WPS6229 Policy Research Working Paper 6229 Exporter Dynamics Database Tolga Cebeci Ana M. Fernandes Caroline Freund Martha Denisse Pierola The World Bank Development Research Group Trade and Integration Team October 2012 Policy Research Working Paper 6229 Abstract This paper introduces the Exporter Dynamics Database. rates are highly correlated with entry rates and both are The database includes exporter characteristics and negatively correlated with survival rates, average exporter measures of exporter growth based on firm-level customs size, and diversification. (iv) The number of exporters information from 38 developing and seven developed and the entry and exit rates in a country-product countries, primarily for the period between 2003 and group are partially driven by country and product- 2010. The measures are available at different levels of group effects; however, the average size of exporters in a aggregation, including: a) country-year, b) country-year- country-product group is not. Although the first three product, and c) country-year-destination. Several new facts can be explained by models incorporating firm stylized facts about exporter behavior across countries heterogeneity and uncertainty, the fourth fact is more emerge from the database. (i) Larger or more developed difficult to explain with existing models. Several findings economies have more exporters, larger and more are confirmed in this database, including the importance diversified exporters, and lower entry and exit rates than of large multi-product firms. This database can be a smaller or developing economies. (ii) In the short run, valuable tool to improve the understanding of the micro- expansions along the intensive margin (exporter size) foundations of export growth, by providing new insights contribute more to export growth than expansions along about exporter characteristics and dynamics. the extensive margin (number of exporters). (iii) Exit This paper is a product of the Trade and Integration Team, Development Research Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at tcebeci@worldbank.org, afernandes@worldbank.org, cfreund@worldbank.org and mpierola@worldbank.org. 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 views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Exporter Dynamics Database * Tolga Cebeci 1 Ana M. Fernandes 2 Caroline Freund 3 Martha Denisse Pierola 4 JEL Classification codes: C81, F14. Keywords: exporter dynamics, exporter growth, firm-level data. * This database was conceived and initiated in 2009 by Caroline Freund and Martha Denisse Pierola with support of the governments of Norway, Sweden and the United Kingdom through the Multi-Donor Trust Fund for Trade and Development. We also acknowledge the generous financial support from the World Bank research support budget and the Knowledge for Change Program (KCP), a trust funded partnership in support of research and data collection on poverty reduction and sustainable development housed in the office of the Chief Economist of the World Bank (www.worldbank.org/kcp). We are grateful to Mario Gutierrez-Rocha and Jamal Haidar for their contribution to this project, and to Aaditya Mattoo for useful comments and discussions. We thank Alberto Behar and Madina Kukenova for collaboration with the project early on and to Francis Aidoo, Paul Brenton, Maurizio Bussolo, Huot Chea, Julian Clarke, Alain D’Hoore, Pablo Fajnzylber, Elisa Gamberoni, Sidiki Guindo, Javier Illescas, Mehar Khan, Sanjay Kathuria, Josaphat Kweka, Erjon Luci, Mariem Malouche, Eric Manes, Yira Mascaro, Mamadou Ndione, Zeinab Partow, Stefano Paternostro, Jorge Rachid, Richard Record, Jose Guilherme Reis, Jose Daniel Reyes, Nadeem Rizwan, Liviu Stirbat, Dimitri Stoelinga, Daria Taglioni, Maria Nicolas, Erik von Uexkull, Cristian Ugarte, and Ravindra Yatawara for facilitating the access to the data for developing countries. We thank Lina Ahlin, Martin Andersson, Juan de Lucio, Emmanuel Dhyne, Luc Dresse, Cédric Duprez, Richard Fabling, Jaan Masso, Raul Minguez, Asier Minondo, Andreas Moxnes, Francisco Requena, Lynda Sanderson, Joana Silva, Priit Vahter, and Hylke Vandenbussche for providing us with measures for developed countries. The findings expressed in this paper are those of the authors and do not necessarily represent the views of the World Bank. 1 Consultant, Trade and International Integration Unit, Development Research Group, World Bank (tcebeci@worldbank.org). 2 Senior Economist, Trade and International Integration Unit, Development Research Group, World Bank (afernandes@worldbank.org). 3 Chief Economist, Middle East and North Africa Region, World Bank and CEPR (cfreund@worldbank.org). 4 Economist, Trade and International Integration Unit, Development Research Group, World Bank (mpierola@worldbank.org). I. Introduction A large and growing strand of the recent trade literature - both theoretical and empirical - focuses on firms and how they export. Since the seminal work of Eaton and Kortum (2002) and Melitz (2003), a number of papers have developed models to understand the micro foundations of export growth. 5 On the empirical front, issues related to exporter behavior and dynamics in general have been heavily studied. 6 These studies have in common the use of micro datasets on export transactions within firms from a specific country or in a region. While the conclusions from many of the studies should in principle be comparable across countries, to date there has not been a comprehensive effort to produce analytical work covering an extensive set of countries, particularly developing countries. It would be of interest to see if for example, the empirical findings identified in this growing literature generalize across countries in different regions and at different levels of development, and whether cross-country evidence could provide new insights about how firms export and expand. To fill this gap, we gather exporter-level customs information from 38 developing and 7 developed countries around the world to build the “Exporter Dynamics Database� (henceforth referred to as the Database), containing comparable measures of exporter, product and market dynamics and which is available to researchers and policy-makers worldwide. The Database contains measures at different levels of disaggregation: a) country-year, b) country-year-product (HS 2-digit, HS 4-digit, or HS 6-digit) and c) country-year-destination mostly for the period between 2003 and 2010 (with longer time series for some countries). The Database covers 5 See Bernard, Eaton, Jensen, and Kortum (2003), Bernard, Redding, and Schott (2007), Das, Robert, and Tybout (2007), Chaney (2008), Arkolakis (2010), Bernard, Redding, and Schott (2011), Eaton, Kortum and Sotelo (2012) and Redding (2011) for a review. 6 See Eaton, Kortum, and Kramarz (2008) for France; Eaton, Eslava, Kugler, and Tybout (2008) for Colombia; Amador and Opromolla (2008) for Portugal; Iacovone and Javorcik (2008) for Mexico; Andersson, Lööf, and Johansson (2008) for Sweden, Freund and Pierola (2010) for Peru; Manova and Zhang (2012) for China, Masso and Vahter (2011) for Estonia, De Lucio, Mínguez-Fuentes, Minondo, and Requena-Silvente (2011) for Spain, Ekholm, Moxnes, and Ulltveit-Moe (2012) for Norway, Fabling and Sanderson (2012) for New Zealand, among others. See Bernard, Jensen, Redding, and Schott (2007, 2011) for reviews of the literature. 2 different aspects of firm dynamics, firm-product and firm-destination dynamics, as well as exporter growth patterns, concentration, and diversification in the non-oil exporting sector. This paper introduces the Database and uncovers four new stylized facts based on the measures contained therein. The distinguishing feature of these stylized facts is that they could not have been uncovered using any other cross-country source of trade data available so far. We also confirm or generalize facts found in the recent trade literature. The first stylized fact is that more developed and larger economies have a larger export base (number of exporters), larger exporter size, more concentrated export sectors among firms, more diversified exporters (in terms of their portfolio of products and destinations), and lower entry and exit rates, relative to less developed and smaller economies. The survival rates of new exporters are not correlated with the level of development or with the economic size of countries. These patterns are also true for countries with larger export sectors relative to income. This is consistent with more developed and larger economies having more competitive export sectors, with more firms that are globally competitive, and a higher proportion of large firms. The second stylized fact is that across countries, expansions along the intensive margin – i.e., increases in the average exporter size – contribute more to export growth in the short run than expansions along the extensive margin – i.e., increases in the number of exporters. This evidence supports a trade model with heterogeneous firms along the lines of Melitz (2003), where entrants into export markets tend to be marginal firms that have little impact on total exports. The third stylized fact is that across countries, entry and exit rates are high and strongly positively correlated with each other. Both entry and exit rates are negatively correlated with survival rates of new exporters, average exporter size, and diversification at the exporter level (in 3 terms of products and destinations). These results reflect the importance of uncertainty and market structure in the decision to export. The fourth stylized fact is that average exporter size, total exports, and exporter survival at the country-sector level are not explained by country or sector characteristics in an analysis of variance (ANOVA) while the number of exporters, the share of the top 5 percent of exporters, the average number of products and destinations per exporter and entry and exit rates are explained to an important extent by those characteristics. This fact is not as clearly consistent with standard heterogeneous firm models in that the number of firms is explained to a significant degree by country and industry effects, while the average and median size of firms are not. Considering that average size and the number of firms together drive export volumes and both respond to trade costs in similar ways in a heterogeneous firm model with comparative advantage (Bernard, Redding, and Schott 2007), it is puzzling to find that the ANOVA shows that country and sector characteristics explain far better the number of firms than the average or median size of firms. This finding could imply that comparative advantage works primarily via firm size as opposed to the number of firms. 7 The fifth stylized fact confirms for a large number of developing countries the well- known fact from the literature focusing on individual countries that total exports are largely dominated by multi-product multi-destinations exporters, but these account for a small share of the number of exporters. In particular, exporters selling more than four products to more than four markets account for 60 percent of exports on average. But there is also significant variation, with Albania displaying the lowest share in at 13 percent and South Africa the highest, at 82 percent. The tremendous skewness of exports towards large firms has important implications (i) 7 Freund and Pierola (2012) explore why export size behaves differently from the number of firms and show that it is related to the skewed distribution of exporter size. In particular, they explore the importance of superstar firms (top one percent of exporters) in exports, export growth and diversification, and comparative advantage, as well as the origin of these firms 4 for empirical firm-level studies that focus on average effects since the average firm is relatively unimportant in trade, and (ii) for expanding exports because that would entail growth in the number and size of large firms. The sixth stylized fact is consistent with previous literature that shows that bilateral exports increase with the size of the destination market and decrease with distance and with bilateral tariffs. Similar to Bernard et al (2007) for US firms and Mayer and Ottaviano (2008) on European firms, we find that a country’s exports expand in larger markets primarily through the firm-count margin as opposed to the firm-size margin; and decline relatively more in distance and in tariffs because of the firm-count margin. Although this finding could be perceived as standing in contrast to the second stylized fact above that expansion in the average exporter size makes the major contribution to within-country export growth, it is worth noting that these facts emerge from two exercises focusing on different time dimensions. While the second stylized fact indicates that firm size drives export growth more than the number of firms in the short run, the sixth stylized fact indicates that the number of exporters in a given destination responds more sharply to standard determinants of trade such as market size and trade costs in the long run. Given the extreme concentration of exports in large firms (the top 5 percent of firms account on average for 80 percent of exports), average firm size will expand rapidly in countries when these large firms grow, and it is primarily these firms that can generate high aggregate export growth, since the majority of firms are too small to have sizeable aggregate effects (the bottom 75 percent of exporters account for just 3 percent of exports). However, in the long run, the number of firms expands when exports grow because of home- or foreign-market size or trade costs effects. The higher number of firms in a given destination drives down considerably the average 5 firm-destination size by definition, and thus obscures the firm-size effect on exports in a gravity equation. Several of the stylized facts uncovered using the Database are broadly consistent with recent trade models with heterogeneous firms such as those reviewed in Bernard, Jensen, Redding, and Schott (2011) and Redding (2011), while other facts point to the need to extend and modify the existing models to accommodate them. In particular, with respect to dynamics, the high and correlated entry and exit rates point to tremendous uncertainty when firms enter export markets. The rest of the paper is structured as follows. Section II describes in detail the construction of the Database based on customs exporter-level data in each country while Section III lists, defines and presents summary statistics on the measures included in the Database. Section IV presents six stylized facts based on the Database. Section V discusses possible avenues for future research and policy analysis using the Database. II. Constructing the Export Dynamics Database Using Customs Data at the Exporter-Level The measures included in the Database are computed using customs data from 45 countries at the exporter-product-destination-year level. 8 Pooling across the datasets for all countries, we obtain 15 million unique observations at the country-firm-product-destination-year level 9 This is the raw dataset that we use to construct the Database. 8 The providers of the raw datasets for each country were mostly governmental agencies, mainly customs offices. Appendix 1 provides a complete list of the countries included in the Database, the periods for which data is available, and the data sources. For Brazil, Egypt, Estonia, Laos, New Zealand, Norway, Portugal, Spain, Sweden, and Turkey we calculated the measures without having permanent access to the raw customs data at the exporter-product-destination-year level. A few of the data providers authorize the access by researchers outside the World Bank to the raw customs data at the exporter-product-destination- year level. The list of countries for which such data access is authorized is available (and subject to updates) at http://econ.worldbank.org/exporter-dynamics-database. 9 This number of observations covers the countries for which we have access to the raw exporter-level customs data (i.e., it does not account for the observations for Brazil, Egypt, Estonia, Laos, New Zealand, Norway, Portugal, Spain, Sweden, and Turkey). 6 1. Cleaning of Raw Datasets The cross-country raw dataset contain six variables -country of origin, firm codes, country of destination, product, value, and year- that have been reformatted uniformly and subjected to the cleaning procedures explained below. 10 a) Country of origin: For each observation in the cross-country raw dataset, we define the country of origin as the country from which we received raw data from. It should be noted that Depending on the type of trade regime (general versus special) used by the country of origin, exports of a country might include goods originating from any geographical territory of that country or from any territory except free zones/customs. The implication of the trading regime in our cross-country raw database is discussed in more detail in the destination section c) below. b) Firm codes: Firm codes are the numbers that uniquely identify firms across time within each country. We received this information in different formats: a) the actual names of the firms, b) their tax identification number or c) artificial unique codes randomly created by our data providers. One peculiarity of the datasets for Albania, Burkina Faso, Cambodia, Cameroon, Mexico, Uganda, and Yemen is that the firm coding systems changed during the sample period. Hence, a specific firm is represented by different codes before and after 2007 in these countries’ datasets (2008 in Yemen’s dataset). As a result, it is not possible to calculate exporter dynamic measures such as firm entry and exit rates in 2007 and survival rates in 2006 and 2007 for those countries (firm entry and exit rates in 2008 and survival rates in 2007 and 2008 in Yemen). 10 The cross-country raw dataset at country-firm-product-destination-year level includes also quantities exported that will be used to calculate some of the measures in the Database. 7 c) Destination: Destination countries in our cross-country raw dataset follow UN’s guidelines (International Merchandise Trade Measures: Concepts and Definitions, p. 60) which recommend that countries of origin record their destinations as defined by the destination countries themselves. For instance, Bermuda (UK); Hong Kong SAR, China; and Macau (China) are all considered distinct destinations although they are not independent countries. Therefore, each country has a potential set of 247 destinations as of end-2011. 11 The first cleaning operation applied to the country of destination variable relates to the use of special trade regimes by some countries –Bulgaria, Costa Rica, Colombia, Ecuador, Egypt, Jordan, Kuwait, Lebanon, Morocco, Peru, Turkey, Yemen. 12 Customs in these countries record the sales from inland to their own free zones/customs warehouses as exports which results in a larger set of potential destinations. For example, in Colombia’s dataset, the Free Zone “Zona Franca Bogota� appears as a separate destination. Since there is a lack of uniformity across countries in the definition of the special trade regime, we do not consider this type of transactions as exports and hence we drop the observations related to sales to free zones from our dataset. 13 This operation had a minor effect on total export volumes as sales to free zones/customs warehouses are negligible in most cases. 14 The second operation applied to the country of destination variable accounts for the changes in name that some statistical territories have undergone over time due to spatial 11 See http://unstats.un.org/unsd/methods/m49/m49alpha.htm for the current potential set of destinations. 12 The “special trade regime� considers transactions where the goods are sold from the domestic territory only to both third countries and free zones/customs warehouses of the origin country as exports. In contrast, the “General trade regime� considers transactions where goods are sold from any national territory (including free zones) to third countries only as exports (see p. 32 of United Nations, 2008). 13 See p. 34 of United Nations (1998). 14 The export volumes of Turkey, the country most affected by this operation, drop by 2-3 percent. 8 divisions. In particular, the Former Republic of Yugoslavia was divided into Bosnia, Croatia and Serbia in 1996 and Serbia was further divided into Serbia and Montenegro in 2006. Furthermore, some countries recognize Kosovo as an independent state rather than a part of Serbia since 2009. For technical and consistency purposes we treat Serbia, Montenegro and Kosovo as a single destination since disregarding these changes in names would bias some calculations, especially those related to the geographical diversification and market dynamics. 15 For example, destination entry and exit rates in a given year would be overestimated as the appearance of a new destination in the dataset would not represent the true expansion of exporters into a different territory but would instead be due to the change in territory classification. Likewise, destination measures that are calculated using only one year of data (e.g., number of destinations, export volume per destination, etc.) would not be comparable to each other. 16 In sum, after taking into account the operations mentioned above, the number of destinations in our dataset is 246. 17 d) Product: The product classification system we use is the Harmonized System at 6-digit level (HS6). Although most countries record their export transactions at a higher level of disaggregation following their domestic or regional nomenclature (8-12 digits), we aggregate their information to HS 6-digit, since this is the most detailed level comparable internationally. That is, a specific HS code at 6 digits represents the same product in all country datasets in a given year. The cleaning operations applied to the product variable have two components: 15 In the Database at the country-year-destination level the combined destination of Serbia, Montenegro, and Kosovo is coded as the 3-digit country code SRB. 16 While some other territories have undergone changes in names, those changes either occurred before the beginning of our sample or do not involve the merger or separation of states (e.g., Zaire changed its name to Democratic Republic of Congo in 2006) and therefore they do not introduce any biases in our calculations. 17 The list of destinations is available for download at http://econ.worldbank.org/exporter-dynamics-database. 9 i) Elimination of non-existing HS codes: First, we combine all the codes existing under the three different HS classifications (HS1996, HS2002 and HS2007) and end up with a unique aggregated list of 6065 HS 6-digit categories. 18 Next, we eliminate the observations with an HS 6-digit code that is not included in that aggregated list. This elimination accounts for about 0.5 percent of total exports in our cross-country raw dataset. 19 ii) Creation of a time-consistent HS classification: While the HS 6-digit classification allows comparisons across countries in a given year, it has undergone transformations over time. The World Customs Organization (WCO) revises the HS classification on the basis of the value of trade realized under each product during the previous period. Three major revisions took place in years 1996, 2002, and 2007. 20 The modifications introduced in each of these revisions have taken two forms: i) two different codes with low trade volume were converted into a single code and ii) an existing code with an increasing trade volume was split into various codes. For example, code 030269 (other fish, fresh or chilled, excluding fish fillets or other fish meat) which included swordfish and toothfish in the HS2002 classification was split into codes 030267 (swordfish), 030268 (toothfish), and 030269 (for other fish) in the HS2007 classification. These modifications create problems for the tracking of trade volumes for certain products over time. In the example above, exports of swordfish under code 030267 would appear as a new export from 2007 onwards, while in reality they might have already been exported before but were recorded under the code 030269. 18 The number of HS categories included in the original classifications HS 1996, HS 2002, and HS 2007 are 5209, 5224, and 5053, respectively. 19 Most of this elimination results from eliminating observations with a product code belonging to HS Chapter 99 as this is reserved for national use and the HS 6-digit codes under this chapter differ across countries. 20 In addition to these major transformations, there are also smaller modifications introduced at the end of every year. 10 In order to solve these inconsistencies, we went through a process of “consolidation� among HS1996, HS2002, and HS2007 classifications. 21 A similar process was used by Schott and Pierce (2012) to concord 10-digit United States Harmonized System codes between 1989 and 2007 and by Wagner and Zahler (2011) to homologate among 6-digit HS1992, HS1996, and HS2002 classifications. The basic principle of consolidation is to identify the HS codes related to each other (e.g., codes that were split or merged with the modifications introduced by the HS2002 or the HS2007) and to replace them with a single code for the entire period. In the example above, this process results in the replacement of codes 030267 and 030268 by the code 030269 from year 2007 onwards. In this way, the products that are represented by these three codes are all included in code 030269 during the entire period. As a result of this consolidation process 1104 codes are replaced by 402 codes that already exist in the HS lists but whose contents are altered as discussed in Cebeci (2012). Consequently, the number of unique potential HS 6-digit codes in the 1996, 2002, and 2007 classifications of 6065 declines to a final number of 4961 unique potential HS 6- digit codes in the consolidated classification. 22 In the cross-country raw dataset 4767 of those 4961 codes are present. e) Exclusion of oil exports: We eliminate from the cross-country raw dataset observations in HS Chapter 27 (hydrocarbons such as oil, petroleum, natural gas, coal etc.) given that 21 See Cebeci (2012) for the methodology used in the consolidation. The paper along with a list of consolidated codes and concordances are available at http://econ.worldbank.org/exporter-dynamics-database. 22 The number of HS 6-digit codes not affected by the consolidation process is 4559, obtained as the total number of codes 6065 minus the 1104 codes that disappeared and the 402 codes whose content changed. Despite this consolidation of HS codes, for the countries for which the data providers allow the sharing of the raw data, the country input database with the original HS 6-digit classifications is also available upon request. 11 we do not have exporter-level data on that chapter for important oil exporting countries such as Burkina Faso, Cameroon, Iran, Kuwait, or Yemen. 23 f) Value: The unit of measurement of export values in the cross-country raw dataset is the US Dollar (USD). The export values in the raw data were already in USD for many countries. For countries for which export values were provided in local currency, they were converted to USD using an annual official exchange rate series. 24 The exchange rates used differ slightly from those used in the United Nations COMTRADE database that are weighted average annual exchange rates based on monthly exchange rates and monthly trade volumes (as weights). 25 The resulting difference in the exchange rate series creates a small discrepancy (less than 0.5 percent in all cases) between our figures and COMTRADE’s figures for the countries that provided their export values in local currency. Export values in our dataset are Free on Board (FOB) figures, except for El Salvador and Senegal, whose export values represent Cost, Insurance and Freight (CIF) figures rather than just the pure value (FOB) of the good. This difference should be taken into account for cross-country comparisons of measures related to the size of exporters, exports per product or destination but is not expected to affect other measures related to concentration, diversification and firm, product and market dynamics. g) Period: The information in the cross-country raw dataset has a yearly periodicity. 23 Possible reasons for this lack of data are confidentiality reasons or the fact that goods exported through pipeline are not recorded at customs but instead are recorded by other government/private institutions in the countries. 24 The specific series used for this conversion is PA.NUS.FCRF taken from the World Development Indicators (whose original source is the IMF’s International Financial Statistics). The series is based on an annual average of daily exchange rates determined by national authorities or in the legally sanctioned exchange market. 25 http://unstats.un.org/unsd/tradekb/Knowledgebase/Calculation-of-dollar-value-in-trade-statistics-Current-value-or-constant- dollar-value. 12 2. Accuracy of the Cross-Country Raw Dataset In order to have a sense of the reliability of the raw data in each of the countries included in the Database, we apply two filters: a) For the first filter we compare the total values exported (excluding HS 27) calculated from the cross-country raw dataset with the total values exported from the United Nations’ COMTRADE database (excluding HS 27) for every country and year. 26 This comparison yields quite different results for different countries. On the one hand, for Albania, Brazil, Cameroon, Chile, Colombia, Costa Rica, Ecuador, Guatemala, Kenya, Lebanon Mexico, Morocco, Pakistan, Peru, South Africa, Tanzania and Turkey, the ratio of total values exported in the raw dataset to total values exported in COMTRADE is about 100 percent. On the other hand, for Mali and Yemen, the ratios indicate that total values exported in the raw dataset are as low as half of the total values exported in COMTRADE. On the opposite end, for Mauritius total values exported in the raw dataset are on average 30 percent above total values exported in COMTRADE.27 Appendix 2 provides detailed results on these comparisons. b) For the second filter, we focus on the countries and years that would have been left out because of a unfavorable match with COMTRADE (below 60 percent) and we keep those countries (and years) where we observe internal consistency within the export totals calculated from the corresponding exporter-level raw dataset over time. Although both sources of trade data, COMTRADE as well as the exporter-level raw datasets that we use, originate from customs authorities, we are aware of potential difficulties in the processing of the exporter-level 26 Given our understanding about which transactions are included in the raw files, for this comparison we consider gross export figures in COMTRADE (which include re-exports) for Albania, Kenya, Mauritius, Tanzania and Senegal and net export figures (gross exports minus re-exports) for Jordan, Uganda and Yemen. For all other countries, it does not matter which COMTRADE figure we choose as they report either insignificant or no re-exports. 27 For Mauritius we examined the discrepancy further at the country-HS 2-digit-year level and found that the average match percentage benefited significantly from the exclusion of observations belonging to Chapter 49. Similarly, for Cambodia and Macedonia the average match percentage benefits significantly from the exclusion of observations belonging, respectively to Chapter 49 and Chapter 62. 13 raw datasets that may justify the differences sometimes observed between the export totals obtained from the exporter-level raw datasets and the export totals available in COMTRADE. One of the potential reasons for these difficulties relates to the manual registration of export transactions that still takes place in some countries and may result in under-recording of export transactions for some countries and years. As a result of this second filter, we keep in the raw dataset the information for Macedonia, Jordan, Mali, Mauritius, and several years of El Salvador’s data which would be left out due to an unfavorable match with COMTRADE data. We should also note that for countries such as Macedonia for which the match with export totals in COMTRADE in the first few sample years is relatively poor, the match improves over time and this is likely due to a shift from manual to digital registration of export transactions. III. The Exporter Dynamics Database The Database includes a series of measures classified under different categories reflecting basic characteristics of the export base in each country (size of the exporting sector, exporter size and exporter growth rates), concentration/diversification (Herfindahl index, share of top exporters, number of products and destinations per exporter), firm, product and market dynamics (entry, exit and survival rates) and unit prices (per exporter, product, market). The measures are available at different disaggregation levels: a) country-year (file CY.dta with 98 measures), b) country-year–product (HS2, HS4, or HS6) (file CYH2.dta with 113 measures, file CYH4.dta with 113 measures, and file CYH6.dta with 89 measures) and c) country-year-destination (file CYD.dta with 74 measures). The list of measures under each category and for each type of 14 disaggregation level is presented in Table 1 along with the corresponding formula. 28 The Database is publicly available for download free of charge at http://data.worldbank.org/data- catalog/exporter-dynamics-database. 29 For the measures of firm dynamics, firm-product and firm-destination market dynamics, we use the following definitions: • Exportert: any firm that exports in year t; • Entrantt: a firm that does not export in year t-1 but exports in year t; • Exitert: a firm that exports in year t-1 but does not export in year t; • Incumbentt: a firm that exports in both years t-1 and t; • 2-Year Incumbentt: a firm that exports in years t-1, t, and t+1; • Survivort: a firm that does not export in year t-1 but exports in both years t and t+1; • 2-Year Survivort: a firm that does not export in year t-1 but exports in years t, t+1 and t+2; • 3-Year Survivort: a firm that does not export in year t-1 but exports in years t, t+1, t+2 and t+3. To protect the confidentiality of the firms in the raw datasets at the exporter-product- destination-year level, some of the measures in the files with product or destination disaggregation (CYH2.dta, CYH4.dta, CYH6.dta, CYD.dta) are missing when the underlying 28 Table 1 also describes the nomenclature used to name the different measures in the stata files consistently. For example, the variable ‘B1’ refers to the Herfindahl index; the variable ‘A7’ refers to the Export Value per Incumbent exporter, etc. While not shown in Table 1, note that we used suffixes to identify means (i), medians (ii), and standard deviations (iii) of certain measures in the stata files. For example, the variable ‘A7i’ represents the mean of the Export Value per Incumbent exporter in a given country and year; ‘A7ii’ represents the median of the same statistic and ‘A7iii’ represents its standard deviation. 29 Although the files CY.dta, CYH2.dta, CYH4.dta, CYH6.dta, and CYD.dta are all based on the cross-country input dataset excluding oil exports in HS Chapter 27, interested users can request from the authors an alternative version of the files where the measures are calculated including exports in HS Chapter 27 (for the countries which provided such information). 15 country-product-year or country-destination-year cell includes a single firm whose individual information cannot be revealed. 30 Table 1: Definition of Measures LEVEL Country – Year – Country Product CODE MEASURES Country - Year - - Year Destinati HS2 & on HS6 HS4 A BASIC CHARACTERISTICS 1,2,3,4,5 N (Exporters, Entrants, Exiters, Survivors, Incumbents)     6,7,8,9,10 MMS & Q1Q3 (TEV per Exporter, Entrant, Exiter, Survivor, Incumbent)     MMS & Q1Q3 (Growth of Incumbents)t= MMS (ln(TEV of incumbentt in t) – 11     ln(TEV of incumbentt in t-1)) MMS & Q1Q3 (Growth of Survivors)t = MMS (ln(TEV of survivort in t+1) - 12     ln(TEV of survivort in t)) B CONCENTRATION/DIVERSIFICATION 1 Herfindahl Index     2 Share of top 1%, 5%, 25% Exporters in TEV     3 MMS (N. HS6 Products per Exporter)    NA 4 MMS (N. Destinations per Exporter)  NA   5 MMS (N. Exporters per HS6 Product)    NA 6 MMS (N. Exporters per Destination)  NA   C FIRM DYNAMICS 1 Firm Entry Ratet = N. Entrantst / N. Exporterst     2 Firm Exit Ratet = N. Exiterst/ N. Exporterst-1     3 Firm Survival Ratet = N. Survivorst / N. Entrantst     4 Share of Entrantst= TEV of Entrantst in t / TEV in t     5 Firm 2-year Survival Ratet = N. 2-year Survivorst / N. Entrantst     6 Firm 3-year Survival Ratet = N. 3-year Survivorst / N. Entrantst     D PRODUCT DYNAMICS 30 For such cells the number of exporters shows a value of 1 and most measures except those based on exiter firms in Table 1 are missing. The developed countries in our sample applied their own confidentiality rules (often stricter). Due to confidentiality reasons, the information in the CYH6.dta file is not provided for Brazil, Belgium, Egypt, New Zealand, Spain, Sweden, the information in the CYH4.dta files is not provided for Brazil, New Zealand and Sweden, the information in the CYH2.dta and the CYD.dta files is not provided for New Zealand. A file with the percentage of total exports corresponding to the hidden values by country and year, for the developing countries for which we have the raw exporter-level data is available at http://data.worldbank.org/data-catalog/exporter-dynamics-database. 16 MMS (Product Entry Rate of Incumbents)t = MMS (N. products not exported 1 in t-1 but exported in t by incumbentt / N. all products exported by    NA incumbentt in t) MMS (Product Entry Rate of Survivors)t = MMS (N. products not exported in 2    NA t-1 but exported in t by survivort-1 / N. all products exported by survivort-1 in t) MMS (Share of New Products in TEV of Incumbents)t= MMS (EV of 3 incumbentt from products not exported in t-1 but exported in t / TEV of    NA incumbentt in t) MMS (Share of New Products in TEV of Survivors)t= MMS (EV of survivort-1 4    NA from products not exported in t-1 but exported in t / TEV of survivort-1 in t) MMS (Product Exit Rate of Incumbents)t = MMS (N. products exported by 5    NA incumbentt in t-1 but not in t/ N. all products exported by incumbentt in t-1) MMS (Product Survival Rate of 2-year Incumbents)t = MMS (N. products not 6 exported in t-1 but exported in both t and t+1 by 2-year incumbentt /N. all    NA products not exported in t-1 but exported in t by 2-year incumbentt) E DESTINATION DYNAMICS MMS (Destination Entry Rate of Incumbents)t = MMS (N. destinations not 1 exported in t-1 but exported in t by Incumbentt / N. all destinations exported  NA   by Incumbentt in t) MMS (Destination Entry Rate of Survivors)t = MMS (N. destinations not 2 exported in t-1 but exported in t by Survivort-1 / N. all destinations exported  NA   by Survivort-1 in t) MMS (Share of New Destinations in TEV of Incumbents)t= MMS (EV of 3 Incumbentt from destinations not exported in t-1 but exported in t / TEV of  NA   Incumbentt in t) MMS (Share of New Destinations in TEV of Survivors)t= MMS (EV of survivort-1 4 from destinations not exported in t-1 but exported in t / TEV of survivort-1 in  NA   t) MMS (Destination Exit Rate of Incumbents)t = MMS (N. destinations exported 5 by Incumbentt in t-1 but not in t/ N. all destinations exported by Incumbentt  NA   in t-1) MMS (Destination Survival Rate of 2-year Incumbents)t = MMS (N. destinations not exported in t-1 but exported in both t and t+1 by 2-year 6  NA   Incumbentt /N. all destinations not exported in t-1 but exported in t by 2-year Incumbentt) F UNIT PRICES MMS (Unit Price (TEV/Quantity) per Exporter, Entrant, Exiter, Incumbent, 1,2,3,4,5 NA NA   Survivor) Notes: MMS indicate Mean, Median, Standard Deviation; Q1Q3 indicate the 25th and the 75th percentiles; N indicates Number of; EV indicates Export Value; and TEV indicates Total Export Value. 17 Given the increasing interest and analysis on multi-product firms in trade, we provide in addition to the measures in the Database a set of companion matrix tables by country and year showing the distribution of exporters and of total exports by the number of products exported and the number of destination markets served. The matrix tables consider the following categories for the number of products and the number of destinations: 1, 2, 3, 4-10, 11-20, 21 or more. 31 Some examples of the matrix tables are provided in Appendix 4. All matrix tables can be downloaded at http://data.worldbank.org/data-catalog/exporter-dynamics-database. To illustrate the content and diversity of measures included in the Database, Table 2 presents a summary of a representative set of measures in the Database –average and median exporter size, share of top 5 percent firms, number of exporters, number of HS 6-digit products exported per firm, number of destinations served per firm, entry exit and survival rates– as well as total exports per country. We focus on averages per country for the period 2006-2008 which is the period most commonly covered across countries (and captures trade performance before the global financial crisis that started at the end of 2008). 32 Table 3 presents the same set of measures in the Database but for groups of HS 2-digit sectors, where the measures for each sector group are obtained as averages across all countries that export that particular group of HS 2-digit sectors, again focusing on the period 2006-2008. Some interesting cross-country patterns emerge from Table 2. First, there is tremendous variation across countries in the export base (number of exporters) and the average exports per firm. Developed countries tend to exhibit the larger numbers of exporters. Among developing countries, the largest numbers of exporters are found in Turkey and Mexico followed by South 31 These matrix tables are available only for the developing countries for which we have the customs exporter-level data. We are unable to provide matrix tables for the developed countries in our sample. 32 For countries for which data is available for only 1 or 2 years within the period 2006-2008 we compute the average based on those years. For Kuwait and Portugal the data coverage does not include that period. Hence for Table 2 and for Figure 1 as well as for the Figures in Appendix 3 we use averages for the period 2009-2010 for Kuwait and averages for the 2003-2005 period for Portugal. 18 Africa and Brazil (around 20,000) then by Pakistan, Bulgaria, and Iran (with more than 13,000 each) whereas the smallest pools of exporters (around and below 300) are found in Niger and Mali, followed by Burkina, Laos, Yemen, and Cambodia. This pattern seems to mirror the countries’ level of development, and this link will be studied further in Section IV. Average exports per firm are the highest (in the range of 7-8 million USD) for Belgium, Brazil, Chile, and Mexico followed by Cambodia, Sweden and Peru. 33 Albania, Lebanon, Macedonia, Yemen, and Kenya exhibit the lowest average export values per exporter (less than 800,000 USD). The tremendous difference between mean and median exports per firm indicates very skewed exporter size distributions in all countries with some very large exporters driving total exports. 34 For Botswana, Chile, Mali, Peru, South Africa, Mexico, and Malawi the skewness in the exporter size distribution is further confirmed by the very high share of exports – more than 90 percent– accounted for by just the top 5 percent of exporters. Interestingly, for most other developing countries the shares for developing countries are smaller than those for Norway, Sweden, Norway, and New Zealand and are smaller than those for the U.S. and other European countries for which the top 5 percent of firms account for 80 percent or more of trade, as shown by Bernard, Jensen, and Schott (2009) and Mayer and Ottaviano (2008). The average number of products per exporter also exhibits a high degree of heterogeneity across countries, ranging from 2 in Laos to 15 in South Africa. In terms of markets, the average number of destinations per exporter is more similar across countries ranging from 1.4 in Botswana to 4.8 in Cambodia and 6.8 in Belgium. Interestingly, the average or median of these 33 The very large average exporter size in Cambodia is driven by a small number of extremely large formerly state-owned apparel and textiles producers, as gathered from statistics based on the World Bank Enterprise Surveys for Cambodia. 34 Using the same cross-country input dataset used in the construction of the Database, Freund and Pierola (2012) show that exports are dominated by a small group of very large exporters (so-called ‘superstars’). These firms are remarkably larger than the rest, they participate in many sectors and most importantly, they define the productive structures and drive the export growth observed in most countries. 19 two firm-level diversification measure taken across developing countries are quite similar to those taken including also developed countries. Table 2: Measures by Country (2006-2008 Averages) Mean Median Number of Number of Total Exports Number of Exports per Exports per Share of Top Entry Exit Survival Products per Destinations (bn USD) Exporters Exporter Exporter 5% Exporters Rate Rate Rate Exporter per Exporter ('000s USD) ('000s USD) ALB Albania 1.1 1895 550 35 63% 3.0 1.5 39% 33% 47% BEL Belgium 309.1 23204 13312 64 84% 9.3 6.8 31% 28% 40% BFA Burkina Faso 0.5 425 1177 37 85% 3.8 2.4 44% 41% 42% BGD Bangladesh 12.4 6356 1946 277 50% 4.2 3.8 28% 22% 61% BGR Bulgaria 12.9 13804 934 22 83% 6.2 2.4 38% 40% BRA Brazil 165.4 19375 8539 233 82% 22% 23% 54% BWA Botswana 4.6 1715 2666 2 99% 6.6 1.4 42% 40% 39% CHL Chile 60.9 7314 8317 49 94% 4.5 3.4 38% 35% 35% CMR Cameroon 1.7 938 1879 19 82% 4.0 2.8 48% 46% 23% COL Colombia 19.1 9768 1957 58 81% 4.9 2.8 32% 31% 42% CRI Costa Rica 8.7 2931 2970 54 82% 5.6 3.2 29% 26% 48% DOM Dominican Republic 4.5 2709 1708 26 85% 4.7 2.3 44% 43% 40% ECU Ecuador 5.7 3110 1830 25 80% 4.4 2.4 41% 37% 41% EGY Egypt 14.3 8370 1717 65 79% 2.7 25% 27% 51% ESP Spain 229.9 89798 2559 21 86% 4.7 4.0 39% 38% 30% EST Estonia 9.3 4915 1885 109 69% 7.8 2.7 44% 41% 30% GTM Guatemala 6.3 4420 1421 38 78% 7.8 2.5 31% 29% 42% IRN Iran 12.8 13770 940 88 72% 6.0 2.1 47% 51% 41% JOR Jordan 3.4 1869 1804 57 83% 2.7 3.1 38% 32% 49% KEN Kenya 4.0 5057 796 18 81% 7.2 2.5 40% 44% 35% KHM Cambodia 3.4 595 5706 546 44% 8.3 4.8 33% 30% 57% KWT Kuwait * 3.0 3315 915 27 86% 4.4 2.0 53% 53% LAO Laos 0.6 462 1284 42 88% 2.3 1.6 52% 40% 50% LBN Lebanon 3.4 5177 659 38 78% 7.7 3.1 MAR Morocco 15.3 5429 2811 90 74% 6.4 2.5 33% 34% 43% MEX Mexico 226.3 34382 6588 44 91% 6.7 2.1 35% 36% 39% MKD Macedonia 2.2 2926 751 24 83% 4.5 2.2 38% 35% 45% MLI Mali 0.8 305 2729 48 93% 3.8 2.2 43% 39% 45% MUS Mauritius 2.6 2251 1138 17 87% 9.0 2.6 30% 31% 43% MWI Malawi 0.6 631 1077 8 91% 4.2 1.9 52% 61% 25% NER Niger 0.3 160 2160 18 89% 3.9 1.6 NIC Nicaragua 1.3 1236 1031 27 76% 5.9 2.1 36% 34% 47% NOR Norway 39.1 18309 2137 14 93% 5.2 3.4 38% 37% NZL New Zealand 24.6 13276 1853 24 90% 7.5 3.1 29% 29% 42% PAK Pakistan 16.8 15023 1116 62 73% 5.5 3.3 28% 27% 56% PER Peru 25.2 6732 3740 37 92% 7.2 2.6 39% 35% 44% PRT Portugal * 33.5 16217 2064 68 77% 8.5 3.5 30% 29% 45% SEN Senegal 0.9 727 1228 73 71% 6.2 3.2 40% 37% 40% SLV El Salvador 4.2 2554 1648 30 82% 6.9 2.4 31% 30% 44% SWE Sweden 129.5 30126 4299 17 92% 6.5 4.3 29% 28% TUR Turkey 98.7 44570 2204 105 80% 9.6 3.9 32% 29% 55% TZA Tanzania 2.3 1899 1180 17 86% 4.1 2.4 51% 46% 32% UGA Uganda 1.2 938 1289 15 77% 3.8 2.4 47% 38% 29% YEM Yemen 0.4 492 779 49 64% 4.4 2.4 52% 54% ZAF South Africa 58.8 21721 2699 29 92% 15.0 3.6 28% 26% 49% Average - developing countries 21.7 7017 2206 63 81% 5.7 2.6 38% 37% 43% Median - developing countries 4.2 2931 1708 37 82% 5.2 2.5 38% 35% 43% Average - all countries 35.1 10027 2489 61 81% 5.9 2.8 38% 36% 43% Median - all countries 5.7 4420 1830 37 82% 5.6 2.6 38% 35% 43% Note: the figures shown are based on measures in the country-year level dataset (file CY.dta) averaged across the period 2006- 2008 for each country. * indicates exceptions to the sample period: for Kuwait averages are taken across the period 2009-2010 and for Portugal averages are taken across the 2003-2005 period. Total exports are obtained as the number of exporters multiplied by the average exports per exporter. 20 The measures of exporter dynamics also exhibit important variability across countries. Entry rates range from 22 percent in Brazil to more than 50 percent in Malawi and Yemen while exit rates range from 22 percent in Bangladesh to 61 percent in Malawi. 35 First-year survival rates of new exporters vary between 23 percent in Cameroon and 61 percent in Bangladesh. The magnitude of the survival rates in Table 2 suggests an extremely high attrition rate of new entrants after just one year in export markets, particularly in Africa. However, that is not a characteristic of less developed countries since high attrition rates of new entrants are also observed in Spain and Estonia. Table 3: Measures by Sector (2006-2008 Averages) Mean Median Number of Number of Number of Exports per Exports per Share of Top Entry Exit Survival HS 2-Digit Codes HS Section Description Products per Destinations Exporters Exporter Exporter 5% Exporters Rate Rate Rate Exporter per Exporter ('000s USD) ('000s USD) 01-05 Live Animals and Animal Products 108 1109 107 53% 1.9 1.9 49% 48% 38% Vegetable Products (including Animal and 06-15 168 631 40 58% 1.6 1.8 50% 48% 36% Vegetable Fats) 16-24 Foodstuff (Beverages, Spirits, Vinegar, Tobacco etc.) 147 1427 147 63% 1.7 2.3 46% 43% 38% 25-26 Mineral Products (except hydrocarbons) 199 9730 294 76% 1.3 1.6 54% 49% 36% 28-38 Chemicals and Parachemical Products 201 1205 50 73% 1.6 2.0 53% 51% 31% 39-40 Plastics and Articles Thereof 989 408 7 77% 1.9 1.8 54% 51% 33% Wood and Articles Thereof (including Paper & 44-46, 47-49, 94 476 1221 55 73% 1.5 1.7 57% 56% 28% Articles, Furniture) 50-59, 41 Textiles (Including Raw Skins and Leather) 153 413 30 66% 1.5 1.8 58% 57% 27% Apparel (Including Footwear, Headgear, Art. of 60-63, 64-67, 42-43 353 402 35 69% 1.9 1.7 58% 57% 28% Feathers, Fur, Leather Products) 68-70 Glass, Ceramics and Articles of Stone, Cement, etc. 429 212 4 76% 1.5 1.7 60% 58% 26% Precious Metals (Pearls, Jewellery, Coin, Precious 71 227 7503 473 78% 1.4 1.6 50% 47% 33% Stones etc.) 72-83 Base Metal and Articles Thereof 327 2010 26 75% 1.5 1.6 60% 58% 26% Mechanical Machinery (including Clocks and Music 84, 91-92 870 252 8 72% 1.9 1.6 62% 61% 24% Instruments) Electrical Machinery (including Optical, Medical, 85, 90 1183 737 8 76% 2.4 1.8 57% 54% 30% Photographic Instruments) 86-89 Transportation Vehicles 326 1275 41 71% 1.4 1.6 65% 63% 24% 93 Arms and Ammunitions 25 523 39 69% 1.4 1.7 59% 61% 19% Note: the figures shown in the table for each group of sectors are based on measures in the country-year-product (HS 2-digit) level dataset (file CYH2.dta) averaged across HS 2-digit sectors, countries, and the 2006-2008 period. 35 Kuwait also exhibits a high entry rate but it covers a different sample period. 21 Some interesting cross-sectoral patterns emerge from Table 3. 36 On average across all countries, the number of exporters is largest in electrical machinery and smallest in live animals and animal products and in arms and ammunitions. The degree of concentration as measured by the share of the top 5 percent of exporters is largest on average in precious metals and in plastics. Entry and exit rates are highest in transportation vehicles and mechanical machinery (more than 60 percent) and lowest in foodstuff (about 45 percent). IV. The Exporter Dynamics Database: Six Stylized Facts In this section we present four new stylized facts (facts 1 through 4) and we generalize two stylized facts (facts 5 and 6) from earlier work. The distinguishing feature of these stylized facts is that they could not have been discovered using any other cross-country source of trade data available so far. Therefore, they show the usefulness of the Database and illustrate the possibilities of future policy-relevant analysis and research that can be done using the Database. Stylized Fact 1: More developed and larger economies have a larger export base (number of exporters), larger average exporter size, more concentrated export sectors among firms, and more diversified exporters (in terms of their portfolios of products and destinations). In contrast, more developed and larger economies exhibit significantly lower exporter entry and exit rates. The survival rates of new entrants are not correlated with the level of development or the economic size of countries. Similar patterns are obtained for the importance of the export sector in the country. 36 We exclude Kuwait and Portugal from the calculation of the averages across all countries that export that particular group of HS 2-digit sectors shown in Table 3 since those countries do not have data for the period 2006-2008. 22 Figure 1 presents the scatter plot between each of a set of measures in the Database at the country level and the stage of development measured by GDP per capita along with the R- squared from the corresponding regression among the plotted pair of variables. Appendix 3 presents analogous scatter plots between each of a set of measures in the Database and either the economic size of the economy measured by GDP or the importance of the export sector measured by the ratio of exports to GDP. The Database measures as well as the GDP per capita, GDP, and the ratio of exports to GDP values are taken as averages over the period 2006-2008. 37 Figure 1: Correlations of Selected Database Measures with GDP per Capita (averages over 2006-2008) Number of Exporters - GDPpc Mean Exports per Exporter - GDPpc 12 17 ESP TUR BEL Ln Mean Exports per Exporter MEX SWE 10 16 BEL Ln Number of Exporters ZAFBRA BRA CHL PRT NOR PAK IRN BGR NZL MEX EGY COL KHM BGD PER CHL KEN MAR LBN EST SWE GTM PER 15 KWT 8 ECU MKD DOMCRI SLV MUS MAR CRI TZA JOR ALB MLI ZAF BWA ESP BWA NIC NER TUR PRT NOR BGD UGA CMR CMR EGY ECU COL JOR EST NZL SEN SLV DOM MWI KHM GTM LAO YEM 14 BFA UGA LAO SEN 6 TZA BFA PAK MUS MLI MWI NIC IRN BGR KEN YEM KWT NER MKD LBN ALB R2=0.72 R2=0.34 13 4 6 7 8 9 10 11 6 7 8 9 10 11 Ln GDPpc Ln GDPpc Share of Top 5% Exporters - GDPpc Number of Products per Exporter - GDPpc 3 1 BWA CHL Ln Number of Products per Exporter MLI PER ZAF NOR ZAF MWI MEX SWE NZL NER LAO 2.5 MUS Share of Top 5% Exporters TZA DOM ESPKWT BFA BEL CMR MKD JOR BGR CRI BRA KEN SLV COL .8 EGY ECU TUR TUR BEL GTM LBN MUS UGA PRT KHM PRT NIC MAR GTM LBN EST NZL 2 PAK IRN KEN PER SEN SLV EST MAR BWA MEX SWE SEN BGR IRN NIC PAK CRI YEM ALB NOR COL .6 1.5 DOM ESP YEM MKD ECU CHL KWT MWI BGD TZA NER CMR UGA BFA MLI BGD ALB JOR 1 KHM R2=0.40 .4 LAO 6 7 8 9 10 11 6 7 8 9 10 11 Ln GDPpc Ln GDPpc 37 GDP per capita, GDP, and total exports in current USD are taken from the World Development Indicators (WDI) of the World Bank. 23 Number of Destinations per Exporter - GDPpc Entry Rate - GDPpc 2 MWI LAO YEM BEL TZA .5 Ln Number of Destinations per Exporter CMR IRN UGA KHM 1.5 SWE BFA MLI DOM EST TUR ESP BWA BGD ECU ZAF .4 CHL PRT NOR KEN SEN Entry Rate SEN PAK JOR CRI ALB PER ESP LBN NZL JOR BGR MKD CHL NOR CMR COL 1 EGY PER MUS EST NIC KEN MAR GTM MEX TZA UGA BFA YEM SLV ECUBGR DOM KHM MAR MLI MKD MEX TUR NIC IRN SLV COL BEL MWI KWT GTM MUS .3 PRT CRI NZL ZAF SWE .5 LAO BGD PAK NER ALB BWA EGY BRA R2=0.37 R2=-0.44 .2 0 6 7 8 9 10 11 6 7 8 9 10 11 Ln GDPpc Ln GDPpc Exit Rate - GDPpc Survival Rate - GDPpc MWI BGD .6 .6 KHM PAK YEM TUR BRA IRN EGY .5 .5 LAO ZAF JOR CRI CMR NIC ALB Survival Rate TZA MLI PER MKD PRT Exit Rate KEN SLV DOM MAR MUS BFA GTM NZL BFA IRN COL ECU EST .4 .4 MLI LAO BGR BWA SEN DOM BWA MEX BEL UGA ESP SEN ECU NOR MEX CHL PER MKD CHL KEN NIC MAR ALB JOR COLMUS TZA EST .3 .3 KHM SLV ESP GTM TUR PRT NZL UGA BEL SWE PAK EGY ZAF CRI MWI BRA CMR BGD R2=-0.36 R2=-0.02 .2 .2 6 7 8 9 10 11 6 7 8 9 10 11 Ln GDPpc Ln GDPpc Notes: The measures plotted are based on measures in the country-year level dataset (file CY.dta) averaged across the 2006-2008 period for each country for each country except Kuwait for which data is for 2009 and Portugal for which averages are across the 2003-2005 period. GDP per capita is in current USD per inhabitant. The panels in Figure 1 show that more developed economies have a larger number of exporters as well as larger average exporter sizes. 38 More developed economies are also characterized by more concentrated export sectors, in terms of higher shares accounted for by the top 5 percent of exporters. Exporters in more developed economies exhibit a more diversified portfolio in terms of products and destinations. However, more developed economies are less dynamic in terms of exporter churning. In such economies, larger pools of more productive firms may already be operating in the exporting sector and taking advantage of the profitable opportunities in markets abroad, making it less appealing to potential exporters to attempt an entry thus reducing entry rates. If those firms are well established, their probability of exit is also 38 A similar pattern is found for median exporter size. 24 lower. Survival rates of new entrants into export markets exhibit no clear correlation with the level of development. One possibility is that survival rates may depend more heavily on characteristics of the primary destination markets (neighborhood) than of the source markets. Indeed, the three countries with the lowest survival rates are in Africa, and two are landlocked. Table 4 shows that all the correlations discussed so far, with the exception of survival rates, are highly significant. Table 4: Correlations between Database Measures, GDP per Capita, GDP, and the Ratio of Exports to GDP Ln Mean Ln Number of Ln Number of Ln Number of Exports per Share of Top Entry Exit Survival Products per Destinations Exporters Exporter 5% Exporters Rate Rate Rate Exporter per Exporter (mn USD) Ln GDP per capita 0.72 0.35 0.32 0.40 0.38 -0.32 -0.25 -0.02 (0.00) (0.02) (0.03) (0.01) (0.01) (0.04) (0.10) (0.90) Ln GDP 0.90 0.52 0.15 0.41 0.54 -0.48 -0.35 0.16 (0.00) (0.00) (0.31) (0.01) (0.00) (0.00) (0.02) (0.34) Ratio of exports to GDP 0.27 0.53 0.08 0.45 0.43 -0.32 -0.32 -0.03 (0.08) (0.00) (0.59) (0.00) (0.00) (0.04) (0.04) (0.84) Notes: P-value shown in parentheses. The measures used in the correlations are based on measures in the country- year level dataset (file CY.dta) averaged across the 2006-2008 period for each country for each country except Kuwait for which averages are across the period 2009-2010 and Portugal for which averages are across the 2003- 2005 period. Table 4 and the figures in Appendix 3 show that the results are quite similar when the correlations are established with either the economic size of countries or the importance of their export sector instead of their level of development. In terms of magnitudes, the correlations of most measures tend to be higher with economic size than with the level of development or the ratio of exports to GDP. 39 39 Kuwait and Portugal are part of the sample used to compute the correlations in Table 4 despite their data covering a period other than 2006-2008. However, similar correlation patterns are obtained if they are excluded from the sample. 25 Stylized Fact 2: Across countries expansions along the intensive margin – i.e., increases in the average exporter size – contribute more to annual export growth than expansions along the extensive margin – i.e., increases in the number of exporters. To understand whether in the short run exports expand through increases in the average size of exporters (the intensive margin) or through increases in the numbers of exporters (the extensive margin), we consider a simple decomposition of the change in exports between consecutive years for each country in the Database. If c designates a country and t a year, total exports in a given year 𝑋�𝑡 can be written as the product of the number of exporters 𝑛�𝑡 and the average exporter size 𝑠�𝑡 : 𝑋�𝑡 = 𝑛�𝑡 ∗ 𝑠�𝑡 . Thus, the change in exports between years t-1 and t 𝑛�𝑡 ∗ 𝑑𝑠�𝑡 +���� can be written as: 𝑑𝑋�𝑡 = ���� 𝑠�𝑡 ∗ 𝑑𝑛�𝑡 where 𝑑𝑠�𝑡 is the change in the average exporter size between years t-1 and t, 𝑑𝑛�𝑡 is the change in the number of exporters between 𝑛�𝑡 is the average number of exporters across years t-1 and t, and ���� years t-1 and t, ���� 𝑠�𝑡 is the average exporter size across years t-1 and t. The contribution of the intensive margin to a change in exports is given by 𝑛�𝑡 ∗ 𝑑𝑠�𝑡 �𝑑𝑋�𝑡 while the contribution of the extensive margin to a change ���� in exports is given by 𝑠�𝑡 ∗ 𝑑𝑛�𝑡 �𝑑𝑋�𝑡 . ���� Table 5 shows the results from this decomposition using the measures available in the Database at the country-year level (CY.dta) across years 2006 and 2007 and across years 2007 and 2008. 40 While there is an important degree of heterogeneity across countries in the role of the intensive and the extensive margins, expansions in the average size of exporters are typically more important than the addition of new exporters for export growth in the short run, as pointed out by the medians across all countries, which are above 80% in both periods (as seen in the bottom rows). Moreover, of the eleven countries with double-digit export growth in one year, all 40 We chose to exclude from the table values for the 2006-2007 decomposition for Albania, Burkina Faso, Cambodia, Cameroon, Mexico, and Uganda since their firm coding systems changed in 2007 and for the 2007-2008 decomposition for Yemen since its firm coding system changed in 2008. 26 but two (Belgium and Spain in 2007-2008) show dominance of the intensive margin, and in more than half of this group the robust export growth is fully explained by the intensive margin. Table 5: Decompositions of Export Growth across Countries Change in Total Change in Total Contribution of Contribution of Contribution of Contribution of Exports Exports the Intensive the Extensive the Intensive the Extensive Between 2006 Between 2007 Margin in 2006- Margin in 2006- Margin in 2007- Margin in 2007- and 2007 and 2008 2007 2007 2008 2008 (bn USD) (bn USD) ALB Albania 0.27 60.9% 39.1% BEL Belgium 48.45 104.5% -4.5% 20.81 45.2% 54.8% BFA Burkina F. 0.01 -331.2% 431.2% BGD Bangladesh -4.06 103.6% -3.6% 6.01 73.9% 26.1% BRA Brazil 22.53 81.3% 18.7% 36.82 114.0% -14.0% BWA Botswana 0.55 98.5% 1.5% -0.70 120.8% -20.8% CHL Chile 9.86 27.6% 72.4% 2.40 0.3% 99.7% CMR Cameroon 0.37 0.0% 100.0% CRI Costa Rica 1.11 86.6% 13.4% 0.11 455.3% -355.3% DOM Dominican Republic 0.29 -475.8% 575.8% 0.54 248.7% -148.7% ECU Ecuador 0.78 44.3% 55.7% 1.42 73.1% 26.9% EGY Egypt 2.57 99.4% 0.6% 3.74 113.9% -13.9% ESP Spain 35.94 106.7% -6.7% 17.19 41.7% 58.3% EST Estonia 1.71 117.7% -17.7% 1.71 51.2% 48.8% GTM Guatemala 0.85 73.7% 26.3% 0.77 53.0% 47.0% IRN Iran 0.94 232.4% -132.4% 2.85 129.8% -29.8% JOR Jordan 0.20 30.1% 69.9% 1.53 81.4% 18.6% KEN Kenya 0.66 149.9% -49.9% 0.88 123.1% -23.1% KHM Cambodia 1.41 110.9% -10.9% LAO Laos -0.05 295.0% -195.0% 0.56 75.3% 24.7% LBN Lebanon MAR Morocco 2.33 99.2% 0.8% 4.86 99.7% 0.3% MEX Mexico 11.67 148.6% -48.6% MKD Macedonia 0.79 77.2% 22.8% 0.30 83.0% 17.0% MLI Mali -0.32 133.3% -33.3% 0.23 85.9% 14.1% MUS Mauritius -0.15 89.8% 10.2% -0.01 -1921.2% 2021.2% MWI Malawi 0.22 223.8% -123.8% 0.03 123.5% -23.5% NER Niger NIC Nicaragua 0.20 72.9% 27.1% 0.29 88.8% 11.2% NZL New Zealand 3.75 109.3% -9.3% 2.19 90.0% 10.0% PAK Pakistan 1.20 66.4% 33.6% 2.51 87.7% 12.3% PER Peru 3.87 82.7% 17.3% 2.12 16.7% 83.3% SEN Senegal 0.14 90.5% 9.5% 0.29 72.9% 27.1% SLV El Salvador 0.46 55.3% 44.7% 1.02 93.2% 6.8% TUR Turkey 20.10 61.6% 38.4% 22.07 95.0% 5.0% TZA Tanzania 0.26 3.4% 96.6% 0.77 82.0% 18.0% UGA Uganda -0.09 316.5% -216.5% YEM Yemen 0.06 124.3% -24.3% ZAF South Africa 9.62 96.2% 3.8% 12.26 73.6% 26.4% Median - all countries 90.5% 9.5% 84.5% 15.5% Median - countries in both sub-periods 90.5% 9.5% 82.5% 17.5% Notes: the terms of the decomposition shown in the table are based on measures in the country-year level dataset (file CY.dta). 27 The findings in Table 5 are consistent with the predictions from trade models with heterogeneous firms whereby growth in the number of firms contributes relatively little to total exports as compared with growth in average firm size because entrants tend to be marginal firms. 41 Stylized Fact 3: Across countries, entry and exit rates are strongly positively correlated with each other and both are negatively correlated with entrant survival rates, average exporter size, and diversification at the exporter level (in terms of products and destinations). Table 6 shows the matrix of correlation coefficients among several measures in the Database at the country-year level (averaged over the 2006-2008 period) and the corresponding p-values. 42 The table presents also the correlations between the dynamics and diversification measures with total exports. Exporter entry and exit rates are strongly positively correlated in Table 6, implying that countries with high entry rates also have high exit rates. This evidence is similar to that based on industry data for individual countries such as Peru by Freund and Pierola (2010). In a typical year there are naturally high levels of experimentation (entry) accompanied by similar levels of failure (exit) in all countries. This intuition is reinforced by the strong negative correlations of both entry and exit with one-year survival of new entrants into export markets. Entry and exit rates are also strongly negatively correlated with average exporter size. This is not surprising since export markets are more contestable in countries where exporters are relatively smaller, observed churning is likely to be higher. 41 This finding differs, however, from cross-sectional evidence provided by Mayer and Ottaviano (2008) and our stylized fact 5 below on the correlation between the number of exporters and standard gravity variables. We discuss this in more detail below. 42 Kuwait and Portugal are part of the sample used to compute the correlations in Table 6 despite their data covering a period other than 2006-2008. However, similar correlation patterns are obtained if they are excluded from the sample. Moreover, similar patterns are obtained if we use data for individual years within or outside the 2006-2008 period. 28 Another fact emerging from the correlations in Table 6 is that entry and exit rates are strongly negatively correlated with the average number of firms, the average number of products per firm and the average number of destination markets served per firm. This indicates that across countries, less churning takes place in more sophisticated export markets, where there are a lot of firms that sell a wider range of goods to more markets. Table 6: Correlation among Selected Measures in the Database at the Country Level Ln Mean Ln Mean Ln Mean Share of Top Number of Number of Ln Number of Ln Total Entry Rate Exit Rate Survival Rate Exports per 5% of Products per Destinations Exporters Exports Exporter Exporters Exporter per Exporter Entry Rate 1 Exit Rate 0.92 1 (0.00) Survival Rate -0.65 -0.73 1 (0.00) (0.00) Ln Mean Number of Products per Exporter -0.59 -0.38 0.14 1 (0.00) (0.01) (0.42) Ln Mean Number of Destinations per Exporter -0.55 -0.52 0.21 0.47 1 (0.00) (0.00) (0.21) (0.00) Ln Number of Exporters 1 -0.57 -0.43 0.1522 0.5202 0.53 (0.00) (0.00) (0.36) (0.00) (0.00) Ln Mean Exports per Exporter -0.44 -0.45 0.10 0.30 0.54 0.35 1 (0.00) (0.00) (0.53) (0.05) (0.00) (0.02) Share of Top 5% of Exporters 0.11 0.15 -0.40 0.00 -0.21 0.14 0.22 1 (0.50) (0.35) (0.01) (0.99) (0.17) (0.34) (0.16) Ln Total Exports -0.62 -0.51 0.16 0.54 0.63 0.93 0.66 0.20 1 (0.00) (0.00) (0.32) (0.00) (0.00) (0.00) (0.00) (0.20) Notes: P-values in parentheses. The correlations shown in the table are based on measures in the country-year level dataset (file CY.dta) averaged across the 2006-2008 period for each country except Kuwait for which averages are across the 2009-2010 period and Portugal for which averages are across the 2003-2005 period. Stylized Fact 4: The typical exporter size, total exports, and exporter survival rates at the country-sector level are not explained by country or sector characteristics while the share of 29 the top 5 percent of exporters, the average number of products and destinations per exporter, and entry and exit rates are explained to an important extent by those characteristics. We conduct an ANOVA decomposition of the same set of Database measures shown in Table 2 taken at the country and HS 2-digit sector level –averaged across the 2006-2008 period– to determine the explanatory power of country effects and sector effects. 43 The results are shown in Table 7 and indicate that country and sector effects combined explain an important share of the observed variation in the number of exporters, the average number of products and destinations per exporter, the share of the top 5 percent of exporters, the number of exporters (export base), and in exporter entry and exit rates. 44 In contrast, country and sector effects do very little to explain total exports in the country-sector, the average or median size of an exporter and exporter survival rates in the country-sector. The pattern of total exports across sectors in a country reflects comparative advantage. Since total exports in a country-sector are by definition determined by the number of exporters in the country-sector and their average size, the importance of country and sector effects in determining the number of exporters but not the size of exporters implies that comparative advantage works primarily via firm size. There are also interesting differences in the importance of sector versus country characteristics in explaining exporter behavior. Sector characteristics do more to explain the share of top 5 percent and the number of products, while country characteristics explain better the number of exporters, the number of destinations, and entry and exit rates. While many of these differences are to be expected, the importance of sector for the share of top 5 percent implies that market structure drives a good part of concentration as opposed to country rules and 43 The ANOVA decomposition is a type of statistical hypothesis testing, whereby the observed variance of a given variable is partitioned into components attributable to different sources of variation. The ANOVA provides a statistical test of whether the means of the variable for different groups are all equal. 44 Kuwait and Portugal are part of the sample used for the ANOVA decomposition in Table 7 despite their data covering a period other than 2006-2008. However, similar findings are obtained if they are excluded from the sample. 30 regulations. In contrast, the importance of country characteristics for entry and exit suggests that fixed costs and uncertainty vary more at the country level than at the sector level. Table 7: ANOVA Decomposition at Country and HS 2-digit Sector Level Mean Median Number of Number of Number of Exports per Exports per Share of Top Entry Exit Survival Total Exports Products per Destinations Exporters Exporter Exporter 5% Exporters Rate Rate Rate Exporter per Exporter (mn USD) (mn USD) Country 10% 24% 3% 2% 28% 14% 37% 30% 29% 14% HS 2-digit Sector 9% 18% 7% 5% 24% 46% 12% 17% 17% 16% Residual 81% 58% 90% 93% 54% 41% 51% 55% 55% 70% Note: The variance decomposition shown is based on measures in the country-year-product (HS 2-digit) level dataset (file CYH2.dta) averaged across the 2006-2008 period for each country except Kuwait for which averages are across the 2009-2010 period and Portugal for which averages are across the 2003-2005 period. Stylized Fact 5: There is a tremendous skewness in exporters as total exports of developing countries are largely dominated by multi-product multi-destinations exporters but these account for a very small share of the number of exporters. The recent trade literature shows a dominant role for firms that export many products to many destinations in explaining trade flows (see, for example, Bernard, Jensen, Redding, and Schott, 2009 on the US; Eaton, Kortum and Kramarz, 2008 on France; and Amador and Opromolla, 2008 on Portugal). As mentioned in Section 3, a component of the Database is a set of matrix tables showing for each developing country and year the distribution of exporters and of total exports by the number of HS 6-digit products exported and the number of destination markets served that can provide an insight onto the importance of multi-product multi- destination exporters across developing countries and over time. Using the information from developing countries over the period 2006-2008, we calculate for each country the average share of exporters and of total exports accounted for by single- product single-destination firms and by firms exporting to more than four countries and to more than four destinations and show them in Table 8. Single-product single-destination firms 31 represent more than a third of exporters on average across all countries. They represent an even higher share over 40 percent in most African countries as well as in Albania and Mexico. The percentages shown in Table 8 are quite close to the 40 percent reported by Bernard, Jensen, Redding, and Schott (2009) for the U.S. 45 However, these single-product single-destination firms account for a minimal fraction of total exports, on average less than 3 percent across countries. That fraction is particularly low in Botswana, Costa Rica, and Niger at 0.5 percent or less. Table 8 also shows that firms exporting more than four products to more than four destinations represent a relatively small percentage of exporters, 12 percent on average across all countries. Those percentages do exhibit a substantial degree of heterogeneity across countries ranging from about 3.5 percent in Albania and Botswana to 21.5 percent in South Africa, 22.4 percent in Bangladesh, and 29.4 percent in Cambodia. These multi-product multi-destination firms account for a large share of total exports in all countries, more than 60 percent on average. In Cambodia, Costa Rica, and South Africa that share is actually close to 80 percent whereas in Albania and Niger it is less than 30 percent. 46 This high degree of concentration of total exports in the hands of a small number of multi-product multi-destination exporters can be rationalized by the model of multi-product multi-destination exporters developed by Bernard, Redding, and Schott (2011), where firms face fixed costs to export each product and serve each market. Only higher ability firms are able to generate variable profits to cover those fixed costs and thus supply a wider range of products to each market. 45 Note, however, that the U.S. percentage is based on products defined at the HS 10-digit level. 46 Focusing on three country examples, Appendix 4 shows further that only 1.2% of exporters in Tanzania, 2.3% in Colombia, and 2.2% in Mexico export more than 11 products to more than 11 destinations and these very rare exporters account for about 32% of total exports in Tanzania, 26% in Colombia, and 45% in Mexico. 32 Table 8: Share of Single-Product Single-Destination and Multi-Product Multi-Destination Exporters Share of Exporters Accounted for by: Share of Total Exports Accounted for by: Firms Exporting Firms Exporting Single-Product Single-Product More than 4 More than 4 Single-Destination Single-Destination Products to More Products to More Firms Firms than 4 Destinations than 4 Destinations ALB Albania 45.4% 3.4% 8.4% 13.4% BFA Burkina Faso 41.2% 13.4% 2.8% 66.5% BGD Bangladesh 26.5% 22.4% 2.0% 63.9% BGR Bulgaria 37.2% 11.2% 1.4% 74.8% BWA Botswana 38.9% 3.5% 0.4% 53.0% CHL Chile 38.5% 14.4% 0.8% 75.1% CMR Cameroon 39.1% 9.3% 3.6% 55.2% COL Colombia 33.1% 12.9% 3.0% 60.1% CRI Costa Rica 27.5% 18.3% 0.7% 79.4% DOM Dominican Republic 37.9% 10.4% 1.6% 60.3% ECU Ecuador 37.5% 8.8% 4.6% 54.6% EGY Egypt 33.9% 12.3% 2.6% 51.9% GTM Guatemala 27.7% 12.8% 1.5% 56.7% IRN Iran 34.3% 6.9% 5.7% 47.5% JOR Jordan 39.1% 13.8% 2.4% 52.7% KEN Kenya 35.3% 12.8% 2.6% 58.7% KHM Cambodia 25.5% 29.4% 1.1% 78.7% LBN Lebanon 31.3% 19.9% 2.6% 70.8% MAR Morocco 28.2% 12.6% 3.0% 51.9% MEX Mexico 40.1% 9.1% 1.2% 63.4% MKD Macedonia 35.2% 11.1% 1.7% 76.2% MLI Mali 35.8% 11.3% 2.4% 60.9% MUS Mauritius 26.6% 15.1% 3.4% 62.7% MWI Malawi 42.5% 5.6% 1.6% 57.9% NER Niger 41.3% 8.1% 0.5% 29.9% NIC Nicaragua 34.2% 8.6% 4.2% 53.4% PAK Pakistan 25.2% 18.6% 2.2% 68.9% PER Peru 29.8% 11.9% 3.8% 73.0% SEN Senegal 35.5% 19.2% 3.6% 64.0% SLV El Salvador 30.1% 14.3% 3.6% 72.5% TZA Tanzania 42.8% 10.2% 3.1% 75.1% UGA Uganda 42.9% 9.3% 3.7% 38.5% YEM Yemen 38.6% 11.1% 6.8% 54.9% ZAF South Africa 25.0% 21.5% 1.3% 82.3% Average 34.8% 12.8% 2.8% 60.5% Median 35.4% 12.1% 2.6% 60.6% Note: the shares are based on several cells in the matrix tables for all countries averaged over the period 2006-2008. Stylized Fact 6: Bilateral exports increase with the size of the destination market and decrease with distance and with bilateral tariffs. Most or all of these effects come from the number of exporting firms serving a destination market (extensive margin) rather than from the average exports per firm in those markets (intensive margin). 33 The gravity model of trade that relates bilateral trade flows between two countries to their economic size and variable trade costs is one of the most successful empirical models in economics (Anderson, 2011). Most studies focus on aggregate bilateral trade flows between countries but a few recent studies have begun to delve into the role of firms for the gravity equation but focusing on single countries only. 47 An exception is Mayer and Ottaviano (2008), who examine data from a handful of European countries. Using measures in the Database at the country-year-destination level, we provide estimates of a gravity equation for our large sample of countries (and their trading partners) that allow us to examine whether the effects of the classical determinants of bilateral trade –economic size and proxies for trade costs (distance and tariffs) along with the level of development– operate through firm export participation or through the average value exported per firm. We decompose exports from country i to partner country j for any given year (ignoring the year subscript) into the product of 𝑛𝑖𝑗 the number of country i firms exporting to country j (extensive margin) and 𝑠𝑖𝑗 = 𝑋𝑖𝑗 �𝑛𝑖𝑗 the average exports per firm for firms that export from country i to country j (intensive margin): 𝑋𝑖𝑗 = 𝑛𝑖𝑗 ∗ (𝑋𝑖𝑗 �𝑛𝑖𝑗) . The measures corresponding to the three elements in this decomposition are available in the Database at the country-year- destination level (CYD.dta) and are used in turn as dependent variables in the gravity equations whose estimates are shown in Table 9. Data on bilateral distances is taken from CEPII described in Mayer and Zignago (2011), and data on bilateral tariffs from Kee, Nicita, and Olarreaga (2009). 48 Since OLS estimation is used in Table 9, the coefficient on an independent variable in 47 For example Bernard, Jensen, Redding, and Schott (2007) and Lawless (2010) examine bilateral U.S. exports, Bastos and Silva (2012) examine bilateral Portuguese exports, and Eaton, Kortum, and Kramarz (2011) examine bilateral French exports. 48 We calculate the average bilateral tariff faced by origin country A when exporting to destination country B as the simple average of the applied tariffs imposed by B on all HS 6-digit products it imports from A in 2008 (the only year for which we have available data). 34 column (1) is equal to the sum of the coefficients on that same independent variable in columns (2) and (3), and the same is true for the other two sets of three columns (4)-(6) and (7)-(9). 49 The estimates in column (1) of Panel A show that bilateral exports increase significantly with GDP of both exporter and destination country and decrease significantly with distance, a finding that mimics those in all prior gravity studies. Columns (2) and (3) of Panel A show that both the number of exporters as well as average exports per firm increase significantly with exporter and destination country GDP but decrease significantly with distance. Importantly, most of the negative effect of distance on bilateral exports operates through the number of exporters: the coefficient of -1.223 in column (2) of Panel A accounts for 76 percent of the total effect of distance on bilateral exports. Most of the positive effects of exporter country size operate through the extensive margin but for the positive effects of destination country size, extensive and intensive margins play more equal roles. The estimates in column (4) of Panel A show that bilateral exports decrease significantly with the average tariff imposed by the destination country. Column (5) of Panel A shows that the number of exporters decreases significantly with tariffs while column (6) of Panel A shows that the average exporter size does not vary significantly with tariffs. Thus, the entire negative effect of bilateral tariffs on bilateral exports operates through the number of exporters. Columns (7)-(9) of Panel A include both distance and bilateral tariffs and provide qualitatively similar results to those when only one of the measures of trade costs are included. Also, Panel B of Table 9 shows the results from estimating gravity regressions where GDP of the exporter and the importer country are replaced by exporter and importer country fixed effects. 49 Since this gravity equation is meant to illustrate the types of analysis that can be done using the Database with the destination disaggregation level, the estimation is done by OLS and zero trade flows are not included in the sample. Future work using the Database can address selection problems and employ the novel estimation techniques for gravity equations proposed for example by Santos-Silva and Tenreyro (2004). 35 The predominance of the number of exporters in accounting for the negative effect of distance and of bilateral tariffs on bilateral trade is maintained in Panel B. Table 9: Gravity Equation Estimates Panel A. Including Exporter and Importer GDP Dependent Variable is: Ln Mean Ln Mean Ln Mean Ln Total Ln Number of Ln Total Ln Number of Ln Total Ln Number of Exports per Exports per Exports per Exports Exporters Exports Exporters Exports Exporters Firm Firm Firm Bilateral Bilateral Bilateral Bilateral Bilateral Bilateral Bilateral Bilateral Bilateral (1) (2) (3) (4) (5) (6) (7) (8) (9) Ln GDP of exporter 1.169*** 0.855*** 0.314*** 1.039*** 0.763*** 0.277*** 1.203*** 0.887*** 0.316*** (0.027) (0.016) (0.019) (0.031) (0.019) (0.019) (0.026) (0.014) (0.019) Ln GDP of destination 0.845*** 0.498*** 0.347*** 0.731*** 0.410*** 0.321*** 0.841*** 0.494*** 0.347*** (0.019) (0.011) (0.012) (0.021) (0.013) (0.013) (0.018) (0.010) (0.012) Ln bilateral distance -1.616*** -1.223*** -0.393*** -1.634*** -1.240*** -0.394*** (0.046) (0.028) (0.028) (0.045) (0.027) (0.028) Ln bilateral tariffs -1.196*** -1.118*** -0.078 -1.252*** -1.160*** -0.0915* (0.084) (0.053) (0.051) (0.076) (0.047) (0.049) Observations 2780 2780 2780 2780 2780 2780 2780 2780 2780 R-squared 0.553 0.615 0.251 0.426 0.469 0.212 0.594 0.698 0.252 Panel B. Including Exporter and Importer Fixed Effects Dependent Variable is: Ln Mean Ln Mean Ln Mean Ln Total Ln Number of Ln Total Ln Number of Ln Total Ln Number of Exports per Exports per Exports per Exports Exporters Exports Exporters Exports Exporters Firm Firm Firm Bilateral Bilateral Bilateral Bilateral Bilateral Bilateral Bilateral Bilateral Bilateral (1) (2) (3) (4) (5) (6) (7) (8) (9) Ln bilateral distance -2.118*** -1.535*** -0.582*** -2.052*** -1.487*** -0.565*** (0.052) (0.034) (0.032) (0.053) (0.035) (0.033) Ln bilateral tariffs -3.395*** -2.470*** -0.925*** -1.134*** -0.831*** -0.302*** (0.776) (0.559) (0.234) (0.260) (0.184) (0.117) Exporter fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Importer fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 2980 2980 2980 2980 2980 2980 2980 2980 2980 R-squared 0.742 0.834 0.439 0.585 0.637 0.393 0.744 0.837 0.44 Notes: Robust standard errors in parentheses; *** and * indicate significance at the 1 and 10 percent percent confidence levels, respectively. The dependent variables used in the regressions are measures in the country-year-destination level dataset (file CYD.dta) in the year 2008 for each country. The real GDP variable used is in PPP terms. Our evidence that significantly more exporters serve larger destination markets is consistent with trade models with heterogeneous firms. These models show that as the size of the export market increases, firms with lower productivity are able to generate sufficient variable profits to cover the fixed costs of exporting and thus the number of exporting firms increases (Bernard, Redding, and Schott, 2011). Distance is typically used in gravity equations to proxy for 36 variable transportation costs, which vary with the amount exported. But distance could also proxy for fixed trade costs related for example to informational networks (Lawless, 2010). Either way, our evidence that larger distance to trading partners as well as higher bilateral tariffs imposed by trading partners have an inhibiting effect in firm participation in exports is consistent with theory. Trade costs – fixed or variable – increase with distance, implying that lower productivity firms do not generate sufficient profits to cover the fixed costs of exporting and thus the number of exporting firms decreases to more distant destinations. While distance is not an actionable policy variable per se, it can be influenced by improvements in transportation or reductions in other trade barriers. Our findings suggest that improvements in transportation or reductions in other trade barriers would likely affect bilateral exports more through the number of exporters than through their average size. Regarding tariffs, which are a variable trade cost per se, our findings suggest that their reduction by partner countries affects bilateral exports primarily through the number of exporters. The results presented here based on a cross-section appear to be at odds with stylized fact 2 on export growth. In particular, our cross-country evidence shows that exporting-country size, destination-country demand, and trade costs largely affect aggregate exports via the number of exporters. In contrast, export growth within countries over time is largely driven by average exporter size and not by the number of exporters. This is likely to be related to the extraordinarily skewed distribution of firm size (recall that the top 5 percent account for over 80 percent of exports on average and the bottom third of the distribution, the single-product single- destination firms, account for less than 3 percent of exports). This implies that aggregate export growth is very dependent on firm growth at the top of the distribution. In contrast, when the export base expands, these are likely to be very small firms that enter, reducing average size, all 37 else equal. This means that shocks that expand both firm size and the number of firms, such as improved market access or lower trade costs, will increase average overall exporter size by a much smaller amount than incumbent exporter size since the entrants are primarily very small exporters. Thus, gravity-type regressions on average size and number of exporters will understate the importance of size effects on aggregate exports. Examining export growth over time, it is in countries where the shocks affect large firms most significantly that exports will grow most rapidly and this will tend to push up the average exporter size more than the number of exporters. This is precisely what we see in stylized fact 2. From a policy perspective, it also suggests that there may be important levers that can be used to enhance exports, such as attracting large multinationals. V. Concluding Remarks While the literature on exporter dynamics is rapidly growing, there has not been a comparison across a large number of countries at differing stages of development. The Exporter Dynamics Database presented in this paper addresses that limitation by providing policymakers and the research community with a set of measures of exporter characteristics and dynamics that allow for cross-country comparisons. The ultimate objective of this Database is to provide researchers with more data that will expand our understanding of how exporting happens and generate policy implications for countries seeking to expand their exports. The Database compiles measures for 45 countries (most of them developing countries) covering the universe of annual firm-level export transactions primarily for the period between 2003 and 2010. This paper presents six stylized facts that arise from the analysis of measures in the Database and their cross-country comparisons. The data point to wide variation in the export 38 base and characteristics of the exporting sector across countries, though some strong general patterns emerge. The facts show that exporting country size, stage of development, and sectoral characteristics explain a number of patterns at the firm level. They also highlight the tremendous skewness of exporter size and the importance of firm growth in aggregate export growth. While some of our stylized facts are consistent with facts shown by previous studies on individual countries, others reveal features that are not explained by current trade models with heterogeneous firms and thus, require further research to understand them fully. The measures in the Database will allow the examination of several interesting cross- country questions, cross-country cross-sector questions, and within-country questions. The measures can also be used as controls in estimations that require exporter characteristics at the country-industry level. In particular, as the measures in the Database offer the first opportunity to study exporter characteristics and dynamics on a global basis, some of the facts described above open the door to questions such as: How can countries attract more large multi-product firms? What determines entrant survival? How is comparative advantage related to the typical exporter characteristics in an industry? And many others that will hopefully be addressed in future research to be conducted using the Database. 39 References Amador, J. and L. Opromolla (2008). “Product and Destination Mix in Export Markets,� Banco de Portugal, Working Paper 17. Anderson, J. (2011). “The Gravity Model,� Annual Review of Economics (3): 133-160. Andersson, M., H. Lööf and S. Johansson (2008). “Productivity and International Trade: FirmLevel Evidence from a Small Open Economy,� Review of World Economics 144: 774-801. Arkolakis, C. (2010). “Market Penetration Costs and the New Consumers Margin in International Trade,� Journal of Political Economy 118: 1151-1199. Bastos, P. and J. Silva (2012). “Networks, Firms, and Trade.� Journal of International Economics 87: 352-364. Bernard, A., J. Eaton, J. Jensen and S. Kortum (2003). “Plants and Productivity in International Trade,� American Economic Review 93: 1268–90. Bernard, A., J. Jensen, and P. Schott (2009). “Importers, Exporters and Multinationals: A Portrait of Firms in the U.S. that Trade Goods� in T. Dunne, J. Jensen and M. Roberts (eds.), Producer Dynamics: New Evidence from Micro Data. University of Chicago Press. Bernard, A., S. Redding, and P. Schott (2007). “Comparative Advantage and Heterogeneous Firms,� Review of Economic Studies, 74: 31-66. Bernard, A., J. Jensen, S. Redding, and P. Schott (2007). “Firms in International Trade,� Journal of Economic Perspectives 21: 105-130. Bernard, A., J. Jensen, S. Redding, and P. Schott (2011). “The Empirics of Firm Heterogeneity and International Trade,� NBER Working paper No. 17627. Bernard, A., S. Redding and P. Schott (2007). “Comparative Advantage and Heterogeneous Firms,� Review of Economic Studies 74: 31–66. Bernard, A., S. Redding and P. Schott (2011). “Multi-Product Firms and Trade Liberalization,� Quarterly Journal of Economics 126: 1271-1318. Cebeci, T. (2012). “A Concordance among Harmonized System 1996, 2002 and 2007 Classifications,� World Bank mimeo available at http://econ.worldbank.org/exporter-dynamics- database. Chaney, T. (2008). “Distorted Gravity: The Intensive and Extensive Margins of International Trade,� American Economic Review 98: 1707-1721. 40 Das, S., M. Roberts and J. Tybout (2007). “Market Entry Costs, Producer Heterogeneity, and Export Dynamics,� Econometrica 75: 837-873. De Lucio, J., R. Mínguez-Fuentes, A. Minondo and F. Requena-Silvente (2011). “The Extensive and Intensive Margins of Spanish Trade,� International Review of Applied Economics 25: 615- 631. Eaton, J., M. Eslava, M. Kugler and J. Tybout (2008). “Export Dynamics in Colombia: Transactions Level Evidence,� Borradores de Economia No. 522, Banco de la Republica, Colombia. Eaton, J. and S. Kortum (2002). “Technology, Geography, and Trade,� Econometrica 70: 1741- 1779. Eaton, J., S. Kortum and F. Kramarz. (2008). “An Anatomy of International Trade: Evidence from French Firms,� Econometrica 79: 1453-1498. Eaton, J., S. Kortum and F. Sotelo. (2012). “International Trade: Linking Micro and Macro,� NBER Working Paper 17864. Ekholm, K., A. Moxnes and K. Ulltveit-Moe (2012). “Manufacturing Restructuring and the Role of Real Exchange Rate Shocks: A Firm-Level Analysis,� Journal of International Economics 86: 101-117. Fabling, R. and L. Sanderson (2012). “Whatever Next? Export Market Choices of New Zealand Firms� Papers in Regional Science 91: 137-159. Freund, C. and M. Pierola (2010). “Export Entrepreneurs Evidence from Peru,� Policy Research Working Paper No. 5407, The World Bank. Freund, C. and M. Pierola (2012). “Export Superstars,� Policy Research Working Paper No. 6222, The World Bank. Iacovone, L. and B. Javorcik (2010). “Multi-Product Exporters: Product Churning, Uncertainty, and Export Discoveries� Economic Journal 120: 481-499. Kee, H., A. Nicita and M. Olarreaga (2009) “Estimating Trade Restrictiveness Indices,� Economic Journal 119: 172-199. Lawless, M. (2010). “Deconstructing Gravity: Trade Costs and Extensive and Intensive Margins,� Canadian Journal of Economics 43: 1149-72. Melitz, M. and G. Ottaviano (2005). “Market Size, Trade, and Productivity.� National Bureau of Economic Research Working Paper 11393. Manova, K. and Z. Zhang (2012). “Export Prices across Firms and Destinations,� Quarterly Journal of Economics 127: 379-436. 41 Masso, J. and P. Vahter (2011). “Exporting and Productivity: the Effects of Multi-Market and Multi-Product Export Entry,� University of Tartu, Faculty of Economics and Business Administration Working Paper No. 83. Mayer, T. and G. Ottaviano (2008). “The Happy Few: The Internationalisation of European Firms,� Intereconomics: Review of European Economic Policy 43: 135-148. Mayer T. and S. Zignago (2011). “Notes on CEPII’s distances measures (GeoDist),� CEPII Working Paper No. 2011-25. Redding, S. (2011). “Theories of Heterogeneous Firms and Trade,� Annual Review of Economics forthcoming. Santos-Silva, J. and S. Tenreyro (2006). “The Log of Gravity,� Review of Economics and Statistics 88: 641-658. Schott, P. and J. Pierce (2012). “Concording U.S. Harmonized System Categories over Time,� Journal of Official Measures 28: 53-68. Wagner, R. and A. Zahler (2011). “New Exports from Emerging Markets: Do Followers Benefit from Pioneers?,� MPRA Paper No. 30312, University Library of Munich, Germany. United Nations (1998). International Merchandise Trade Statistics: Concepts and Definitions. United Nations (2008). International Merchandise Trade Statistics: Supplement to the Compilers Manual. 42 Appendix 1: Countries Included in the Database, Periods Available and Sources of Data a) Latin America and the Caribbean (LAC) Period Country Source available 1 Brazil 1997-2010 Government (Receita Federal) 2 Chile 2003 - 2009 Private company (Veritrade) 3 Colombia 2007 - 2009 Private company (Veritrade) Government (Promotora del Comercio 4 Costa Rica 1998 - 2009 Exterior de Costa Rica - PROCOMER) Dominican Government (Dirección General de 5 Republic 2002 - 2009 Aduanas) 6 Ecuador 2006 - 2009 Private company (Veritrade) Government (Dirección General de 7 El Salvador 2002 - 2009 Aduanas) Government (Superintendencia de 8 Guatemala 2003 -2010 Administración Tributaria) 9 Mexico 2000 - 2009 Government (Secretaría de Economía) Government (Dirección General de 10 Nicaragua 2002 - 2011 Servicios Aduaneros) Government (Superintendencia Nacional 11 Peru 1997 - 2008 de Administración Tributaria - SUNAT) b) Europe and Central Asia (ECA) Period Country Source available 12 Albania 2004 - 2009 Government (Ministry of Finance) 13 Bulgaria 2001 - 2006 Government (National Customs Agency) 14 Macedonia 2001 - 2010 Government (Customs Administration) Government (Turkish Statistical 15 Turkey 2002-2010 Institute) 43 c) Sub-Saharan Africa (SSA) Period Country Source available Government (Botswana Unified Revenue 16 Botswana 2003 - 2010 Service) Government (Direction Générale des 17 Burkina Faso 2005 - 2010 Douanes) 18 Cameroon 1997 - 2009 Government (Douanes Camerounaises) 19 Kenya 2006 - 2009 Government (Kenya Revenue Authority) Government (Malawi Revenue 20 Malawi 2006 - 2008 Authority) Government (Direction Générale des 21 Mali 2005 - 2008 Douanes) Government (Mauritius Revenue 22 Mauritius 2002 - 2009 Authority) Government (Direction Générale des 23 Niger 2008 - 2010 Douanes) Government (Direction Générale des 24 Senegal 2000 - 2010 Douanes) Government (South Africa Revenue 25 South Africa 2001 - 2009 Service) Government (Tanzania Revenue 26 Tanzania 2003 - 2009 Authority) 2000 - 2010 Government (Uganda Revenue 27 Uganda (except 2006) Authority) d) East Asia and the Pacific (EAP) Period Country Source available Government (General Department of 28 Cambodia 2000 - 2009 Customs and Excise of Cambodia) 29 Laos 2005 - 2010 Government (Lao PDR Customs) e) South Asia (SA) Period Country Source available Government (National Board of 30 Bangladesh 2005 - 2011 Revenue) Government (Federal Bureau of Pakistan 31 Pakistan 2002 - 2010 and Pakistan Customs Department) 44 f) MENA Period Country Source available Non-profit Organization (Economic 32 Egypt 2006 - 2010 Research Forum) Government (The Islamic Republic of 33 Iran 2006 - 2010 Iran Customs Administration - IRICA) Government (Ministry of Planning and 34 Jordan 2003 - 2010 International Cooperation) Government (Lebanese Customs 35 Lebanon 2008 - 2010 Administration) Government (Administration des 36 Morocco 2002 - 2010 Douanes et Impôts Indirects) Government (Yemen Customs 37 Yemen 2006 - 2010 Authority) g) Others Period Country Source available Researcher: Emmanuel Dhyne (Bank of Belgium), Luc Dresse (Bank of Belgium 1997 - 2010 Belgium), Cedric Duprez (Bank of Belgium), Hylke Vandenbussche 38 (Université Catholique de Louvain) Researcher: Jaan Masso (University of Estonia 1997 - 2009 Tartu) and Priit Vahter (University of 39 Birmingham) Government (General Administration of Kuwait 2009 - 2010 Customs, Office of Public Scrutiny, 40 Statistics, and Conservation) Researcher: Lynda Sanderson (The New Zealand 1999-2010 Treasury) and Richard Fabling (Motu 41 Economic and Public Policy Research) Researcher: Andreas Moxnes Norway 1997 - 2006 42 (Dartmouth College) 43 Portugal 1997 - 2005 Researcher: Joana Silva (World Bank) Researchers: Juan de Lucio (High Council of Spanish Chambers of Commerce), Raul Minguez (High Council of Spanish Chambers of Commerce), Asier Minondo (Deusto Business School), Francisco Requena, 44 Spain 2005 - 2009 (University of Valencia) Researcher: Martin Andersson (Lund University) and Lina Ahlin (Lund 45 Sweden 1997 - 2006 University) 45 Appendix 2: Country-Year Comparison, Comtrade Data Total Exports Customs Data Total Exports WITS Match Total Exports Customs Data Total Exports WITS Match Country Year Country Year (US$ millions) (US$ millions) Ratio (US$ millions) (US$ millions) Ratio ALB Albania 2004 649 586 111 BRA Brazil 2006 138,000 127,000 109 ALB Albania 2005 712 639 111 BRA Brazil 2007 161,000 147,000 109 ALB Albania 2006 789 758 104 BRA Brazil 2008 197,000 179,000 110 ALB Albania 2007 1,050 996 105 BRA Brazil 2009 153,000 139,000 110 ALB Albania 2008 1,320 1,230 107 BRA Brazil 2010 202,000 178,000 114 ALB Albania 2009 1,140 961 118 BWA Botswana 2003 2,270 3,800 60 BEL Belgium 1997 139,000 - BWA Botswana 2004 3,240 3,510 92 BEL Belgium 1998 144,000 - BWA Botswana 2005 4,270 4,430 96 BEL Belgium 1999 144,000 174,000 83 BWA Botswana 2006 4,430 4,500 99 BEL Belgium 2000 141,000 176,000 80 BWA Botswana 2007 4,990 5,060 99 BEL Belgium 2001 142,000 182,000 78 BWA Botswana 2008 4,290 4,930 87 BEL Belgium 2002 160,000 206,000 78 BWA Botswana 2009 2,860 3,440 83 BEL Belgium 2003 190,000 242,000 79 BWA Botswana 2010 4,280 4,680 92 BEL Belgium 2004 230,000 289,000 80 CHL Chile 2003 19,000 21,200 90 BEL Belgium 2005 247,000 311,000 79 CHL Chile 2004 29,600 32,000 92 BEL Belgium 2006 270,000 341,000 79 CHL Chile 2005 36,800 40,400 91 BEL Belgium 2007 318,000 402,000 79 CHL Chile 2006 53,500 57,600 93 BEL Belgium 2008 339,000 433,000 78 CHL Chile 2007 63,400 67,100 94 BEL Belgium 2009 274,000 344,000 80 CHL Chile 2008 65,800 64,700 102 BEL Belgium 2010 321,000 375,000 86 CHL Chile 2009 47,600 53,200 90 BFA Burkina Faso 2005 348 332 105 CMR Cameroon 1997 1,140 - BFA Burkina Faso 2006 458 - CMR Cameroon 1998 1,170 - BFA Burkina Faso 2007 493 453 109 CMR Cameroon 1999 1,040 - BFA Burkina Faso 2008 501 470 107 CMR Cameroon 2000 860 836 103 BFA Burkina Faso 2009 622 795 78 CMR Cameroon 2001 872 839 104 BFA Burkina Faso 2010 1,310 1,290 102 CMR Cameroon 2002 915 919 99 BGD Bangladesh 2005 8,910 9,280 96 CMR Cameroon 2003 1,130 1,170 97 BGD Bangladesh 2006 13,100 11,600 113 CMR Cameroon 2004 1,230 1,320 93 BGD Bangladesh 2007 9,070 12,900 70 CMR Cameroon 2005 1,230 1,230 100 BGD Bangladesh 2008 15,100 - CMR Cameroon 2006 1,360 1,360 100 BGD Bangladesh 2009 15,100 - CMR Cameroon 2007 1,750 1,750 100 BGD Bangladesh 2010 18,900 - CMR Cameroon 2008 2,120 2,100 101 BGD Bangladesh 2011 24,000 - CMR Cameroon 2009 1,710 1,720 100 BGR Bulgaria 2001 4,770 4,660 102 COL Colombia 2007 18,500 18,500 100 BGR Bulgaria 2002 5,650 5,220 108 COL Colombia 2008 19,800 19,800 100 BGR Bulgaria 2003 7,170 7,100 101 COL Colombia 2009 16,700 16,600 100 BGR Bulgaria 2004 9,190 8,850 104 CRI Costa Rica 1998 4,840 5,130 94 BGR Bulgaria 2005 10,300 10,100 102 CRI Costa Rica 1999 6,230 6,260 99 BGR Bulgaria 2006 12,900 12,900 100 CRI Costa Rica 2000 5,530 5,450 101 BRA Brazil 1997 52,800 52,700 100 CRI Costa Rica 2001 4,830 4,680 103 BRA Brazil 1998 51,100 50,800 101 CRI Costa Rica 2002 5,060 4,900 103 BRA Brazil 1999 48,800 47,600 103 CRI Costa Rica 2003 5,890 5,770 102 BRA Brazil 2000 55,800 54,200 103 CRI Costa Rica 2004 6,060 5,950 102 BRA Brazil 2001 58,300 56,200 104 CRI Costa Rica 2005 6,730 7,120 95 BRA Brazil 2002 60,400 57,500 105 CRI Costa Rica 2006 7,930 7,210 110 BRA Brazil 2003 73,000 69,400 105 CRI Costa Rica 2007 9,040 8,880 102 BRA Brazil 2004 96,700 92,300 105 CRI Costa Rica 2008 9,140 9,650 95 BRA Brazil 2005 120,000 111,000 107 CRI Costa Rica 2009 8,390 8,790 96 46 Total Exports Customs Data Total Exports WITS Match Total Exports Customs Data Total Exports WITS Match Country Year Country Year (US$ millions) (US$ millions) Ratio (US$ millions) (US$ millions) Ratio DOM Dominican Rep 2002 3,390 5,130 66 JOR Jordan 2004 1,470 3,180 46 DOM Dominican Rep 2003 3,730 5,350 70 JOR Jordan 2005 1,940 3,550 55 DOM Dominican Rep 2004 3,550 5,720 62 JOR Jordan 2006 2,750 3,990 69 DOM Dominican Rep 2005 3,960 5,860 68 JOR Jordan 2007 2,960 4,380 68 DOM Dominican Rep 2006 4,170 6,080 69 JOR Jordan 2008 4,490 6,040 74 DOM Dominican Rep 2007 4,460 6,400 70 JOR Jordan 2009 4,540 4,930 92 DOM Dominican Rep 2008 5,000 5,890 85 JOR Jordan 2010 4,880 5,780 84 DOM Dominican Rep 2009 4,540 4,370 104 KEN Kenya 2006 3,240 3,240 100 ECU Ecuador 2006 4,730 5,170 91 KEN Kenya 2007 3,900 3,900 100 ECU Ecuador 2007 5,510 5,510 100 KEN Kenya 2008 4,780 4,480 107 ECU Ecuador 2008 6,930 7,030 99 KEN Kenya 2009 4,270 4,270 100 ECU Ecuador 2009 7,090 6,810 104 KHM Cambodia 2000 1,050 1,390 75 EGY Egypt 2006 11,400 - KHM Cambodia 2001 1,250 1,500 84 EGY Egypt 2007 13,900 - KHM Cambodia 2002 1,460 1,920 76 EGY Egypt 2008 17,700 14,400 123 KHM Cambodia 2003 1,750 2,120 83 EGY Egypt 2009 16,100 17,200 93 KHM Cambodia 2004 2,140 2,800 77 EGY Egypt 2010 18,800 18,700 100 KHM Cambodia 2005 2,380 3,020 79 ESP Spain 2005 181,000 184,000 98 KHM Cambodia 2006 2,890 3,570 81 ESP Spain 2006 200,000 204,000 98 KHM Cambodia 2007 2,920 3,530 83 ESP Spain 2007 236,000 242,000 98 KHM Cambodia 2008 4,330 4,360 99 ESP Spain 2008 253,000 261,000 97 KHM Cambodia 2009 4,950 4,990 99 ESP Spain 2009 207,000 213,000 97 KWT Kuwait 2009 2,840 3,420 83 EST Estonia 1997 1,860 2,750 68 KWT Kuwait 2010 3,220 - EST Estonia 1998 2,150 3,120 69 LAO Laos 2006 459 - EST Estonia 1999 2,710 2,880 94 LAO Laos 2007 406 - EST Estonia 2000 3,530 3,660 96 LAO Laos 2008 962 - EST Estonia 2001 3,580 3,840 93 LAO Laos 2009 1,090 - EST Estonia 2002 3,820 4,110 93 LBN Lebanon 2008 3,410 3,470 98 EST Estonia 2003 5,140 5,390 95 LBN Lebanon 2009 3,440 3,450 100 EST Estonia 2004 5,840 6,230 94 LBN Lebanon 2010 4,210 4,220 100 EST Estonia 2005 6,790 7,650 89 MAR Morocco 2002 7,490 7,540 99 EST Estonia 2006 7,580 8,480 89 MAR Morocco 2003 8,410 8,510 99 EST Estonia 2007 9,280 10,300 90 MAR Morocco 2004 9,330 9,430 99 EST Estonia 2008 11,000 12,100 91 MAR Morocco 2005 10,500 10,500 100 EST Estonia 2009 7,880 8,730 90 MAR Morocco 2006 12,100 12,000 101 GTM Guatemala 2003 4,170 2,420 172 MAR Morocco 2007 14,400 13,900 104 GTM Guatemala 2004 4,610 2,690 171 MAR Morocco 2008 19,300 19,500 99 GTM Guatemala 2005 4,970 5,080 98 MAR Morocco 2009 13,300 13,600 98 GTM Guatemala 2006 5,480 2,920 188 MAR Morocco 2010 17,000 17,200 99 GTM Guatemala 2007 6,320 6,550 97 MEX Mexico 2000 150,000 150,000 100 GTM Guatemala 2008 7,090 7,200 99 MEX Mexico 2001 146,000 146,000 100 GTM Guatemala 2009 6,620 6,920 96 MEX Mexico 2002 146,000 146,000 100 GTM Guatemala 2010 7,800 8,080 97 MEX Mexico 2003 146,000 146,000 100 IRN Iran 2006 11,200 10,500 106 MEX Mexico 2004 164,000 165,000 100 IRN Iran 2007 12,100 - MEX Mexico 2005 182,000 182,000 100 IRN Iran 2008 15,000 - MEX Mexico 2006 211,000 211,000 100 IRN Iran 2009 16,100 - MEX Mexico 2007 228,000 229,000 100 IRN Iran 2010 21,000 24,300 86 MEX Mexico 2008 240,000 241,000 100 JOR Jordan 2003 1,140 2,270 50 MEX Mexico 2009 198,000 199,000 100 47 Total Exports Customs Data Total Exports WITS Match Total Exports Customs Data Total Exports WITS Match Country Year Country Year (US$ millions) (US$ millions) Ratio (US$ millions) (US$ millions) Ratio MKD Macedonia 2001 532 1,110 48 NOR Norway 1999 21,500 22,800 95 MKD Macedonia 2002 534 1,090 49 NOR Norway 2000 21,000 21,600 97 MKD Macedonia 2003 659 1,290 51 NOR Norway 2001 22,200 22,600 98 MKD Macedonia 2004 953 1,600 60 NOR Norway 2002 22,800 23,400 97 MKD Macedonia 2005 1,230 1,880 66 NOR Norway 2003 26,300 26,400 100 MKD Macedonia 2006 1,590 2,180 73 NOR Norway 2004 29,700 30,000 99 MKD Macedonia 2007 2,380 3,190 75 NOR Norway 2005 33,400 33,500 99 MKD Macedonia 2008 2,670 - NOR Norway 2006 39,100 39,300 99 MKD Macedonia 2009 1,670 2,660 63 NZL New Zealand 1999 11,800 12,200 97 MKD Macedonia 2010 2,280 - NZL New Zealand 2000 12,600 12,400 101 MLI Mali 2005 571 1,070 53 NZL New Zealand 2001 13,100 12,900 101 MLI Mali 2006 960 1,520 63 NZL New Zealand 2002 13,700 13,500 102 MLI Mali 2007 637 1,430 44 NZL New Zealand 2003 15,900 15,600 102 MLI Mali 2008 872 1,890 46 NZL New Zealand 2004 19,500 19,200 102 MUS Mauritius 2002 2,290 1,750 131 NZL New Zealand 2005 20,700 20,200 102 MUS Mauritius 2003 2,790 1,860 150 NZL New Zealand 2006 21,400 20,600 104 MUS Mauritius 2004 2,570 2,000 128 NZL New Zealand 2007 25,100 24,600 102 MUS Mauritius 2005 2,800 1,560 180 NZL New Zealand 2008 27,300 27,200 100 MUS Mauritius 2006 2,670 2,330 114 NZL New Zealand 2009 22,700 22,600 101 MUS Mauritius 2007 2,510 2,230 113 NZL New Zealand 2010 28,600 28,200 101 MUS Mauritius 2008 2,500 2,400 104 PAK Pakistan 2002 6,770 - MUS Mauritius 2009 2,100 1,770 119 PAK Pakistan 2003 11,000 11,600 95 MWI Malawi 2006 470 662 71 PAK Pakistan 2004 12,200 12,600 97 MWI Malawi 2007 692 868 80 PAK Pakistan 2005 13,800 15,300 90 MWI Malawi 2008 726 878 83 PAK Pakistan 2006 15,100 16,000 95 NER Niger 2008 346 429 81 PAK Pakistan 2007 16,400 16,200 101 NER Niger 2009 376 621 61 PAK Pakistan 2008 18,900 18,700 101 NER Niger 2010 432 476 91 PAK Pakistan 2009 17,200 16,600 104 NIC Nicaragua 2002 545 622 88 PAK Pakistan 2010 20,300 19,800 103 NIC Nicaragua 2003 596 596 100 PER Peru 1997 6,200 - NIC Nicaragua 2004 753 751 100 PER Peru 1998 5,310 5,420 98 NIC Nicaragua 2005 847 852 99 PER Peru 1999 5,740 5,680 101 NIC Nicaragua 2006 1,050 752 140 PER Peru 2000 6,500 6,460 101 NIC Nicaragua 2007 1,250 1,180 106 PER Peru 2001 6,490 6,410 101 NIC Nicaragua 2008 1,540 2,530 61 PER Peru 2002 7,190 7,180 100 NIC Nicaragua 2009 1,430 1,380 103 PER Peru 2003 8,410 8,360 101 NIC Nicaragua 2010 2,010 1,820 110 PER Peru 2004 12,200 12,000 101 NIC Nicaragua 2011 2,440 - PER Peru 2005 15,600 15,500 101 NOR Norway 1997 21,700 22,400 97 PER Peru 2006 21,900 21,900 100 NOR Norway 1998 22,400 22,900 98 PER Peru 2007 25,800 25,600 101 48 Total Exports Customs Data Total Exports WITS Match Total Exports Customs Data Total Exports WITS Match Country Year Country Year (US$ millions) (US$ millions) Ratio (US$ millions) (US$ millions) Ratio PER Peru 2008 27,900 28,400 98 TUR Turkey 2002 32,900 33,000 99 PER Peru 2009 24,600 24,600 100 TUR Turkey 2003 43,600 44,200 99 PRT Portugal 1997 18,400 23,000 80 TUR Turkey 2004 58,600 59,100 99 PRT Portugal 1998 18,500 23,800 78 TUR Turkey 2005 66,900 67,800 99 PRT Portugal 1999 23,700 24,000 99 TUR Turkey 2006 78,000 79,000 99 PRT Portugal 2000 23,400 23,700 99 TUR Turkey 2007 98,100 99,200 99 PRT Portugal 2001 23,300 23,600 99 TUR Turkey 2008 120,000 121,000 99 PRT Portugal 2002 24,900 25,300 98 TUR Turkey 2009 95,000 96,300 99 PRT Portugal 2003 30,500 31,100 98 TUR Turkey 2010 106,000 107,000 98 PRT Portugal 2004 35,200 43,300 81 TZA Tanzania 2003 1,240 1,130 110 PRT Portugal 2005 34,800 36,500 95 TZA Tanzania 2004 1,480 1,320 112 SEN Senegal 2000 262 596 44 TZA Tanzania 2005 1,670 1,500 111 SEN Senegal 2001 345 643 54 TZA Tanzania 2006 1,830 1,690 108 SEN Senegal 2002 493 262 188 TZA Tanzania 2007 2,090 1,950 107 SEN Senegal 2003 616 922 67 TZA Tanzania 2008 2,850 3,050 94 SEN Senegal 2004 677 1,060 64 TZA Tanzania 2009 2,950 2,960 100 SEN Senegal 2005 804 1,160 69 UGA Uganda 2000 345 343 101 SEN Senegal 2006 708 261 271 UGA Uganda 2001 393 390 101 SEN Senegal 2007 848 1,250 68 UGA Uganda 2002 350 405 86 SEN Senegal 2008 1,140 1,430 79 UGA Uganda 2003 470 423 111 SEN Senegal 2009 1,280 1,580 81 UGA Uganda 2004 637 529 120 SEN Senegal 2010 1,200 1,650 73 UGA Uganda 2005 819 661 124 SLV El Salvador 2002 3,040 2,930 104 UGA Uganda 2007 1,240 1,050 118 SLV El Salvador 2003 3,180 3,060 104 UGA Uganda 2008 1,150 1,280 90 SLV El Salvador 2004 3,400 3,240 105 UGA Uganda 2009 976 969 101 SLV El Salvador 2005 3,530 3,370 105 UGA Uganda 2010 1,090 1,060 102 SLV El Salvador 2006 3,580 3,640 98 YEM Yemen 2006 322 322 100 SLV El Salvador 2007 4,030 3,890 104 YEM Yemen 2007 385 406 95 SLV El Salvador 2008 5,050 4,470 113 YEM Yemen 2008 445 451 99 SLV El Salvador 2009 4,050 3,750 108 YEM Yemen 2009 434 441 99 SWE Sweden 1997 76,000 79,700 95 YEM Yemen 2010 477 - SWE Sweden 1998 78,800 83,400 94 ZAF South Africa 2001 24,800 22,900 108 SWE Sweden 1999 78,400 74,000 106 ZAF South Africa 2002 22,700 20,300 112 SWE Sweden 2000 80,200 84,100 95 ZAF South Africa 2003 29,300 28,500 103 SWE Sweden 2001 71,000 73,800 96 ZAF South Africa 2004 37,100 36,600 101 SWE Sweden 2002 75,100 80,600 93 ZAF South Africa 2005 42,800 42,100 102 SWE Sweden 2003 92,100 99,200 93 ZAF South Africa 2006 48,300 47,600 102 SWE Sweden 2004 112,000 119,000 94 ZAF South Africa 2007 57,900 57,200 101 SWE Sweden 2005 116,000 124,000 94 ZAF South Africa 2008 70,200 66,800 105 SWE Sweden 2006 130,000 139,000 93 ZAF South Africa 2009 43,100 47,800 90 49 Appendix 3: Database Measures, GDP and Export Share Appendix 3 Figure 1: Selected Database Measures and GDP Number of Exporters - GDP Mean Exports per Exporter - GDP 12 17 ESP TUR BEL Ln Mean Exports per Exporter SWE MEX 10 16 BEL Ln Number of Exporters ZAF BRA CHL BRA PRT NOR BGR NZLPAK IRN MEX EGY COL KHM BGD PER CHL LBN ESTKEN MAR SWE GTM PER 15 KWT 8 MKD ECU MUS SLV CRI DOM CRI MAR ALB MLI BW A ZAF BW ATZA JOR ESP NIC NER PRT NOR TUR EST CMR BGD COL UGA CMR JOR ECU NZL EGY SEN SLV DOM MWI KHM GTM YEM 14 LAO BFA LAO UGA SEN 6 BFA TZA MLI MWI NIC MUS PAK BGR IRN KEN KWT NER MKD YEM LBN ALB R2=0.90 R2=0.52 13 4 22 24 26 28 22 24 26 28 Ln GDP Ln GDP Share of Top 5% Exporters - GDP Number of Products per Exporter - GDP 3 1 BWA CHL Ln Number of Products per Exporter MLI PER NOR ZAF ZAF MWI SWE MEX NZL NER LAO 2.5 MUS Share of Top 5% Exporters TZA DOM KWT ESP BFA BEL MKD JOR CMR CRIBGR BRA SLV KEN COL .8 LBN ECU EGY TUR MUS BEL TUR UGA GTM PRT PRT NIC KHM MAR EST LBN GTM NZL 2 PAK IRN KEN PER SEN SLV MEX EST BW A MAR SWE SEN BGR IRN NIC CRI PAK ALB YEM NOR COL .6 1.5 DOM ESP MKD YEM ECU BGD KWT CHL MWI TZA CMR NERBFA MLI UGA BGD ALB JOR 1 KHM R2=0.41 .4 LAO 22 24 26 28 22 24 26 28 Ln GDP Ln GDP Number of Destinations per Exporter - GDP Entry Rate - GDP 2 MWI LAO YEM BEL TZA .5 Ln Number of Destinations per Exporter CMR IRN UGA KHM 1.5 SWE BFA EST DOM MLI TUR ESP BW A BGD ECU ZAF .4 PRT CHL NOR SEN KEN Entry Rate SEN JOR CRI PAK ALB PER CHL ESP LBN NZL MKD JOR BGR NOR CMR COL 1 MUS EST EGY PER NIC KENGTM MAR MEX BFA UGA YEM BGR ECU SLV TZA KHM DOM MAR MLI MKD MEX TUR NIC IRN SLV COL BEL MWI KWT GTM MUS .3 PRT CRI NZL ZAF SWE .5 LAO BGD PAK NER ALB BW A EGY BRA R2=0.54 R2=-0.52 .2 0 22 24 26 28 22 24 26 28 Ln GDP Ln GDP 50 Exit Rate - GDP Survival Rate - GDP MWI BGD .6 .6 KHM PAK YEM TUR BRA IRN EGY .5 .5 LAO ZAF JOR CRI CMR NIC ALB Survival Rate TZA MLI MKD PER PRT Exit Rate KEN SLV DOM MUS MAR BFA GTM NZL COL BFA ECU IRN BW A EST .4 .4 LAO MLI BGR SEN DOM BEL ESP BW A MEX UGA SEN ECU NOR MEX CHL PER CHL KEN NIC MKDALB MAR MUS JOR TZA COL EST .3 .3 KHM SLV ESP GTM UGA NZL PRT BEL TUR SWE CRI EGY PAK ZAF MWI BRA CMR BGD R2=-0.40 R2=0.16 .2 .2 22 24 26 28 22 24 26 28 Ln GDP Ln GDP Notes: The measures plotted are based on measures in the country-year level dataset (file CY.dta) averaged across the 2006-2008 period for each country for each country except Kuwait for which averages are across the period 2009-2010 and Portugal for which averages are across the 2003-2005 period. GDP per is in current USD. Appendix 3 Figure 2: Selected Database Measures and Ratio of Exports to GDP Number of Exporters - Exports/GDP Mean Exports per Exporter - Exports/GDP 12 17 ESP TUR BEL Ln Mean Exports per Exporter MEX SWE 10 16 BEL Ln Number of Exporters BRA ZAF BRA CHL NOR PRT IRN PAK NZL BGR MEX COL EGY KHM BGD PER CHL LBNKEN MAR EST SWE GTM PER 15 KWT 8 ECU DOM SLV MKD CRI MUS CRI ALBTZA JOR MLI ESPMAR ZAF BWA BWA NIC NER NORTUR PRT UGA CMR COLECU BGD CMR NZL EST SEN DOM JOR EGY SLV MWI KHM GTM YEM 14 BFA LAO UGA LAO SEN 6 BFA TZA PAK MUS MLI MWI NIC IRN BGR KWT KEN NER YEM MKD LBN ALB R2=0.26 R2=0.54 13 4 0 .2 .4 .6 .8 0 .2 .4 .6 .8 Exports/GDP Exports/GDP Share of Top 5% Exporters - Exports/GDP Number of Products per Exporter - Exports/GDP 3 1 BWA CHL Ln Number of Products per Exporter NOR MLI PER ZAF ZAF MWI MEX SWE NZL NER LAO 2.5 MUS Share of Top 5% Exporters TZAESP KWT BFA DOM BEL CMR JOR MKD CRI BGR COLBRA KEN SLV .8 ECU EGY TUR TUR BEL LBN GTM MUS UGA PRT PRT KHM NIC MAR LBN GTM EST KEN NZL 2 IRN PAK PER SEN SLV MEX EST MAR SWE BWA IRNSEN NIC BGR PAK CRI YEM ALB NOR COL .6 1.5 DOM ESP YEM KWT ECU BGD MKD CHL CMR TZA MWI NERUGA BFA MLI BGD ALB JOR 1 KHM R2=0.45 .4 LAO 0 .2 .4 .6 .8 0 .2 .4 .6 .8 Exports/GDP Exports/GDP 51 Number of Destinations per Exporter - Exports/GDP Entry Rate - Exports/GDP 2 YEM LAO MWI BEL TZA .5 Ln Number of Destinations per Exporter IRN CMR UGA KHM 1.5 SWE BFA DOM MLI EST ESP TUR BWA BGD ECU ZAF .4 NOR PRT CHL SEN KEN Entry Rate SENPAK JOR CRI ALB ESP PERMKD LBN NZL CHL BGR NOR JOR CMR COL 1 EGY PER MUS EST NIC KEN MAR GTM MEX YEM BFA UGA TZA ECU SLV BGR DOM MAR KHM MLI MEX MKD TUR IRN NIC COL SLV BEL KWT MWI GTM MUS .3 PRT NZL CRI ZAF SWE .5 LAO PAK BGD NER ALB BWA EGY BRA R2=0.44 R2=-0.28 .2 0 0 .2 .4 .6 .8 0 .2 .4 .6 .8 Exports/GDP Exports/GDP Exit Rate - Exports/GDP Survival Rate - Exports/GDP MWI BGD .6 .6 PAK KHM YEM BRA TUR IRN EGY .5 .5 LAO JORZAF CRI CMR TZA ALB NIC Survival Rate MLI PRT PERMKD Exit Rate KEN SLV DOM MAR MUS BFA COL GTM NZL BFA IRN ECU EST .4 .4 LAO MLI BWA BGR SENDOM BEL ESP MEX BWA UGA SENNOR ECU MEX CHL PER CHL KEN ALB NIC MKD MAR JOR MUS TZA COL EST .3 .3 SLV KHM ESP TUR GTM NZL PRT UGA BEL SWE EGY PAK ZAF CRI MWI BRA CMR BGD R2=-0.28 R2=-0.02 .2 .2 0 .2 .4 .6 .8 0 .2 .4 .6 .8 Exports/GDP Exports/GDP Notes: The measures plotted are based on measures in the country-year level dataset (file CY.dta) averaged across the 2006-2008 period for each country for each country except Kuwait for which averages are across the period 2009-2010 and Portugal for which averages are across the 2003-2005 period. 52 Appendix 4: Joint Distribution of Exporters and of Total Exports across Number of Products and Destinations in Tanzania, Colombia and Mexico Appendix 4 Table 1: Joint Distribution of Exporters Number of Destinations 1 2 3 4 to 10 11 to 20 21 or more Total 1 42.3 4.6 1.3 2.2 0.2 0.0 50.6 2 7.8 5.5 2.3 2.0 0.2 0.0 17.9 Number of Products 3 2.0 2.2 1.7 1.5 0.1 0.0 7.5 4 to 10 5.4 3.0 1.8 5.5 0.7 0.4 16.8 11 to 20 1.3 0.6 0.2 1.5 0.5 0.0 4.2 21 or more 0.6 0.3 0.3 1.1 0.5 0.2 3.0 Total 59.5 16.3 7.5 13.9 2.2 0.5 100.0 Panel B. Colombia Number of Destinations 1 2 3 4 to 10 11 to 20 21 or more Total 1 32.0 4.5 1.4 2.3 0.5 0.1 40.7 2 8.5 4.1 1.6 2.0 0.2 0.1 16.4 Number of Products 3 4.1 2.3 1.4 1.7 0.2 0.0 9.8 4 to 10 8.1 4.1 2.5 6.0 1.2 0.2 22.1 11 to 20 1.9 0.8 0.6 2.2 0.9 0.2 6.5 21 or more 1.0 0.4 0.4 1.6 0.8 0.4 4.6 Total 55.5 16.2 7.8 15.7 3.7 1.0 100.0 Panel C. Mexico Number of Destinations 1 2 3 4 to 10 11 to 20 21 or more Total 1 40.3 1.8 0.5 0.5 0.1 0.0 43.2 2 10.6 2.5 0.6 0.6 0.1 0.0 14.4 Number of Products 3 5.5 1.5 0.6 0.6 0.1 0.0 8.3 4 to 10 11.0 3.3 1.9 3.1 0.5 0.1 19.8 11 to 20 3.4 0.9 0.6 1.7 0.5 0.2 7.1 21 or more 2.8 0.8 0.4 1.7 0.9 0.6 7.2 Total 73.5 10.7 4.6 8.2 2.1 0.8 100.0 Note: Each cell in a panel of the table represents the share of firms exporting a given number of products (shown in the row) to a given number of destinations (shown in the column) in 2007. Appendix 4 Table 2: Joint Distribution of Total Exports Panel A. Tanzania Number of Destinations 1 2 3 4 to 10 11 to 20 21 or more Total 1 4.2 0.8 0.7 1.8 2.0 0.0 9.5 2 0.5 1.7 1.8 2.4 2.0 0.0 8.4 Number of Products 3 0.1 0.3 0.4 1.8 0.0 0.0 2.7 4 to 10 0.7 2.0 0.8 8.8 3.7 9.9 25.9 11 to 20 0.5 0.1 0.2 6.9 4.8 0.0 12.5 21 or more 0.2 0.8 0.3 12.6 22.4 4.7 41.1 Total 6.2 5.7 4.3 34.2 34.9 14.6 100.0 Panel B. Colombia Number of Destinations 1 2 3 4 to 10 11 to 20 21 or more Total 1 2.7 1.2 0.6 2.3 1.8 1.0 9.5 2 2.0 1.8 0.5 2.3 0.6 4.0 11.1 Number of Products 3 1.7 1.1 0.3 1.3 0.2 1.6 6.3 4 to 10 2.7 1.2 0.9 7.4 14.4 5.8 32.5 11 to 20 0.6 0.3 0.4 2.6 6.3 3.2 13.4 21 or more 2.0 0.5 0.4 8.0 8.7 7.6 27.0 Total 11.7 6.1 3.1 23.9 31.9 23.3 100.0 Panel C. Mexico Number of Destinations 1 2 3 4 to 10 11 to 20 21 or more Total 1 1.1 0.2 0.1 0.2 0.0 0.0 1.6 2 1.1 0.2 0.1 0.3 0.1 0.0 1.8 Number of Products 3 0.7 0.2 0.1 0.2 0.1 0.1 1.4 4 to 10 3.4 1.5 0.6 2.4 2.1 1.2 11.3 11 to 20 4.2 0.9 0.4 2.1 2.0 1.1 10.7 21 or more 13.9 4.6 2.4 10.4 14.0 27.9 73.1 Total 24.4 7.6 3.8 15.7 18.3 30.2 100.0 Note: Each cell in a panel of the table represents the share of total exports in 2007 accounted for by firms exporting a given number of products (shown in the row) to a given number of destinations (shown in the column). 53