International Trade and Investment.

• Author: Feenstra, Robert C.
• Position: National Bureau of Economic Research's Program on International Trade and Investment

Robert C. Feenstra [*]

NBER's Program on International Trade and Investment (ITI) aims to cover all areas of research dealing with the movement of goods and factor inputs across borders, with an emphasis on empirical research. This report summarizes research conducted in the ITI Program since mid-1997. The starting point for such research is always the explanation of trade patterns: what causes countries to export some goods and import others, and how well can we explain these patterns? In addition, what is the impact of these trade patterns on the incomes earned by particular groups and on the country over all? After reviewing these topics, I turn to a discussion of the policies that countries use to limit international trade flows and the international organizations that govern these policies.

Explaining Trade Patterns

The Hecksher-Ohlin-Vanek (HOV) model, which links a country's endowments of land, labor, and capital to its exports and imports of these factors as embodied in goods, is the leading explanation for trade patterns. Past research has been marked by a series of successes (the HOV model predicts trade well) and failures (it does not predict much better than a coin toss), leading to yet another round of investigation. Three researchers in the ITI program extend the empirical investigation of the HOV model in new directions, resulting in more plausible explanations for trade patterns than have been obtained previously. Donald R. Davis and David E. Weinstein loosen the assumptions of the HOV model to allow for different technologies across countries, different factor prices across countries, and a more general structure of demand. [1] At the same time, they carefully construct datasets for the United States, Japan, and other countries that are consistent internally and with respect to external trade flows. The results of this research substantially close the gap between predicted trade flows (based on the factor endowment of countries) and actual trade flows (as measured by the factors embodied in trade). James Harrigan's work also confirms the importance of factor endowments in explaining trade patterns, while allowing for different technologies. [2]

One feature that is left out of the HOV model, but is surely important in many industries, is increasing returns to scale. The theoretical implications of increasing returns for international trade patterns have been known for some time, because of the work of Paul R. Krugman, Elhanan Helpman, and others. Recent empirical work has caught up to the theory and offers the first tests of the importance of increasing returns for trade patterns. One implication is that industries with particularly strong increasing returns to scale will want to "agglomerate," locating in the same region or country. This means that a country with strong demand for some product will attract industries producing it, and the country's demand-side bias will cause it to be an exporter rather than an importer of that good. Krugman calls this the "home market effect." Davis and Weinstein, [3] as well as James A. Markusen, Andrew K. Rose, and I [4] empirically investigate his theoretical prediction. We show that the home market effect can arise from models even without increasing returns to scale and that it is much more pronounced in industries producing differentiated rather than homogeneous goods. Other work on the geographical location of industries (across or within countries) has been done by Richard E. Baldwin [5] and by Gordon H. Hanson. [6]

The Gravity Equation

Closely related to the home market effect is the use of the "gravity equation" to explain trade patterns. The gravity equation states that the trade flow between two countries will be proportional to the product of their GDPs, so that countries of sizes ten and one (for example, the United States and Canada) should have roughly as much trade between them as do two countries of size 3.3 (such as Germany and England). This equation can be derived from an increasing returns trade model, but current research has shown that it can also be obtained from models with homogeneous goods. [7] One such justification, developed by Jonathan Eaton and Samuel S. Kortum, considers a Ricardian trade model with random technological differences across countries and transportation costs. [8] In this framework, there is a probability that any particular country will have the best technology in each good, and therefore will export it to neigh boring countries. This gives rise to a gravity equation that incorporates country size as well as transportation costs and technological spillovers across countries.

The gravity equation represents a useful benchmark against which other explanations for international trade can be assessed. One alternative explanation explicitly takes account of traders: goods are exchanged only if two agents in different countries want to trade them. Considering that countries have such diverse languages, customs, and institutions, the process of "matching" buyers and sellers is bound to create some frictions. James E. Rauch has investigated this empirically, and finds that ethnic networks work to promote trade. [9] He developed the theoretical basis for these findings with Alessandra Casella. [10] In related work, Robert Z. Lawrence confirms the importance of information and search costs in the trade flows of multinationals and particularly in their response to changes in exchange rates. [11]

A final direction of current research on trade patterns is to use plant-level rather than industry-level data. This is important because of the abundant evidence that plants are not the same within each industry, but rather differ dramatically in their productivity. In particular, Andrew B. Bernard and J. Bradford Jensen have found that the most productive plants become exporters. [12] This finding is significant because it contradicts the idea that promoting exports might lead to more efficient firms; on the contrary, Bernard and Jensen find that productivity causes exports, but not the reverse, at the level of the plant. This empirical finding can be explained in Eaton and Kortum's theoretical model [13] in which firms have random productivities, or alternatively in Marc Melitz's model which allows for heterogeneous firms within each industry. [14] In Melitz's model, opening trade leads to a rationalization of plants (with the less efficient exiting), so that trade implies productivity gains at the level of the industry. Nina Pavcnik and James A. Levinsohn also have done work documenting the importance of firm heterogeneity. [15]

Immigration and Capital Flows

In addition to studying the movement of goods across borders, much research is devoted to explaining the movement of labor and capital. Hanson and Matthew J. Slaughter have investigated in detail the effects of cross-state migration in the United States and of immigration from Mexico. [16] They argue that cross-state flows of workers have relatively little impact on local wages but are absorbed mainly through changes in the mix of goods produced in that state. Whether this is also true for immigration from other countries, including Mexico, is an important policy question. Hanson investigates enforcement measures at the Mexican-U.S. border on wages in adjoining U.S. cities and finds a minimal impact. [17] In contrast, Daniel Trefler forcefully argues that immigration into the United States has had a significant downward effect on the wages of less-skilled workers, a topic to which we return below. [18]

Movements of capital between countries, or foreign direct investment (FDI), are explained by many of the same features that affect trade (factor endowments, transportation costs, and increasing returns). In addition, though, because FDI is defined as the ownership of capital abroad, there must be some reason that firms wish to own the facility used for production rather than simply exporting to the other country. Markusen has developed tractable models of the multinational enter prise [19] and applied data to test their main hypotheses. [20] One particularly interesting hypothesis looks at whether the activities of multinational enterprises mainly fit a "horizontal" model (with firms spreading into the same activities in other countries) or a "vertical" model (in which firms go abroad to invest in other, complementary production activities). At this time the evidence strongly supports the "horizontal" model. [21] Both Baldwin [22] and Joshua Aizenman [23] have developed other theoretical models of FDI.

In addition to determining the reason for FDI, we are also interested in its impact on local wages, employment, and trade. Bruce A. Blonigen uses information on foreign plants in the United States to study their local impact. [24] He finds that foreign plants: typically grow faster than U.S. plants of similar size; lead to a larger positive impact on local wages; and generate some shifts in local government budgets away from public education and towards transportation and public safety. He also investigates the extent to which FDI and imports are substitutes (with inflows of capital reducing imports) or complements (with FDI leading to higher imports). He finds evidence to support both hypotheses, [25] as does Linda S. Goldberg using data from Latin America. [26]

The longest-standing contributions to our understanding of FDI have been made by Robert E. Lipsey, Director of NBER's New York office. His...