Antràs Property-Rights Theory of the Multinational Firm

The Antràs property-rights theory of the multinational firm is a foundational model in international trade, organizational economics, and business and economics theory introduced by Pol Antràs in his 2003 Quarterly Journal of Economics article, "Firms, Contracts, and Trade Structure," and extended to heterogeneous firms by Pol Antràs and Elhanan Helpman in their 2004 Journal of Political Economy article, "Global Sourcing." Roughly one-third to one-half of all international trade takes place within the boundaries of multinational corporations (intrafirm trade between parent headquarters and foreign affiliates), yet classical Heckscher–Ohlin and Helpman–Krugman monopolistic-competition trade models determine the volume and factor content of trade while leaving firm boundaries indeterminate. By embedding the Grossman–Hart–Moore property-rights theory of incomplete contracts and relationship-specific investments into a Dixit–Stiglitz monopolistic-competition trade equilibrium, Antràs (2003) accounted for two systematic empirical regularities in U.S. customs and Bureau of Economic Analysis (BEA) data: (1) across manufacturing industries, the share of intrafirm imports in total U.S. imports rises strongly with the capital intensity (and R&D intensity) of the exporting industry; and (2) across trading partners, the share of intrafirm imports rises with the capital-labor endowment ratio of the exporting country. When contracts are incomplete and both the final-good producer (headquarters, contributing capital K with output elasticity η) and the component supplier (contributing labor L with elasticity 1 − η) make non-contractible relationship-specific investments, vertical integration (V, foreign direct investment) strengthens the headquarter firm's ex-post bargaining threat point (δ^α > 0) at the expense of the supplier's incentive. Consequently, when production is capital-intensive (η > η^*), protecting the headquarter's capital investment dominates and firms choose vertical integration (FDI); when production is labor-intensive (η < η^*), motivating the supplier's labor investment dominates and firms choose arm's-length foreign outsourcing (O).

Prior to Antràs (2003), theories of the multinational enterprise (such as the Dunning OLI paradigm and Markusen's knowledge-capital model) treated internalization primarily as a mechanism to protect proprietary intangible knowledge from dissipation. However, BEA and U.S. Census data on related-party (intrafirm) vs. non-related-party (arm's-length) trade revealed two cross-industry and cross-country patterns:

1. Industry-Level Pattern (Capital-Intensive Goods Are Traded Intrafirm): Across 23 manufacturing industries in the BEA survey, the share of U.S. imports occurring between related parties (intrafirm trade) is strongly positively correlated with the industry's capital intensity (log(K/L)) and R&D intensity. Capital-intensive sectors (chemicals, pharmaceuticals, motor vehicles, machinery) exhibit intrafirm import shares of 55% to 75%, whereas labor-intensive sectors (apparel, footwear, textiles, furniture) are imported predominantly at arm's length from independent foreign suppliers (intrafirm shares below 10% to 15%).
2. Country-Level Pattern (North–North vs. North–South Trade): Across 28 exporting countries, the share of U.S. imports conducted within multinational firm boundaries rises strongly with the exporting country's capital-labor endowment ratio (K_j / L_j).

Antràs (2003) explained both patterns by combining two postulates: (1) international input transactions are governed by incomplete contracts where ownership of the input (vertical integration V vs. outsourcing O) allocates residual control rights; and (2) final-good parent firms (H) have a comparative advantage in supplying relationship-specific capital (K), whereas local suppliers (S) supply relationship-specific labor (L).

Preferences, Revenue, and Cobb–Douglas Production Technology

Consider an industry producing differentiated varieties ω ∈ [0, n] under Dixit–Stiglitz monopolistic competition. Consumers maximize the CES sub-utility index Y = [∫₀ⁿ y(ω)^α dω]^1/α with α ∈ (0, 1), so the elasticity of substitution between varieties is σ = 1 / (1 − α) > 1. Each variety faces inverse demand p(ω) = A y(ω)^α−1, where A > 0 is an industry demand shifter, generating total sales revenue R(ω) = p(ω)y(ω) = A y(ω)^α.

Production of each differentiated final variety y requires a customized, relationship-specific intermediate input x (y = x), produced jointly by two risk-neutral parties:

1. Final-Good Producer / Headquarters (H): Supplies relationship-specific capital inputs K(ω) (specialized machinery, tooling, blueprints, and chemical equipment) at rental cost r per unit of capital.
2. Intermediate Supplier / Operator (S): Supplies relationship-specific labor and managerial inputs L(ω) at wage rate w per unit of labor.

Output of the customized component y(ω) is governed by a constant-returns-to-scale Cobb–Douglas production function with capital intensity parameter η ∈ (0, 1):

Substituting y(ω) into the Dixit–Stiglitz revenue function R = A y^α yields the joint relationship revenue as a function of the two non-contractible investments (K, L):

Because the investments K and L are customized to variety ω, their quality attributes cannot be verified by an outside court ex ante. Hence, H and S choose K and L non-cooperatively, and then bargain over the resulting revenue R(K, L) ex post once the input has been produced.

Ownership Regimes, Outside Options, and Generalized Nash Bargaining

Prior to investing (t = 0), headquarters H chooses the ownership structure (organizational mode) k ∈ {V, O}, which allocates the residual rights of control (legal ownership of the intermediate input x) following Grossman and Hart (1986) and Hart and Moore (1990):

• Supplier's Outside Option (Always Zero): If ex-post bargaining breaks down, the supplier S cannot sell the customized input x to any third party without H's downstream assembly assets, so the supplier's disagreement payoff is 0 under both ownership modes (O_S^V = O_S^O = 0).
• Headquarters' Outside Option Under Outsourcing (k = O): Under arm's-length outsourcing (O), the independent supplier S legally owns the intermediate input x. If ex-post bargaining fails, S withholds the input and H earns outside option 0 (O_H^O = 0).
• Headquarters' Outside Option Under Vertical Integration (k = V): Under vertical integration (V, where S is an affiliate manager of H), headquarters H legally owns the intermediate input x. If ex-post bargaining fails, H has the legal right to fire the supplier manager S and seize the produced input x, recovering a fraction δ ∈ (0, 1) of usable output (δy). Substituting δy into the revenue function R = A(δy)^α gives H's disagreement payoff under vertical integration:

In the ex-post Nash bargaining game (where H has primitive bargaining weight β ∈ (0, 1), with β = 1/2 in Antràs 2003), each party receives its outside option plus its bargaining share β (or 1 − β) of the quasi-rent surplus R − O_H^k − O_S^k:

1. Ex-Post Revenue Shares Under Vertical Integration (k = V): Since O_H^V = δ^αR and O_S^V = 0, headquarters H captures share β_V of total revenue R, and the integrated supplier S captures 1 − β_V:

1. Ex-Post Revenue Shares Under Arm's-Length Outsourcing (k = O): Since both outside options are zero (O_H^O = O_S^O = 0), headquarters H captures share β_O = β, and the independent supplier S captures share 1 − β_O = 1 − β.

Comparing the two regimes yields the Property-Rights Ranking of Bargaining Shares:

Vertical integration gives headquarters H a larger ex-post slice of revenue (β_V > β_O), whereas outsourcing gives the supplier S a larger ex-post slice of revenue (1 − β_O > 1 − β_V).

Non-Cooperative Investment Choices (K_k, L_k) Under Hold-Up

Given organizational mode k ∈ {V, O} with headquarter revenue share β_k ∈ {β_V, β_O}, at the investment stage H chooses capital K to maximize β_kR(K, L) − rK, and S simultaneously chooses labor L to maximize (1 − β_k)R(K, L) − wL:

Because β_k < 1 and 1 − β_k < 1, both H and S underinvest relative to the first-best benchmark. Moreover, raising β_k increases H's capital investment K(β_k) while reducing S's labor investment L(β_k). Solving the two first-order conditions simultaneously and noting that ex ante (t = 0) headquarters H extracts ex-post rents from S via an upfront lump-sum participation transfer, H chooses the ownership mode k ∈ {V, O} that maximizes total joint operating profit Π(β_k, η) = R(K_k, L_k) − rK_k − wL_k, which is proportional to the organizational efficiency multiplier Ψ(β_k, η):

Optimal Headquarter Share β^*(η) and the Cutoff Theorem

Define the relative profitability ratio of vertical integration (V) over outsourcing (O): Θ(η) ≡ Ψ(β_V, η) / Ψ(β_O, η). Taking the derivative of ln Θ(η) with respect to capital intensity η shows that Θ′(η) > 0 for all η ∈ (0, 1), with Θ(0) < 1 and Θ(1) > 1. By the Intermediate Value Theorem, there exists a unique capital-intensity cutoff η^* ∈ (0, 1) where Θ(η^*) = 1:

Theorem 1 (Antràs's Capital-Intensity Cutoff Theorem for Firm Boundaries, 2003): There exists a unique threshold capital intensity η^* ∈ (0, 1) (with η^* ∈ (1/2, 1) when β = 1/2) such that:

1. Capital-Intensive Industries (η > η^*): Θ(η) > 1, so Vertical Integration (k = V, Foreign Direct Investment / Intrafirm Trade) dominates outsourcing. Because marginal product depends heavily on headquarters' capital investment K, ownership of the input is assigned to headquarters (β_V > β_O) to mitigate H's underinvestment.
2. Labor-Intensive Industries (η < η^*): Θ(η) < 1, so Arm's-Length Outsourcing (k = O) dominates vertical integration. Because marginal product depends heavily on the supplier's labor and effort L, ownership is assigned to the independent supplier (1 − β_O > 1 − β_V) to give the supplier stronger incentives.

Cross-Country Intrafirm Trade Shares in General Equilibrium

To explain the cross-country correlation between exporting countries' capital-labor ratios (K_j / L_j) and the share of intrafirm imports, Antràs (2003) embedded this incomplete-contracts organizational choice into a general-equilibrium 2 × 2 × 2 Helpman–Krugman trade model with a capital-intensive Sector 1 (η₁ > η^*, organized via vertical integration V) and a labor-intensive Sector 2 (η₂ < η^*, organized via outsourcing O). In a factor-price-equalization world trading equilibrium among countries j with factor endowments (K_j, L_j), a country j with a higher capital-labor endowment ratio K_j / L_j produces and exports a larger number of varieties in the capital-intensive, vertically integrated Sector 1, and a smaller number of varieties in the labor-intensive, outsourced Sector 2. Consequently, the share of country i's total imports from country j that consists of intrafirm imports, S_i−f^i,j, is an increasing function of the exporting country's capital-labor ratio K_j / L_j—matching cross-country U.S. customs regressions at the 1% significance level.

Antràs–Helpman (2004) Global Sourcing with Melitz Firm Heterogeneity

In "Global Sourcing" (2004), Pol Antràs and Elhanan Helpman integrated the Antràs (2003) incomplete-contracts model with the Melitz (2003) heterogeneous-firm trade model:

1. Two-Way Sorting by Location (North N vs. South S) and Ownership (V vs. O): Each final-good entrepreneur in the North (N) draws a Melitz productivity parameter θ from a Pareto distribution G(θ) and chooses both where to source components—domestically in the high-wage North (l = N, wage w_N) or offshore in the low-wage South (l = S, wage w_S < w_N)—and whether to integrate (k = V) or outsource (k = O), facing fixed organizational costs f_V^S > f_O^S > f_V^N > f_O^N.
2. Productivity Cutoffs in Headquarter-Intensive vs. Component-Intensive Sectors:
   • In headquarter-intensive sectors (η > η^*), because both offshoring to the low-wage South (w_S < w_N) and vertical integration (β_V > β_O) raise the slope of operating profit with respect to θ^α/(1−α) while requiring higher fixed costs, firms sort into four contiguous productivity tiers: (1) lowest viable productivity (θ ∈ [θ_O^N, θ_V^N)) chooses Domestic Outsourcing (N, O); (2) low-medium productivity chooses Domestic Vertical Integration (N, V); (3) high-medium productivity chooses Foreign Outsourcing (S, O); and (4) highest productivity (θ ≥ θ_V^S) chooses Foreign Vertical Integration / Multinational FDI (S, V).
   • In component-intensive (labor-intensive) sectors (η < η^*), vertical integration lowers variable operating profit while costing higher fixed overhead, so no firm integrates: lower-productivity firms outsource domestically (N, O) and higher-productivity firms outsource offshore (S, O).

Why does vertical integration reduce the supplier's investment incentive in the Antràs (2003) model?

Following the Grossman–Hart–Moore property-rights theory, vertical integration (k = V) gives the parent firm (H) legal ownership of the customized input x, allowing H to fire the supplier manager (S) and retain a fraction δ ∈ (0, 1) of output if bargaining fails. Because H's outside option rises from 0 to δ^αR, H's ex-post revenue share rises from β_O = 1/2 to β_V = (1 + δ^α)/2 > 1/2, which shrinks the supplier's ex-post revenue share from 1/2 down to (1 − δ^α)/2 < 1/2. Anticipating a smaller share of ex-post revenue, an integrated supplier invests less in relationship-specific labor effort L than an independent supplier under outsourcing.

Why are capital-intensive goods traded within multinational firms rather than outsourced at arm's length?

In capital-intensive industries (η > η^*), joint production relies primarily on specialized machinery, technology, and equipment (K) supplied or co-financed by the headquarters firm (H), whereas the local supplier's labor input (L) accounts for a smaller share 1 − η of value added. Because underinvestment by H is much more damaging to total surplus than underinvestment by S when η > η^*, the optimal organizational form assigns residual property rights to H (vertical integration / FDI) so H captures the larger bargaining share β_V > β_O.

How does the Antràs–Helpman (2004) model extend Melitz (2003)?

Whereas the Melitz (2003) model explains which firms export vs. serve only the domestic market while treating all firms as single integrated production units, Antràs and Helpman (2004) combined firm-level productivity heterogeneity (θ) with incomplete contracts over intermediate inputs (η). Their model predicts how firms within the same headquarter-intensive industry sort by productivity across four global sourcing modes (domestic outsourcing, domestic integration, foreign outsourcing, and foreign vertical integration / FDI).

1. Antràs, P. (2003). "Firms, Contracts, and Trade Structure." Quarterly Journal of Economics, 118(4), 1375–1418. https://doi.org/10.1162/003355303322552829
2. Antràs, P., & Helpman, E. (2004). "Global Sourcing." Journal of Political Economy, 112(3), 552–580. https://doi.org/10.1086/383099
3. Grossman, S. J., & Hart, O. D. (1986). "The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration." Journal of Political Economy, 94(4), 691–719. https://doi.org/10.1086/261404
4. Antràs, P., & Chor, D. (2013). "Organizing the Global Value Chain." Econometrica, 81(6), 2127–2204. https://doi.org/10.3982/ECTA10813
5. Nunn, N., & Trefler, D. (2008). "The Boundaries of the Multinational Firm: An Empirical Analysis." In E. Helpman, D. Marin, & T. Verdier (Eds.), The Organization of Firms in a Global Economy (pp. 55–83). Harvard University Press. https://doi.org/10.4159/9780674038547-003