Trade Pattern Persistence

Preliminary: Do Not Quote without Permission Trade Pattern Persistence* James Cassing Department of Economics University of Pittsburgh Steven Husted Department of Economics University of Pittsburgh December 2002 *We would like to thank without implicating David DeJong and Jean-Francois Richard for helpful comments and Glenson France for expert research assistance. An earlier version of this paper was presented at the Fall 2002 European Trade Study Group meetings. 1. Introduction In 1979, Jimmy Carter was President of the United States, the Berlin Wall was intact as was the Soviet Union, the Tokyo Round was not implemented and the Uruguay Round was a decade away. Between 1980 and 2000, about 125 free trade arrangements were negotiated and implemented, the EU expanded twice and adopted a common currency, and there was a capital market disruption in 1997 of historic proportions involving the former Soviet Union and many countries of East Asia.1 Through all of this the world economy grew by more than 70% and world trade grew by a remarkable 175%. Yet, in this paper we provide evidence that the pattern of international trade for many countries appears to have remained relatively stable: principal trading partners do not change, and trade shares exhibit remarkable constancy. This trade pattern stability over a long period of time is not what economists usually predict. For example, Yarbrough and Yarbrough (2000) echo the intuition of many economists when they write in their popular textbook: The evolution of comparative advantage over time implies that production and trade patterns will change over time as well, creating changes in the distribution of income and some difficult dilemmas for policymakers. These changes are evident even over a fairly short period of time. But, examination of the market for aggregate imports for almost 100 countries over the last two decades of the twentieth century calls this fluidity of trade patterns into question. Another way to think about the presumptions of the profession regarding trade pattern stability is the skepticism that has often met evidence of very low price and income elasticities in studies of trade behavior. Clearly, since 1980, relative prices have changed a lot and the terms of trade of most countries have fluctuated considerably (see Table 1). Yet again, we will demonstrate 1 Figures on the number of regional trade agreements created over this period are taken from the WTO web page, Regionalism: Facts and Figures, http://www.wto.org/english/tratop_e/region_e/regfac_e.htm. that trade shares have remained relatively constant. This would seem to support estimates of low trade elasticities found in studies such as Warner and Kreinin (1983). [insert Table 1 about here] Furthermore, growth rates have varied substantially between countries, which should affect trade shares (see Table 2). Similarly, technological progress has made the world a different place and this should also impact trading patterns, especially to the extent that technology changes have not been uniform across countries. Finally, ocean freight and port charges per short ton of cargo have fallen 50% since 1945, air transport costs are down by even more, and communications costs have essentially disappeared. These changes, one would think, might have an affect on trade patterns, but the data are not so conclusive. [insert Table 2 about here] More specifically, in this paper we exploit a newly manageable data set from the IMF in order to quantify the extent to which trade patterns have changed over the last two decades. The trade data focus on bilateral trade patterns for 93 countries and are quite robust in cutting across countries of diverse economic, political, and cultural traits during a fairly turbulent period of economic history. In Section 2, we describe the data and develop some measures of trade pattern change. We report evidence on the characteristics of countries that hold primary market shares in national import markets and demonstrate that the shares of these countries and other principal suppliers remain relatively constant over the sample period. Then, in Section 3, we present some preliminary thoughts on what is going on and what this might portend for various theories of international trade. Finally, Section 4 offers some conclusions. 2 2. Empirics In this section, we present evidence on the stability of trade patterns for a large number of countries over the past two decades. In developing this evidence, we use data from the International Monetary Fund’s Direction of Trade Statistics March 2002 CD-ROM. This database provides figures on the values of merchandise exports and imports by trade partners for 186 countries. In our analysis, we focus on bilateral import totals valued in c.i.f. terms. The database reports these values measured in U.S. dollars for all countries. Our study utilizes annual data from 92 countries and 1 territory (Hong Kong) over the period 1980-2000.2 The countries chosen for this study cover all geographic regions of the world and employ most economic systems. We chose not to include data from countries such as the states of the former Soviet Union that did not exist at the start of the sample period or countries such as Kuwait and Sierra Leone, where there were breaks in economic data because of wars or other disruptions. A complete list of countries included in our analysis can be found in Appendix 1. Our analysis focuses on the behavior of the nominal market shares of the countries that supply goods to the countries in our sample. We measured these as the ratio of nominal imports to total imports (as reported in the database) times 100. Because of the large number of countries in the database, we present details for only those countries that enjoyed at least 2 percent of a given country’s market in 1980. Even with this limitation, we ended up with 926 bilateral trade patterns to analyze, or roughly ten trading partners for each country in the sample. 2.1 Bilateral trade: Primary trade partners Using this market share data, we focus on a variety of questions related to bilateral trade 2 Data prior to 1980 are not available on CD-ROM. 3 patterns. The first issue we consider is related to primary trade partners. To clarify the points we would like to make, consider Figure 1. This provides a graph of the import market shares for Tunisia over the sample period. The behavior in this graph typifies most of the patterns we found in our sample. [Insert Figure 1 about here] In particular, throughout the sample period, one country (here France) held the largest share of the market. Its market share averaged about 25 percent. For more than two thirds (66 of 93) of the 93 countries in our sample, the same country had the greatest market share at both the start and the end of the sample period. In addition, the average market share of the primary trade partner was 27 percent in 1980 and 26 percent in 2000. The fact that France held the largest share of Tunisia’s import market is readily explained. France is a high income, industrialized country located near Tunisia. Perhaps most important, it governed a colonial protectorate in the country for more than 60 years, and, as a result, French is an official language of the country. As it turns out, attributes such as proximity, former colonial relationships, and being the geographically nearest high income industrialized country appear to be extremely important in explaining market share dominance.3 This is confirmed in Table 3 where we show that these characteristics are found in more than two thirds of the primary market share holders in our sample. [Insert Tables 3 & 4 about here] Table 4 summarizes how often various countries held leading market shares at the start and 3 Note, in cases such as Mexico, where the United States is the primary trade partner, we classified it as a neighbor rather than the nearest high income country. 4 end of the sample period.4 As Table 4 demonstrates, a small number of high-income countries, including the United States, France, Germany, Japan, and the United Kingdom were most likely to hold the dominant share of a country’s market at both the start and end of the period. Moreover, as is exemplified in Figure 1, high-income countries tend to maintain market shares in every country in our sample. This, of course, is consistent with the fact that these countries consistently lead world rankings as major exporters in value terms. What may be surprising is the global universality of their market penetration, and the relative stability of this penetration over time.5 We turn now to the question of market share stability. 2.2 Bilateral trade: Market share stability Figures 2-5 provide graphs of the evolution of these market shares for Austria, France, Pakistan, and Uruguay. As the figures indicate there are remarkable differences in these patterns from one country to the next. Nonetheless, it is clear that the shares in Figures 2 and 3 exhibit considerably more stability than those in Figures 4 and 5. [insert Figures 2-5 about here] To our knowledge, no previous work has compared the behavior of market shares over time for a large set of countries.6 We do not have strong priors as to what patterns one should expect to find in these shares. Nonetheless, as we noted in the introduction, given all of the changes experienced in the world economy over the past two decades we would not have found it unusual to observe considerable instability in trade patterns and partners. In the remainder of this section, 4 More complete detail on the characteristics of dominant trade partners for the countries in our sample can be found in Appendix 2. 5 As will be discussed below, an exception is Japan which has lost market share in a number of markets, especially in the 1990s. 6 Wall (2002) examines trade share behavior for Japanese exports to various markets between 1986 and 1997. 5 we will try to make the case that stability, rather than variation, in trade shares is a strong and recurring phenomenon. There is no good metric to quantify the behavior we are trying to study. Consequently, we have chosen a simple procedure to make our first pass at the data. In particular, we regressed each series of trade shares on a constant and a time trend. As the plots in Figures 2-5 indicate, simple linear models should approximate well the behavior of many share patterns, and we found that to be the case throughout. Our purpose in this exercise was to determine how often we would find that the shares were trendless. Our first thought in approaching this question was to see how often we could reject the null hypothesis that the coefficient on the time trend was zero. We also considered testing the joint hypothesis that all of the time trend coefficients were zero. In examining the output from our statistical work, however, it became clear that this would not work well in our context. In particular, if shares were virtually constant, the t statistic on the estimated trend would approach infinity, even as the estimated slope is near zero. Indeed, we found numerous examples of trend coefficient estimates with values almost identically zero, but with t statistics much larger than two. 7 This same problem also confounded our ability to test the joint hypothesis that all of the slopes for a given country were zero.8 Given these problems with standard hypothesis test procedures, the remainder of this section is devoted to distributions of point estimates. [insert Table 5 about here] Table 5 provides detail on the point estimates from the regression exercise. In order to better interpret the results we multiplied each of the estimates by 21 in order to determine the estimated change over the sample period in each of the trade shares. As the table shows, the predicted change 7 Only 272 of 926 estimated slope coefficients had t values small enough to not reject the null that the slope equaled zero. 8 We rejected the null that all slopes equaled zero in every case. 6 did not exceed 2 percent in absolute value for almost half (449 of 927) of the possible trade patterns. As the table also shows, falls in shares were estimated for well over half the cases. The median projected decline in market share over the sample period was 1.69 percent. In more than 15 percent of the cases, estimates of trade share falls exceeded 5 percent over the sample period. Many, if not all, of the largest declines were associated with trade shares for petroleum exporting countries. This pattern is easy to explain and points to a problem with the fact that data availability required us to begin our analysis in 1980. That year coincided with the second world oil shock. In subsequent years, oil prices (and probably oil consumption) fell dramatically leading to significant falls in market shares. These declines are apparent in Figures 2-5. As was the case with primary trade partners, a small number of high income countries held market shares of at least two percent for virtually all countries in the sample over the sample period. And, as with the case of all shares, the shares of these countries were relatively stable. These points are illustrated by the information in Table 6. [insert Table 6 about here] As the table shows, the United States held at least 2 percent of the market in 88 of 92 possible cases. Both the median and the mean of the projected market share change over the sample period were between zero and a fall of one percent. Germany, the United Kingdom, and Japan also held at least two percent of the markets in 80 countries or more. Only the mean and median projected change in market shares for Japan exceeded 2 percent in absolute value. France, Italy, and the Netherlands all held significant market shares in more than half of the sample countries, and the average expected market share changes for all were less than one percent. Only Saudi Arabia, which held at least two percent of the market in 30 countries in 1980 had large (greater than five percent) 7 average projected changes in market share. As noted already, this pattern held true for virtually all oil exporters in virtually all countries. The major point of this section is that trade shares have remained quite stable for most countries over the past two decades, despite significant changes in the economic and political structure of the world trading relationships. There is clearly more work to be done in this analysis and we are beginning these efforts. Our first goal is to derive a metric that allows us to compute and then try to explain how much and why trade patterns have changed on a country-by-country basis. Our plan is to try to develop a test to see if the annual trade shares we observe for a country look to be draws from a common distribution. It is possible to develop a statistic that represents the average distance between each of the draws and to use this statistic in subsequent tests of possible models that predict relative stability in trade patterns. Since we are not ready at this time to present these tests, in what follows we discuss several conjectures from trade theory that might explain the stability that we have found in the data. 3. Theory and Speculation Three themes seem to recur in our data. First, country size seems to matter and, in particular, trade shares of most nations with the United States and most other high income countries are especially stable. Second, distance matters. Countries have very stable trade shares with their immediate neighbors. Finally, historical ties seem to matter in the sense that trade shares between former colonies and the mother-country are quite stable. A curiosity of the data is that while trade shares remain fairly stable, the composition of trade changes quite a bit. For example, in 1979, office machines, computers, parts (SITC 751, 753, 759), constituted only 1.4% of the share of trade and was ranked 16th by category. In 1999, this group was 8 6% of trade and ranked number one as the most traded commodity group.9 Yet, nations do not appear to be acquiring market share at the expense of other nations. This change in the pattern of trade seems to generalize. That is, the commodity composition of trade changes, but the trade shares between countries do not. This raises some questions that we address here, but offer only speculation as answers. Specifically, there seems to be a sort of “hysteresis” in the pattern of trade. The world changes a lot in terms of prices, technology, and so on, but trading partners and trade shares remain constant. This “evidence” speaks in favor of some trade theories as being more or less important. 3.1 Standard static trade theories The constancy of bilateral trade shares over time can inform us about the robustness of various trade theories. For example, the standard, variable-proportions – Heckscher-Ohlin – model predicts that, if tastes are identical and homothetic, then differential (balanced) growth rates among trading partners will lead to changing trade shares (see Appendix 3). Yet, while growth rates do vary quite a lot between countries (recall Table 2.), trade shares are relatively constant. While this hardly constitutes a test of the H-O model, it is suggestive that one of the model’s predictions is questionable. Since technological progress is analytically similar to growth, differential rates of technical progress across countries should similarly lead to changing bilateral trade shares. Such results, we think, would apply equally to a specific factors model and, in light of the Jones-McKenzie Theorem, to a Ricardian model with many goods and countries. (Specifically, in the multi-good Ricardian model, the conditions for country j to specialize in good j – “j-j specialization” – depend upon a comparison of a multiplicative string of one set of labor-output 9 See Table 1.3 in Steven Husted and Michael Melvin, International Economics, 6th edition, (Addison Wesley Longman, Boston) 2003. 9 coefficients with another such string. But, differential technical progress – some lower labor-output coefficients for some countries -- will change the pattern of specialization and so necessarily the share patterns in trade.) Yet, economic change in the world does not appear to lead to changes in bilateral trade shares. Finally, the wave of new preferential trade agreements (PTAs) and new accessions to old PTAs since 1979 should entail trade creation and trade diversion. Thus, bilateral trade shares should be affected. Here we find mixed experience regarding trade shares in our data. For instance, Mexico’s share of North American trade has grown rapidly since the formation of NAFTA, but trade shares of EU countries in Austria have remained virtually constant since Austria’s entrance into the EU in 1995. 3.2 Geography and trade One recently advanced view of the world that is consistent with our findings is the “economic geography” paradigm, long around, of course, but most familiar to trade economists through Paul Krugman’s Gaston Eyskens lectures at the Catholic University of Leuven in Belgium. That paradigm renders country size, proximity, and cultural ties -- which do not change so rapidly -crucial to the bilateral shares of trade, but not to the commodity composition of bilateral trade. This seems consistent with our findings. First, the economic geography model focuses on transport networks and scale economies. Firms and industries cluster geographically, and this creates persistent production nodes even as the product lines shift. Transport networks get put into place and, once the costs are sunk, can determine which countries trade with which other countries. For example, if there are three 10 countries in the world – A, B, and C – and there are railroads between A-B and B-C, but not A-C, then trade shares might remain fairly constant to the extent that transport costs are important. This implication of economic geography models is consistent with our finding that distance is an important determinant of why trade shares remain constant between trading partners of the world. Also, if “distance” is taken to include cultural distance, our findings corroborate a stability in the trade between former colonies and their colonial mother-countries. Finally, the trade shares of large countries, including France, Germany, Japan, the United Kingdom, and the United States, seem to remain particularly constant. This is consistent with Krugman (1991). Large countries (regions) exploit scale economies and transport systems in order to maintain an industrial base even as the products themselves change. Thus, for example, the industrial belt of the Midwest in the United States continued to prosper even as the product line changed over time. Consequently, the region was the stable exporter of industrial products for a very long time, even though the products themselves were changing. This is roughly consistent with our findings of constant bilateral trade flows. 4. Conclusions The world would appear to have changed in most dimensions quite a lot since 1979. Certainly the economic and political landscapes are wildly different. Yet, when we measure which countries trade with which other countries the world looks remarkably constant. We cannot find much change in bilateral trade shares among countries. This finding is all the more striking because the commodity composition of trade does appear to change. While our approach has been simply to report the stability of bilateral trade, we also conjecture that the data support the “economic geography” paradigm wherein distance, history, and 11 country size are important determinants of international trade patterns 12 5. References Krugman, Paul, Geography and Trade, 1991, (MIT Press, Cambridge). Wall, Howard J., “Has Japan Been Left Out in the Cold by Regional Integration?,” Review, Federal Reserve Bank of St. Louis, September/October 2002, pp. 25-36. Warner, Dennis and Mordechai E. Kreinin, “Determinants of International Trade Flows,” The Review of Economics and Statistics, Vol. 65, No. 1. (Feb., 1983), pp. 96-104 Yarbrough & Yarbrough, The World Economy: Trade and Finance, fifth ed., 2000, (Harcourt College Publishers, Chicago). 13 Figure 1: Tunisia: Major Import Suppliers 30.00 25.00 20.00 15.00 10.00 5.00 0.00 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 IMPORTS FROM FRANCE IMPORTS FROM ITALY IMPORTS FROM GERMANY IMPORTS FROM SAUDI ARABIA IMPORTS FROM UNITED STATES IMPORTS FROM GREECE IMPORTS FROM SPAIN IMPORTS FROM BELGIUM-LUXEMBOURG IMPORTS FROM NETHERLANDS IMPORTS FROM UNITED KINGDOM 14 2000 15 16 17 18 Table 1: Characteristics of Annual Terms of Trade Changes: Selected Countries, 1980-2000* United States Canada Australia Japan New Zealand Finland Germany Greece Ireland Italy Netherlands Spain Denmark Norway United Kingdom Kenya China,P.R.:Hong Kong India Korea Pakistan Singapore Thailand Israel Jordan Brazil Colombia avg. % chg. 0.53 -0.33 -0.79 3.50 0.51 0.57 0.31 -1.35 -0.09 1.11 0.24 1.10 0.57 1.37 0.01 -1.25 0.00 2.21 -1.16 5.34 -1.19 -1.91 1.13 0.90 -1.08 -0.15 s.d. % change 2.72 2.90 5.49 8.68 3.61 4.45 4.51 3.86 2.59 4.57 1.85 7.00 2.04 11.60 1.77 11.31 1.45 10.96 5.58 29.43 1.85 5.23 2.96 7.10 13.17 11.50 max % change 5.39 5.61 14.46 33.71 10.98 12.45 15.07 4.60 4.39 15.85 3.09 20.85 6.59 35.28 2.67 17.25 3.20 23.47 8.85 122.64 2.09 10.83 5.54 14.34 24.67 21.83 min % change -5.14 -6.34 -9.60 -8.10 -5.23 -4.98 -6.59 -9.55 -4.64 -7.59 -3.54 -17.74 -2.56 -24.77 -5.47 -18.22 -2.77 -12.95 -13.63 -15.85 -4.50 -12.88 -3.52 -11.53 -21.71 -22.87 *Source: International Monetary Fund, International Financial Statistics CD-ROM, December 2002. Terms of trade constructed as the ratio of export unit values to import unit values. 19 Table 2 Average Annual Growth Rates: Selected Countries, 1990-2001* Algeria Argentina Australia Austria Bangladesh Belgium Benin Bolivia Brazil Burkina Faso Burundi Cameroon Canada Central African Republic Chad Chile China Colombia Congo, Rep. Costa Rica Cote d'Ivoire Denmark Dominican Republic Ecuador El Salvador Finland France Germany Ghana Greece Guatemala Guinea Haiti Honduras Hong Kong India Indonesia Ireland Israel Italy Jamaica Japan 2 3.7 4 2.1 4.9 2.1 4.8 3.8 2.8 4.9 -2.2 2.1 3 2.1 4.6 6.4 10 2.7 -0.1 5.1 3.1 2.5 6 1.7 4.5 3 1.8 1.5 4.2 2.3 4.1 4.1 4.1 3.1 3.9 5.9 3.8 7.6 5.1 1.6 0.6 1.3 Jordan Kenya Korea Madagascar Malawi Malaysia Mali Mauritania Mexico Morocco Mozambique Nepal Netherlands New Zealand Niger Nigeria Norway Pakistan Panama Papua New Guinea Paraguay Peru Philippines Portugal Rwanda Senegal Singapore South Africa Spain Sri Lanka Sweden Switzerland Tanzania Thailand Togo Tunisia Turkey Uganda United Kingdom United States Uruguay Venezuela *Source: World Bank: World Development Report 2003 20 4.8 2 5.7 2.4 3.7 6.5 4.1 4.2 3.1 2.5 7.5 4.9 2.8 2.9 2.6 2.5 3.5 3.7 3.8 3.6 2 4.3 3.3 2.7 0.8 3.9 7.8 2.1 2.6 5.1 2 0.9 3.1 3.8 2.2 4.7 3.3 6.8 2.6 3.5 2.9 1.5 Table 3 Characteristics of Leading Import Suppliers* neighbor colony nearest HIC oil rest 1980 24 21 16 7 25 2000 31 16 15 4 27 * Notes: nearest HIC = geographically nearest high income country; oil = oil exporter; colony = importing country is a former colony of exporter 21 Table 4 Leading Import Supplier Count 1980 Australia Brazil Cameroon Canada Cote d’lvoire France Germany India Iran Iraq Japan Kenya Netherlands Portugal South Africa Saudi Arabia Sweden Trinidad & Tobago United Arab Emirates United Kingdom United States USSR 2000 4 2 1 1 1 14 10 1 1 1 12 1 1 1 2 4 1 1 1 6 26 1 Argentina Australia Brazil Canada Cote d’lvoire France Germany India Italy Japan Kenya Kuwait Nigeria Portugal Peoples Republic of China South Africa Singapore Spain Sweden United Arab Emirates United Kingdom United States 22 1 4 3 1 1 17 11 2 1 11 1 1 1 2 3 4 1 1 1 2 3 21 Table 5 Empirical Results Distribution of estimated share growth (coefficients on time trend × 21) 5% or more between 2% & 5% between 0 and 2% between -2% and 0% between -2% and -5% -5% or less No. 60 82 162 288 191 143 23 Percent of total 6.5% 8.9% 17.5% 31.1% 20.6% 15.4% Table 6 Market Share Trend Patterns for Selected Countries United States Germany United Kingdom Japan France Italy Netherlands Belgium-Luxembourg Saudi Arabia 24 # 88 84 83 80 64 56 54 34 31 median -0.81381 -1.45281 -1.59655 -2.53008 -0.64854 -0.08782 -0.3731 0.318948 -5.67525 mean -0.26414 -1.48571 -2.74939 -3.32239 -2.3053 0.71017 -0.30761 0.241521 -6.95684 Appendix 1 Sample Countries Algeria Argentina Australia Austria Bangladesh Belize Benin Bolivia Brazil Brunei Burkina Faso Burundi Cameroon Canada Central African Rep. Chad Chile Hong Kong Peoples Rep. of China Colombia Costa Rica Cote d'Ivoire Cyprus Denmark Djibouti Fiji Finland France Gabon Gambia Germany Ghana Greece Guatemala Guinea Guinea-Bissau Guyana Haiti Honduras Iceland India Indonesia Ireland Israel Italy Jamaica Japan Jordan Kenya Korea Malawi Malaysia Mali Mauritania Mauritius Mexico Morocco Mozambique Nepal The Netherlands New Zealand Niger Nigeria Norway Oman Pakistan Papua New Guinea Paraguay Peru The Philippines Portugal Qatar Saudi Arabia Senegal Singapore Spain Sri Lanka Suriname Sweden Switzerland Tanzania Thailand 25 Tunisia Turkey Uganda United Arab Emirates United Kingdom United States Uruguay Vanuatu Venezuela Zambia Zimbabwe Country Appendix 2 Characteristics of Leading Market Share Holders 1980 2000 Leader Relation % Leader Relation Algeria Argentina Australia Austria Bangladesh Belize Benin Berkina Faso Bolivia Brazil Brunei Burundi Cameroon Canada Cent. Afr. Rep Chad Chile China-HK China-PRC Colombia Costa Rica Cote d'Ivoire Cyprus Denmark Djibuti Fiji Finland France Gabon Gambia Germany Ghana Greece Guatemala Guinea Guinea-Bissau Guyana Haiti Honduras Iceland India Indonesia Ireland Israel Italy France US US Germany US US France France US US Japan Iran France US France Cameroon US Japan Japan US US France UK Germany France Australia Germany Germany France UK Netherlands UK Germany US France Portugal Trinidad US US USSR US Japan UK US Germany colony nearest HIC neighbor nearest HIC colony colony nearest HIC nearest HIC oil colony neighbor colony neighbor nearest HIC nearest HIC nearest HIC nearest HIC nearest HIC colony colony neighbor colony neighbor colony colony neighbor colony nearest HIC colony colony neighbor nearest HIC nearest HIC neighbor neighbor 23 23 22 41 14 36 23 39 28 19 24 20 38 68 62 40 29 23 27 39 34 40 15 18 25 31 13 16 58 27 11 22 14 35 29 31 28 53 42 10 13 31 51 16 17 26 France Brazil US Germany India US PRC France Brazil US Singapore France France US France France US PRC Japan US US France UK Germany Saudi Arabia Australia Germany Germany France PRC France Nigeria Italy US France Portugal US US US Germany US Japan UK US Germany Colony Neighbor Neighbor Neighbor nearest HIC Colony Neighbor nearest HIC Neighbor Colony Colony Neighbor Colony Colony nearest HIC Neighbor nearest HIC nearest HIC nearest HIC Colony Colony Neighbor Oil Neighbor Colony Neighbor Oil nearest HIC Colony Colony nearest HIC nearest HIC nearest HIC Neighbor Neighbor % 31 20 20 44 11 50 28 27 24 23 33 13 33 64 31 33 20 43 18 34 41 19 11 21 18 49 15 17 65 21 10 19 13 34 19 31 33 54 57 12 8 16 33 18 18 Jamaica Japan Jordan Kenya Korea Malawi Malaysia Mali Mauritania Mauritius Mexico Morocco Mozambique Nepal Netherlands New Zealand Niger Nigeria Norway Oman Pakistan Papua-NG Paraguay Peru Philippines Portugal Qatar Saudi Arabia Senegal Singapore Spain Sri Lanka Suriname Sweden Switzerland Tanzania Thailand Tunisia Turkey UAE Uganda UK Uruguay US Vanuatu Venezuela Zambia Zimbabwe Average Share US US Saudi Arabia Saudi Arabia Japan S. Africa Japan Cote d'Ivore France S. Africa US France UAE India Germany Australia France UK Sweden Japan US Australia Brazil US US Germany Japan US France Japan US Japan US Germany Germany UK Japan France Iraq Japan Kenya US Brazil Canada Australia US Saudi Arabia S. Africa nearest HIC oil oil neighbor neighbor colony oil neighbor colony oil neighbor neighbor neighbor colony colony neighbor neighbor neighbor nearest HIC colony colony nearest HIC neighbor colony colony oil neighbor neighbor neighbor neighbor nearest HIC oil neighbor 32 17 17 18 27 37 23 29 34 13 62 25 13 41 22 18 38 20 17 20 14 41 27 30 24 12 18 20 34 18 13 13 32 18 28 16 21 25 15 17 29 12 17 16 25 48 19 25 27 US US Germany UAE Japan S. Africa Japan Cote d'Ivore France S. Africa US France S. Africa India Germany Australia France UK Sweden UAE Kuwait Australia Brazil US Japan Spain Japan US France Japan France Japan US Germany Germany Japan Japan France Germany Japan Kenya US Argentina Canada Australia US S. Africa S. Africa nearest HIC 45 19 12 Oil 11 Neighbor 20 43 21 Neighbor 17 Colony 26 15 neighbor 73 colony 25 neighbor 37 neighbor 33 neighbor 16 neighbor 22 colony 20 colony 10 neighbor 16 neighbor 32 oil 12 neighbor 50 neighbor 30 nearest HIC 30 nearest HIC 19 neighbor 25 11 21 colony 27 18 neighbor 18 10 nearest HIC 31 17 neighbor 29 9 24 colony 27 14 7 neighbor 43 13 neighbor 24 neighbor 19 neighbor 26 nearest HIC 36 57 neighbor 42 26 * Notes: nearest HIC = geographically nearest high income country; oil = oil exporter; colony = importing country is a former colony of exporter 27 Appendix 3 to be written 28