Preliminary – Do not quote Comments welcome Shuo.jan.dave The Link between Trade and Income: Export Effect, Import Effect, or Both? Shuo Zhang Jan Ondrich J. David Richardson1 Department of Economics Syracuse University March 2003 Abstract A new framework is developed to evaluate how cross-country differences in export openness and import openness in 1990 affected the level of real per capita income. Familiar and novel instruments are used to extract the exogenous components of total trade (exports plus imports) and of net exports (exports minus imports), which in turn imply distinct export and import effects. We build on an existing literature (FrankelRomer and others) that uses aspects of a country’s geography as instrumental variables for total trade openness. We build on a country’s demography and net wealth abroad to develop a novel instrument for net export openness. Our new estimates reveal that export openness alone correlates with income cross-sectionally, not import openness. 1 Department of Economics, 110 Eggers Hall, Syracuse, New York 13244. All responsibility for errors remains with the authors. I. Introduction The doctrine of mercantilism views trade, at least in part, as a zero-sum game. Trade is considered favorable if exports bring in more money than what is paid out for imports. Adam Smith and David Ricardo challenged the mercantilism theory by arguing that overall openness is what matters – nations grow more prosperous through the process of specialization and trade through imports as well as exports. Many developing countries have replaced their strategy of government-protected import substitution industrialization with a market-focused export oriented strategy. They seek to promote economic growth by exporting more and more of their products; this has the feel of neomercantilism. This paper re-poses the question whether trade “raises” a country’s income per person (or per worker); and if it does, what are the channels through which trade affects income, exports or imports? Numerous theoretical and empirical studies have attempted to answer these questions. Theoretical support for the positive link between different sorts of openness and economic growth were provided by Romer (1986) and Lucas (1988). Barro and Sala-i-Martin (1995) and Romer (1992) emphasized how countries that are more open have a greater ability to absorb (import) technological advances, which ultimately leads to higher real per capita income. Krugman (1974), Rodrik (1995), and Rodriguez and Rodrik (2000) on the other hand, cast doubt on the impact of openness on growth. Warner (2003) casts doubt on their doubts and views our central question as a frontier issue.2 “Proposition [to be assessed] 10. Passive liberalization [balanced export and import openness] works but only assures moderate growth; fast growth requires more active liberalization-promoting exports rather than just creating an enabling environment for open trade.” Warner (2003), pp. 20. 2 2 Empirical studies construct measures of openness variable based on exports 3 , imports4, or the sum of the exports and imports5. Levine and Renelt (1992) argue that “all findings using the share of exports in GDP could be obtained almost identically using the total trade or import share. Thus, studies that use export indicators should not be interpreted as studying the relationship between growth and exports per se but rather as studying the relationship between growth and trade defined more broadly.” (p.959) Grossman and Helpman (1991) stated that technological spillovers could come via imports as easily as exports. Lawrence and Weinstein (2001) argue that imports, not exports, contribute importantly to the productivity growth of Japan and Korea. Frankel and Romer (1999), on which chapter one of this dissertation is based (Zhang 2002a), alleviates many of the conceptual and econometric barriers to these issues by showing how geographical characteristics provide an arguably good instrument for a country’s intrinsic openness. Yet they remark that their trade and income investigation cannot separate the import effect and export effect. Wei (with Wu 2001, 2002a, 2002b) consciously follows Frankel’s and Romer’s lead in a study of how globalization affects Chinese city-level growth and inequality. Wei conflates export and import influences by selecting Chinese cities’ distanced to two major Chinese ports as his instrument for a city’s natural openness. 3 See Krueger (1978), Balassa (1978), Feder (1982) for some of the earliest studies, and the collection of Asian-Miracle (re)considerations in Stiglitz and Yusuf (2000) for more recent references. Micro-data studies that correlate firm performance with export openness are summarized in Lewis and Richardson (2001). 4 See Ram (1990), Masumdar (2000), Mody and Yilmax (2002). Micro-data studies that correlate firm performance with import openness include Ozler and Yilmax (2000), Pavcnik (2000), and Levinsohn (2003). 5 For example, Franker and Romer (1999) and Harrison (1996) are based on the assumption that the coefficients on exports and imports are the same. 3 No one to our knowledge has yet figured a way to do what seems initially the most natural thing. That is to construct a measure of export openness, then an arguably independent measure of import openness, and investigate whether one has a different effect on income than the other, ceteris paribus. That is the principal objective of this paper -- to identify the separate influences of export openness and import openness on income levels after controlling for endogeneity. In particular, we develop a framework that is slightly different from the Frankel and Romer (1999) income determination model by considering an additional “net trade effect” on income levels. When combined with Frankel and Romer’s “total-trade” effect, the two effects together imply separable export and import effects. The extended income model is intended to contribute to three empirical challenges related to trade and income. The first is the Frankel-Romer determination of instruments for total trade. The second is the determination of instruments for net trade. Thirdly, the model and data employed permit the identification of the impact of exports and imports on income separately. Our concern about endogeneity in this chapter differs slightly from our concern in the previous chapter. When considering exports and imports separately, the expected bias from ignoring endogeneity is, respectively, caused by the unobserved income determinants that are correlated with exports and imports. This paper (Zhang 2002b) is one part of my three dissertation papers (Zhang 2002a, Zhang 2002b, Zhang 2003). Zhang (2002a) explores the sensitivity of the Frankel and Romer (1999) empirical relationship between country’s total-trade openness and income level to heteroscedasticity and sample selection in their first-stage bilateral instrumenting regressions. The results support their hypothesis that trade has a 4 significant and positive, yet relatively small impact on income. The present chapter tries to differentiate Frankel’s and Romer’s export openness from import openness. Export openness plays the dominant role. Unfortunately, the unique instrumenting techniques for “net trade” in this chapter cannot be implemented bilaterally, so there remains some doubt about the exact onformity of our conclusions to Frankel and Romer’s. Zhang (2003) employs established and new panel econometrics techniques to further examine the robustness of the findings presented in Frankel and Romer (1999), Zhang (2002a), Zhang (2002b) and other literature. The rest of the paper is as follows. Section II describes the models. Section III provides data definitions and sources. Section IV reports the empirical results and Section VI contains the conclusions of the paper. II. Empirical Models Extensive research has been devoted to investigating factors that influence the per-capita income levels of countries. Recently, an enormous literature has developed on the influence of trade openness. 6 The major shortcoming of many of these empirical studies is their inability to separate the impact of exports and imports. Some focus on one to the neglect of the other; others focus on openness measures that force them to have equal weight. In particular, by regressing income on total trade, many studies embody an underlying assumption that exports and imports contribute equally to income growth. In this section, we develop a simple model based on one of them, the Frankel-Romer study, 6 See notes 3 through 5 above. Plus, Dollar (1992), Sachs and Warner (1995), and Edwards (1998) find significant relationship between the lower barriers to trade and higher growth. Both on theoretical and empirical grounds, Rodriguez and Rodrik (2000) cast doubt on the robustness of this linkeage. 5 by separating the total trade share into export share and import share in order to distinguish between export effect and import effect. (1) ln( yi ) 0 1 ( X i ) 2 (M i ) 3 (Z i ) i where y i represents per capita income. X i and M i are exports and imports scaled by real GDP.7 Z i stands for other control variables. As written, the “true” effects of export openness and import openness, respectively, are 1 and 2 . Yet in practice, researchers usually omit one of the openness terms or force the coefficients to be identical on the two measures of openness (calling the sum of the two “overall” or total “openness”). In principle, this could cause omittedvariable bias in the remaining coefficients of interest or specification bias, due to constraining two coefficients to have the same value. Yet the natural temptation to “just run equation (1)” usually creates misleading inferences because export openness and import openness are indeed highly correlated, as any general-equilibrium thinker knows and as Levine and Renelt observe (above), and because both are acknowledged to be endogenously related income per capita, the focus variable on which they are thought to “operate.”8 Confronting these challenges, what’s a researcher to do? Both exports and imports contribute to income growth. Exports promote specialization and exploit economies of scale. An increase in demand for country’s output raises real income. Imports allow countries to take advantage of other countries’ technology embodied in imported inputs. Through intensive involvement in international competition, countries become more productive. But income also has a causal effect on 7 The scaling introduces special estimation challenges, discussed below. Focusing on the cases of Japan and Korea, Lawrence and Weinstein (2001) found that imports rather than exports are the beneficial conduit of faster productivity growth. However, directly using the import and 8 6 exports and imports. More supply of output leads potentially to more exports. Higher income facilitates consumption of both domestic and foreign products which ultimately raises imports. Therefore, we cannot directly regress income on exports and imports as shown in equation (1). Standard regression is inconsistent in the presence of mutual endogeneity. Instrumental variable estimation provides a theoretically appealing way to handle the endogeneity problem. The important practical question posed in this paper is how to find two sets of instrumenting variables that can not only capture the exogenous components of exports and imports but also distinguish the export effect from the import effect. Country A’s exports to country B is country B’s imports from country A. There is a tendency both within countries and across them for exports and imports to co-vary for general-equilibrium reasons. Thus it is difficult in practice for instrumental-variable groups to ideally capture the distinction. This paper proposes an alternative method to distinguishing the effects of imports and exports. The challenge of endogeneity in standard models is maintained by algebraically re-arranging equation (1). Letting T be the total trade (exports plus imports) divided by real GDP and E be the net trade (exports less imports) divided by real GDP, to control for scale effects, equation (1) transforms to (2) ln( y i ) 0 1 (Ti ) 2 ( Ei ) 3 (Z i ) i We argue that finding good instruments for T and E in order to estimate equation (2) is much easier than finding good instruments for X and M in order to estimate equation (1). In Figure 1, the average export share X from 1970 to 1998 is plotted against the average export variables on the right hand side of the regression function as they did may have led to inconsistent estimates due to the effect of reverse causation. 7 import share M from 1970 to 1998. The World Development Indicator 2000 data cover 174 countries.9 The figure exhibits a strong positive relationship between export share and import share (with the correlation of 0.85). In Figure 2, we plot the sum of the export share and the import share along the vertical axis. Along the horizontal axis, we plot the difference between the export share and the import share. The chart shows no evident relationship between the two variables in the long run (the correlation is equal to -0.13.) Frankel and Romer (1999) have already provided good instrument candidates for the total trade share. The major task of the next section is to present the Frankel-Romer model with the heteroscedasticity correction, propose instruments for the net trade share variable by capturing cross-country differences in borrowing and lending behavior, and investigate the impact of total trade and net trade on income after instrumenting for both endogenous independent variables. Since Ti X i M i and Ei X i M i , the estimation will shed implicit light on the distinctive export effect and the distinctive import effect. A. Bilateral Trade Regression Trade promotes growth through increased specialization, efficient resource allocation, diffusion of international knowledge, and heightened domestic competition (Sachs and Warner 1995). On the other hand, countries that produce more output tend to trade more with the rest of the world. To correct for the simultaneity bias, Frankel and Romer (1999) proposed the geographical characteristics of countries as instruments for total trade. They argued that geography is a powerful determinant of bilateral trade as 9 The observations on Netherlands Antilles (X = 495%, M = 486%, T = 981%, E = 9%) are dropped for graphical purpose. 8 well as overall trade. Furthermore, countries’ geographical features are not affected by their incomes, or by government policies and other factors that influence income. This section applies the extended version of the Frankel-Romer bilateral trade model as appeared in Frankel and Rose (2002). We also employ the updated bilateral trade data set which “is estimated to cover at least 98 percent of all trade.” (Frankel and Rose 2002, pp. 462) The first stage regression is based on the international trade extension of the gravity model, namely, that trade volume between two trading partners rises with an increase of national incomes or a reduction of the distance between them: (3) ln t ij ln( ij / GDPi ) a' X ij eij a0 a1 ln( dist ij ) a2 ln( pop j ) a3 ln( areai area j ) a4 (languageij ) a5 (borderij ) a6 (landlockedij ) eij . In equation (3), ij denotes the bilateral trade between countries i and j (measured as exports plus imports), a is the vector of coefficients, Xij is the vector of the covariates, distij measures the great circle distance between the principle cities of countries i and j, pop and area represent population and area respectively, languageij is a dummy variable which takes the value of 1 if people in country i and country j speak the same language, landlockedij is the number of landlocked countries within the country pair, and borderij is the dummy variable for a common border between two countries. Countries far away from each other tend to trade less. Since country size is inversely related to proximity, the impact of area is expected to be negative. Larger trading partner’s population implies higher demand for the domestic country’s export. Thus, the sign for popj should be positive. Countries are expected to trade more with each other if they have the same 9 official language or share a border. If countries have access to the ocean, shipping costs are significantly reduced which makes it less difficult to trade. Due to the possibility of heteroscedasticity in large cross-country analysis, this paper conducts the Breusch-Pagan test on equation (3). The test statistic rejects the homoscedasticity assumption with a high level of confidence. To counteract the problem of heteroscedasticity, weighted least square (WLS) estimation is used based on the procedure proposed by Harvey (1976).10 Assuming the variance of the disturbance term eij in equation (3) has the following hypothesized specification: (4) ij2 exp( ' X ij ). The estimator of is: (5) X ijX ij ' i j ~ 1 X i j ij ln( eˆij2 ) based on the estimation (6) ln( eˆij2 ) ' X ij ij where eˆij2 is the residual resulting from the OLS regression of equation (3) and ij ln( eˆij2 / ij2 ) . The values of both the dependent and independent variables are divided by the square root of the predicted variance of the disturbance term, which corresponds to the pattern of the residuals. This normalizes the residuals so that they are homoscedastic. The next step is to obtain the predicted value of the dependent variable in levels. Simply exponentiating ln tˆij underestimates the expected value of t ij because 10 We also adopt the robust ordinary least square (ROLS) estimation to correct for heteroscedascity by applying the Huber-White sandwich estimator of variance in place of the traditional variance calculation (Huber 1967, White 1982). This approach allows the correlation of errors for any one country, while errors 10 (7) E (t ij | X ij ) ij exp( a ' X ij ) where ij is the expected value of exp( eij ) , which is greater than unity. Thus, the bilateral trade share is predicted as: (8) tˆij ˆij exp( â ' X ij ). Assuming ij has the following functional form (9) ˆij exp( ˆ ' X ij ) , the following nonlinear regression is conducted to estimate the value of : (10) t ij exp( ' X ij ) exp( â ' X ij ) u ij . The value of ˆ ij is obtained from equation (9).11 Therefore, the predicted value of the overall trade share for country i becomes: (11) Tˆi ˆij exp( â ' X ij ) i j B. Net Trade Regression While aspects of the time-series relationship between net exports and income are well-established, e.g., its cyclicality, the cross-section relationship is not. What kind of national and international variables determine which countries have overall trade surpluses and which have overall trade deficits? What variables determine bilateral net exports between two countries? And are those cross-sectional determinants suitable for between countries remain uncorrelated. ROLS generates similar results as we have derived using OLS and WLS. 11 Detailed explanation of the adjustment procedure is presented in Zhang (2002a). 11 first-stage instruments in the current study, or do they too vary cross-sectionally with income levels? This section aims to answer. To begin, current account balance is consisted of three parts: (1) balance of trade in goods and services E expressed as the difference between exports and imports; (2) net rents, interest, profits and dividends represented as rB where B (-B) is the net foreign credit (debt) that earns (pays) interest at some appropriate interest rate r; (3) transfer payments TP such as foreign workers’ remittances. In the open economy, current account balance is also determined by the saving-investment gap (S – I). Changes in official reserves R may offset the surplus or deficit in the overall balance of the capital account KA and the current account CA. CA E rB TP S I R KA . (12) All variables in the above equation are divided by GDP.12 In principle, equation (12) describes both overall and bilateral current-account balances. As we explain in the next section, bilateral empirical implementation of our approach to (12) unfortunately founders on data limitations (few nations or global data-collection agencies publish more than a handful of the variables in equation (12) on a bilateral basis, including the current account itself.) It is clear that we can re-write equation (12) to conceive of E as determined crosssectionally by S, I, r, B, remittances, and reserve changes. Though S, I, and possibly r are also correlated with income levels, transfer payments and reserve changes are less likely to be. And there are at least some deeper determinants of S and E itself that might qualify as good instrument candidates, especially in financially open economies. Even if all the variables proposed and GDP are reasonably measured, the resulting ratio has the “division bias”. 12 12 With respect to S and I, in a closed economy with zero international capital movement, saving is equal to investment. Economic growth depends on both saving and investment directly. As countries open up, however, domestic investment is financed by the worldwide pool of savings, while domestic saving seeks the highest returns in the global capital market. When domestic saving is not sufficient to finance domestic investment, such as in the US, the difference is made up by foreign capital inflow. When domestic saving exceeds domestic investment, such as in Japan, the extra capital is invested abroad. Higher saving is associated with capital outflow and lower saving indicates capital inflow. Thus, while investment remains as a direct determinant of countries’ income level in the open market scenario, saving is related to income through the channel of global borrowing and lending, and domestic saving alone is arguably detached from its closed-economy influence on income. That in turn makes it or its deeper determinants reasonably good candidates for net-trade (E) instruments. Yet before this strategy becomes persuasive, there is an important question to answer: How integrated is the international capital market? Using the data for 16 OECD countries from 1960-1974, Feldstein and Horioka (1980) initially answered “not very.” They cross-sectionally regressed national gross domestic investment ratios on gross domestic saving ratios and presented a high saving-investment coefficient. They concluded that the strong sectional correlation between investment and saving is likely to be a result of low capital mobility and thus low capital market integration. Their skeptical conclusion is challenged by various studies. The major theoretical argument is that the high correlation may be explained by the real short-run shock (productivity, growth, globalization, government behavior, etc.) toward both saving and investment 13 (Obstfeld 1986; Obstfeld 1995; Obstfeld and Rogoff 1995; Baxter and Crucini 1993; Coakley, Kulasi and Smith 1998). Based on the empirical study of extended sample and improved approaches, recent cross-country and time-series study discovered a low saving-investment coefficient, in accord with the belief in high capital mobility. Moreover, almost all commentators agree that however “low” it is, the degree of international capital mobility has increased significantly, especially during the last 20-25 years. The widespread capital control removal in the industrial countries enhanced financial integration. The financial sector reforms and the opening up of the capital account to private capital inflows help to predict lower and lower correlation between saving and investment, even in developing countries. (With less diversified production and export structures, oil-exporting countries and small countries tend to have a low saving-investment correlation.) High capital mobility allows us to use saving or factors that determine saving as instruments for net trade or current account balance. 13 Yet the relationship between saving and net trade or current account balance reflects a complex interaction of households, firms, and governments both at home and oversea. A random shock specific to the net trade balance could potentially affect a country’s saving ratio. For instance, governments tend to adopt contractionary fiscal policy to prevent sustained large capital flows when the countries are in trade deficit. In order to avoid these new endogeneity problems, we propose a country’s demographical characteristics as more fundamental 13 Glick and Rogoff (1995) showed that country-specific shocks rather than global productivity shocks are important determinants of current account fluctuations. Empirical work extended dynamic optimizing models proposed by Ghosh (1995) and Ghosh and Ostry (1995) to the open economy context (Razin 1995, Obstfeld and Rogoff 1998). From a saving-investment perspective, Debelle and Faruqee (1996) investigated the determinants of current account using the structural approach. Given certain investment level, countries with a higher saving rate tend to be net exporters and net importers with a lower saving rate. 14 and cleaner instruments for net trade – recognizing that differences of saving between two countries can be very well explained by their differences in demography. Demography matters in both age structure and in steady-state population growth. With regard to structure, both theoretical and empirical literature proposes the empirical linkage between net trade and the age structure of the population. Life cycle models in an intertemporal approach suggest that net trade balance is the outcome of forwardlooking dynamic saving and investment decisions. People save during the earning span at productive age and dissave when they are young or old. Thus, aggregate national savings will be relatively high if the size of the dependent population is low comparing to the size of the working-age population. High savings build up domestic and foreign assets, reflected as a large net trade surplus. In the cross-sectional context, real growth of output (GDP) is expected to have a positive impact on the net trade balance due to the fast growing supply in the global market. We use the population growth rate as the proxy for the real growth of output to capture the impact. Population growth rate may also serve as the determinant of saving. Balanced population growth leads to the growth of income, which generates greater gap between the target consumption levels of the current working generation and the dissaving of retirees from a less prosperous generation. (Hussein and Thirlwall 1999, Knight and Scacciavillani 1998) Consequently, the increased population growth rate leads to higher savings. The unbalanced population growth impact, which changes the demographic composition of the society, is captured by the dependency ratio. Saving ratio is independent of the level of per capita income but is a function of the population growth rate as shown in the life-cycle hypothesis of Ando and Modigliani (1963). 15 Urbanization is also a deeper fundamental determinant of saving, like demography, and less likely than saving itself to be subject to feedback influences from exogenous shocks to income. The influence of the urban population ratio on saving is mixed. The trend toward urbanization leads to lower private saving rate because of the precautionary saving motive. (Edward 1996, Loayza, Schmidt-Hebbel and Serven 2000) On the other hand, urbanization exerts a positive and statistically significant impact on national saving (Rodrik 1998). An important determinant of net trade balance is the net income from abroad. Negative current account balance increases foreign indebtedness. A country can run a steady-state trade deficit equal to the investment income. The deterioration in investment income requires improvement in the trade balance. Therefore, negative net foreign liability requires countries to run trade surplus while positive net foreign asset enables countries to run trade deficits. International reserves flow into the home country when country exports and flow out of the home country when country imports. The volume of imports largely depends on the amount of international reserves available to finance imports, especially for developing countries. A country can run a steady-state trade deficit equal to its decline in official reserves. Negative net export and negative changes in reserves go together. Net trade balance is also influenced by a nation’s relative price of tradeable output to non-tradeable output (compared in turn to the comparable world price ratio). The higher is a nation’s relative tradeables price (by world standards), the larger will be its relative output of both exportables and import substitutes, output that gets exported in the first instance and displaces imports in the second (Obstfeld and Rogoff 1996, pp. 199- 16 257). Yet there is no obvious reason why a nation’s income level varies systematically vis-à-vis other nations with this relative price, because every nation produces both tradeables and non-tradeables and earns income from both.14 Current transfers, such as workers’ remittances, are an important source of foreign exchange for many countries. Positive transfers of resources increase country’s real exchange rate and thereby hurt country’s competitiveness in the world markets by reducing the range of home goods for exports (Obstfeld and Rogoff 1996, pp. 255.) The net export determination equation is shown as the following: (13) Eˆ i ˆ0 ˆ1 (depi ,t ) ˆ2 ( popg i ,t ) ˆ3 (oil i ,t ) ˆ4 (urpop i ,t ) ˆ5 (nii ,t ) ˆ6 (dres i ,t ) ˆ7 (relp i ,t / relp i ,t 5 ) ˆ8 (nct i ,t ) where depi,t is calculated as the proportion of people under 14 or over 65 in country i. popgi,t stands for the population growth rate. oili,t takes the value of 1 if country i depends heavily on oil revenues as her main source of income. The dummy variable for oil exporting countries (primarily Gulf States) is included due to their unique demographic trends such as the intra-regional and international migration. The dummy variable for oil exporting countries is expected to be positive since these countries typically have a more favorable current account position on average (Chinn and Prasad 2000.) urpopi,t is defined as the percentage of the total population living in urban area. nii,t is the stock of net income from abroad as ratio of GDP, measured in 1990. It 14 One could object that several of these variables, e.g., net remittances or returns on cross-border capital placement are co-determined endogenously with net exports (E) as part of the deeper fundamentals of current account behavior – how the current account sum of all of them, E, remittances, capital income, responds to shifts and shocks to output, investment prospects, and government spending needs (Obstfeld and Rogoff 1999, pp. 74-116). We agree, but can think of no feasible measures of global and countryspecific productivity and fiscal shocks that could serve as deeper, more fundamental instruments than those we choose. Furthermore, we think this objection affects net capital income and flows of official reserves more than it affects remittances. 17 captures the term rB in equation (12). dresi,t denotes changes in net official reserves as a ratio of GDP. Price relative (relp i ,t / relp i ,t 5 ) is measured as the ratio of ratios of export prices to GDP deflators from 1985 to 1990. ncti,t represents current transfers (from abroad), such as migrants’ remittances, divided by GDP. C. Income Regression Frankel-Romer study estimated the impact of total trade on per capita income controlling for the size of each country. To incorporate the net trade extension, this paper estimates the income equation (14) ln( yi ) 0 1 (Ti ) 2 ( Ei ) 3 ln( popi ) 4 ln( area i ) u i . where ui is a proxy for all the uncertain factors that may also affect per capita income level. Since both Ti and Ei are likely to be correlated with the error term, this paper utilizes the instrumental variable (IV) estimation technique, substituting these two variables with their predicted values from section A and section B. The first stage regression to estimate total trade share is based on the bilateral observations. Ideally, we would also hope to obtain the bilateral determinants for net trade balance between country pairs. To estimate the net trade balance on a bilateral basis, we need to obtain observations on a bilateral basis as well, for example, relative price of the tradeables. However, variables of this kind are not available. Due to the limit of good instrument candidate, we employ a set of instruments for each country’s overall net trade balance while aggregating the bilateral trade observations as instruments for each country’s total trade. 18 The size and significance of the total trade and net export variables determines if exports or imports improve prediction of per capita output. (15) ln( yˆ i ) ˆ0 (ˆ1 ˆ2 )( X i ) (ˆ1 ˆ2 )( M i ) ˆ3 ln( popi ) ˆ4 ln( areai ) where ˆ1 ˆ 2 measures the predicted partial effect of exports on the per capita income and ˆ1 ˆ 2 measures the predicted partial effect of imports on the per capita income, holding country size constant. The relative importance of exports and imports depends on the sign, magnitude, and the joint significance of ˆ1 and ̂ 2 estimated for equation (14). For instance, if ˆ1 0, ˆ 2 0 and ˆ1 ˆ 2 , both exports and imports raise income. If ˆ1 0, ˆ 2 0 and ˆ1 ˆ 2 , exports but imports raise income. If ˆ1 and ̂ 2 are jointly insignificant, the regression produces imprecise estimates. The estimation result is shown in Section V. III. Data The bilateral trade regression is based on the data set obtained from the Andrew Rose Web site. This data set covers the bilateral trade share, distance, population, area, common language dummy variable, border dummy variable, and landlocked dummy variable for 186 countries in year 1990. The original source for the trade data is the World Trade Database and United Nation’s International Trade Statistics Yearbook. The population and real GDP per capita data come from Penn World Table 5.6. Area data are from the World Reference Atlas. The information of the distance and common language dummy variable is from the CIA’s web site. 19 The net trade data set including net trade share, population aged 0-14, population aged 65 and above, total population, population growth rate, net income, changes in net reserves, GDP at market prices, official exchange rate, and current transfers is taken from World Bank’s World Development Indicator 2000. Net income from abroad includes the net labor income and net property entrepreneurial income components of the SNA. Changes in net reserves are the net change in a country’s holdings of international reserves resulting from transactions on the current, capital, and financial accounts. Current transfers take three forms. The net current transfers from abroad comprise transfers of income between residents of the reporting country and the rest of the world that carry no provisions for repayment. The net current transfers are recorded in the balance of payments whenever an economy provides or receives goods, services, income, or financial items without a quid pro quo. The third measure comes from the net workers’ remittances as recorded by IMF Balance of Payments Statistics. The relative price of tradeables is calculated based on the data from the World Table. It refers to the 1990 to 1985 ratio of export-to-GDP-deflator price indexes. Since the export prices index is in 1987 current US dollars while the GDP deflator is in 1987 local currencies, we create an exchange rate index for year 1990 and year 1985 to adjust the prices into the same currencies. The exchange rate index for 1990 is 100 times the ratio of the 1990 exchange rate and 1987 exchange rate. The data on official exchange rate is taken from the World Development Indicator 2000. Oil country list includes 11 OPEC member countries (Algeria, Libya, Nigeria, Indonesia, Iran, Iraq, Kuwait, Qatar, Saudi Arabia, the United Arab Emirates), additional Gulf Cooperation Council countries (Oman, Bahrain), and other non-OPEC oil producing countries (Russia, Norway, Mexico). 20 Data on trade, gross domestic product (GDP), population, and area for the income regression are also extracted from the Andrew Rose’ web site. This data set consists of the year 1990 observation for 210 “countries” from the World Development Indicators 1998, merged with data from the Penn World Table 5.6. Summary statistics and detailed description of all variables are provided in the appendix. IV. Estimation results A. Bilateral Trade Regression This paper runs the bilateral trade regression by conducting the Breusch-Pagan tests on the ordinary least square (OLS) estimate. The resulting test statistic of 88.38 is statistically significant at the p = 0.000 level. That is, the null hypothesis of homoscedasticity is rejected. OLS results are adjusted using the approach proposed by Harvey (1976), where the square of the weight ( ij2 ) is constructed as the exponential function of the predicted ln eˆij2 . All the variables in the regression are weighted by the inverse of the variance. WLS regressions reveal a positive and statistically significant relationship between bilateral trade and country’s geographic characteristics. All else being equal, large distance leads to less trade. If the official languages in country i and country j are the same, trade rises by 55 percent. Two countries that share a border trade 62 percent more than countries pairs which do not. Landlocked countries tend to trade less due to the high transportation cost. 41% of the variation in the bilateral trade share is explained 21 by the estimated bilateral trade equation. Frankel-Romer results are confirmed by using a larger sample.15 B. Net Trade Regression To understand how the results may be driven by the specification and possible omitted variable bias, the net export regressions are estimated by including variables to control for countries’ demographic characteristics and balance of payment features. Table 2 provides the result of the net trade regression based on equation (13). The results of the first specification indicate that an increase in the age dependency ratio leads to a reduction in net trade. The impact of population growth rate is positive but not significant. Controlling for country’s demography, oil-exporting countries tend to have a higher net trade share relative to non-oil-exporting countries. The second specification adds the urban population ratio variable. Urban population ratio is positively related to the net trade balance at 1% significance level. Demography appears to explain approximately 15 percent of the per capita GDP level in 1990. These estimates imply that a country’s demography affects her net trade. In specification (3), (4) and (5), we suggest a different set of net trade determination factors. As expected, the net income from abroad variable has a significant and negative impact on net export share. Changes in reserves and the relative price of tradeables do not appear to play a major role in determining the trade balance, possibly because of the inability of international trade flow to adjust to changes in the market 15 Data on bilateral trade are only available for 62 countries in the Frankel and Romer (1999) study. Based on the coefficients estimated from the gravity model, Frankel and Romer impute the bilateral trade share for country pairs whose recorded trade share is missing. The quality of the instruments and the precision of 22 condition due to the non-market forces in some planned economics. Current transfers take three forms. All three measures indicate that transfers affect the net trade balance negatively. A one percentage point difference in the current transfers is associated with an approximately one percentage point difference in the net trade share balance. From specification (6) to (11), we combine the two sets of instruments. In most cases, lower dependency ratio, higher urban population ratio, lower net income from abroad, and lower net current transfers are significantly associated with an improved trade balance. The binary variable of oil-exporting countries is associated with the net trade balance at the 1% significance levels, while we do not find a significant relationship between trade balance and population growth rate, changes in reserves, as well as relative price of tradeables. 14% to 80% of the variation in the net export share is accounted for by the joint predictive power of the net trade regression according to different specification results. C. Income Regression In running the regression of equation (14), endogeneity problem arises since the per capita income may affect both the total trade value and net trade value. Endogeneity renders biased OLS estimates. Higher income raises both export and import. The OLS estimation would be biased upward. As increased levels of the income promote domestic production of tradeables while stimulate higher import demand, the ultimate effect on the net trade balance is uncertain. OLS estimates for the net export coefficient may bias either upward or downward. In order to cope with the simultaneity between openness the estimated effects are brought into questions, especially if the gravity relation is systematically different for countries in the sample than for countries added through imputation. 23 and income, the income equation is estimated using the instrumental variables procedure. The objective of this section is to test whether, after controlling for the size of the economy, total trade and net trade balance contributes to explaining the cross-country differences in per capita income. Before running the income regression, it is worthwhile to investigate the quality of the instruments suggested in section A and B. In the correlation matrix presented in Figure 3, a positive relationship between the selected variables and their predicted values is suggested. Next to the correlation matrixes, we present the correlation tables. Overall, it shows that total trade share is positively linked with the aggregated exponents of estimated bilateral log equations (around 0.53 to 0.77.) Regarding the net exports share, we find that the actual values are in general positively correlated with predicted values with the correlation between 0.51 and 0.90. Considerable information about countries overall trade and trade balance is provided by the predicted values. The scatter plot matrix in Figure 3 helps visually identify the positive correlations between total trade share, net trade share, and their predictions. Additional evidence of the correlation is displayed in the correlation tables. The correlation between the total trade share and the net trade share, as well as the correlation between the predicted total trade share and the predicted net trade share, are weak and nonlinear, implying that the multi-collinearity problem does not occur in the IV regressions. Both the predicted total trade share and the predicted net trade share are proved to be effective instruments with no apparent outliers. Table 3 presents the two-stage least squares estimation results based on the eleven specifications of the first stage net trade balance estimation. The primary focus of this 24 table is the coefficient estimates of the total trade share and net trade share. When controlling for the size of the economy, we find that higher levels of both the overall trade share and the net trade share are typically linked with more per capita income. Size of countries, as measured by population and area, has insignificant effect. This suggests that the main influence of geography on income is through the channel of international flow of goods and services. Table 3 shows that the regressions account for 28 to 60 percent of the cross-country variation in per capital real GDP. The main results from Table 3 are summarized in Table 4, where we record the coefficients estimates for T and E. The F statistics reject the joint hypothesis that both coefficients in Row 1 and 2 are zero. Coefficients of X in Row 3 are the sum of the Row 1 coefficients of T and Row 2 coefficients of E. Coefficients of M in Row 4 are the difference of the Row 1 coefficients of T and Row 2 coefficients of E. The relative standard errors for the sum/difference are computed and recorded in the parenthesis. The size of the coefficients for X changes moderately but remains positive and highly significant. The coefficients for M are significantly most of the time. Table 4 leads to two main conclusions: first, comparable and equally open countries that differ between themselves in X-openness and M-openness have different per capita GDP; those that are 1 percent more X-open than M-open have approximately 0.1 higher GDP per person. Second, increasing the import trade share by 1 percentage point is associated with a reduction of the per capita GDP by approximately 0.1 percent. In other words, countries with higher net export share are also countries with higher income per person, ceteris paribus.16 16 Very similar results emerge implicitly from Miller’s and Upadhavy’s (2000) study of the effect of openness on a country’s total factor productivity (intimately related, of course, to its income per person). 25 For ordinary least squares to generate the best linear unbiased estimator, it is assumed that all variables on the right hand side of equation (14) are purely exogenous. This paper examines the hypotheses using two Hausman tests. Hausman specification test follows the method outlined in Hausman (1978). The test statistics reject the null that IV and OLS estimates are equal. The second, and arguably better, Hausman test simply adds the predicted endogenous variables into the income equation. T-test results reject the hypothesis that the additional regression parameters equal zero, indicating no misspecification in our model. In other words, both total trade and net trade are shown to be endogenous. Figure 4 graphs the confidence ellipse around the mean of our estimated coefficients on 1 and 2 at the 95% confidence level. The orientation of the ellipses shows that ˆ1 and ˆ2 are negatively correlated. Most of the points lie inside the northeast panel and far beyond the Y-coordinate. To test the conjecture that income is related to trade, this paper runs an F-test to find the joint significance of the coefficients on total trade share and net trade share. The test statistics offers support for the conjecture that income is associated with both total trade and net trade. VI. Conclusion Rather than staying closed, many countries gain substantial amounts of income through their intensive involvement in international trade. This phenomenon is widely recognized and has received considerable attention from analysts who wish to learn whether countries should enhance or detract from trade. Although many recent empirical Their “trade orientation” measure, based on the countries deviations from Purchasing-Power-Parity exchange rates proxies for import shares. 26 studies attempt to capture the trade-income nexus using cross-sectional and time series data, they lack the power to distinguish between the impact of exports on income and the impact of imports on income. What causes different levels of per capita income in various countries, export openness or import openness? This study explores empirical evidence on this issue by disaggregating both exports and imports. In particular, the conventional wisdom views that trade (both exports and imports) are beneficial for economic growth is challenged. Using instrumental variable estimation with the cross-country variations in overall trade explained by geographic factors and the cross-country variations in net trade balance explained by demography as well as balance of payment features, we investigate the impact of both exports and imports on income. A significantly positive effect of export share on per capita income level confirms the findings of earlier studies in which export matters in determining the cross-country variation in income. Imports, rather, affect growth negatively. Several studies in the literature help explain the phenomena. One explanation is that import pattern matters. Countries import primarily consumption goods grow slower than countries with a large proportion of their imports as intermediate goods. Among the latter group of countries, importing primarily from follower countries leads to less technological improvement than importing primarily from leaders (Keller 2000). A more popular argument is that importing from established foreign firms may discourage the development of domestic infant industries, especially in the third world countries. Therefore, reducing the imports of infant industry products may encourage the internalization of costs of natural resources and help increase the total factor productivity, hence countries’ overall per capita income level. 27 The growth of exports provides productivity and technology benefits to national economies. There are a number of channels through which the rapid expansion in exports can contribute to the increased economic output. First, export growth offers greater economies of scale, encourages specialization, and increases production efficiency. The improved production techniques and management styles acquired by the export sectors generate positive externalities to non-export sectors. Unlike import substitution economies, export-oriented countries tend to have relatively less price distortions and more efficient allocation of resources. Second, increased exports ease the foreign exchange constraint. By providing a source of foreign exchange for countries that wish to import intermediate inputs that embody domestically unavailable technology, export expansion leads to cost reduction, increased efficiency and greater access to international market. Third, exports enhance diffusion of knowledge through learning by doing. Intensive interaction with the international market encourages communication of ideas, intensify competition, and stimulate domestic imitation and innovation. Thus, outward-oriented strategy, with the emphasis on exports, is increasingly preferred by the policymakers. 28 Appendix Table A1: Summary Statistics of the Total Trade Share Regression General Statistics Variable bilateral trade share distanceij populationj areai areaj languageij borderij landlockedij Obs. 10940 13264 11034 14746 14750 14746 14746 Mean 0.005 4667.31 57986.14 8.86e+11 0.12 0.02 0.19 Std. Dev. 0.025 2735.97 168141.90 6.41e+12 0.32 0.14 0.41 Min 5.31e-09 19.43 40 1883 0 0 0 Max 0.793 12351.26 1133683 2.08e+14 1 1 2 Variable Description Variable bilateral trade share distanceij populationj areai areaj languageij borderij landlockedij Description Bilateral trade value divided by real GDP The great circle distance between the principle cities of country i and j Country j’s population in 000’s Product of real areas. 1 for common official language between country i and j 1 for common land border between country i and j Number of landlocked countries in pair (0, 1 or 2) Data source: Frankel and Rose (2002). Table A2: Summary Statistics of the Net Trade Share Regression General Statistics Variable Net trade share Age dependency ratio Population growth rate Oil-exporting countries Urban population ratio Net income from abroad Changes in reserves Relative prices Net current transfers from abroad Net current transfers Remittances Obs. 145 179 196 206 196 168 148 116 96 142 91 Mean -7.01 40.01 1.89 0.08 50.69 -1.40 -1.24 1.00 3.57 5.60 1.41 Std. Dev. 17.63 6.71 1.64 0.27 23.75 11.51 4.05 0.40 10.78 10.45 4.90 Min -199.46 26.98 -4.87 0 5.2 -30.62 -19.87 0.14 -22.82 -26.86 -10.70 Max 26.84 52.11 14.12 1 100 74.74 20.95 3.05 81.29 64.03 27.24 Variable Description Variable Net trade share Age dependency ratio Description 1990 Net trade divided by real GDP 1990 Population aged 0-14 + population aged 65 and above (% of total) 29 Population growth rate Oil-exporting countries Urban population ratio Net income from abroad Changes in reserves Relative prices Net current transfers Net current transfers from abroad Remittances 1990 Population growth (annual %) 1 for OPEC, Abab-OPEC, and non-OPEC countries 1990 Urban population (% of total) 1990 net income from abroad (% of GDP) Net changes in reserves (% of GDP) 1990 to 1985 ratio of export-to-GDP-deflator price indexes 1990 net current transfers (% of GDP) 1990 net current transfers from abroad (% of GDP) 1990 Workers’ remittances (% of GDP) Data source: Frankel and Rose (2002). World Development Indicators 2000. International Monetary Fund Balance of Payments Statistics World Bank Table. Table A3: Summary Statistics of the Income Regression General Statistics Variable Per capita GDP Total trade share Net trade share Population Area Obs. 115 145 145 115 210 Mean 4913.78 80.67 -7.01 39177.84 631418.7 Std. Dev. 4945.33 63.57 17.63 134104.40 1813066 Variable Description Variable Per capita GDP Total trade share Net trade share Population Area Description Real per capita GDP chain index Total trade divided by real GDP Net trade divided by real GDP Population in 000’s Area in sq. km. Data source: Frankel and Rose (2002). World Development Indicators 2000. 30 Min 399 13.62 -109.47 40 2 Max 18054 538.674 26.84 1133683 1.71e+07 Bibliography Ando, A., and Franco Modigliani. 1963. “The Life-Cycle Hypothesis: Aggregate Implications and Tests.” American Economic Review: 55-84. Balassa, Bela. 1978. “Exports and Economic Growth: Further Evidence.” Journal of Development Economics 5:181-189. Barro, J. Robert and Xaview Sala-I-Martin. 1995. Economic Growth. New York: McGarw-Hill. Baxter, M. and M. J. Crucini. 1993. “Explaining Saving-Investment Correlations.” American Economic Review 83: 416-436. Chinn, Menzie and Eswar S. Prasad. 2000. “Medium-term Determinants of Current Accounts in Industrial and Developing Countries: An Empirical Exploration.” Cambridge, Massachusetts: National Bureau of Economic Research Working Paper No. 7581. Coakley, J., F. Kulasi, and R. Smith. 1998. “The Feldstein-Horioka Puzzle and Capital Mobility: A Review.” International Journal of Finance and Economics 3: 169-88. Debelle, Guy, and Hamid Faruqee. 1996. “What Determines the Current Account? A Cross-Sectional and Panel Approach.” International Monetary Fund Working Paper No. 96/58. Dollar, David. 1992. “Outward-Oriented Developing Economies Really Do Grow More Rapidly: Evidence from 95 LDCs, 1976-85.” Economic Development and Cultural Change: 523-44. Edwards, Sebastian. 1996. “Why Are Latin America’s Savings Rates So Low? An International Comparative Analysis.” Journal of Development Economics 51(1): 5-44. Edwards, Sebastian. 1998. “Openness, Productivity, and Growth: What do We Really Know?” Economic Journal 108: 383-98. Feder, Gershon. 1982. “On Exports and Economic Growth.” Journal of Development Economics 12(1): 59-73. Frankel, A. Jeffrey and David Romer. 1999. “Does Trade Cause Growth?” American Economic Review 89: 379-399. Frankel, A. Jeffrey and Andrew K. Rose. 2002. “Estimating the Effect of Currency Unions on Trade and Output.” Quarterly Journal of Economics. 31 Ghosh, Atish. 1995. “International Capital Mobility Amongst the Major Industrialised Countries: Too Little or Too Much?” The Economic Journal 105: 107-28. Ghosh, Atish R., and Jonathan D. Ostry. 1995. “The Current Account in Developing Countries: A Perspective from the Consumption-Smoothing Approach.” World Bank Economic Review 9: 305-333. Glick, Reuven and Kenneth Rogoff. 1995. “Global versus Country-Specific Productivity Shocks and the Current Account.” Journal of Monetary Economics 35: 159-92. Grossman, Gene and Elhanan Helpman. 1991. Innovation and Growth in the Global Economy, Cambridge: Massachusetts Institute of Technology Press. Gruen, David, Terry O’Brien, and Jeremy Lawson, eds. 2002. Globalization, Living Standards and Inequality. Canberra: Reserve Bank of Australia and Australian Treasury. Proceedings of a Conference held in Sydney, May 27-28, 2002. Harrison, Ann. 1996. “Openness and Growth: A Time-Series, Cross-Country Analysis for Developing Countries.” Journal of Development Economics 48(2): 419-47. Harvey, C. Andrew. 1976. “Estimating Regression Models with Multiplicative heteroscedasticity.” Econometrica 44(3): 461-65. Hausman, A. Jerry. 1978. “Specification Test in Econometrics.” Econometrica 46(6): 1251-70. Huber, P. J.. 1967. “The Behavior of Maximum Likelihood Estimates under Nonstandard Conditions.” in Proceedings of the Fifth Berkeley Symposium in Mathematical Statistics and Probability (1), University of California Press, Berkeley, CA, pp. 221-233. Hussein, A. Khaled and A. P. Thirlwall. 1999. “Explaining Differences in the Domestic Savings Ratio Across Countries: A Panel Data Study.” Journal of Development Studies 36(1): 31-52. Keller, Wolfgang. 2000. “Do Trade Patterns and Technology Flows Affect Productivity Growth?” The International Bank for Reconstruction and Development. The World Bank. Knight, Malcolm and Fabio Scacciavillani. 1998. “Current Accounts: What is Their Relevance for Economic Policymaking?” International Monetary Fund Working Paper No. 98/71. Krueger, O. Anne. 1978. “Foreign Trade Regimes and Economic Development: Liberalization Attempts and Consequences.” Cambridge Massachusetts: Ballinger Publishing Co. for the National Bureau of Economic Research. 32 Krugman, Paul. 1994. “The Myth of Asia’s Miracle.” Foreign Affairs: 62-78. Lawrence, Z. Robert and David E. Weinstein. 2001. “Trade and Growth: Import-led or Export-led? Evidence from Japan and Korea.” In Stiglitz and Yusuf (2001). Levine, Ross and David Renelt. 1992. “A Sensitivity Analysis of Cross-Country Growth Regression.” American Economic Review 82(4): 942-63. Levinsohn, Jim. 2003 forthcoming. Responses to Globalization: The U.S. Textile and Apparel Industries. Washington, D.C.: Institute for International Economics. Lewis, Howard III and J. David Richardson. 2001. Why Global Commitment Really Matters! Washington, D.C.: Institute for International Economics, September. Loayza, Norman, Klaus Schmidt-Hebbel and Serven Lui. 2000. “What Drives Private Savings Across the World?” The Review of Economics and Statistics. Lucas, E. Robert. 1988. “On the Mechanics of Economic Development.” Journal of Monetary Economics 22(1): 3-42. Mazumdar, Joy. 2001. “Imported Machinery and Growth in LDCs.” Development Economics 65: 209-24. Journal of Michaely, Michael. 1977. “Exports and Growth: An Empirical Investigation.” Journal of Development Economics 4: 49-54. Miller, Stephen M. and Mukti P. Upadhyay. 2000. “The Effects of Openness, Trade Orientation, and Human Capital on Total Factor Productivity.” Journal of Development Economics 63: 399-423. Mody, Ashoka and Kamil Yilmaz. 2002. “Imported Machinery for Export Competitiveness.” World Bank Economic Review 16: 23-48. Obstfeld, Maurice. 1986. “Capital Mobility in the World Economy: Theory and Measurement.” Carnegie Rochester Conference Series on Public Policy 24: 55-104. Obstfeld, Maurice. 1995. “International Capital Mobility in the 1990s.” In Kenen, P. ed. Understanding Interdependence, Princeton University Press 1995. Obstfeld, Maurice and Kenneth Rogoff. 1995. “The Intertemporal Approach to the Current Account.” Cambridge, Massachusetts: National Bureau of Economic Research Working Paper No. 4893. Obstfeld, Maurice and Kenneth Rogoff. 1996. “Foundations of International Macroeconomics.” Cambridge, Massachusetts: Massachusetts Institute of Technology Press. 33 Ozler, Sule and Kamil Yilmax. 2000. “Investment in Imported Machinery for Export Cometitiveness.” Manuscript, August 7. Pavcznik, Nina. 2000. “Trade Liberalization, Exit, and Productivity Improvements: Evidence from Chilean Plants.” Cambridge, Massachusetts: National Bureau of Economic Research Working Paper No. 7582, August. Ram, Rati. 1990. “Imports and Economic Growth: A Cross Country Study.” Economia Internazionale 43: 45-66. Razin, Assaf. 1995. “The Dynamic-Optimizing Approach to the Current Account: Theory and Evidence.” In P. Kenen (ed.) Understanding Interdependence: The Macroeconomics of the Open Economy. NJ: Princeton. Rodriguez, Frankcisco and Dani Rodrik. 2000. “Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-National Evidence.” Macroeconomics Annual 2000, Volume 15, Ben Bernanke and Kenneth Rogoff, eds.. Massachusetts Institute of Technology Press for NBER, Cambridge, MA: pp. 261-325. Rodrik, Dani. 1995. “Trade Policy and Industrial Policy Reform.” In Jere Behrman and T. N. Srinivasan eds. Handbook of Development Economics 3B, Amsterdam: North Holland. Rodrik, Dani. 1998. “Saving Transitions.” Paper Prepared in the World Bank Research Project entitled “Saving Across the World”. Rodrik, Dani. 1994. “Getting Interventions Right: How South Korea and Taiwan Grew Rich.” 20th Panel Meeting of Economic Policy, NBER Working Paper No. 4964. Romer, M. Paul. 1986. “Increasing Returns to Long Run Growth.” Journal of Political Economy 94(5): 1002-37. Romer, M. Paul. 1992. “Two Strategies for Economic Development: Using Ideas and Producing Ideas.” World Bank Annual Conference on Economic Development, Washington D. C.. The World Bank. Sachs, Jeffrey, and Andrew Warner. 1995. “Economic Reform and the Process of Global Integration.” Brookings Papers on Economic Activity: 1-95. Stiglitz, Joseph E. and Shahid Yusuf, eds. Washington, D.C.: The World Bank. Rethinking The East Asian Miracle. Warner, Andrew M. 2003. “Once more Into the Breach: Economic Growth and Global Integration.” Manuscript, January. 34 Wei, Shang-Jin and Yi Wu. 2001. “Globalization and Inequality: Evidence from Within China.” Cambridge, Massachusetts: National Bureau of Economic Research Working Paper No. 8611. Wei, Shang-Jin. 2002a. “Is Globalization Good for the Poor in China?” Finance & Development: 26-9, September 2002. Wei, Shang-Jin. 2002b. “China as a Window to the World: Trade Openness, Living Standards and Income Inequality.” In Gruen et al.: 109-17. White, H.. 1982. “Maximum Likelihood Estimation of Misspecified Models.” Econometrica 50: 1-25. Zhang, Shuo. 2002a. “Further Investigation of the Link Between Trade and Income.” Working Paper, Department of Economics, Syracuse University. Zhang, Shuo. 2002b. “The Link between Trade and Income: Export Effect, Import Effect, or Both?” Working Paper, Department of Economics, Syracuse University. Zhang, Shuo. 2003. “A Panel Study of the Openness-Growth Linkage.” Working Paper, Department of Economics, Syracuse University. 35 Table 1: The Bilateral Trade Equation, 1990 Dependent variable: ln (bilateral trade share) ln (distanceij) -1.21*** (0.03) ln (populationj) 1.01*** (0.02) -0.29*** ln (areai areaj) (0.01) languageij 0.55*** (0.08) borderij 0.62*** (0.15) landlockedij -0.70*** (0.06) Constant -0.58* (0.33) Number of Observations 8104 R2 0.41 SE of regression 2.28 Standard errors recorded in parentheses. The symbols *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. 36 Table 2: The Net Export Equation, 1990 Dependency ratio Population growth rate Oil-exporting countries (1) -0.73** (0.28) 0.93 (1.59) 14.39*** (4.64) Urban population ratio (2) -0.34 (0.36) 0.60 (1.59) 12.16** (4.79) (3) Dependent variable: net export share (4) (5) (6) -0.65*** (0.19) 1.52 (1.14) 8.10*** (2.39) -0.16 (0.16) -0.72*** (0.27) -4.08* (2.66) -1.00*** (0.13) Relative price Net transfers from abroad Net current transfers -0.79*** (0.10) -0.35* (0.19) -0.43 (1.64) -1.09*** (0.16) -0.37 (0.38) -1.97 (3.17) -0.69*** (0.19) -0.45* (0.24) -2.36 (2.10) -0.70*** (0.13) -1.14*** (0.10) Net remittances Number of Observations R2 SE of regression (8) -0.38 (0.32) -0.90 (2.03) 10.41** * (3.65) 0.13* (0.07) Net income from abroad Changes in reserves Constant (7) -0.06 (0.15) -0.89 (0.82) 4.26** (2.09) -0.64*** (0.14) -0.16 (0.17) 0.23 (1.35) -0.79*** (0.26) 0.15 (0.35) -1.27 (2.78) (9) -0.43** (0.22) 1.51 (1.11) 6.86*** (2.40) (10) 0.09 (0.18) -1.02 (0.82) 3.56* (2.12) (11) 0.13 (0.41) -2.23 (2.10) 9.28** (3.61) 0.08** (0.04) -0.58*** (0.19) -0.45* (0.24) -2.10 (2.05) -0.68*** (0.13) 0.05* (0.03) -0.59*** (0.14) -0.16 (0.17) 0.47 (1.35) 0.15* (0.07) -0.67** (0.26) 0.15 (0.34) -0.52 (2.74) -1.03*** (0.10) 22.30** (9.94) 134 -0.64 (16.08) 134 0.83 (2.71) 76 -2.31 (1.73) 113 -1.14*** (0.26) -5.90* (3.31) 74 0.14 14.43 0.16 14.33 0.80 7.19 0.78 6.85 0.65 10.19 Robust standard errors recorded in parentheses. The symbols *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. 21.01*** (6.57) 70 0.69 5.42 -1.02*** (0.10) 2.04 (5.34) 106 -0.82*** (0.29) 12.80 (10.55) 67 7.91 (8.87) 70 -6.84 (7.49) 106 -0.84*** (0.28) -13.85 (17.12) 67 0.73 5.50 0.49 8.46 0.71 5.28 0.73 5.45 0.52 8.27 Table 3: Trade, Net Trade, and Income, 1990 Total trade share Net trade share Ln (population) Ln (area) Constant Number of Observations R2 RMSE (1) 0.004 (0.007) 0.147*** (0.044) 0.00 (0.10) -0.06 (0.13) 8.83*** (2.17) 102 0.56 0.74 (2) 0.004 (0.007) 0.150*** (0.035) 0.00 (0.11) -0.06 (0.13) 8.93*** (2.04) 102 0.65 0.66 Dependent variable: ln (per capita real GDP) (3) (4) (5) (6) (7) 0.019 0.024 0.107** 0.007 0.017 (0.016) (0.017) (0.041) (0.010) (0.012) 0.020 0.046** -0.076 0.102*** 0.068*** (0.057) (0.016) (0.072) (0.038) (0.016) 0.06 0.12 1.14** 0.03 0.12 (0.18) (0.18) (0.48) (0.12) (0.14) 0.20 0.17 0.36 -0.02 0.10 (0.23) (0.14) (0.35) (0.18) (0.20) 4.01 3.32 -14.00 7.96** 4.65* (4.65) (3.42) (9.87) (3.16) (2.59) 65 94 61 64 92 0.28 0.35 0.28 0.45 0.45 0.91 0.87 0.91 0.80 0.81 Robust standard errors recorded in parentheses. The symbols *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. 38 (8) 0.097* (0.055) 0.035 (0.068) 0.98* (0.51) 0.32 (0.37) -11.10 (11.79) 59 0.52 0.76 (9) 0.006 (0.010) 0.112*** (0.035) 0.02 (0.12) -0.04 (0.17) 8.38*** (2.98) 64 0.50 0.76 (10) 0.017 (0.011) 0.072*** (0.016) 0.11 (0.14) 0.10 (0.12) 4.82* (2.57) 92 0.48 0.79 (11) 0.096* (0.054) 0.039 (0.058) 0.97* (0.50) 0.31 (0.36) -10.82 (11.49) 59 0.60 0.69 Table 4: The impact of export share and import share on per capita GDP, 1990 Total trade share Net trade share F test Export share Import share (1) 0.004 (0.007) 0.147*** (0.044) 19.37 (0.000) 0.151*** (0.037) -0.143*** (0.048) (2) 0.004 (0.007) 0.150*** (0.035) 25.88 (0.000) 0.154*** (0.03) -0.146*** (0.040) (3) 0.019 (0.016) 0.020 (0.057) 2.88 (0.064) 0.039 (0.046) -0.001 (0.070) (4) 0.024 (0.017) 0.046** (0.016) 6.31 (0.003) 0.076*** (0.021) -0.022 (0.026) Coefficient estimates (5) (6) 0.107** 0.007 (0.041) (0.010) -0.076 0.102*** (0.072) (0.038) 3.35 13.19 (0.042) (0.000) 0.031 0.109*** (0.064) (0.030) 0.183* -0.095** (0.099) (0.046) Robust standard errors recorded in parentheses. The symbols *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. 39 (7) 0.017 (0.012) 0.068*** (0.016) 12.43 (0.000) 0.085*** (0.017) -0.050** (0.022) (8) 0.097* (0.055) 0.035 (0.068) 5.19 (0.009) 0.132*** (0.046) 0.062 (0.115) (9) 0.006 (0.010) 0.112*** (0.035) 13.79 (0.000) 0.117*** (0.029) -0.106** (0.043) (10) 0.017 (0.011) 0.072*** (0.016) 13.40 (0.000) 0.089*** (0.018) -0.055** (0.022) (11) 0.096* (0.054) 0.039 (0.058) 6.11 (0.009) 0.135*** (0.040) 0.056 (0.104) Figure 1: Export Share vs. Import Share, Mean 1970 – 1998 200 180 BHS SGP 160 Export (% of GDP) 140 MDV SUR HKGABW 120 100 BHR MLT GUY ATG PAN SWZ LCA MYS BEL SYC SVK VCT SVN IRL MNG BLZ SLBKNA KWT BRB MUS GAB BWA CZE JAM COG OMN NLD MKD EST NAM GNQ FJI AGO BGR CYP GMB VUT IRQSAU LTU DMA LBRTKM GRD HRV LBY TTO CHEPNG MDA TGO DJI HUN YEM NOR MRT JOR CIV LVA CRI NGA TUN ZMB AUT UKR ISL VNM DNK STP HND KOR SWE TJK SENISR THA FIN WSM ERI LKA TON KEN CAN NZL PHL DOM ZAF KAZ DEU IDN CHL PRY PRT VEN GBR ZWE ECU BLR ROM SLV MWI NICBTN DZA SYR KHM BEN CMR BOL GIN FRA POL RUS MAR AZE URY GHA ITA KGZ EGY ZAR MEX ESP NER IRN AFG CAF SOM SLE GTM CPV COM MLI CHN UZB LAO MDG AUS PER TZA GRC ARM GEO ALB TUR COL TCD NPL HTI PAK LBN JPN UGA MOZ GNB ETH ARG BDIBFA RWA BRA USA SDN IND BGD 80 60 40 20 KIR LSO 0 0 20 40 60 80 100 120 Import (% of GDP) 140 160 180 200 Figure 2: Total Trade Share vs. Net Trade Share, Mean 1970 – 1998 350 BHS SGP Total Trade (% of GDP) 300 MDV SUR ABW 250 200 BHR GUY MLT ATG SWZ KIR 150 HKG LSO GNQ ERI 100 LBN 50 PAN LCA KNAVCT SYC SLB MYS SVK BEL MKD MNGSVN NAMBLZ BRB MUS DMA GMB JAM IRL COG CZE EST BWA JOR DJIGRD CYP FJI NLD VUT BGR KWT WSM YEM MDA LTU HRV OMN PNGAGOSAUGAB TON STP MRTTGO TKM LBR CHE HUN BTN LVA CRI LBY IRQ TTO ISR TUN SOM VNM NOR CIV UKR ZMB NIC HNDAUT NGA SEN ISL CPV LKA DNK TJK KOR THASWE MWI KHM PRT BEN COM KEN PHL DOM SLV FIN AZE PRY NZL KAZ SYR EGY CAN MAR ZWE GNB ARM BLR ROM KGZ GBR GIN CHL DEU IDN BOL TZA ZAF MLI CAF GHA ECU NER CMR VEN ALB DZA POL FRA AFG MOZ BFALAO GEOSLE URY ITA RUS ZAR TCD GTM ESP MDG MEX GRC IRN RWA BDIHTI NPLTUR UZB PAK AUS PER CHN UGA ETH SDN COL JPN BGDUSA ARG BRA IND 0 -120 -110 -100 -90 -80 -70 -60 -50 -40 -30 Net Trade (% of GDP) 40 -20 -10 0 10 20 Figure 3: Correlation Matrix for Specification (1) to (11) (1) total_trade 50 0 NOG A MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IC H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C G T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 T E Tˆ T E Tˆ Ê 1.0000 0.1344 0.5277 1.0000 0.0844 1.0000 0.1683 0.5134 0.2866 1.0000 T E Tˆ Ê T E Tˆ Ê 1.0000 0.1344 0.5277 1.0000 0.0844 1.0000 0.2180 0.5579 0.3060 1.0000 T E Tˆ Ê T E Tˆ 1.0000 -0.2253 0.7473 1.0000 -0.1731 1.0000 Ê -0.1790 0.8983 -0.0890 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR BH S D EU KW G C R O C MRG MB H TM PR U T N C N DBLZ IT BG R JPAN ITA RG O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q ESP N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE IA G MW H TH O MN ILYR A JC C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G AG AR ETH IR BFA H G N C ZMB O BO ZAF ID N MR L PN SAU TG SD SO MO N M Z C MLI AN U C SA AF AG O BR AU TC N ZAR A ER S D BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD ID SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR C AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN C AF U SA AG OG TC N ER D AU ZAR BR SO AN KIR SU R predicted_total_trade 40 Q AT 20 0 BH R NKW O RT ID N VEN SAU MEX D ZA NOG A MN IRKO N R H KG JCG PN C D AN EU C HKTLD EBEL N A ITA S H N FIN R C D NU UAU SA N ZL FR A RESP O M C H PR ISL TH H U BH LR T NAYP MU S MLT S G BR PO SW E C BG BR BMYS ALB TU R LKA R ISR YLM BR A C R LBN IG AR CU O G O LPN IN ZAF D TU N G AB IR MB LPAN TTO FJ I PER MAR EC U EG PH Y L VN M JG AM MD PR G Y BO LA SO M LSO PAK SLV N AM SD N ZW N PL BTN E LAO G N B SM SLB C SLE AF BW G H MR N D T SW C MR SEN TG O BG BD TZA D IW C G O N BLZ G QJAT H MW TI C G PV IN IN VU O RZ MD V ZAR AG O TC G N SYR TM C ER D IV IC UETH C MO MLI O YEM M Z R W BFA A BEN ZMB KEN SG P SU R LSO LBN BH R KW NNO T R ID VEN SAU MEX D ZAP SG N GA O MN IR N KO R C J D AN PN HH EU E N LD AU TKG AU ITA S C H N G R FIN C D N K BEL MLT ESP U N SA ZL FR A R O M MU PR TH H C T ISL A BH S U HB S LL G BR SW PO E C YP BG BR R ALB LKA ISR TU MYS U R R YAB BR ALN C R IR D O C AR PAN M O L G IN D ZAF TU N G MB IG G TTO FJ IEC SU R MAR PN PER U EG PH YY L JPL VN AM MD PR G BO LM SO M SLV NB AM N BTN SD ZW N E WJMO SLB GQ N LAO ETH C AF B SLE W AO G SW MR H N A Z T TG SEN O C G NSM BD BG IPAK BLZ D CMR GV C O PV R VU T MW H G TI IN ID ZAR AG MD N IC TC N ER D C TM IVO SYR CTZA ZO MLI YEM U M G A BFA R W A ZMB BEN KEN AR E BH R KW T N OR ID N VEN SAU MEX DOIR ZA Q N A MN IRGN KO R C AN JHU PN DRG EU BYU NN LDTC H E AU AU S ITA CC N FIN G D U SA ZL ESP FR AM R O CN N TH H ISL C LBG A H L PR U BH N TCS SBR B G BR SW PO E LC CKYP MUBEL RZE TU U MYS LKA R R Y ISR BR A G U Y C R IIIR AR C D O GD O L PAN ZAF IN MMR TU G AB L TTO SU R FJ DN IJMB JAM PN PER MAR G EC UG EG PH Y LEM VN M MD PR G Y BO L MN SO G M PAK SLV SD ZW N BTN N PL ETH LAO G SLB N C AF SLE MR T H A N C MR SEN TG C TZA O BD G G BG N ITI BLZ Q D G MW IN H JH IB O AG ZAR O TC N ER C D IV SYR N G IC TM MO MLI YEM U G Z A C ORDMO BFA R LBR W A ZMB BEN KEN SG P H KG predicted_net_trade MLT -20 0 200 400 600 0 100 200 Ê 300 (2) total_trade 50 0 NOG A MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DTBLZ ESP PN JE C G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D PR ER TH C T MB R W A LKA R G O H M A YAAM SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR BH S D EU KW G C R O C MRG MB H TM PR U T N C N DBLZ IT BG R JPAN ITA RG O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q ESP N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE IA G MW H TH O MN ILYR A JC C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G AG AR ETH IR BFA H G N C ZMB O BO ZAF ID N MR L PN SAU TG SD SO MO N M Z C MLI AN U C SA AF AG O BR AU TC N ZAR A ER S D BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D J OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NN O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR SLB MAR BEN TU EC R U BD ID SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB C MR YEM Y G MD MEX G D N GA ZMB BFA ETH IR AR C N H O G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN CPR AF U SA AG OG TC N ER D AU ZAR BR SZA AN KIR SU R predicted_total_trade Q AT 20 0 KW T BH R VEN NSAU OR MEX MN D ZA IRID NO N H KG N GNALD BELMLT D EU S ISL C D AN N JAU PN G N KO ISR ZL BH S C H LK U RBR YRE SW U SA AR ESP G LBN FR A ITA AU T H BR A FIN G R C C O H LC ULM NE PO BG R G AB TU R PER TTO RD O M O C YP PR TIR LMYS EC UL MU JPAN OSR ZAF BR B CLTU R IN CBO H N MAR PH J AM ALB PR Y EG Y FJ CBW O GAI N IC SLV SYR ABLZ H MR N G D T MB IN D LKA C MR C PV SEN C PAK AF CTH IV G SO H M A G TM MD ZW G E G ZMB NG QSW Z MD V SLE NSM AM SD VN N TG O H BEN TI PN ZAR AG LSO C OYEM IN M G N B TZA W LAO MO ZMO BG MLI DG SLB TC D KEN N N MW PL BTN UETH G AER BFA BD I IVU T R W A SG P LBN SU R LSO KWRT BH VEN SAU N OR MEX O MN D SG ZAP IRID N HNKG N G A BEL N LD EU AU S MLT ISL C D AN N ISR G KO N JD BR PN ZL R BH S C H LK U R Y SW E U SA ESP AR FR A AU ITA TG H BR A FIN GBG RC CR H C U OE N LTTO PO LRAB G TU R PER R O M SU DC CO YP M TU N MYS PR T PAN IR J O R PR MU S EC U BR ZAF BL R I BO L C H N MAR PH L J AM ALB Y EG Y FJ IT O G N SLV TH BLZ AC G MR H MB BW N D ASYR LKA IN D CQIC PV C MR SEN C PAK AF C IV SO M G H A G TM G NSM ZMBZW MD G E N AM SLE SD VN N M TG O H BEN PN TI G SW ZAR Z AG O V M G IN GCTZA N B MD WMO ZO YEM LAO SLB MLI BG D TC ETH D VU T KEN N N MW PL ER I BTN U BFA BD RI G WAA AR E KW T BH R VEN SAU N OR MEX IR Q OZA MN D IR N ID N NG A D EU BEL AU S ISL CAR AN DNS NLD K JU PN ZL KO G BR RBH ISR CN H B LITA R Y SW E U SA G ESP FR CR AJZE AU TC H E D BR A FIN G CITTO CR L CN O LR H UM N PO BG LM RG G AB TU PER YU O SU RO D CL YP MU SBR B TU MYS PAN TIIR EC JULPR ON RAM ZAF MN G C R BO L C H N MAR PH J PR Y EG Y FJ IMB CCG O G U N Y IC SLV TH SYR A BLZ MR T G N D IN D LKA MR MMR SEN C AF PAK C IV SO G M H A LBR G TM MD ZMB ZW G G NH EB Q SLE SD VN N M TG PN BEN G HN TI AG ZAR O G IN C O MO G TZA YEM LAO Z MLI SLB BG D TC ETH D KEN NMO ER MW N BTN U G APL BFA IIA RBD W SG P H KG MLT predicted_net_trade -20 0 200 400 600 0 100 200 300 (3) total_trade 50 0 NOG A MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR BH S D EU KW G C R O C MRG MB G TM PR U T NBLZ H C N D IT BG R JPAN ITA RESP O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G M N Q N O FIN PO L SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE I G MW H TH O A MN ILYR A JC C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G AG AR ETH IR BFA H G N C ZMB O BO ZAF ID N MR L PN SAU TG SD SO MO N M Z C MLI AN U C SA AF AG O BR AU TC N ZAR A ER S D 0 ID N PN G SW C D H N LK FIN N MR ZL O R MYS TH A MU PAN S BR C IR ESP U C ITA O A VEN R N AN L Y FR ZAF G SEN BR A N LD SD U N AU SA EC N PL K ISL EN A S U U JE T AM JC MEX PN H KO N C R H E BG U BO G D A LU FJ IATG IN D D PH ZA H LR N N IC ETH PR C M R T IR ID T L LC N D TU LKA R R D G PAK SLV R CG TC G TM D MA KN AA SLB MD DNOGBMVU G T C PV W SM KIR SU R SG P BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB D O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD I SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR C AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN C AF U SA AG OG TC N ER D AU ZAR BR SO AN predicted_total_trade SG P ID N PN G SW C D H N E LK FIN N C MYS N MR O R MU TH A PAN S ESP IR C BR ITA C U AN N O R A Y VEN SEN G FR BR N ZAF LD A N SD KEN JPL U AU AM AU ISL SA S TL U KO MEX C J PN H H R E NL U BG G FJ BO ATG D IEC L N IC PH IN H L D U ZA LC PR C ETH MR R A T IZL T H N D DG LKA TU R SLV G PAK R CD DG MA D TM KN ARTC SLB MD G DO M GN BT VU C PV W SM ID N PN G C SW H L E D NLD K HES FIN C N MR N ZL MYS O R TH PAN A BR C IR VEN C AN U A O ESP R ITA Y ZAF SEN G FR BR AU U SD SA KEN EC S N ISL N U PL JA AM C MEX JH PN KO N RD BO U G L BG FJ AL ITM D IN ZA PH D N LIC U N MR ETH T PR C RSLV TC IN IRAU L TC MU H N D TU LKA R PAK R TC D G SLB MD DGN OB M G SG P KN A -50 predicted_net_trade -100 KIR KIR KIR -150 0 200 400 600 0 100 200 300 41 1.0000 (4) total_trade 50 N GA 0 O MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR S D EU KW G C R O C MRBH G MB G H TM PR U T N C N DBLZ IT BG R JPAN ITA RESP O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE I G MW H TH O A MN ILYR A J C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G AG AR C ETH IR BFA H G N C ZMB O BO ZAF ID N MR L PN SAU TG SD SO MO N M Z C MLI AN U C SA AF AG O BR AU TC N ZAR A ER S D KIR SU R BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB D O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD I SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB C MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN C AF U SA AG OG TC N ER D AU ZAR BR SO AN predicted_total_trade 50 BH R 0 C O IV CSLE OC G LSAU AB N G A G ATG ZAR TTO C H LK EC U N ZL SW VEN SYR MYS AR C AU U FIN ID PO G MR D R N S N N Y O LR R LC ZAF C AN ISL IR L FJ PAN I SA BR PER D G ITA N A ZA BR ER TH MEX FR A N JE AM LD BO ESP H H N AU L U N T SD IN U SA D N SEN U JC PN KO A D MU MA PH K EN C G L R R ISYC D O C PN M H G E YP N IR G PL TM N ETH H N BR DA KN B A MW IC TU PR R TU Y N BG MAR D ZMB PAK BEN LKA MR BLZ T MD G H G A T G H R TI C TG C AF MLI TZA MLT EG SLV YG O TC D BFA G NISR MB BD IB SLB N ICJTO R VU C PV BH R CIV GAB C C OOG LTTO SAU SLE ZAR C H LSYR ATG U N GA N ZL SW MYS SG E P VEN LC AU FIN C A ID AR D U N PO MR N S N R O L K Y R C ISL AN IR ZAF FJ PAN IEC N TH G ER PER BR ITA BR A D ZA A JPL MEX FR AM N A ESP BO C AU H H LLD TL NG SD SEN IN U SA D DG MA U MU G KO J A S PN R RMD DG C KEN PH R L IU D PN C C O YP H M G E N SYC TM N KN AZMB ETH H BR N B MW ID PR TU TU Y N R BG MAR D LKA PAK MR BEN BLZ T PR HIR G T A TG G H R O TI C C AF ISR MLI MLT SLV EG TC DY BFA GTZA N B G BD I MB J SLB OIC R N VU T C PV W SM SG P W SM -50 -100 KIR 0 BH R C O G IV GC AB O N G ALLU SAU SLE ZAR TTO C H EC N ZL VEN SW MYS SYR E SG P AU AR ID C MR U N S PO G FIN R O Y LR C ZAF AN ISL LK TC MU FJ PAN ITM N BR D ER PER TH ZA A G A ITA BR MEX FR JA AM NN LD C BO L N ESP H U NIIR AU U SD SA IN N SEN D U JH G PN KO A RD KEN PH LPR RD PN D G O H E SBR B KN A IR N N PL G SYC ETH H N D MW ICM PR TU Y TU R ZMB MAR BG MR BEN PAK T LKA BLZ MD G H G A TCOC YP H TI TG R C AF ISR MLI TZA EG YG SLV TC D BFA GN BMB G BD OR SLB NJIIC KIR 200 400 MLT predicted_net_trade KIR 600 0 100 200 T E Tˆ Ê T E Tˆ 1.0000 -0.1820 0.7105 1.0000 -0.1185 1.0000 -0.1445 0.9028 -0.0775 1.0000 T E Tˆ Ê 1.0000 -0.3336 0.6590 1.0000 -0.1956 1.0000 -0.1748 0.8000 -0.1799 1.0000 T E Tˆ Ê 1.0000 0.0752 0.7719 1.0000 0.0959 1.0000 0.2711 0.7534 0.1697 Ê 300 (5) total_trade 50 0 NOG A MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR BH S D EU KW G C R O C MRG MB G H TM PR U T N C N DBLZ IT BG R JPAN ITA RESP O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE IA G MW H TH O MN ILYR A JC C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G A AR ETH IR BFA H G N C ZMB O G BO ZAF ID N MR L PN SAU G T SD SO MO N M Z C MLI AN UAU C SA AG BR TC N ZAR AAF ER SD O KIR SU R BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD ID SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR C AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN CNAF U SA AG OG TC ER D AU ZAR BR SO AN predicted_total_trade 50 BH R 0 BH R C CCIV OOG LGAB ZMB N GA ATG SLE JBO AM PAN TTO N ZL MR C MR TLN N KEN SYC ER SD ID N VEN GGR NRTC B C AF SW E MD LC SEN PER G A N O R BR D ZA A TG D MEX O G H ITA SAU A ESP FR A SLB DG N LD D KO IN O AU M D TSYR JD O PH U BLZ SA L TU BEN TM BFA MA MAR G LKA R C PR PAK C T H E TU RR KN AMLI SLV BG D MLT EG VU T Y C PV W SM IVO G C OCG LC AB N G ZMB AAM ATG SLE TTO JESYC PAN N ZL C MR MR T BO N KEN ER L SD ID VEN N C G SW SYR AF BG PER MD SEN N G O R LC BR D A ZA MEX TC D TG OD G ID TA H SAU A ESP FR R LD IN KO O AU M R T U SA PH LN BLZ JMA OAA R G BEN N BFA MLI MAR D G R LKA PAK PR CTU HTSLB EKN TU R BG SLV DTM EG Y MLT C PVVU T W SM CC OO GL C IV GG AB N ZMB APAN SLE J AM TTO SYC N ZL MR C MR T N BO ER KEN L SD ID VEN N N C AF G SW SYR N B E MD PER SEN G O R BR D ZA A TC MEX D SAU G H A ITA ESP FR ARTCNOAU SLB LDTC H E IN D KO D O M RTG U SA PH JR LPR BLZ O BEN TU G N TM MLI BFA MAR LKA R PAK TU EGBG Y D SLV BH R KN A MLT predicted_net_trade -50 KIR -100 0 KIR 200 400 KIR 600 0 100 200 T E Tˆ Ê 300 (6) total_trade 50 0 NOG A MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT C O TZA G NM B C PV D MA SO STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR BH S D EU KW G C R O C MRG MB TM PR U T NBLZ H C N D IT BG R JPAN ITA RG O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q ESP N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE IA G MW H TH O MN ILYR A JC C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G AG AR ETH IR BFA H G N C ZMB O BO ZAF ID N MR L PN SAU TG SD SO MO N M Z C MLI AN U C SA AF AG O BR AU TC N ZAR A ER S D KIR SU R BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD ID SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR C AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN C AF U SA AG OG TC N ER D AU ZAR BR SO AN predicted_total_trade 20 0 ID N NLO R VEN CO C H L MEX N ZL C AU S FIN DAN NN DITA ZA MYS SW EKA LD KO TH AU TAM PN G ISL U H EC R N Y URH MU S JCIR PN JN ESP N FR ZAF A BR G A BR IR LPAN U SA H U C C C MR H DIE LNLR IN D FJ I PR T LKA SD N SEN GBO R C PH T TU KEN RMR UETH N G A IC PAK DPL OGMNSLB MD BG DTM GSLV TC D GN B -20 IDSG NP VEN CNOOLR CD H LK MEX NMYS ZL C AN AU S FIN N D ZA SW E ITA N LD TH KO AU A R TL PN PAN G ISL MU C S U EC H R NY U JG AM J PN ESP IR N FR ZAF A BR BR IR A U SA H C H C R IU C HR N DEN BO FJ IN D IMR PR T LKA SD SEN NL G C PH MR TL KEN TU R U ND ETH G PL N IC SLV PAK OA MD BG DM SLBTC GDG TM GN B SG P VU T C PV W SM ID N VEN CNOO LR CN HZL L DNK CMEX AN AU S FIN D ZA MYS SW E ITA NAU LDT TH KO A RJAAM PN G PAN ISL C EC U N R U Y JH PN IR N ESP ZAF FR BR A G BR U H URD NIIR L C MU HES C CSA MR H N BO L IN DLKA FJ IR TC SD N G PH LPR MR TSEN TU R ETH UKEN G NA IC PAK DN OPL M SLV MD G BG SLB GDTM TC D GN B SG P predicted_net_trade VU T C PV W SM -40 0 200 400 600 0 100 200 300 42 T E Tˆ Ê 1.0000 (7) total_trade 50 N GA 0 O MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR S D EU KW G C R O C MRBH G MB G H TM PR U T N C N DBLZ IT BG R JPAN ITA RESP O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE I G MW H TH O A MN ILYR A J C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G AG AR C ETH IR BFA H G N C ZMB O BO ZAF ID N MR L PN SAU TG SD SO MO N M Z C MLI AN U C SA AF AG O BR AU TC N ZAR A ER S D BH R 20 0 -20 KIR SU R BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB D O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD I SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB C MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN C AF U SA AG OG TC N ER D AU ZAR BR SO AN NLSAU GO AG C C IV C ID ON R G AB NO TTO VEN SLE D ZA MEX N C ZL H LK FIN SW D N E EC U AU ZAR PO ISL S LN IR LMYS U ITA RBR Y C G AN AR G FR A LD FJ I H U ZAF ESP BR JC C PN MR A SYR TH U SA H N AU TAM IR N PER KO R JNA IN D C H E MU S BO N ER L BR BPAN SD N C R SEN PH L CIG YP D O M G TM UTU A PN KEN N PL R LKA PR T BG G R D H C TU N IY DBLZ N ETH MW MAR MR T MLT PAK MD G H TI G H A ZMB BEN C AF TG O ISR MLI TZA EG Y TC SLV BFA G NDB BD I G MB SLB N ICJTO R VU C PV BH R SAU N GA OLG C CC NIV O RAB ID NG VEN SLE D ZA MEX N C ZL H LTTO SW FIN D N ELN K U AU ZAR ISL IR PO S L ITA U R Y G C BR AN AR FJ FR N IEC LD A H U MYS ZAF SG P C J BR PN MR A TH A SYR U AU SA H TNG IR N KO PER R JESP AM PAN IN D C H MU S N ER BR SD N C R ILBE SEN PH CBO YP L D O M TM U PN G A KEN N PL TU R PR PR LKA Y T G BG TU H RBLZ N D C NG ETH MW ID MAR MR MD G H TI GPAK HO AT ZMB BEN MLT C TG AF ISR MLI TZA EG TC SLV DY G N BFA B BDG I MB N VU T J SLB OIC R SG P BH R SAU NC GNO AL R C O G GVEN AB ID NIVO TTO SLE D ZA MEX CN H ZL LU SW FIN E D NLD EC ZAR AU S PO ISL LU IR LK U R ITA Y C AN G BR AR G FJ FR H NN ZAF MYS ESP BR C JH MR A PN TH SYR A U C SA N AU TC MU IR N PER KO RIJAAM PAN IN D H E SBR B N BO ER L SD N R I C YP SEN PH LCMTM D OR G PN U G G A KEN N PL TU PR Y LKA PR TC BG TU G NRD ETH MW ID MR MAR T PAK MD G HH BLZ G H ATI ZMB BEN CMLI AF TG O TZA EG Y ISR TC D BFA G N B SLV BD G I MB SLB NJIC OR SG P MLT predicted_net_trade C PV W SM -40 predicted_total_trade W SM 0 200 400 600 0 100 200 T E Tˆ Ê T E Tˆ 1.0000 0.0114 0.7075 1.0000 0.0111 1.0000 0.0257 0.8242 0.0246 1.0000 T E Tˆ Ê 1.0000 -0.0641 0.6312 1.0000 -0.0595 1.0000 0.1637 0.6058 0.0271 1.0000 T E Tˆ Ê 1.0000 0.0752 0.7719 1.0000 0.0959 1.0000 0.2753 0.7712 0.1787 Ê 300 (8) total_trade 50 0 NOG A MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR BH S D EU KW G C R O C MRG MB G H TM PR U T N C N DBLZ IT BG R JPAN ITA RESP O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE IA G MW H TH O MN ILYR A JC C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G A AR ETH IR BFA H G N C ZMB O G BO ZAF ID N MR L PN SAU G T SD SO MO N M Z C MLI AN UAU C SA AG BR TC N ZAR AAF ER SD O BH R 20 0 -20 KIR SU R BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD ID SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR C AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN CNAF U SA AG OG TC ER D AU ZAR BR SO AN predicted_total_trade BH R N OR ID N CMEX OC NLSAU GO AG IV C DVEN ZA G NTTO ZLAB ITA N JETAM LDPAN ESP SW KO R FR AU SLE BR A U SA MD BO MR G LA T PER G RAF C HE SD N C G MR N BZMB C IN D O M PR N SEN ER TC KEN LKA D PH SYR LTSLB TU N TU G H R A TG O MLT G PAK MAR BFA BG DTMY BLZ MLI SLV EG BEN J OR C PVVU T N OR ID NL C IV O MEX N GA C VEN C ZA OG D SAU G AB TTO N ZL ITA JESP AM N LD SW E KO R FR AU T SLE BR PAN A U SA MD MR BO G TLA PER R C C H E SD N G NZMB BG CT MR C AF D IN O M D PR N SEN ER TC LKA KEN D SLBMLT PH TU NLR SYR TG G TU H OTM A G PAK MAR BFA BLZ BG MLI SLV EG YD BEN J OR VU T C PV W SM BH R N OR IDCNO L N MEX G A C IV C O G DVEN ZA SAU GNAB TTO ZLITA NAU LDT SW ESP E KO RJAAM FR SLE BR A PAN U SA MR BO MD L TD GN PER CHE SD ZMB N CER MR G B AF IN D OG MR TC NC SEN TC KEN D LKA SLB PH SYR LPR TU N O G TU H R A TG G PAK MAR BFA BG DTM MLI SLV EG YBLZ BEN J OR MLT predicted_net_trade W SM -40 0 200 400 600 0 100 200 T E Tˆ Ê 300 (9) total_trade 50 0 NOG A MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT C O TZA G NM B C PV D MA SO STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR BH S D EU KW G C R O C MRG MB TM PR U T NBLZ H C N D IT BG R JPAN ITA RG O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q ESP N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE IA G MW H TH O MN ILYR A JC C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G AG AR ETH IR BFA H G N C ZMB O BO ZAF ID N MR L PN SAU TG SD SO MO N M Z C MLI AN U C SA AF AG O BR AU TC N ZAR A ER S D KIR SU R BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD ID SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR C AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN C AF U SA AG OG TC N ER D AU ZAR BR SO AN predicted_total_trade 20 0 -20 VEN NLN OLR C ID O C H MEX N ZL AU S C D AN N SW EKLD D ZA ISL N UITA FIN R Y KO G BR BR A N PAN EC AU UR T MYS J IR PN ESP FR A U SA JN ZAF IR L S PN MU HN C U H E TH L AAM CBO C IG NR D C HMR SEN FJ PH TU RMR GD R PR CLN TT SD IN N KEN IC I LKA UETH G DPL AO M SLV N PAK G BG MD DTM G TC D SLB GN B VU T C PV SG P SG P VEN ID NOL CNO H LR MEX NCD ZL AU S C AN N SW EK D ZA ISL N LD FIN MYS U R KO R BR IR ITA PAN N AU EC TY U ESP JBR PN FR AA U SA JG AM ZAF IR MU PN G C H H N TH BO LEL C C R IU H NAS D H N SEN FJ IMR PH MR TU R T PR R C SD KEN INT NL D N IC G LKA U D ETH G OA M SLV NBG PAK PL G TM MD G D SLBTC D GN B SG P VEN O ID N LR CC HO L NSW ZL AU S CMEX K E DNNLD DAN ZA ISL U FIN MYS RPAN Y KO RJAAMAU T G BR BR IR A N ITA U JEC PN ESP FR U SA ZAF PN G URD NIIR L C MU HES BO TH L A CH MR HLCN C NFJ IR TC PH MR TU G PR SD KEN IN NTSEN D NR IC ETH UG DNLKA A OGMTM PAK MDSLB BG G PL D SLV TC D GN B predicted_net_trade VU T C PV W SM W SM -40 0 200 400 600 0 100 200 300 43 T E Tˆ Ê 1.0000 (10) total_trade 50 N GA 0 O MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR S D EU KW G C R O C MRBH G MB G H TM PR U T N C N DBLZ IT BG R JPAN ITA RESP O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE I G MW H TH O A MN ILYR A J C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G AG AR C ETH IR BFA H G N C ZMB O BO ZAF ID N MR L PN SAU TG SD SO MO N M Z C MLI AN U C SA AF AG O BR AU TC N ZAR A ER S D BH R 20 0 -20 KIR SU R BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB D O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD I SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB C MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN C AF U SA AG OG TC N ER D AU ZAR BR SO AN SAU N G O AG CO C LC IV N G O AB R VEN TTO C H LK MEX SLE N SW D ZL N E U ID RBR ISL N Y D ZA AR G G AU EC S U FIN IR LMYS ZAR PO LN LD FR C ITA AN A BR A ESP SYR FJ I H U JPER U CIR PN SA ZAF MR N AU KO RH JNTAM PAN BO LM CG H N TH C IN D N SEN ER MU PH TM L BR B S SD N O C RYAIE YP R KEN UTU G A PR H TU NC DG N N PL PN BG G R D PR C MR T MLT ETH MAR LKA PAK MW ZMB ITBLZ H G TI H A BEN ISR MD G C AF TG O MLI EG Y TZA SLV TC BFA G NDB BD I N GSLB MB IC VUJTO R BH R SAU CIV OG GAB N GA C CN OO LTTO R VEN C H LK MEX SW N SLE D ZL N ELG ID ISL U N R Y D ZA GAU BR AR S U FIN IR ZAR N PO LD LP C FR ITA AN SG A BR ESP IEC H U U PER C J MYS SA PN ZAF MR IR N AU TNSYR KO RA JFJ AM PAN BO TH C C A H H IN D N MU SEN S PH G BR TM L SD D C O R N M ILBEN CER YP TU R KEN U PR G A Y TU H N NG D N PN PL G BG PR R D T C MR BLZ MAR ETH MLT LKA ZMB PAK MW I G H HO TI AT BEN ISR MD G C AF TG MLI EG Y TZA SLV TC G N BFA B D BDG I MB N J SLB OIC R VU T C PV SG P C PV W SM -40 predicted_total_trade BH R SAU C N O G G AO C IV L R TTO GVEN AB NO CNEC HZL LUBR MEX SLE SW E DIR NLD ID U N ISL R Y D ZA AR G G AU S FIN LK ZAR PO LU C AN FR ITA ANN BR A ESP SYR FJ H UBO ZAF C PER SA JMR PN MYS IR N AU TC H E KO RIJ AM PAN L CER H TH N AG IN DO NSD SEN MU SBR B PH LCMTM D N R I C YP TU R PR UKEN G Y AH TU NRD PN G N PL BG G PR TC MR T BLZ ETH MAR LKA ZMB MW PAK ID G H H A TI BEN ISR G CMD AF TG O MLI EG Y SLV TZA TC D BFA GN B BD IIC NJG OMB R SLB SG P MLT predicted_net_trade W SM 0 200 400 600 0 100 200 T E Tˆ Ê T E Tˆ 1.0000 0.0114 0.7075 1.0000 0.0111 1.0000 0.0407 0.8295 0.0376 1.0000 T E Tˆ Ê 1.0000 -0.0641 0.6312 1.0000 -0.0595 1.0000 0.2058 0.6414 0.0752 Ê 300 (11) total_trade 50 0 NOG A MN VEN SYR TTO G AB MD HV KG EC U C O G N O R U PO R AG KW Y LN T AR D ZAF G ZA D N K IR LMYS C O L BH LD SPAN C H SAU LO H D MR H N EU U N BH R BR ID A ISL AU N TBLZ BEL ATG JC PN ZAR C IV PER ZW FR C ITA N VN SW AN ZL C A E M BW H E A SLE BR B U AU MEX SA KO S R IR FIN N IN BO G D BR L TU G H TM R N DA ESP PN JE C AM G YP G MR IN T BEN BG R FJ I SAZ D MAR MW PH KEN O ISR M L IG H ALB TI SEN C R ISYC SD N TU N MU BG G PAK N R D ER TH C T MB R W A LKA R G O H PR M A Y SW UETH G BTN LC N EG PL Y MD G TG O VC T LAO TC D C AF SLV YEM MLT BFA ZMB MLI ABW VU BD IM GN RAM DT CTZA OPV G N C D MA SO MB STP SLB NSM IC MO Z TO N W G NKN QJ OAR SG P SU R net_trade -50 LBN -100 LSOKIR MLT MLT H KG H KG SG P SG P KN A KN A BR B BH R BEL MU S C H SYC E CTLD YP AUN IR L KIR DTTO NJK SLV TG AM O ISR BH S D EU KW G C R O C MRG MB G H TM PR U T N C N DBLZ IT BG R JPAN ITA RESP O M G BR FR A LKA TU N BG D SYR N FJ IC MYS IO R N H PL KO TI SEN RE D O G N Q N O FIN PO LM SW BTN U G R Y B R W PAK ZW PH A ISL E TU BEN EC MAR R U SLB BD SLE IA G MW H TH O MN ILYR A JC C PN EG O L Y IN LAO D N K G EN ZL IN U G VEN A PER VN C C H IV M C MR PR G YEM AB MD D MEX TZA ZA N G G A AR ETH IR BFA H G N C ZMB O G BO ZAF ID N MR L PN SAU G T SD SO MO N M Z C MLI AN UAU C SA AG BR TC N ZAR AAF ER SD O 20 0 -20 KIR SU R BR BHBR BEL MU S SYC CHE CAU YP T N DIRLD NLTTO K SLV TG J O AM ISR BH ST D EU KW M G R C G MB PR G H TM T N C H BLZ R N D JN OCIC ROR BG RIIU ITA PAN O M G BR FR TU LKA N BG D FJ MYS N SEN H PL KO TI RA G NSLB O M ESP N O R FIN PO LY BTN SW GQN B U R R PAK PH W ZW ISL A L ELSYR MAR BEN TU EC R U BD ID SLE G TH MW H A IIV O MN EG J Y C PN O LAO KEN G IN N IN D ZL U G A VEN VN PER C SU H M RAB MR YEM PR Y G MD MEX G D ZA N GA ZMB BFA ETH IR C AR C N H G MR PN BO ID SAU ZAF G T L N MO SOTZA ZM SD N MLI C AN CNAF U SA AG OG TC ER D AU ZAR BR SO AN BH R predicted_total_trade BH R N OR VEN CMEX O LSAU C O C TTO DID ZA N G AG N ZL NIV G AB SW E LD N ESP FR ITA AR J AM BR A PER U SA KO BO LZMB SLE MR AU TT PAN C CD MR AF G R O SEN SYR C E MLT SD N MD G N BM TC PH KEN PR D LHTBLZ G TM TU R TU N G H A N ER IN D TGSLB OJ O R MAR SLV PAK BEN BFA MLI BG DLKA EG Y C PV VU T N OR MEX VEN C OOLG SAU CIV TTO D N GA NC ID NZA SW EG AB NZL LD JESP FR AM ITA A BR UPER SA KO BO LA MR SLE TR ZMB AU PAN T MLT CG AF C MR RC C D O M SEN SD NH MD G G N TC B PR KEN PH D T L G TM TU TU N RESYR G H A N ER BLZ IN D TG O J SLB O RMLI MAR SLV LKA PAK BEN BFA BG EG YD C PV VU T W SM BH R N OR MEX VEN C O L SAU C O G C IV TTO N DN G ZA A ID N ZL G AB SW E N LD ESP FR ITA BR AKO PER U SA RJAAM BO L MR SLE T PAN ZMB C CAF MR GMR C AU TC H E D OG SYR SD NSEN G G N TC KEN D PH R T TM TU TU R N G H ALPBTG NMD ER IN D SLB JBLZ OR O MAR LKA PAK BEN BFA MLI BG EG Y D SLV MLT predicted_net_trade W SM -40 0 200 400 600 0 100 200 300 44 T E Tˆ Ê 1.0000 Figure 4: Estimated Total Trade Share Coefficient vs. Estimated Net Trade Share Coefficient for Specification (1) to (11) (95% Confidence Interval Ellipse) (2) (1) .02 .02 .015 .01 Estimated ptt1 Estimated ptt1 .015 .005 0 .01 .005 0 -.005 -.005 0 .05 .1 Estimated pnt1 .15 .2 .05 .1 Estimated pnt1 .15 .2 0 .05 .1 Estimated pnt1 .15 .2 0 .05 .1 Estimated pnt1 .15 .2 (4) Estimated ptt1 (3) Estimated ptt1 0 .02 .015 .01 .005 .02 .015 .01 .005 0 0 -.005 0 .05 .1 Estimated pnt1 .15 -.005 .2 (6) (5) Estimated ptt1 Estimated ptt1 .02 .015 .01 .005 0 -.005 .02 .015 .01 .005 0 -.005 0 .05 Estimated pnt1 .1 .15 .2 45 (8) .02 Estimated ptt1 Estimated ptt1 (7) .015 .01 .005 0 .02 .015 .01 .005 0 -.005 -.005 0 .05 .1 Estimated pnt1 .15 .2 0 .05 .1 Estimated pnt1 .15 .2 (10) (9) Estimated ptt1 Estimated ptt1 .02 .015 .01 .005 0 .02 .015 .01 .005 -.005 0 -.005 0 .05 .1 Estimated pnt1 .15 .2 .1 Estimated pnt1 .15 .2 0 Estimated ptt1 (11) .02 .015 .01 .005 0 -.005 0 .05 46 .05 .1 Estimated pnt1 .15 .2