WPS3889 Comparative Advantage, Demand for External Finance, and Financial Development* Quy-Toan Do Andrei A. Levchenko The World Bank International Monetary Fund Abstract The differences in the levels of financial development between advanced and developing countries are large and persistent. Theoretical and empirical literature has argued that these differences are the source of comparative advantage and could therefore shape trade patterns. This paper points out the reverse link: financial development is influenced by comparative advantage. We illustrate this idea using a model in which a country's financial development is an equilibrium outcome of the economy's productive structure: financial systems are more developed in countries with large financially intensive sectors. After trade opening demand for external finance, and therefore financial development, are higher in a country that specializes in financially intensive goods. By contrast, financial development is lower in countries that primarily export goods which don't rely on external finance. We demonstrate this effect empirically using data on financial development and export patterns in a panel of 96 countries over the period 1970-99. Using trade data, we construct a summary measure of a country's external finance need of exports, and relate it to the level of financial development. In order to overcome the simultaneity problem, we adopt a strategy in the spirit of Frankel and Romer (1999). We exploit sector-level bilateral trade data to construct, for each country and time period, a predicted value of external finance need of exports based on the estimated effect of geography variables on trade volumes across sectors. Our results indicate that financial development is an equilibrium outcome that depends strongly on a country's trade pattern. JEL Classification Codes: F02, F14, O16, O19. Keywords: trade patterns, demand for external finance, financial development, gravity model. World Bank Policy Research Working Paper 3889, April 2006 The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Policy Research Working Papers are available online at http://econ.worldbank.org. We are grateful to Daron Acemoglu, Abhijit Banerjee, Thorsten Beck, Olivier Blanchard, Simon Johnson, Aart Kraay, Roberto Rigobon, Alan Winters and workshop participants at MIT and the World Bank for helpful comments. The views expressed in this paper are those of the authors and should not be attributed to the International Monetary Fund, the World Bank, their Executive Boards, or their respective managements. Correspondence: International Monetary Fund, 700 19th St. NW, Washington, DC 20431. Email: qdo@worldbank.org; alevchenko@imf.org. 1 Introduction A quick glance at levels of ...nancial development across countries reveals large di¤erences. Figure 1 plots for developing and advanced countries the ratios of private credit to GDP and trade openness to GDP starting in 1970. The average share of private credit to GDP is more or less three times higher in advanced countries than in developing ones throughout the period. On the other hand, trade volume as a share of GDP grew faster in developing countries, which have now surpassed the advanced ones. What explains persistent ...nancial underdevelopment? Can we say something about the relationship between ...nancial development and trade openness? The literature has often emphasized the idea that ...nancial development is an endowment. La Porta, Lopez-de-Silanes, Shleifer and Vishny (1998) provide empirical evidence that a country's legal origin is a strong and arguably exogenous determinant of a country's ...nancial development. When it comes to institutions more broadly, Acemoglu, Johnson and Robinson (2001) document that the quality of institutions is largely determined by settler mortality rates during the colonial period. Applying these insights to international trade immediately suggests a pattern of comparative advantage: countries endowed with better ...nancial systems will specialize in goods that rely on external ...nance in production. Indeed, this idea has been formalized theoretically by Kletzer and Bardhan (1987), Baldwin (1989), and Ju and Wei (2005), and has found empirical support in a number of studies (e.g. Beck, 2002, 2003, Becker and Greenberg, 2003, Svaleryd and Vlachos, 2005, and Manova, 2005). The purpose of this paper is to show the reverse link: ...nancial development itself depends on trade patterns. We argue that ...nancial development is endogenous, and that it will be determined in part by demand for external ...nance in each country. Comparative advantage in trade will a¤ect a country's production pattern, and in turn its demand for external ...nance. Countries specializing in ...nancially dependent goods will have high demand for external ...nance and thus a high level of ...nancial intermediation. On the other hand, the ...nancial system will be less developed in countries that specialize in goods not requiring external ...nance. In this paper, we ...rst illustrate this point using a very simple model in which goods di¤er in their reliance on external ...nance. Comparative advantage implies that after trade opening, the ...nancially intensive sector expands in one coun- try and disappears in the other. This change in production patterns in turn has implications for equilibrium ...nancial development in the trading countries. We then demonstrate this e¤ect empirically. For a panel of 96 countries and 30 years, we use industry-level export data and information on each industry's reliance on external ...nance from Rajan and Zingales (1998) to build a measure of the external ...nance need of exports. This measure, constructed following the methodology of Almeida and Wolfenzon (2005), summarizes the demand 2 for external ...nance that comes from a country's export pattern. We then use a comprehensive dataset on ...nancial development ...rst introduced by Beck, Demirguc-Kunt, and Levine (2000) to show that a country's ...nancial development is strongly and robustly a¤ected by the external ...nance need of its exports. The e¤ect we ...nd is economically signi...cant. Our most conservative coe˘ cient estimates imply that moving from the 25th to the 75th percentile in the distribution of external ...nance need of exports is associated with an increase in ...nancial development of about 0.33 standard deviations, or a 12 percentage point increase in private credit to GDP. A key feature of this paper is the way it addresses the simultaneity problem arising in this exercise. We require an instrument for a country's export pattern. In order to construct it, we expand the geography-based methodology of Frankel and Romer (1999). These authors use the gravity model to predict bilateral trade volumes between each pair of countries based on a set of geographical variables, such as bilateral distance, common border, area, and population. Summing up across trading partners then yields, for each country, its "natural openness:"the overall trade to GDP as predicted by its geography. Because we need an instrument for trade patterns rather than total trade volumes, our point of departure is to estimate the Frankel and Romer gravity regressions in each industry. Following their methodology, we can then obtain the predicted trade volume as a share of GDP not just in each country, but also in each sector within each country.1 Doing so allows us to construct each country's predicted external ...nance need of exports, based on its predicted trade shares in each sector. We then use it as an instrument for the actual external ...nance need of exports. As a further extension of the Frankel and Romer approach, we perform this exercise for each ...ve-year period between 1970 and 1999, giving a time dimension to our instrument. The model we use to illustrate our point has two sectors, one of which relies on external ...nance. The size of the ...nancial system, that is, the amount of borrowing and lending that occurs in the economy, is naturally a function of total output in the ...nancially intensive sector. An additional feature of our theoretical setup is that the quality of the ...nancial system is a function of its size. A larger ...nancial sector leads to the greater ease with which entrepreneurs are able to ful...ll the need for external ...nance. This is because when entrepreneurs start ...nancially intensive projects and engage the country's ...nancial system, they add liquidity. They become potential providers of external ...nance for fellow entrepreneurs, reducing the likelihood of ...nancial distress. Each en- trepreneur who invests in the ...nancially intensive sector hence generates a positive spillover by increasing ...nancial depth.2 Opening to trade will a¤ect demand for external ...nance in both trading 1This strategy is adapted from di Giovanni, Levchenko, and Ranciere (2005). 2In modeling the market for external ...nance and the positive e¤ect of ...nancial system size on its quality, we abstract from the informational and enforcement frictions that are often invoked in this context. One can clearly adopt this approach as well, and think of the quality of the ...nancial system in terms of how well it can overcome 3 countries. In particular, the ...nancial system deepens in a country that increases production of the ...nancially dependent good. In the other country the ...nancially dependent sector shrinks, leading to a deterioration in the size and quality of the country's ...nancial system. The assumptions underlying our model ...nd support in empirical studies which relate the size of ...nancial systems to their quality. Levine and Schmukler (2005) ...nd evidence of a causal link between market size and ...nancial depth: when looking at domestic market liquidity in emerging economies, they ...nd that when some ...rms decide to raise ...nance abroad, the remaining domestic ...rms'trading liquidity is adversely a¤ected. Note also that in most empirical studies of ...nancial development, the positive association between size and quality is implicit. The quality of a ...nancial system ­...nancial development ­is often proxied by measures of market size such as ratios of private credit to GDP or stock market capitalization to GDP. This paper is not the ...rst to explore the e¤ect of trade on ...nancial development. Rajan and Zingales (2003) argue that trade opening, especially when combined with openness to capital ows, weakens the incentives of incumbent ...rms to block ...nancial development in order to reduce entry and competition. Furthermore, the relative political power of incumbents may decrease with trade as well. Thus, these authors argue that trade has a bene...cial impact on ...nancial development. Braun and Raddatz (2005) explore the political channel further. They demonstrate that in countries where trade liberalization reduced the power of groups most interested in blocking ...nancial development, the ...nancial system improved. If, on the other hand, trade opening strengthened those groups, external ...nance su¤ered. This paper can be thought of as complementary to Rajan and Zingales (2003) and Braun and Raddatz (2005). While these two studies are about how trade a¤ects the supply of external ...nance, this paper is instead about the demand side. It is also important to note that trade may a¤ect ...nancial development through a variety of other channels. Newbery and Stiglitz (1984) argue that trade, by a¤ecting price elasticity, can potentially increase uncertainty and income volatility. Financial development could then be fostered by increased demand for insurance, though Broner and Ventura (2006) show that the outcome is sensitive to assumptions about the nature of asset market frictions.3 While a Newbery and Stiglitz- type of argument invokes the role of ...nancial markets for insuring risk in consumption, in this paper the ...nancial system plays a role on the production side. Thus, in contrast to the consumption insurance view, our focus in on the di¤erential impact of trade across countries as a function of the these distortions and achieve the e˘ cient level of lending. A positive link between the size of the ...nancial markets and their ability to resolve such frictions has been modeled, for example, by Acemoglu and Zilibotti (1999). 3Rodrik (1998) shows that more open countries have larger governments to help them deal with increased uncer- tainty that is associated with openness. Svaleryd and Vlachos (2002) provide empirical evidence that countries with better developed ...nancial systems are more likely to be open to trade, and argue this is because a better ...nancial system allows a country to better cope with increased uncertainty. Tangentially, these authors also provide some evidence that the ...nancial system improves after trade opening. 4 pattern of comparative advantage. The rest of the paper is organized as follows. Section 2 presents a stylized model of an economy in which the quality of a ...nancial system and its size are jointly determined. We then open the economy to trade and look at the changes in the ...nancial system size and quality as a function of comparative advantage. In Section 3, we discuss our empirical methodology and construct a measure of external ...nance need of exports, as well as an instrumental variable that will allow us to identify the causal impact of trade on ...nancial development. The data used in this paper are described in Section 4. Our estimation results are presented in Section 5, and Section 6 concludes. 2 The Model 2.1 The Environment Consider an economy with 1 factor, L (labor) and 2 goods: a ...nancially dependent good F and a simple good A. The time horizon consists of the interval t 2 [0;1], and consumption takes place at t = 1. Utility is Cobb-Douglas in the two goods: U (cF;cA) = cFc1A : (1) Let good A be the numeraire, and pF be the relative price of good F in terms of A. Utility maximization implies the following relationship between consumption and the relative price: pF = cA : (2) 1 cF There is a potentially in...nite number of entrepreneurs that can produce either A or F. Entrepre- neurs make the decision to enter either of the two intermediate goods sectors at t = 0. Production in the two sectors then occurs continuously in the interval t 2 [0;1]. Good A is produced with a linear technology that requires one unit of L to produce one unit of A. Pro...t maximization in that sector implies that the price of A is equal to the wage w: pA = w = 1. Good F relies on external ...nance. Setting up a production unit of good F requires one unit of L. A project in the F sector consists of a continuous ow of returns (Rt)t . In each time interval 2[0;1] [t;t + dt], the project experiences a liquidity shock ~ tdt of the following form: 1 ~ w=prob: 2 t= ; w=prob: 1 2 where is a positive constant. We assume that shocks are i.i.d. across time and ...rms, and cannot be saved. If in the interval [t;t + dt], the liquidity shock is positive, or the liquidity need is ful...lled, then the project yields a ow of returns Rdt; otherwise it returns 0 in that instant.4 4If there is an instant at some t 2 [0; 1] in which the project returns 0, it is not liquidated completely: the next instant it may produce again. 5 Agents with a liquidity need can borrow to ful...ll it. At each time interval [t;t + dt], there exists a spot credit market in which entrepreneurs with excess liquidity lend to ...nancially distressed agents at the prevailing interest rate rt. Debt contracted in the time interval [t;t + dt] is a claim on t = 1 returns. As we assume spot credit markets, rt is determined by demand and supply of liquidity: if the aggregate liquidity shock is positive, then there is excess supply of ...nance and interest rates drop to zero. On the other hand, when there is a negative aggregate liquidity shock, lenders capture the entire bene...t of re...nancing the project so that rt dt = pFRdt. In the latter case, there are some projects with unful...lled liquidity needs which yield zero return that instant. How can we determine the total output in the F sector? Let be the share of the labor force L employed in the F sector. Then the total number of ...rms in that sector is L, and we index those ...rms by i 2 f1;:::; Lg.5 The cumulative output in this sector depends on how many projects are liquidated in each interval [t;t + dt], and therefore on aggregate liquidity in each instant. Let be t the fraction of projects that are liquidated in the time interval [t;t + dt]. It is given by: : 0 PL PL 1 ~i if ~i < 0 = L t t (3) t 8 < i=1 i=1 otherwise PL The sum of all the shocks across ...rms in the F sector, ~i,gives the aggregate liquidity in this t i=1 economy at time t. If it is positive, no projects are liquidated. If it is negative, the fraction of projects that are liquidated depends on the magnitude of the negative aggregate shock. Assuming projects are liquidated at random when aggregate liquidity is negative, the cumulative output realized by R0 1 each ...rm in sector F is given by R [1 ( L)], where ( L) dt. Pro...t maximization by t entrepreneurs in sector F therefore implies that the price of good F equals unit cost: pFR [1 ( L)] = w = 1: (4) Our model captures the positive relationship between the ...nancial system's size and its quality. The equilibrium value ( L) is the fraction of time that a ...rm is unable to ful...ll the need for external ...nance and thus loses output.6 Thus we can think of 1 ( L) as the quality of the ...nancial system. This quality depends positively the size of ...nancially intensive sector . As the number of entrepreneurs in the F sector increases, the probability of a negative aggregate shock of a given magnitude is lower, thus making liquidation a more unlikely event. The following lemma formalizes this property of ...nancial system's quality. Lemma 1: the quality of the ...nancial system ( L). The function ( L) is decreasing in , with lim 1 !0 ( L) = and lim ( L) = 0: 2 !1 5Here and in the rest of the paper, we ignore integer constraints on L for simplicity. 6In our setup, the value of ( L) will be appreciably greater than zero only if the number of ...rms L is not too large. Thus, in our model, L should be thought of not as the number of workers, but as the number of large enterprises that the labor force in this economy can potentially sta¤. 6 Proof: see Appendix. 2.2 Autarky Equilibrium We can now analyze the equilibrium in the closed economy. The equilibrium production structure is characterized by a single variable, , which is the share of the labor force employed in sector F. A value of pins down the total production of the two goods, and market clearing implies that consumption equals output: cF = R [1 ( L)] L (5) and cA = (1 )L: (6) Equations (2) through (6) de...ne the autarky equilibrium. The assumptions we made lead to a simple expression for the allocation of production: A = : (7) We can then derive the volume of external ...nance that occurs in this economy. At each instant t 2 [0;1], let k be the number of ...rms that receive a positive shock, and thus L k be the number of ...rms that receive a negative shock. If k > L k, the amount of lending in that instant is L k. If k < L k, the amount of lending in that instant is k: the economy is liquidity-constrained. Thus, the expected value of lending at each t, and thus the overall value of lending over the period between t = 0 and t = 1 is: L Private Credit = XkP 2 (k) + 1 X L ( L k)P(k); L +1 2 where k is a binomial random variable with probability and the total number of draws L. Applying 1 2 the law of iterated expectations, the expression above simpli...es to: 1 Private Credit = L; 2 which shows that in this simple model, the amount of external ...nance is linear in the size of the externally dependent sector. 2.3 Trade Equilibrium Suppose that there are two countries, the North and the South. They are endowed with LN and LS units of labor, respectively, and exhibit a Ricardian productivity di¤erence in the F sector: RN > RS. We assume that the parameter values are such that the North is the only country to 7 produce the F good in the trade equilibrium. As we will see below, this outcome will obtain as long as the North is large enough, and/or the F good is small enough in the consumption bundle. This means that in order to pin down the trade equilibrium production structure, all we need to solve for is the share of labor force employed in the F sector in the North, N. Equilibrium is de...ned by a version of equation (2) in which cF and cA are now overall world consumption values, equation (4) for a given N , and the trade versions of the good market clearing conditions: cF = RN 1 N N L N N L (8) and cA = (1 N )LN + LS: (9) These four equations lead to a simple expression for equilibrium allocation of resources: LN + LS N = (10) LN as long as N 1. It is immediate from this expression that this condition will be satis...ed if LN is large enough, or is small enough. For example, if the two goods have an equal share of consumption basket, = , and the two countries have the same factor endowments, LN = LS, 1 N 2 is exactly 1. What is happening to private credit? It is clear that there is no longer any borrowing or lending in the South. Furthermore, as S = 0, the value of ( S S L ) in the South is at the maximum: the quality of the ...nancial system deteriorates as the marginal entrepreneur does not have any opportunity to insure against shocks through external ...nance. In the North, comparing (7) and (10) it is immediate that there is more borrowing and lending after trade opening: N > A . This in turn implies that the quality of the ...nancial system improves as well: N N L < A N L . As more ...rms enter the F good production, the fraction of time external ...nance needs of ...rms are unful...lled decreases. 3 Empirical Methodology The main point of the paper is that to the extent ...nancial development is an outcome of supply and demand for external ...nance, a country's trade patterns will a¤ect its ...nancial development. Countries whose trade specialization implies that they produce and export ...nancially dependent goods will experience a higher level of ...nancial development than countries that produce goods for which it is not important to rely on external ...nance, all else equal. This is especially true of conventional measures of ...nancial development, such as private credit to GDP, which are equilibrium quantities. In order to demonstrate this point empirically, we must construct a summary measure of how ...nancially dependent is a country's export pattern. 8 3.1 The External Finance Need of Exports We start with the standard Rajan and Zingales (1998) classi...cation of industries according to their dependence on external ...nance. The Rajan and Zingales measure is de...ned as capital expenditure minus cash ow, divided by capital expenditure, and is constructed based on US ...rm-level data. We use the version of the variable assembled by Klingebiel, Kroszner, and Laeven (2005), in which industries are classi...ed according to the 3-digit ISIC Revision 2 classi...cation. The Rajan and Zingales external dependence measure is reproduced in Table 1. We combine this industry-level information with data on the structure of a country's exports to develop a measure of a country's external ...nance need of exports (hereafter EFNX) by following the approach of Almeida and Wolfenzon (2005). In particular, we construct the following variable for each country and period of time: EFNXct = X!X I (11) ictEDi; i=1 where c indexes countries, t time periods, i industries, !X is the share of exports in sector i in total ict manufacturing exports from country c in time period t, and EDi is the Rajan and Zingales measure of dependence on external ...nance. Summing up across sectors in each country and year implies that our index is at country level, but potentially varies over time. Armed with this variable, we would like to estimate the following equation: FinDevct = + EFNXct + Zct + c + t+ "ct (12) The left-hand side variable, FinDevct is a measure of a country's level of ...nancial development. We condition on the vector of controls Zct, country ...xed e¤ects c, and time ...xed e¤ects t. Our hypothesis is that the e¤ect of ...nancial content of exports, EFNXct, on ...nancial development is positive ( > 0). 3.2 Instrumentation Strategy It is immediate that we have an important simultaneity problem: a country's trade pattern is surely inuenced by its ...nancial development, as documented by Beck (2003), for instance. Thus, in order to estimate the causal relationship going from trade to ...nancial development, we must develop an instrument for our main right hand side variable, namely the external ...nancing need of exports. In order to do this, we expand the geography-based approach of Frankel and Romer (1999). These authors constructed predicted trade as a share of GDP by ...rst estimating a gravity regression on bilateral trade volumes between countries using only exogenous geographical explanatory variables, such as bilateral distance, land areas, and populations. From the estimated gravity equation, these 9 authors predicted bilateral trade between countries based solely on geographical variables. Then for each country they summed over trade partners to obtain the predicted total trade to GDP, or "natural openness." Our objective is to ...nd an instrument for a measure of export patterns, not aggregate trade openness. Thus, we must extend the Frankel and Romer approach accordingly. Namely, we apply their methodology to exports at sector level, following di Giovanni, Levchenko, and Ranciere (2005). For each industry i and time t, we run the Frankel and Romer regression: LogXicdt = + landlockedcd + (13) 1ldistcd + 2lpopct + 3lareac + 4lpopdt + 5 laread + 6 7bordercd + 8bordercd ldistcd + 9bordercd popct + 10bordercd areac + 11bordercd popdt + 12 bordercd aread + 13bordercd landlockedcd + "cd; where LogXicdt is the log of exports as a share of GDP in industry i, from country c to country d, at time t. The right-hand side consists of the geographical variables. In particular, ldistcd is the log of distance between the two countries, de...ned as distance between the major cities in the two countries, lpopct is the log of population in year t, lareac log of land area, landlockedcd takes the value of 0, 1, or 2 depending on whether none, one, or both of the trading countries are landlocked, and bordercd is the dummy variable for common border. The right-hand side of the speci...cation is identical to the one Frankel and Romer (1999) use. Note that we will be estimating a separate gravity equation for each sector and time period. All of the right-hand side variables except population, however, are non-time varying, as would be expected of geographical characteristics. Thus, to the extent that we will observe variation in predicted exports in an industry over time, it will be driven purely by changing estimated coe˘ cients in the equation (13) from period to period. Having estimated equation (13) for each industry and time period, we then obtain the predicted logarithm of industry i exports to GDP from country c to each of its trading partners indexed by d, LogXicdt. In order to construct the predicted overall industry i exports as a share of GDP from \ country c, we take the exponential of the predicted bilateral log of trade, and sum over the trading partner countries 1 through C, exactly as in Frankel and Romer (1999): Xict = b XeLogX C \icdt : d=1 d6=c That is, predicted total trade as a share of GDP for each industry, country, and time period is the sum of the predicted bilateral trade to GDP over all trading partners.7 Thus, we in e¤ect modi...ed and extended the Frankel and Romer methodology in three respects. First, and most importantly, 7An important question is how to deal with cases of zero bilateral trade. Since we take logs of trade values, our gravity estimation procedure ignores zeros. Thus, we generate predicted values of trade only when the actual value is positive. One interpretation of our procedure is that it "predicts"zero trade when it observes zero trade. 10 we construct the Frankel and Romer predicted trade measures by industry. Second, we do it over time. And ...nally, rather than looking at total trade, we look solely at exports. Armed with a working model for predicting exports to GDP in each industry, it is straightforward to construct the instrument for external ...nancing need of exports, based on predicted export patterns rather than actual ones. That is, our instrument will be, in a manner identical to equation (11): Here, the predicted share of exports in industry ibictEDi: EFNXct = \ X!X I i=1 , in country c and time t, !X , is constructed from the predicted exports to GDP ratios, Xict in a straightforward manner: !X = bictb bict XIibXict : Xict =1 normalized by the samebGDP, and thus they cancel out when we take the predicted export share. Note that even though Xict is exports in industry i normalized by a country's GDP, every sector is b We proceed by describing the data sources in the next section. We provide a snapshot of our data, focusing on the patterns of external ...nancing needs of exports that we obtain. Then, in the following section we document stages of constructing our instrument, and present OLS and 2SLS regression results for both a cross-section of countries and a panel of 5-year averages going back to the 1970's. 4 Data Description International trade ows come from the World Trade Database described in Feenstra et al. (2005). This database contains bilateral trade ows between some 150 countries, accounting for 98% of world trade. Trade ows are reported using the 4-digit SITC Revision 2 classi...cation. Since our variable of interest, EFNXct, is constructed using information on total exports from each country in each industry, we ...rst aggregate bilateral ows across countries to obtain total exports for each country and manufacturing sector. We then convert the trade ows from SITC to 3-digit ISIC Revision 2 classi...cation.8 This allows us to combine the trade data with the information on external dependence from Rajan and Zingales. For the purposes of estimating the gravity equation (13), we retain information on bilateral trade, converting it once again into the 3-digit ISIC Rev. 2 classi...cation. We merge bilateral trade data with geography variables taken from Centre d'Etudes Prospectives et d'Informations Internationales 8The conversion is based on the concordance found on the International Trade Resources website maintained by Jon D. Haveman: http://www.haveman.org. 11 (CEPII). The CEPII database contains information on bilateral distances between the major cities for each pair of countries, whether two countries share a border, as well as information on land area and whether a country is landlocked.9 Population data is taken from World Bank's World Development Indicators for the period 1970-1999. Exporter and importer population is the only variable in our gravity speci...cation that changes over time. The data on ...nancial development comes from the database originally compiled by Beck, Demirguc- Kunt, and Levine (2000). We use a version that has been checked for accuracy by Loayza and Ranciere (2005). Following the standard in the literature, our preferred indicator of ...nancial devel- opment is the ratio of credit by banks and other ...nancial institutions to the private sector as a share of GDP ("private credit"). The controls in our estimation include overall trade openness (imports plus exports as a share of GDP) and PPP-adjusted GDP per capita income, both of which come from the Penn World Tables (Heston, Summers and Aten, 2002). Finally, we use information on countries'legal origin as de...ned by La Porta et al. (1998), extended to include the socialist legal system. The ...nal sample includes 96 countries, and is an unbalanced panel of 5-year averages from 1970-74 to 1995-99. Appendix Table A1 presents the data on the external ...nancing need of ex- ports, EFNXct, for our sample of countries for the most recent 5-year period, 1995-99. Aside from EFNXct, the table contains information on the top two export sectors, the share of the top two sectors in the overall manufacturing exports, overall trade openness, private credit, as well as the sample means of these variables. It is clear that while there is some correlation between per capita income and the ...nancial content of trade, income is far from a perfect predictor of EFNXct. The top two countries ranked according to the ...nancial content of trade in this period are Malaysia and Philippines, only then followed by Singapore, Japan, and Switzerland. Only the bottom two countries have the ...nancial content of trade that is negative in this period, Malawi and Zimbabwe. In these countries, the main export industry is Tobacco, which has a negative external ...nance dependence according to the Rajan-Zingales classi...cation. We plot our estimates of the external ...nance need of exports against log of PPP-adjusted per capita income in Figure 2. It is clear that while there is a positive relationship between income and our variable of interest, it is far from close. The correlation between the two variables is less than 0.4. Figure 3 plots the external ...nance need of exports against overall trade openness. There is little relationship between the two variables, and thus, as expected, we are measuring something distinct from trade openness when we construct our measure of the external ...nancing need of exports. Finally, Figure 4 plots ...nancial development against EFNXct. There is a positive relationship 9The dataset is available online at http://www.cepii.fr/anglaisgraph/bdd/distances.htm. 12 between the two variables, though it is not extremely close. The correlation coe˘ cient between them is above 0.5, with a Spearman rank correlation of 0.42 in the period 1995-99. We turn to a regression analysis of the relationship between these two variables after presenting the stages of constructing the instrument. 5 Results 5.1 Sector-Level Gravity Estimation In order to build the instrument, we estimate equation (13) for each industry and 5-year period between 1970-74 and 1995-99. Because all in all we have to run some 170 regressions, presenting the full regression output would be impractical. Thus, we summarize the results for the most part graphically. In the entire sample of our regressions, the smallest number of observations is 773, the largest is 6877, with the mean of 3677. The R-squared's range from 0.14 to 0.56, with the mean of 0.32. Because the right-hand side variables are the same in all regressions, our empirical strategy would only work if the estimated coe˘ cients di¤er signi...cantly across sectors. Thus, the ...rst important question we must answer is whether or not there is much variation in the estimated coe˘ cients. Figure 5 plots, for each of the 13 right-hand variables, the range of coe˘ cient estimates across sectors and years. There is clearly quite a bit of dispersion in virtually all of the coe˘ cients. For the period 1995-99 for example, the distance coe˘ cient ranges from -0.8 to 0.04 (though the latter is not signi...cantly di¤erent from zero). This variation in gravity coe˘ cients across industries is quite typical of sector-level gravity studies, which have focused for the most part on the distance coe˘ cient (see, among others, Rauch, 1999, Hummels, 2001, Evans, 2003, and Chaney, 2005). We are also hoping to construct an instrument that varies meaningfully over time. Because none of our regressors except population changes over time, any time variation in the instrument will come from changes in the coe˘ cient estimates for each sector. We check whether our estimates have this feature in Figure 6. It plots, for each sector, the evolution of the coe˘ cient on the log of bilateral distance. Solid dots indicate the point estimates, while hollow dots are the point estimates plus and minus two standard errors.10 It is clear that we do have time variation, and its extent varies across sectors. Nonetheless, in almost every sector, among the coe˘ cients for the individual time periods there is a pair that is signi...cantly di¤erent from each other. Another notable feature of this Figure is that the changes are not monotonic over time: the distance coe˘ cient within a sector often falls in some periods and rises in others. Distance coe˘ cient changes in gravity models over time have been examined elsewhere in the literature, though the conclusions di¤er across studies. While some 10Similar plots for every other coe˘ cient are available upon request. 13 (e.g. Frankel, Stein, and Wei, 1997, Eichengreen and Erwin, 1998, and Soloaga and Winters, 2001) ...nd the distance coe˘ cient increasing over time for various samples of countries and time periods, others (e.g. Coe et al., 2002, Brun et al., 2005) reach the opposite conclusion. Our strategy does not rely on a particular direction of change in the coe˘ cient: all we require are changes over time, and, preferably, di¤erentially across sectors. Note also that both sector-level and across-time studies of gravity coe˘ cients have been primarily about the coe˘ cient on distance. Our approach exploits sector and time variation in all of our estimated coe˘ cients. Finally, we use our estimates to generate predicted exports as a share of GDP in each sector, as outlined in Section 3. Using that, we construct the predicted external ...nance need of exports. Figure 7 plots it against the actual EFNX for the period 1995-99, along with a 45-degree line. We can see that while there is a strong positive relationship between the two, it is not at all one-to- one. In particular, our procedure clearly underestimates the external ...nance need of exports for countries in which it is unusually high, and overestimates it for countries where it is low. This is comforting for us, as it indicates that our approach is not so mechanical that is reproduces the actual values perfectly. Appendix Figure A1 plots the actual and predicted values of EFNX for our entire sample of countries over time. When it comes to time variation in the actual and predicted values of EFNX, the picture is broadly similar: the predicted value most of the time follows a similar trend as the actual EFNX, though it is usually atter. 5.2 Financial Development Results 5.2.1 Cross-sectional Speci...cations We start with the cross-sectional OLS regression. We estimate equation (12) using the averages of the left-hand side and all of the controls for the entire time period, 1970-99.11 The results are presented in Table 2, with White robust standard errors in parentheses. Column 1 reports the bivariate relationship between ...nancial development and simple trade openness. While trade openness is signi...cant at 10% level, the relationship is not close, with the R-squared of 0.05. When instead we use EFNXc, as is done in Column 2, the R-squared is 0.28, and the variable of interest is signi...cant at the 1% level, with a t-statistic of 4.1. Column 3 includes both the trade openness and the external ...nance need of exports. The coe˘ cient on EFNX is virtually unchanged. Columns 4 and 5 attempt to control for other determinants of ...nancial development. We ...rst include the legal origin dummies from La Porta et al. (1998), and then per capita income. The latter is meant to capture a country's overall level of development. While in both of these speci...cations the coe˘ cient on EFNXc is about one third smaller, it nonetheless remains signi...cant at the 1% level. Finally, 11Note that since we have an unbalanced panel, our procedure results in averaging over di¤erent numbers of years for di¤erent countries. 14 column 5 includes both the legal origin dummies and per capita income on the right-hand side. The coe˘ cient on our variable of interest is further reduced somewhat, but preserves its signi...cance at 1% level. Note that with all of the controls included in our speci...cation, the adjusted R-squared is 0.63, only about double the R-squared of the bivariate regression with only EFNXc as the independent variable. Endogeneity is clearly a ...rst-order issue in our estimation. As has been shown in several empirical studies, a country's level of ...nancial development a¤ects trade patterns, and thus will a¤ect the external ...nance need of exports as we construct it. We deal with the simultaneity problem by adopting an instrumental variables approach we described in Section 3. We estimate a two-stage least squares (2SLS) regression, using predicted external ...nance need of exports (EFNXc) as an \ instrument for actual EFNXc. Table 3 reports the results. The top panel contains the full results of the second stage of the regression, while the bottom panel reports the coe˘ cient on EFNXc \ from the ...rst stage. Column 1 reports a bivariate regression with EFNXc on the right-hand side. The 2SLS coe˘ cient is signi...cant at 1% level. It is about two thirds higher in magnitude than the OLS coe˘ cient. Columns 2 through 5 follow the sequence of Table 2. We ...rst include overall trade openness into the regression, and see that the coe˘ cient of interest is virtually unchanged. Including the legal origin controls reduces the coe˘ cient a bit, while controlling for per capita income lowers it further. In the most stringent speci...cation, which includes openness, legal origin indicators, and per capita income, the coe˘ cient of interest is about half the magnitude of the coe˘ cient in column 1. It is nonetheless highly signi...cant, with the p-value of 2.3%. Examining the bottom panel of the Table, we can see that in the ...rst stage, the coe˘ cient on the predicted external ...nance need of trade is very close, and slightly above, 1. The coe˘ cient on EFNXc is always signi...cant at the 1% \ level. The results are economically signi...cant but not implausibly large. Using the most conservative coe˘ cient estimates, the OLS results imply that moving from the 25th to the 75th percentile in the external ...nance need of exports raises the ratio of private credit to GDP by roughly 10 percentage points. This is equivalent to about 0.3 of the standard deviation of private credit, or to moving from the 25th to the 50th percentile in the distribution of private credit in our sample. The most conservative 2SLS estimate implies that the same movement in EFNXc leads to a predicted change in private credit over GDP of about 19 percentage points, or 0.56 of a standard deviation of private credit observed in our sample. 15 5.2.2 Panel Speci...cations The cross-sectional results clearly point to an important role of trade in the development of a country's ...nancial system. We would like to go beyond the cross-section, however, and exploit the time series dimension of our data. To this end we estimate the full panel version of equation (12) on a sample of non-overlapping ...ve-year averages of all the variables from 1970-74 to 1995-99. In order to identify our e¤ect from the time variation in the variable of interest, all our speci...cations include a full set of country and time ...xed e¤ects. Furthermore, we cluster the standard errors at country level throughout, to address the problem of time series correlation in our variables (see Bertrand, Duo, and Mullainathan, 2004). This is the most conservative clustering available to us with this dataset. Table 4 presents the results. Columns 1 through 4 report the OLS exercise. We ...rst demonstrate, in Column 1, that overall trade openness does not a¤ect ...nancial development when we control for country and time ...xed e¤ects. Column 2 reports a speci...cation in which only EFNXct is included in the regression aside from the battery of ...xed e¤ects. The coe˘ cient of interest is signi...cant at 1% level with the t-statistic of 3.1. Including trade openness, as in Column 3, hardly changes the coe˘ cient. However, when we control for per capita income, the coe˘ cient is reduced by about one third, similarly to the cross-sectional regressions. Nonetheless, it remains signi...cant at the 1% level. Note that the use of ...xed e¤ects results in the adjusted R-squared of between 0.87 and 0.9, indeed the R-squared of the regression with no independent variables aside from the ...xed e¤ects is 0.86. Thus, while the cross-sectional variation across countries accounts for the overwhelming majority of the variation in ...nancial development, we can still detect the e¤ect of the time variation in the external ...nance need of exports quite clearly in our regressions. Columns 5 through 7 report the results of the 2SLS exercise. Once again, we instrument for EFNXct with predicted EFNXct, the main di¤erence being that now both the actual and the \ predicted values of the external ...nance need of exports vary over time. One possible di˘ culty we face is that all of the gravity regressors in equation (13) aside from population do not vary over time. Thus, to the extent that EFNXct changes from period to period, it will do so primarily due \ to changes in the estimated coe˘ cients on the gravity regressors for the various sectors over time. As we discuss above, our gravity coe˘ cients do change over time, giving us variation in predicted EFNXct. Furthermore, to sweep out the country component, we always include the full set of \ country ...xed e¤ects in the ...rst stage regressions, and the standard errors we report are clustered at the country level as well. The 2SLS results support what we found with OLS. The top panel reports the second stage coe˘ cients. These are generally about one third larger than the OLS coe˘ cients, and signi...cant 16 at 5% level. When we control for per capita income, the t-statistic is 2.17, with a corresponding p-value of 3.3%. The bottom panel reports the coe˘ cient of interest from the ...rst stage regression. The instrument is highly signi...cant, and the coe˘ cient is quite close to 1, though in contrast to the cross-sectional results, it is lower than 1. The quantitative e¤ect of our variable of interest as estimated in the panel speci...cations is similar to the cross-sectional magnitudes. The most conservative OLS coe˘ cient implies that moving from the 25th to the 75th percentile of EFNXct results in an increase in private credit over GDP of 8.2 percentage points, or about 0.22 of a standard deviation of private credit to GDP observed in our data. The 2SLS coe˘ cients imply a change in ...nancial development of 12 percentage points of GDP, or 0.33 standard deviations. 5.2.3 Robustness We check robustness of our results by i) dropping outliers; ii) dropping groups of countries; and iii) using alternative measures of ...nancial development as the dependent variable. Tables 5 and 6 present the results. In both of these Tables, we only report the coe˘ cients and standard errors on EFNX, and in each case use the instrumental variables speci...cation with the most stringent set of controls. The top half of each table contains the cross-sectional 2SLS results when controlling for openness, income, and legal origin. The bottom half presents the panel results when controlling for openness and income, country and time ...xed e¤ects, and clustering of the standard errors at the country level. Column 1 of Table 5 reports the results of estimating our equations while dropping the top three and bottom three countries in the distribution of EFNX. Compared to the full sample, the estimated coe˘ cients are actually larger and more signi...cant. In order to check whether the results are driven exclusively by the developed countries, the next column estimates our equations on non-OECD countries only.12 While the coe˘ cients are somewhat lower, both the cross-section and panel estimates retain their signi...cance level. The economies sometimes called "Asian tigers"have experienced some of the fastest growth of both trade and ...nancial development in the period we are considering. Column 3 excludes the Asian tigers, to check that the results are not driven by these particular countries.13 It is clear that the results are not due to Asian tigers. In fact, the coe˘ cient estimates from this subsample are virtually identical to the full sample coe˘ cients. The next two columns drop ...rst the Latin American and Caribbean countries, and then the sub-Saharan African countries. The results are not sensitive to the exclusion of these continents, in fact the estimated 12OECD countries in our sample are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, and United States. We thus exclude the newer members of the OECD, such as Korea and Mexico. 13In our sample, we consider Asian tigers to be: Indonesia, Korea, Malaysia, Philippines, and Thailand. 17 coe˘ cients are higher and more signi...cant when sub-Saharan Africa is excluded. Finally, because the major oil exporters of the Middle East may occupy a special place in the world trading system, the last column of Table 5 drops these countries, leaving the results once again unchanged. All in all, the panel results are insensitive to the subsample used, as could be expected given that all of our panel speci...cations include country e¤ects. When it comes to the cross-sectional estimates, we ...nd that all of the subsample coe˘ cients are actually higher than the full sample coe˘ cients, with the exception of the non-OECD sample. Table 6 presents the results of using alternative measures of ...nancial development.14 Colulmn 1 uses the ratio of liquid liabilities to GDP instead of private credit. Both the cross-sectional and panel results are strong, and in the cross-section the signi...cance level actually improves to 1%. Column 2 uses the ratio of stock market value to GDP instead. While the cross-sectional results are signi...cant at the 1% level, the panel estimates are not. Clearly, to the extent that EFNX explains the di¤erences in the size of countries'stock markets, it does so across countries, and not within countries over time. It is important to note that the sample size is noticeably reduced when we use this measure, especially along the time series dimension. Thus, we simply may not have observations going back far enough in time to make identi...cation o¤ the time series variation. Column 3 presents the results of using the stock market turnover ratio as the dependent variable. It is de...ned as the value of total shares traded divided by the average real market capitalization. Unlike stock market value to GDP, which is a measure of market size, turnover is a measure of stock market activity. The results we obtain are similar. While the cross-sectional estimate is signi...cant at the 1% level, we cannot identify the e¤ect from the time series. Finally, we would like to use a measure of the quality of the ...nancial system rather than its size. Column 4 reports the outcome of using the net interest margin as the dependent variable. The net interest margin is de...ned as the accounting value of banks net interest revenue as a share of its interest-bearing assets.15 This variable is available only post-1996, and thus we cannot estimate a panel speci...cation. The cross-sectional 2SLS estimate, however, is signi...cant at the 5% level, suggesting that there may be an e¤ect on the quality of the ...nancial system as well as its size. 6 Conclusion It is often argued that institutional quality in general and ...nancial development in particular are shaped largely by exogenous events in the past. It is then natural to think of the ...nancial system as an endowment, and therefore di¤erences in ...nancial development as sources of comparative advan- 14All of the alternative measures come from the most recent version of the Beck, Demirguc-Kunt, and Levine (2000) database. 15Unlike all of the other measures, a low value of net interest margin indicates a high quality of the ...nancial system. 18 tage in trade. This paper takes a di¤erent view by asking instead: will trade patterns in turn a¤ect countries'...nancial development? This is an important question. There is a great deal of evidence that ...nancial development is a key determinant of economic growth (see Levine, 2005, for a survey). On the other hand, the debate about the e¤ect of trade on growth is far from settled.16 This paper demonstrates that trade a¤ects ...nancial development directly, a channel for the relationship between trade and growth which has not previously been explored. We ...rst illustrate our main idea by building a model in which ...nancial development ­both the ...nancial system size and its quality ­is determined by demand for external ...nance in production. After trade opening, the country which produces and exports ...nancially dependent goods experiences ...nancial deepening, as demand for external ...nance inside that country increases. On the other hand, the country which imports ...nancially dependent goods will see its ...nancial system deteriorate, making access to ...nance more di˘ cult for domestic ...rms. We then demonstrate this e¤ect empirically by constructing a measure of a country's external ...nance need of exports, and relating it to ...nancial development in a large panel of countries. The magnitude of the e¤ect we obtain is appreciable, but not very large. Thus, we do not conclude from our exercise that trade volumes or trade patterns are the primary determinant of ...nancial development. Admittedly, other variables, such as history, legal systems, institutions, openness to capital ows, or the overall level of development are other signi...cant determinants. Another important caveat when it comes to interpretation is that our measure of external ...nance need of exports is positive except in very rare cases. Thus, our empirical results do not imply that trade has a negative e¤ect on private credit. Rather, what we show is that the demand for external ...nance coming from exports di¤ers a great deal across countries, and has an appreciable impact on observed levels of ...nancial development. 7 Appendix Proof of Lemma 1: t is a random variable with the following probability distribution: 0 with probability L t= 2 L =0 k ; and this implies that: 8 L+1 < ( ) 1 1 2k with probability 1 L 0 k Int L 1 L 2 Xkfor Int 2 L k 2 1 L 1 t ( L) E = 2 L Int L 1 2 16Recent papers that argue for a positive impact of trade on growth include, but are not limited to, Frankel and Romer (1999) and Alcala and Ciccone (2004). For the opposing view, see Rodrik and Rodriguez (2000), Rodrik, Subramanian, and Trebbi (2004), and Rigobon and Rodrik (2004). 19 and it is easy to check that (1) = 1=2 and lim !1 ( L) = 0: 8 References 1. Acemoglu, Daron, Simon Johnson, and James Robinson (2001) "The Colonial Origins of Com- parative Development,"The American Economic Review, 91, 1369-401. 2. Acemoglu, Daron, and Fabrizio Zilibotti (1999) "Information Accumulation in Development," Journal of Economic Growth, 4, 5-38. 3. Alcalá, Francisco and Antonio Ciccone (2004) "Trade and Productivity,"Quarterly Journal of Economics 119, 613-46. 4. Almeida, Heitor and Daniel Wolfenzon (2005), "The E¤ect of External Finance on the Equi- librium Allocation of Capital ,"Journal of Financial Economics 75, 133-164. 5. Baldwin, Richard (1989) "Exporting the capital markets:Comparative advantage and capital market imperfections."In: Audretsch, D., Sleuwaegen, L., Yamawaki, H. (Eds.), The Conver- gence of International and Domestic Markets. North-Holland, Amsterdam. 6. Beck, Thorsten (2002) "Financial Development and International Trade. Is There a Link?" Journal of International Economics, 57, 107-131. 7. Beck, Thorsten (2003) "Financial dependence and international trade,"Review of International Economics, 11, 296-316. 8. Beck, Thorsten, Asli Demirguc-Kunt, and Ross Levine (2000) "A New Database on Financial Development and Structure"World Bank Economic Review, September 2000, 597-605. 9. Becker, Bo, and David Greenberg (2003) "The Real E¤ects of Finance: Evidence from Ex- ports,"mimeo, University of Chicago. 10. Bertrand, Marianne, Esther Duo, and Sendhil Mullainathan (2004) "How Much Should We Trust Di¤erences-in-Di¤erences Estimates?,"Quarterly Journal of Economics, 119, 249-75. 11. Braun, Matias, and Claudio Raddatz (2005) "Trade Liberalization and the Politics of Financial Development,"World Bank Working Paper 3517. 12. Broner, Fernando and Jaume Ventura "Globalization and Risk Sharing,"mimeo, CREI-UPF. 13. Brun, Jean-Francois, Celine Carrere, Patrick Guillaumont, and Jaime de Melo (2005) "Has Distance Died? Evidence from a Panel Gravity Model," World Bank Economic Review 19, 99-120. 20 14. Chaney, Thomas (2005) "Distorted Gravity: Heterogeneous Firms, Market Structure, and the Geography of International Trade,"mimeo, University of Chicago. 15. Coe, David, Arvind Subramanian, and Natalia Tamirisa, with Rikhil Bhavnani (2002) "The Missing Globalization Puzzle,"IMF Working Paper 02-171. 16. Demirguc-Kunt, Asli, and Ross Levine (2001) "Financial Structure and Economic Growth: A Cross-Country Comparison of Bank, Markets, and Development," Cambridge, MA: MIT Press. 17. di Giovanni, Julian, Andrei Levchenko, and Romain Ranciere (2005) "Trade and Growth at Industry Level,"unpublished. 18. Do, Quy-Toan, and Andrei Levchenko (2004) "Trade and Financial Development", World Bank Working Paper 3347. 19. Eichengreen, Barry, and D.A. Irwin (1998) "The Role of HIstory in Bilateral Trade Flows," in Je¤rey Frankel (ed.), The Regionalization of the World Economy, Chicago: University of Chicago Press. 20. Evans, Carolyn (2003) "The Economic Signi...cance of National Border E¤ects,"The American Economic Review, 93(4), 1291-1312. 21. Feenstra, Robert C., Robert E. Lipsey, Haiyan Deng, Alyson C. Ma, and Hengyong Mo (2005) "World Trade Flows: 1962-2000,"NBER Working Paper No. 11040. 22. Frankel, Je¤rey, and David Romer (1999) "Does Trade Cause Growth?"American Economic Review, 89, 379-399. 23. Frankel, Je¤rey, with Ernesto Stein, and Shang-Jin Wei (1997) "Regional Trading Blocs in the World Economic System,"Washington DC: Institute of International Economics. 24. Heston, Alan, Robert Summers, and Bettina Aten, (2002) "Penn World Table Version 6.1," Center for International Comparisons at the University of Pennsylvania (CICUP). 25. Hummels, David (2001) "Towards a Geography of Trade Costs,"mimeo Purdue University. 26. Ju, Jiandong, and Shang-Jin Wei (2005) "Endowment versus Finance: A Wooden Barrel Theory of International Trade,"IMF Working Paper No. 05/123. 27. Kletzer, Kenneth, and Pranab Bardhan (1987) "Credit markets and patterns of international trade."Journal of Development Economics, 27, 57­70. 21 28. Klingebiel, Daniela, Randall Kroszner, and Luc Laeven (2005) "Financial Crises, Financial Dependence, and Industry Growth,"World Bank Working Paper 2855. 29. La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny (1998) "Law and Finance,"Journal of Political Economy, 106, 1113-1155. 30. Levine, Ross (2005) "Finance and Growth: Theory and Evidence," in Philippe Aghion and Steven Durlauf, eds. Handbook of Economic Growth. The Netherlands: Elsevier Science. 31. Levine, Ross and Sergio Schmukler (2005) "Internationalization and Stock Market Liquidity," NBER Working Paper 11894. 32. Loayza, Norman, and Romain Ranciere (2005) "Financial Development, Financial Fragility, and Growth,"forthcoming, Journal of Money, Credit, and Banking. 33. Manova, Kalina (2005) "Credit Constraints in Trade: Financial development and export com- position,"mimeo, Harvard University. 34. Newbery, David, and Joseph Stiglitz (1984) "Pareto inferior trade,"Review of Economic Stud- ies, 51, 1­12. 35. Rajan, Raghuram, and Luigi Zingales (1998) "Financial dependence and growth,"American Economic Review, 88, 559­586. 36. Rajan, Raghuram, and Luigi Zingales (2003) "The great reversals: the politics of ...nancial development in the 20th Century,"Journal of Financial Economics, 69, 5-50. 37. Rauch, James E., 1999, "Networks versus Markets in International Trade,"Journal of Inter- national Economics, 48, 7-35. 38. Rigobon, Roberto, and Dani Rodrik (2004) "Rule of Law, Democracy, Openness, and Income: Estimating the Interrelationships,"forthcoming, Economics of Transition. 39. Rodriguez, Francisco, and Dani Rodrik (2000) "Trade Policy and Economic Growth: A Skep- tic's Guide to the Cross-National Evidence," (with Francisco Rodríguez), Macroeconomics Annual 2000, eds. Ben Bernanke and Kenneth S. Rogo¤, MIT Press for NBER, Cambridge, MA. 40. Rodrik, Dani (1998) "Why Do More Open Economies Have Bigger Governments?"Journal of Political Economy, 106, 997­1032. 22 41. Rodrik, Dani, Arvind Subramanian and Francesco Trebbi (2004) "Institutions Rule: The Primacy of Institutions over Geography and Integration in Economic Development,"Journal of Economic Growth, 9, 131-165. 42. Soloaga, Isidro and L. Alan Winters (2001) "Regionalism in the nineties: What e¤ect on trade?"North American Journal of Economics and Finance, 12, 1­29. 43. Svaleryd, Helena, and Jonas Vlachos (2002) "Markets for risk and openness to trade: how are they related?"Journal of International Economics, 57, 369­395 44. Svaleryd, Helena, and Jonas Vlachos (2005) "Financial markets, the pattern of industrial spe- cialization and comparative advantage: Evidence from OECD countries,"European Economic Review, 49, 113-144. 23 Figure 1: Financial Development and Trade Volumes, 1970-1999. Advanced Countries Developing Countries 2 1. .5 1 .4 .8 .6 .3 .4 .2 .2 1970 1980 1990 2000 1970 1980 1990 2000 Trade/GDP (left axis) Private Credit/GDP (right axis) Figure 2: External Finance Need of Exports and Per Capita Income, 1995-99. 0 10 9-59 MLT 19,stropxE PHL MYS SGP JPN BDI DJI TWN ISR CHE 50 MEX RWACAF KOR GBR IRL HKGUSA SLEGMB THA SWE HUN FRADNK AUT GIN CHN KNABRB FIN AGO COG CRI PAN ESP BEL ITACAN oft IDN JOR LBN POL NER NOR IND PRTBHS COLTUR OMN BGRBRA CYPBMUAUS NZL ETH GNBTCD MLI TGO MOZ GHA CIV MAR ARGGRC entnoCliacnaniF TUNZAFGAB TTO MUS URY MDG BFABENNPL SDNLAO KHM VNMGNQ KENSEN CMR PAK BOLGUYECUROMIRN LKAJAMPERFJI BLZ PNGEGY DOMPRY SYC SAU QAT MNG MRT HTI SYR SLV ISL TZA YEM UGA NGA BGDNICHND ALB GTMDZA CUB VEN CHL BHR NLD KWT 0 ZMB ZWE MWI 0 -5 6 7 8 9 10 Per Capita GDP, 1995-9 24 Figure 3: External Finance Need of Exports and Trade Openness, 1995-99. 0 10 9- 95 MLT 19,st PHL MYS JPN SGP BDI ISR TWN DJI IRL HKG GBR MEXKOR CHE orpxE 50 USA RWA CHNGINITA SLE CAF FRA DNK AUT FIN SWE THAGMBKNA HUN BRB ESP CAN CRI COGAGO BEL IDNNOR BHS PRT oft CYP JOR BRA IND NER POLLBNBMUPANGAB OMN ARG ZAFMAR TUR NZL GHACIV TGO en SDN COLAUSMOZGNBROM IRNBFA HTI EGYTCD PAK ETHGRCBENMLIDOMSAUTUNBGR URYCMR NPL BOL ECUQAT PNG TTO GUY GNQ nt PER VENDZA MDGSLVSYRKHMMNGBLZ KENISLLKAVNMJAM MUSSYC BGD CUB UGA GTMALB TZACHL LAOSENPRYMRT NICFJI YEM NGAKWT HNDNLD BHR Coliac 0 ZMB ZWE aninF MWI 0 -5 3 4 5 6 Log of Trade/GDP, 1995-9 Figure 4: Financial Development and External Finance Need of Exports, 1995-99. CHE 0 20 JPN USA NLD HKG 9-59 THA KOR MYS ZAF 19,PDG/tdie FRAGBR SWE SGP 0 NZL 10 NORESP PANAUT CHN PRT AUS BEL CAN JOR IRL BHRISLMAR IDN ITA CHLSAU TUN ISR KWTTTOBOL FIN DNK PHL Cr NICEGY SLV ZWE BGDLKADOMIND HNDPRY KENBRA atvirP VENSENTUR PAK URYCOL ECU GRC NPL HUN NGAHTI PERETH IRNTGOPOL JAM e GTMPNGCIV ARG CRI ZMBUGASDN DZASYR MDGGHA BFA GMBMEX 0 MWI CMR ROMGABCOGCAF NER SLE RWABDI 00 -1 -50 0 50 100 Financial Content of Exports, 1995-9 25 Figure 5: Estimated Sector-Level Gravity Model Coefficients, 1970-99. log(distance) log(pop_exporter) log(area_exporter) log(pop_importer) .5 0 .2 1 0 .8 5-. 5-. 0 .6 .4 -1 -1 2-. 5. 5. 4-. .2 -1 -1 0 1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000 log(area_importer) landlocked border border*log(distance) .1 1 10 1 0 0 5 .5 1-. 0 -1 0 2-. -2 -50 3-. -15 5-. -3 -1 -1 1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000 border*log(pop_exporter) border*log(area_exporter) border*log(pop_importer) border*log(area_importer) 1 .4 1 .4 .2 .2 .5 0 .5 0 0 5-. 2-.4-.6-. 0 5-. 2-.4-.6-. 1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000 border*landlocked 3 2 1 0 -1 -2 1970 1980 1990 2000 Figure 6: Estimates and Two-Standard Error Bands for the Distance Coefficient, by Sector and over Time. 311: Food products 313: Beverages 314: Tobacco 321: Textiles 322: Wearing apparel 323: Leather products .5 0 5-. -1 .5 -1 324: Footwear 331: Wood products, ex. furnit. 332: Furniture, except metal 341: Paper and products 342: Printing and publishing 351: Industrial chemicals .5 0 5-. -1 .5 -1 352: Other chemicals 353: Petroleum refineries 354: Misc. petr. and coal prod. 355: Rubber products 356: Plastic products 361: Pottery, china, earthenware .5 0 5-. -1 .5 -1 362: Glass and products 369: Oth. non-metal. mineral prod. 371: Iron and steel 372: Non-ferrous metals 381: Fabricated metal prod. 382: Machinery, ex. electric .5 0 5-. -1 .5 -1 1970 1980 1990 2000 1970 1980 1990 2000 383: Machinery, electric 384: Transport equipment 385: Prof. & scient. equip. 390: Other manufacturing .5 0 5-. -1 .5 -1 1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000 26 Figure 7: Actual and Predicted External Finance Need of Exports, Average 1970-99 60 99- 70,st orpxE 40 SGP HKG NLD MUS ROM LBN POL BEL KOR JPNCHE oft VNM IDN JAM LKABGR INDCHNHUNITAAUTBRBISR BHR BGD DOMTUR NGAGRC PAK EGYCOL ECU BRA CRITHA LAOSLVHTIDNKIRLUSA en KWT IRNTUNTGO GTM PRTESPBMUFRAGBRMEX JOR GIN VENPER SYRMARSYC RWA KHM TTOGHAPRY BLZZAF NPLCYP FINANT SWE PANSLE DJILBR GNQ 20 SAUFJIKIRMDGAGOKNA URYARENER KENBEN NOR SEN HNDCIV ARGIRQAFG nt CHLALBCMRBFAMLI NICNZL GMB Coliac ZWENCL TZAMNGBDI DZAYEM LBYISL BOLUGAGUY MOZBHS PNGAUSOMN GNB SDN ETHQAT TCD COG CAN GAB CAF SURMRT GRL SOM MWI ZMB aninF 0 edtic edrP 0 -2 -20 0 20 40 60 Actual Financial Content of Exports, 70-99 27 Table 1: The Rajan and Zingales Measure of External Dependence ISIC code Industrial sector External dependence 311 Food products 0.14 313 Beverages 0.08 314 Tobacco -0.45 321 Textile 0.19 322 Apparel 0.03 323 Leather -0.14 324 Footwear -0.08 331 Wood products 0.28 332 Furniture 0.24 341 Paper and products 0.17 342 Printing and publishing 0.2 351 Industrial chemicals 0.25 352 Other chemicals 0.75 353 Petroleum refineries 0.04 354 Petroleum and coal products 0.33 355 Rubber products 0.23 356 Plastic products 1.14 361 Pottery -0.15 362 Glass 0.53 369 Nonmetal products 0.06 371 Iron and steel 0.09 372 Nonferrous metal 0.01 381 Metal products 0.24 382 Machinery 0.6 383 Electric machinery 0.95 384 Transportation equipment 0.36 385 Professional goods 0.96 390 Other industries 0.47 Source: Klingebiel, Kroszner, and Laeven (2005). External dependence is defined as capital expenditure minus cash flow, divided by capital expenditure, and is constructed based on US firm-level data. 28 and (6) 96 2.796 (5.964) 0.678** (0.276) (2.645) 0.63 values 16.901*** -9.226** (4.577) 33.031* (16.820) -7.613 (12.369) -11.571 (20.545) (28.143) -124.197*** GDP; Dependent of 1%; GDP; average at (5) 96 sharea of Scandinavian, are 2.785 (6.308) 0.939*** (0.322) (2.441) 0.56 18.019*** (26.517) as -143.515*** sharea significant as German, variables *** the institutions (4) 96 imports of 11.043 French, (7.338) 1.074*** (0.363) -7.674 (6.456) 15.9 50.243** (19.202) (11.601) -7.909 0.44 (17.965) -27.199 (27.862) 5%; All at and financial Tables; (1998). other exports (3) 96 of al. text. 11.499 World (7.176) 1.544*** (0.403) 0.31 significant -40.715 (25.344) and et the ** log 1970-1999 is Penn in Porta 10%; banks at by La from detail by (2) 96 in Averages, 1.596*** (0.389) 3.949 (8.195) 0.28 sector income Results, significant* originally described private Log(Trade/GDP) capita (1) 96 the per 14.511* (8.540) 0.05 -17.246 (34.788) to defined exports; sources of real as and Regression parentheses; credit are in is need errors finance PPP-adjusted dummies definitions Origin Credit/GDP of Credit/GDP Cross-Sectional Origin Origin log Legal Origin standard Origin external is Variable Private OLS Legal Private the Legal 2: Legal Legal Robust is Var.: 1970-99. Table Dep. Log(Trade/GDP) EFNX Log(Income) French German Scandinavian Socialist Constant Observations R-squared Notes: variable, EFNX Log(Income) Socialist over 29 and external export (5) 96 1.315** (0.567) 2.466 (5.407) (2.535) 15.063*** -8.839* (5.039) 23.016 (17.703) -8.524 (11.876) -13.756 (19.851) (28.812) sources 1.047*** (0.212) GDP; -121.617*** Dependent of and 1%; GDP; predicted predicted at sharea of Scandinavian, the as is (4) 2.93 96 sharea definitions 1.476** (0.592) (5.914) (2.718) generating 15.847*** (25.783) 1.038*** (0.175) -138.177*** significant as German, EFNX then *** institutions Variable Stage Stage imports and French, 5%; Predicted at and (3) Second First 96 (0.690) 8.496 financial 1970-99. Tables; A: 2.307*** (6.470) -7.221 (7.231) 25.578 (18.665) 8.704 (11.882) -13.27 (17.112) -43.738 (30.971) B: 1.104*** (0.216) (1998). other exports regressions, over al. Panel Panel significant of and World et ** log gravity values 1970-1999 is Penn Porta es,g (2) 96 9.649 10%; banks La 2.493*** (0.608) (6.621) -55.130** (27.507) 1.181*** (0.189) at by from by average Avera sector-level are sector income Results, significant* private Log(Trade/GDP) originally variables (1) 96 capita estimating 2.587*** (0.629) -18.81 (14.303) 1.188*** (0.184) the per the to ressiong exports; defined first of of real as by All Re parentheses; credit are in is need sector. errors text. finance PPP-adjusted dummies constructed and Origin Credit/GDP of the in Credit/GDP Cross-Sectional Origin Origin log exports country Legal Origin standard Origin external is of detail Private EFNX 2SLS Legal Legal EFNX Private the Legal each in 3: Legal Robust is need Var.: Var.: for Table Dep. EFNX Log(Trade/GDP) Log(Income) French German Scandinavian Socialist Constant Dep. Predicted Observations Notes: variable, EFNX Log(Income) Socialist finance shares described 30 is in (7) 529 96 finance country 0.801** (0.370) -6.062 (4.088) (5.964) (0.261) 2SLS yes yes EFNX detail 33.506*** 0.825*** 1%; at PPP-adjusted each in Stage nt GDP; of external Stage orf ed of log is describ (6) Second First 96 yes yes (0.500) -0.535 (5.049) (0.253) 2SLS 529 significa shares A: 1.084** B: sharea predicted 0.862*** *** as the is export sources Panel Panel 5%; and at Log(Income) (5) 547 96 EFNX yes yes institutions predicted 1.142** (0.530) 0.843*** (0.255) 2SLS GDP; of definitions significant ** financial Predicted sharea generating 10%; other as Variable then at (4) OLS 529 96 and 0.9 yes yes regressions, and 0.540*** (0.160) -5.695 (4.024) (5.162) 35.386*** imports 1995-99. banks ... significant* by and Variables regressions, (3) 0.443 529 96 yes yes sector exports 0.730*** (0.236) (4.959) OLS 0.88 1975-97, of 1970-1999 es,g private log gravity parentheses; is Instrumental the 1970-74; (2) OLS 547 96 0.87 yes yes in the Avera 0.739*** (0.238) to are In sector-level level credit averages is Tables; Five-Year (1) 2.459 (5.941) OLS 529 96 0.87 yes yes Log(Trade/GDP) country estimating at World five-year first are Results, Credit/GDP exports; Penn by of clustered from ressiong Private need variables Credit/GDP Re errors the income constructed of Private EFNX finance Panel EFNX variable, All FE 4: capita exports Var.: Var.: FE Standard external per of sector. text. Table Dep. EFNX Log(Trade/GDP) Log(Income) Dep. Predicted Estimation Observations Countries R-Squared Country Time Notes: Dependent the real need and the 31 at East Penn the Africa 1995-99. (6) 2SLS 85 2SLS 479 85 yes yes is each ... Middle 1.518** (0.695) 0.776** (0.356) significant from external for no /North *** the EFNX is income shares 5%; 1975-97, at EFNX capita Predicted export per 1970-74; (5) 72 Africa 395 72 yes yes Sub-Saharan 1.569*** (0.590) 2SLS GDP; 1.304** (0.584) 2SLS significant of real (1998). no ** al. predictedg averages Dummies sharea et as Porta PPP-adjusted generatin La five-year America Origin parentheses; of (4) by then are Estimates 1.568** (0.779) 2SLS 75 0.820** (0.385) 2SLS 403 75 yes yes in log Latin /Caribbean Legal are institutions is and no Estimates Log(Income) level originally variables Panel financial the Tigers B: regressions, of Cross-Sectional Log(Income), country Log(Income) defined other all (3) Asian A: at as B: 1.370** (0.612) 2SLS 91 Panel 0.843** (0.417) 2SLS 499 91 yes yes and Log(Trade/GDP), GDP; are gravity East Panel of Panel no clustered banks by sharea dummies Log(Trade/GDP), errors as sector-level 1970-99. sector Origin (2) OECD 2SLS 74 2SLS 399 74 yes yes over no 1.133** (0.518) 0.535** (0.265) standard imports private B: Legal estimating the and values first text. to Panel by the exports Socialist average in credit of is and are (1) outliers 90 494 90 detail yes yes log no 2.078** (0.874) 2SLS (0.496) 2SLS is constructed in 1.458*** parentheses; in variables Subsamples and errors Credit/GDP, exports the Scandinavian, of of described all Private need A: Outliers standard Log(Trade/GDP) German, sources Credit/GDP finance Panel and robust variable, exports; French, A: of Private sector. Robustness: FE external 5: need and definitions Var.: FE Panel Tables; Dependent Table Sample Dep. EFNX Controls Estimation: Observations EFNX Controls Estimation: Observations Countries Country Time Notes: 1%; finance World predicted country Variable 32 Margin by the exports; capita sector. 1970- defined (4) 2SLS 104 in to of per as and Interest -0.0015** (0.0007) are need real are Net credit constructed averages level is country finance dummies exports each five-year country of for are (3) Market PPP-adjusted Estimates 2SLS 79 0.068 2SLS 274 79 yes yes at external of 0.020*** (0.006) (0.085) Credit/GDP, Origin need shares Stock the log text. Turnover/GDP Dummies Estimates is is clustered Legal variables Private finance export the the in Origin Panel EFNX of B: errors Socialist all detail Cross-Sectional Legal variable, external predicted in (2) Market GDP; B: A: 2SLS 79 Log(Income) Panel 1.067 285 79 yes yes and 0.735*** (0.266) (1.197) 2SLS of standard Stock Value/GDP Panel Panel B: Dependent sharea GDP; predicted of the generating described Development Log(Income), Log(Income) Panel as 1%; is at sharea Scandinavian, then 1970-99. sources as and EFNX over and Financial (1) 115 537 115 yes yes of M2/GDP 1.904*** (0.680) 2SLS 0.549** (0.278) 2SLS institutions parentheses; German, in significant imports values Log(Trade/GDP), Log(Trade/GDP), *** Predicted regressions, errors financial and French, definitions average Measures 5%; at other exports (1998). gravity are Tables; al. Variable Other standard and of et log World variables robust significant banks is Porta sector-level 1995-99. A: ** by the Penn La Variable of ... Robustness: by FE all 6: FE Panel sector from A: estimating 1975-97, Table Dependent EFNX Controls Estimation: Observations EFNX Controls Estimation: Observations Countries Country Time Notes: parentheses; private Log(Trade/GDP) income originally first Panel 74; 33 Appendix Figure A1: Actual and Predicted External Finance Need of Exports, by Country and Time Period Algeria Argentina Australia 20 25 25 20 20 0 15 15 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Austria Bahrain, Kingdom of Bangladesh 40 40 40 20 20 20 0 0 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Belgium Bolivia Brazil 40 40 30 20 25 20 0 20 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Burkina Faso Burundi Cameroon 0 25 10 30 20 20 15 0 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Canada Central African Rep. Chile 40 40 40 30 20 20 20 0 0 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 China,P.R.: Mainland China,P.R.:Hong Kong Colombia 40 60 30 30 40 25 20 20 20 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Legend: : Actual EFNX; : Predicted EFNX 34 Congo, Republic of Costa Rica Côte d'Ivoire 40 40 30 30 20 20 20 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Denmark Dominican Republic Ecuador 40 30 40 20 20 20 0 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Egypt El Salvador Ethiopia 40 25 40 20 20 20 15 0 0 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Finland France Gabon 30 40 40 30 20 20 20 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Gambia, The Germany Ghana 50 30 40 20 0 20 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Greece Guatemala Haiti 40 30 30 30 20 20 20 10 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Legend: : Actual EFNX; : Predicted EFNX 35 Honduras Hungary Iceland 30 30 40 20 20 30 10 20 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 India Indonesia Iran, I.R. of 40 30 30 30 20 20 20 10 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Ireland Israel Italy 60 60 40 40 40 20 20 20 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Jamaica Japan Jordan 40 60 40 30 20 40 20 0 20 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Kenya Korea Kuwait 30 60 30 20 40 20 10 10 20 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Madagascar Malawi Malaysia 25 50 60 20 0 40 20 0 15 -5 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Legend: : Actual EFNX; : Predicted EFNX 36 Mexico Morocco Nepal 60 30 30 40 20 20 20 10 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Netherlands New Zealand Nicaragua 50 30 30 20 20 0 10 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Niger Nigeria Norway 30 30 30 20 20 25 20 10 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Pakistan Panama Papua New Guinea 30 30 40 20 30 20 10 20 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Paraguay Peru Philippines 30 60 25 20 40 20 10 20 15 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Poland Portugal Romania 40 40 40 20 20 20 0 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Legend: : Actual EFNX; : Predicted EFNX 37 Rwanda Saudi Arabia Senegal 50 30 30 20 20 10 0 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Sierra Leone Singapore 40 South Africa 60 60 20 40 40 20 20 0 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Spain Sri Lanka Sudan 40 40 25 20 20 20 0 15 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Sweden Switzerland Syrian Arab Republic 60 60 30 40 40 20 20 20 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Thailand Togo Trinidad and Tobago 50 30 30 20 20 10 0 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Tunisia Turkey Uganda 40 30 30 30 20 20 20 10 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Legend: : Actual EFNX; : Predicted EFNX 38 United Kingdom United States Uruguay 60 30 40 40 20 20 20 10 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Venezuela, Rep. Bol. Zambia Zimbabwe 40 20 40 20 20 0 0 0 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 1969 1974 1979 1984 1989 1994 1999 Legend: : Actual EFNX; : Predicted EFNX 39 page 197.27 97.36 314.94 18.81 72.92 76.12 28.06 61.61 275.38 151.40 23.95 71.86 94.51 77.16 56.37 30.51 109.93 91.34 42.90 68.31 47.08 83.25 43.30 67.93 38.73 48.76 141.73 77.67 51.18 80.26 130.04 91.65 68.23 54.80 119.81 72.41 next Trade/GDP cont'd Private 138.81 49.11 118.93 199.44 225.56 69.31 17.53 21.00 159.83 77.16 186.59 136.50 145.37 119.59 114.72 8.23 12.10 22.31 2.45 45.65 114.77 97.91 4.22 55.93 96.23 58.47 75.18 81.93 80.57 90.49 8.35 14.01 83.37 19.89 71.33 87.90 Credit/GDP Two of Sectors 0.65 0.67 0.67 0.48 0.33 0.50 0.86 0.53 0.41 0.49 0.39 0.41 0.44 0.36 0.36 0.80 0.97 0.35 0.82 0.35 0.35 0.29 0.92 0.46 0.31 0.29 0.32 0.44 0.38 0.57 0.75 0.43 0.27 0.20 0.69 0.26 Share Largest 1995-1999 Average Sector footwear footwear footwear electrical electrical electrical except electrical electrical electrical electrical except electrical except Export Openness, except except electric except electric except electric electric metals except except except electric electric equipment equipment equipment equipment equipment equipment chemicals products refineries except electric Largest apparel, apparel, apparel, chemicals products products and products chemicals products Trade and Second Machinery, Machinery, Machinery, Machinery, Other Machinery, Textiles Transport Wearing Machinery, Machinery, Transport Machinery, Transport Transport Non-ferrous Food Transport Food Machinery, Machinery, Machinery, Textiles Machinery, Wearing Transport Industrial Paper Machinery, Food Petroleum Machinery, Machinery, Wearing Other Food Development, footwear electrical electrical products products electrical electrical electrical products products products electrical products except Sector Financial electric electric except electric except electric electric except electric except electric except electric electric except metals chemicals equipment equipment products equipment equipment equipment equipment Export apparel, equipment equipment chemicals manufactured manufactured products manufactured manufactured products manufactured and manufactured Exports, of Largest Machinery, Machinery, Machinery, Machinery, Machinery, Other Other Machinery, Machinery, Industrial Machinery, Machinery, Machinery, Machinery, Machinery, Food Other Machinery, Other Food Transport Transport Other Paper Machinery, Machinery, Transport Transport Transport Transport Other Wearing Transport Transport Industrial Non-ferrous Needs EFNX 63.74 63.54 59.50 57.34 55.79 53.56 53.02 52.52 50.79 50.75 48.87 48.85 47.72 47.23 46.78 45.85 45.28 42.94 42.80 42.44 42.39 42.20 42.09 41.55 39.95 37.95 36.48 36.42 36.06 35.69 34.70 34.33 31.42 31.01 30.68 30.34 Finance External A1: Kong Rep. of Table Mainland The States Kingdom Leone African Republic Rica Appendix Country Malaysia Philippines Singapore Japan Switzerland Israel Burundi Mexico China,P.R.:Hong Ireland United Korea Thailand Sweden United Rwanda Gambia, Hungary Sierra Denmark France Austria Central Finland China,P.R.: Italy Belgium Canada Spain Panama Congo, Costa Portugal Poland Jordan Norway 40 page 64.26 40.97 92.49 25.53 40.64 35.57 58.93 18.16 72.73 50.18 75.62 48.17 56.56 81.23 45.77 21.83 40.24 67.17 49.43 58.26 89.02 40.42 61.75 108.04 58.08 47.46 83.08 76.68 41.84 72.33 64.73 36.63 98.31 37.44 36.69 44.53 next Trade/GDP cont'd Private 48.01 4.68 8.39 27.50 75.48 33.67 102.92 32.20 9.30 16.98 18.73 121.95 58.03 18.17 34.54 21.65 19.78 25.68 53.32 34.61 62.84 30.67 8.43 25.62 22.87 7.54 25.03 55.44 9.45 15.67 31.24 22.22 51.79 23.87 14.68 43.08 Credit/GDP Two of Sectors 0.26 0.94 0.58 0.40 0.43 0.28 0.55 0.34 0.44 0.44 0.82 0.34 0.54 0.67 0.36 0.57 0.59 0.53 0.52 0.63 0.57 0.54 0.34 0.56 0.78 0.63 0.66 0.78 1.03 0.82 0.62 0.50 0.69 0.79 0.82 0.54 Share Largest 1995-1999 Average Sector footwear furniture footwear footwear footwear footwear products products products except Export except except except except except Openness, electric metals metals chemicals products equipment refineries refineries refineries refineries refineries Largest apparel, steel chemicals equipment chemicals products steel apparel, chemicals chemicals apparel, apparel, apparel, chemicals manufactured Trade products and manufactured manufactured products, and products and products and Second Machinery, Other Industrial Other Non-ferrous Food Paper Transport Other Wearing Other Iron Industrial Wood Food Transport Leather Textiles Non-ferrous Petroleum Industrial Textiles Iron Food Wearing Textiles Textiles Industrial Petroleum Industrial Wearing Petroleum Petroleum Wearing Wearing Petroleum Development, furniture furniture furniture footwear footwear footwear footwear furniture except except except except Financial Sector except except except except metals chemicals chemicals refineries Export apparel, apparel, apparel, apparel, chemicals products, products, Exports, products products products products, products products products products products products products, products products products of Largest Wood Industrial Wood Textiles Food Industrial Food Food Wood Textiles Textiles Non-ferrous Wearing Food Textiles Food Food Wearing Food Food Wearing Food Wearing Textiles Textiles Wood Food Petroleum Textiles Food Food Textiles Industrial Textiles Textiles Textiles Needs Finance EFNX 30.03 27.48 26.81 26.55 26.38 25.12 24.43 24.30 23.96 23.88 23.26 23.12 22.36 21.96 20.87 20.59 20.26 20.12 19.80 19.79 18.97 18.97 18.80 18.67 18.58 18.14 17.91 17.64 17.47 17.40 17.02 17.00 16.97 16.93 16.53 16.41 External: Tobago Republic (continued)2 Faso of and Zealand Africa d'Ivoire Arabia I.R. Table Country Indonesia Niger Gabon India Australia Colombia New Brazil Ghana Turkey Togo South Morocco Côte Greece Argentina Ethiopia Dominican Bolivia Ecuador Tunisia Uruguay Romania Jamaica Nepal Cameroon Paraguay Saudi Burkina Senegal Kenya Iran, Trinidad Pakistan Haiti Egypt 41 the is 72.05 59.58 66.97 79.54 30.52 58.14 97.85 47.03 113.20 113.72 43.24 33.14 93.87 193.62 78.88 30.37 52.99 72.03 83.65 64.72 72.43 Trade/GDP manufacturing Credit/GDP total the Private Private 55.70 37.49 9.42 26.01 21.46 64.16 28.76 10.60 37.56 171.00 18.00 3.96 52.24 58.56 11.98 27.74 4.59 6.99 30.41 6.92 50.92 in Credit/GDP share country; each highest Two of Sectors from GDP. . 0.88 0.84 0.69 0.69 0.67 0.63 0.81 0.59 0.73 0.86 0.74 0.81 0.90 0.77 0.65 0.89 0.92 0.93 0.51 0.87 of second Share Largest and exports sharea as highest 1995-1999 the imports manufacturing products with plus Average Sector footwear footwear total coal except except sectors the Export and in exports the Openness, metals metals refineries chemicals chemicals are total Largest apparel, apparel, is products steel sectors Trade products products products products products petroleum and two Sectors and Second Non-ferrous Wearing Petroleum Textiles Food Food Wearing Industrial Food Food Food Textiles Misc. Non-ferrous Leather Textiles Industrial Textiles Iron Textiles largest Trade/GDP Export the of GDP. Largest to Development, footwear footwear footwear footwear share Second total except except except except the Financial Sector and institutions is metals metals metals refineries refineries refineries refineries refineries refineries Export apparel, apparel, apparel, apparel, financial Exports, products products Largest Sectors of Two other Largest Food Textiles Textiles Wearing Non-ferrous Non-ferrous Textiles Petroleum Wearing Petroleum Wearing Food Petroleum Petroleum Petroleum Wearing Petroleum Non-ferrous Tobacco Tobacco exports. and Needs of Largest need the banks Finance EFNX 15.43 15.38 14.43 14.41 13.74 12.16 12.12 11.65 11.16 10.11 10.08 9.78 9.56 9.45 9.38 8.50 7.37 3.64 -6.74 -31.34 27.00 of by finance Share sector External: of external private country. Bol. the the is Republic each to Rep. (continued)2 Kingdom Arab Average EFNX from credit of Salvador Lanka Table Country Iceland El Syrian Sri Peru Chile Honduras Venezuela, Nicaragua Netherlands Guatemala Uganda Kuwait Bahrain, Nigeria Bangladesh Algeria Zambia Zimbabwe Malawi Sample Notes: exports ratio 42