Does Ease of Communication Increase Trade? Commonality

Gravity model explanations of trade volumes frequently include dummy variables to account for the commonality of language among trading partners. In this paper we use a data set for the number of people in a country who speak English as a first language or English as a second language (Crystal, 1997) as an indicator of the ease with which trade with the United States occurs. Controlling for commodity fixed effects we use SITC three digit industry data centered on 1995 United States bilateral trade with 33 countries to determine the effect of the degree of language commonality on bilateral trade. Both English as a first language and English as a second language are found to be more important for exports than for imports. This is true for all three digit industries as well as when the specific industry groups identified in Rauch (1999) are considered Which products are traded among countries and the volume of trade in these products are two questions that have captured the attention of international trade economists for more than two centuries. Ricardo, Mill, Heckscher, Ohlin, and others have focused on determining which products would be exported and which would be imported. In the 1960s interest shifted to determining the volume of trade and the work of Tinbergen (1962) and Linnemann (1966) are early examples of such studies. Models of these two aspects of trade, however, are quite different. Models that determine which products are traded focus on the relative factor endowments of countries, whereas models of the volume of trade focus on the relative economic size and well being of countries. We focus on the determinants of the volume of trade, which are usually captured in the framework of a gravity model. If one assumes identical and homothetic preferences, then in a gravity model the relative economic size of trading partners determines the potential volume of trade. Economic size is generally measured by some combination of country GDP, population, or per capita GDP. In a gravity model, the difficulty associated with the exchange of goods between trading partners diminishes the potential volume of trade relative to that implied by their economic size in a frictionless world. Theory indicates that a country demands goods in the world market in direct proportion to its share of world income, controlling for the difficulty of obtaining those goods from another country. The stage of a country's development may as well impact the volume of trade independently of the absolute size of its GDP. Degree of development is usually measured by per capita income.1 1 Per capita income also reflects the capital labor ratio in a country, assuming identical constant returns to scale production technology in all countries. If one assumes increasing returns to scale technology, then a higher per capita income may result from larger size as well as a higher capital labor ratio, depending on the nature of the increasing returns to scale. Impediments to trade such as transport costs, tariffs, legal differences and a group of factors that generate positive transactions costs diminish the volume of trade that takes place between any two countries. A major barrier to trade is difficulty in communication. Because people in different countries as a rule speak different languages, exchange of goods may be impeded by costs associated with surmounting language barriers. We are concerned with the effect of language on the volume of trade, i.e., the commonality or lack of commonality of language for residents of trading partners. In particular, we shall examine the importance of English speakers in a trading partner country in determining that country's volume of trade with the United Sates. We shall proceed by discussing in section 2 how the existing literature provides the theoretical underpinnings for our analysis, by presenting the data in section 3, and by discussing the empirical findings in section 4. We shall conclude with a brief summary of our findings and a consideration of avenues for future research. 2 The Basic Approach Theoretical foundations for the gravity model are surveyed in Frankel (1997), Baier and Bergstrand (2001), and elsewhere in this issue by Feenstra (2002). Therefore, we shall provide only the necessary background to motivate our research. Early applications of the gravity model were viewed with skepticism, to say the least, by those who used more structurally based models to analyze international trade. Explaining the volume of trade is sometimes more useful than explaining the factor content of trade or testing the Heckscher-Ohlin model. The work of Anderson (1979) and Bergstrand (1985, 1989) provided a sound microeconomic foundation for gravity model analysis of trade flows.2 The result has been wider acceptance and more frequent 2 2 Helpman and Krugman (1985) derive the gravity model within an imperfect competition model. application of the gravity model to questions regarding the volume of international trade among countries. Initially, the gravity model was used to explain aggregate bilateral trade volumes, but more recently this model has been applied to disaggregated data to determine if particular product groups are impacted differentially by the factors that determine the volume of trade.3 The general equilibrium theoretical framework used by Anderson (1979), Bergstrand (1985 and 1989), and Gould (1994) derives export and import flows as a function of the standard trading partner variables: GDP and either population or per capita income. Additionally, the general equilibrium model introduces relative prices, including tariffs, and a vector of variables that either increase or decrease the cost of conducting international trade. For example, distance between trading partners, contiguity of trading partners, the commonality of language, culture and laws, whether the trading partners belong to a preferential trading area (PTA) or to a common PTA. Rauch (1999) develops a theory of networks that explains how many of these factors impact the volume of bilateral trade.4 This article focuses on the importance of the commonality of language in facilitating the flow of international trade. Many gravity model applications have used dummy variables to capture the benefits of trading partners speaking a common language. The usual result is that a common language results in greater trade volume between two countries.5 3 5 Frankel (1997, pp. 74-76) reviews the literature on this topic. Chiswick and Miller (1998) use a language index to explain the relative difficulty for immigrants from particular countries become integrated into the United States economy. 4 Head and Ries (1998), Gould (1994), Dunlevy and Hutchinson (1999) and Hutchinson and Dunlevy (2001), Rauch and Trindade (2002), and Wagner, Head, and Ries (2002) address the effects of immigrants and the formation of networks that enhance the volume of trade between the host country and the country of origin. Frankel, Stein and Wei (1998) provide an excellent review and analysis of the other major factors within the context of a gravity model. 3 See for example Bergstrand (1989, 1990), Gould (1994), Feenstra, Markusen and Rose (1998) Dunlevy and Hutchinson (1999), Hutchinson and Dunlevy (2001), Baier and Bergstrand (2001). Boisso and Ferantino (1997) construct a measure of linguistic similarity based on the proportion of a population that speaks a particular language as a first language. The index approximates the probability that two randomly chosen people in an importing and exporting country speak the same language. Their measure of linguistic distance did not significantly affect the level of trade between two countries nor did it affect the over all level of trade. Hutchinson (2001) uses an index, developed by Chiswick and Miller (1998), to extend Gould (1994) to analyze the importance for trade flows that results from the difficulty for a native English speaker in learning a foreign language. Chiswick and Miller (1998) develop a "linguistic distance" index that indicates the relative difficulty of learning various foreign languages if one is a native English speaker. Controlling for the standard gravity model variables, including the stock of migrants from a particular country, Hutchinson finds that United States bilateral trade between 1970 and 1986 was lower with a country that had a dominant language which was more difficult to learn. This effect was especially important for U. S. imports of consumer manufactures. Here we exploit a new source of information regarding the commonality of language. David Crystal, a linguist, focuses on the growth and spread of the English language to various other countries in the world. For a set of 74 countries Crystal (1997) estimates the number of people in each country who speak English as their first language and the number who speak English as their second language. We use these data in a gravity model to estimate the effect on bilateral trade of the proportion of the population in a country who speak English as a first or as a second language. We hypothesize that the larger the proportion of the population who speak English as a first language the greater the volume of bilateral trade with the United States. The same should be true for the proportion who speaks English as a second language. 4