Infrastructure, Multinational Affiliate Production, and the Pattern of Trade

Abstract

In a two-country general equilibrium model with endogenously determined domestic and multinational firms, it is shown that public infrastructure development can have diverging implications for horizontal multinational affiliate firm production and trade, depending on the type of infrastructure invested in. Infrastructure investments with strong productive or local transport effects (i.e. schools or local roads) lead to greater domestic firm production and exports, fewer imports, and more foreign multinational affiliate firm production in the country making the investment. On the other hand, infrastructure projects that lower international trade and transaction costs (i.e. shipping ports or airports) lead to more domestic firms in both countries, a greater volume of bilateral exports in both directions, and less multinational affiliate production. Further, the effect of different types of infrastructure investment on income and welfare of the open economies is explored.

Key Words:

• Infrastructure
• multinational corporations
• international trade
• development
• FDI

JEL Classifications:

• F2
• F1
• R1
• R3
• H4

Notes

1 Martin and Rogers (1995) show how capital flows change due to changes in transport infrastructure in a two-country context with agglomeration externalities. The model here differs by incorporating productive effects of infrastructure as well as a multi-plant framework for multinational production. The latter of which is crucial for understanding how infrastructure influences the proximity-concentration tradeoff for multi-plant horizontal firms.

2 This is not shown in Table 1 but comes from the BEA statistics on inward US investment, implying that combined European affiliate sales that were shipped back to their home countries or third markets was only about 10% of total affiliate sales.

3 For example, the percentage change in telephone mainlines per capita was calculated as the change in telephone mainlines per capita for the period 1991–1995 against the telephone mainlines per capita average for the period 1996–2000.

4 In Table 2 and the discussion to follow, the ratio of US affiliate sales/US exports are majority owned US affiliate sales in a country divided by US domestic firm exports to a country.

5 Within group averages are weighted by GDP to control for market size.

6 Following Markusen and Venables (1998), Yeaple (2003), and others the assumption is made that there are zero trade costs in the Y sector. This allows us to focus on how changes in infrastructure affect the imperfectly competitive industrialized X sector.

7 The model will be presented in terms of country i, with country j being defined analogously.

8 By definition, the location of the purchase of the firm-specific fixed-cost identifies the multinational or domestic firms’ headquarters location.

9 For a domestic firm in equation 3(3a) (a) local distribution costs would entail transport costs to deliver the product domestically for units or domestic transport costs to the dock for units to be exported. It is assumed that includes the international cost of transport and delivery in the foreign market.

10 Labor (L) is assumed to be a composite of high-skilled labor and equipment (such as a computer, lab equipment, or machinery requiring specific skills) where each firm uses the composite factor for the development of a blueprint (requiring F units) and for the technical production process (c_w units per unit of output).

11 Capital (K) can be thought of as a composite of low-skilled labor and resources (such as concrete, steel, etc.) for building a production facility.

12 For example, it is straightforward to think of the constant unit input requirement for X production to entail two components; a constant unit production requirement, c_p, and a constant unit domestic distribution requirement, c_d, such that c_w = c_p + c_d.

13 See Markusen (2002, p. 81) for a detailed derivation of the Cournot markups.

14 Note that new infrastructure development in country i implies that , where is the initial stock of infrastructure development in country i before investment.

15 For ease of intuition of the impacts of infrastructure on affiliate sales and exports, the conditions in equation (12)(11) are conditional on a firm's headquarters location. In the full general equilibrium model to follow, this is relaxed to allow for free geographic choice in headquarters location, similar to Markusen and Venables (1998).

16 This implies that factor prices and the price of X are identical in both countries.

17 Country i relative labor abundance implies w_i < w_j and r_i > r_j, while country j relative labor abundance implies w_i > w_jand r_i < r_j.

18 It should be noted that if factor price differences are severe enough, two other possible equilibria exist: domestic firms only in country i (country j specializes in good Y), and domestic firms only in country j (country i specializes in good Y). Since this paper is focused on the effects of infrastructure on horizontal multinational activity (rather than vertical activity), we do not consider these two equilibria.

19 In the partial equilibrium, existing firms choose their firm type and output given fixed initial equilibrium prices and incomes. In the general equilibrium, free entry drives profits to zero and prices and incomes endogenously adjust.

20 See for example, Bougheas, Demetriades, and Morgenroth (1999) who explore the length of motorways and the overall level of public capital on trade volumes; Limão and Venables (2001) who look at a measure of road and telecommunications infrastructures on transport costs and trade flows; Freund and Weinhold (2004) who examine the effect of the internet on bilateral trade flows; Clark et al. (2004) who look at the effects of ports on transport costs and bilateral trade volumes; or Wilson, Mann, and Otsuki (2005) who look at port and services infrastructures on trade flows.

21 Note that in all three cases, country j infrastructure is held constant while country i infrastructure changes.

22 After the regime switch, notice that county j headquartered multinational affiliate output (H_ji) increases (as predicted by the partial equilibrium result of Proposition 2) but then starts to decline. This decline is a second-order effect brought about by the increasing relative profitability of country i domestic firms over country j headquartered multinationals firms.

23 Note that there are no ‘shut-down’ costs associated with a multinational firm closing a plant in a foreign country. Adding it to the model would delay the point at which firms switch to producing solely as a domestic firm but would not qualitatively change the results.

24 Again, notice that once the regime switch occurs, country j multinational affiliates (H_ji) enter country i and capture market share as they take advantage of the improved productivity in country i. This initially imposes a second-order competitive effect on country i domestic output, but as productivity expands, country i's domestic firm's profit advantage increases relative to country j multinational affiliate producers.

25 Note that income, as given in equation (4)(3b) , is measured in units of the numeraire good, Y.

26 Remember σ < 1, so there are decreasing returns to infrastructure investment.

27 I must give credit to an anonymous referee for making this point.

28 Again, welfare for country i will increase up to the point at which the marginal benefit of an additional unit of infrastructure development equals its social marginal cost.