > 0.
Introduction
HFDI Proximity/Concentration
Testing the model (2)
Table: Estimation results, Yeaple 2003
Source: Yeaple (2003)
VFDI Input Costs