Machinery production networks in Latin America: A quantity and quality analysis

Chang, Mateus Silva Article Machinery production networks in Latin America: A quantity and quality analysis Latin American Economic Review Provided in Cooperation with: Centro de Investigación y Docencia Económica (CIDE), Mexico City Suggested Citation: Chang, Mateus Silva (2017) : Machinery production networks in Latin America: A quantity and quality analysis, Latin American Economic Review, ISSN 2196-436X, Springer, Heidelberg, Vol. 26, Iss. 1, pp. 1-22, https://doi.org/10.1007/s40503-017-0054-5 This Version is available at: https://hdl.handle.net/10419/195245 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Machinery production networks in Latin America: a quantity and quality analysis Mateus Silva Chang 1 Received: 18 October 2016 / Revised: 28 July 2017 / Accepted: 25 October 2017 / Published online: 13 November 2017 ÓThe Author(s) 2017. This article is an open access publication Abstract This paper investigates the effects that the increase in the importation of machinery parts and components and the changes in the supplier composition had in the trade of final products and parts and components inside Latin America. In our analysis, we consider these effects according to two dimensions: a quantity one that captures whether there was an intensification of trade and a quality one that captures changes in the sophistication of the traded goods. The research employs disaggregated trade data obtained from UN Comtrade for 17 Latin American countries between 1996 and 2011. We find evidence that an increase in the importation of parts and components from Latin America had positive impacts on both the quantity and quality dimensions. Subregional heterogeneities revealed that, in general, imports from East Asia had positive effects on the quantity dimension, nurturing the expansion of machinery production networks inside Latin America, and on the quality dimension, increasing the sophistication of the products traded inside Latin America, especially for Mercosur member exports. Imports from North America had positive quantity effects, especially for exports of countries from the Andean Community, Central America, Chile, and Mexico. Keywords Machinery trade Fragmentation International production networks Latin America JEL Classification F14 F15 F23 &Mateus Silva Chang [email protected] 1 Graduate School of Economics, Keio University, 2-15-45 Mita, Minato-ku, Tokyo 108-8345, Japan 123 Lat Am Econ Rev (2017) 26:9 https://doi.org/10.1007/s40503-017-0054-5 1 Introduction In the past few decades international trade has increased exponentially and production fragmentation was one of the main causes. The fragmentation process, referred as ‘‘trade in tasks’’ by Grossman and Rossi-Hansberg (2008), has led to an increase in global integration, generating a web of economic interactions commonly denoted as international production networks. In the beginning, this change involved mostly trade among rich nations, but the real ‘‘revolution started when supply chain trade gained importance among high-tech and low-wage nations between 1985 and 1995’’ (Baldwin and Lopez-Gonzalez 2015). Production fragmentation opened new possibilities of economic growth to developing countries, allowing their engagement in the production process of manufactured goods that they were not able to produce. The expansion of production networks changed the rules of the economic development game, facilitating developing countries’ access to networks, global markets, capital, knowledge, and technology (OECD 2013). Previously, a country had to climb every single step in the industrial development ladder, mastering all production processes, to manufacture a given good. However, the advent of production networks offered the possibility of skipping steps in the catch-up procedure through the acquisition of knowledge and technology from third countries and the specialization in one or few steps of the production process. Understanding these changes and their consequent implications is crucial to draw policies that integrate a country in this new production structure and allow it to explore the best possibilities for guaranteeing sustainable economic growth and development. The empirical literature about production fragmentation is very rich, with many studies focusing on the regions where production fragmentation is more developed: East Asia, the European Union, and North America. 1 Although the demand is growing for the analysis of the current situation and the effects that this new trend can have on Latin America, the literature is still incipient. A few papers provide some information about the status of production fragmentation in Latin America based on descriptive analysis. Aminian et al. (2009) compared the economic integration process in East Asia and Latin America, analyzing the characteristics and intensity of intra-bloc and inter-bloc trade. They used a revealed comparative advantage (RCA) index to identify the share of traded manufacture parts and components with comparative advantage in the intra-bloc trade. Curran and Zignago (2013) studied the regionalization of trade in South America from 1994 to 2007, differentiating the trade flows by the end use of the products and the level of embodied technology. They concluded that the trade agreements have not extensively affected the regional trade level and that trade of intermediate products was still very low, indicating that regional production networks were still under-developed. Calfat et al. (2011) investigated the participation of Argentina, Brazil, Guatemala, and Nicaragua in fragmented world 1 For papers on production networks in the mentioned regions please refer to Ng and Yeats (2003), Ando and Kimura (2005), Athukorala and Yamashita (2006), Kimura et al. (2007), Yokota (2008), Kohpaiboon and Yamashita (2011), Ando and Kimura (2013a) and Ando and Kimura (2013b). 9Page 2 of 35 Lat Am Econ Rev (2017) 26:9 123 production. They concluded that Brazil was the only country with a consolidated participation in fragmented production. Fung et al. (2015a) used manufacturing trade data, classified according to the Standard International Trade Classification (SITC), to compare production sharing in Latin America, North America, and East Asia from 1985 to 2006. They identified the existence of a relatively thick production network involving the trade of parts of motor vehicles, telecommunication equipment, and electronic components. However, it was concentrated on Mexico’s trade with US and Canada, while Brazil also played a smaller role. Fung et al. (2015b) used the same data and methodology to compare Brazil, China, and Mexico’s participation in production networks. They analyzed the international trade patterns for the period 1990–2010, identifying that China’s global presence in the trade of parts and components increased. Although Mexico concentrated its trade of parts and components with the US, the data showed that China has become a major source of parts and components to Mexico and Brazil. The authors highlight the increasing importance of a Pan-Pacific link and a possible creation of a China– Brazil–Mexico production network. Florensa et al. (2015) produced the first paper that used a quantitative framework to analyze economic integration and production fragmentation in Latin America. Using trade data classified according to the Broad Economic Categories (BEC), the authors analyzed the impact that changes in import of intermediate goods from different world regions had in the development of Latin America’s regional trade. They found evidence of increasing regional production networks and the importance of other regions as suppliers of intermediate products, with special attention on China. Given the importance of this topic to the development literature and that production networks in Latin America are still understudied, this work contributes to the literature shedding light on the evolution of Latin American machinery industry. The first reason to focus on this industry is that machinery final products have a high level of complexity and use of a large number of parts and components, being the most developed manufacturing industry in terms of production fragmentation. Consequently, when the industry is more fragmented, a country will have more opportunities to engage in the production network. The second reason is that the high level of fragmentation and the availability of disaggregated trade data allow us to classify this industry into four different sectors. This division permits the development of a finer correspondence between parts and components and final products for a specific sector, reducing the bias that more aggregated data can generate. Finally, the share of machinery in Latin American exports and imports of manufactured products in 1996 was approximately 49.5 and 55.8%, respectively, while in 2011, the shares increased to 55.5 and 56.7%, respectively. Therefore, machinery is the most important industry in the region’s manufacturing trade. 2 The contribution of this article is threefold. First, to the best of our knowledge, this is the first article to analyze the quality effects that the changes in the structural composition of suppliers of parts and components have had in the development of 2 The shares were calculated based on the harmonized system (HS) classification. The manufactured goods range from HS 28 to HS 92, while machinery products range from HS 84 to HS 92. Lat Am Econ Rev (2017) 26:9 Page 3 of 35 9 123 regional machinery production networks in Latin America. Second, different from the previous papers on Latin American production networks that use aggregated intermediate manufacture data, we focus our analysis on a specific industry and use disaggregated data. Third, in this paper we adopt a model similar to Florensa et al. (2015), but estimate it using the PPML method instead of the ordinary least squares (OLS) to control the zero trade values and the data heteroscedasticity. The study is organized as follows. Section 2includes a descriptive analysis of Latin America’s participation in the machinery trade. In Sect. 3, we present the data and in Sect. 4we describe the empirical methodology employed. Section 5shows the results of the quantity analysis, while Sect. 6contains the results for the quality analysis. The final considerations of the paper are available in Sect. 7. 2 Machinery international trade and Latin America In this section, we used trade data—classified according to the Harmonized System (HS) disaggregated to the six-digit level—to analyze the machinery international market and Latin America’s participation in it in 1996 and 2011. We also considered the changes in trade patterns from a trade margin and product sophistication perspective to identify modifications in Latin American countries’ trade basket composition. 2.1 Descriptive analysis: traded values Production networks, in particular the machinery ones, are constituted mainly of geographical agglomerations that form regional blocs of production. Three main regional blocs are recognized for machinery production: the East Asian region, the European Union, and NAFTA. In general, machinery trade studies focus on these main areas, ignoring the situation in Latin America and the rest of the world. The main reason for this fact is illustrated by Fig. 1, which shows the data for world imports and exports of machinery. The aforementioned main areas are responsible for the bulk of the traded values, while Latin America and the rest of the world are responsible for a very small share. 3 Another important difference is that the traded value of parts and components is larger than the final product in the three main regions, while in Latin America, the traded value of the final product is higher. This evidence suggests a lower engagement of Latin America in machinery production networks. The idiosyncrasies of Latin America, a heterogeneous region composed of countries of different sizes and governments with different political and economic orientation, localized in a vast territory full of geographical barriers, such as the Andes Mountains and the Amazon Forest, are possible reasons for its low level of engagement in production networks. Another reason, identified by Moreira et al. 3 The figures and table in this subsection present a conservative estimate of the situation in Latin America, given that Mexico trade data are accounted for in the NAFTA region. From the next subsection on, Mexico’s data will be considered as part of Latin America. 9Page 4 of 35 Lat Am Econ Rev (2017) 26:9 123 (2013), is the quality of the local infrastructure that penalizes the trade, increasing the freight costs or simply making it impracticable at competitive prices. Nevertheless, Fig. 2reveals that for the period 1996–2011 parts and components traded value increased relative to final products. Latin American export data show an increase in traded parts and components from 46.6 percentage points to 54.4 percentage points. Although import data reveal an increase in traded parts and components share of just 0.5 percentage points, in terms of values this change represents a substantial increase, given that imported machinery parts and components value is higher than the exported machinery total value. Even though traded values are still smaller than the three main blocs, the increase in import of parts and components indicate that Latin America is slowly adhering to production networks in the machinery industry. Table 1displays the compound annual growth rate of the machinery trade, revealing that Latin America had the highest growth rate for total machinery export and the second highest for machinery import. In fact, considering just the trade of parts and components, the region had the highest growth rates, corroborating the idea that participation in machinery production networks is growing in this region. According to fragmentation theory, the core of a production network is concentrated in regional agglomerations, given the reduced costs incurred in shorter distance transport freights, reduced lead time, and the possibility of faster coordination for the whole network (Harrigan and Venables 2004; Kimura et al. 2007). Nonetheless, the development of the internet and other communication and coordination technologies, as well as the decrease in freight cost, has led to a decrease in the service link costs (Jones and Kierzkowski 2005), allowing for a growth in interactions between regional blocs. As a result, the East Asian region increased its role as a supplier of machinery parts and components to other region’s production networks. Bearing this fact in mind, in Fig. 3we observe the 0 500,000 1,000,000 1,500,000 2,000,000 Exports1996 Imports1996 Exports2011 Imports2011 Exports1996 Imports1996 Exports2011 Imports2011 Exports1996 Imports1996 Exports2011 Imports2011 Exports1996 Imports1996 Exports2011 Imports2011 Exports1996 Imports1996 Exports2011 Imports2011 EAST ASIA EU27 NAFTA LA ROW Final Products Parts and Components Fig. 1 Total values of exports and imports per region in 1996 and 2011 (in million US$). Source Chang and Kimura (2015) Lat Am Econ Rev (2017) 26:9 Page 5 of 35 9 123 composition of parts and components suppliers for all five regions to verify changes and patterns. As expected, except for NAFTA in 2011, the intra-bloc import of machinery parts and components is dominant in the three main regions. We also verify an increase of the East Asian share in all regions. Data for Latin America reveal a few things: first, the Latin American share as a supplier is irrelevant in all regions except Latin America; second, its intra-bloc share is smaller than the shares of the three main regions; and third, NAFTA and European Union shares decreased 46.6% 37.4% 54.4% 37.9% 53.4% 62.6% 45.6% 62.1% 0 50,000 100,000 150,000 200,000 Exports 1996 Imports 1996 Exports 2011 Imports 2011 Parts and Components Final Products Fig. 2 Latin America‘s total value of machinery products trade in 1996 and 2011 (in million US$). Source Author’s calculation, using data available from the UN Comtrade Table 1 Compound annual growth rate of machinery trade from 1996 to 2011 Parts and components (%) Final products (%) Total (%) East Asia Exports 7.1 6.4 6.7 Imports 6.6 4.6 5.7 EU Exports 4.5 3.7 4.1 Imports 4.2 3.6 3.9 NAFTA Exports 1.2 2.3 1.8 Imports 2.6 4.0 3.4 LA Exports 10.1 7.9 9.0 Imports 7.1 6.9 7.0 ROW Exports 5.1 5.7 5.4 Imports 6.9 7.9 7.6 The higher compound annual growth rates are highlighted using bold text Source Author’s calculation, using data available from the UN Comtrade 9Page 6 of 35 Lat Am Econ Rev (2017) 26:9 123 1.3 1.5 16.2 17.4 36.9 42.2 1.0 3.0 44.5 35.8 0% 20% 40% 60% 80% 100% 1996 2011 NAFTA Rest of the world European Union East Asia Lan America NAFTA 1.3 1.9 32.9 22.5 17.1 37.5 12.9 12.5 35.8 25.6 0% 20% 40% 60% 80% 100% 1996 2011 LATIN AMERICA Rest of the world European Union East Asia Lan America NAFTA 1.5 1.3 15.3 13.6 58.0 72.6 0.2 0.9 25.1 11.7 0% 20% 40% 60% 80% 100% 1996 2011 EAST ASIA Rest of the world European Union East Asia Lan America NAFTA 4.7 5.1 63.7 63.4 15.5 22.1 0.3 0.8 15.7 8.6 0% 20% 40% 60% 80% 100% 1996 2011 EUROPEAN UNION Rest of the world European Union East Asia Lan America NAFTA 6.1 6.0 63.3 55.1 10.2 27.9 0.2 0.3 20.2 10.7 0% 20% 40% 60% 80% 100% 1996 2011 REST OF THE WOLRLD Rest of the world European Union East Asia Lan America NAFTA Fig. 3 Composition of suppliers of machinery parts and components in 1996 and 2011. Source Author’s calculation, using data available from the UN Comtrade Lat Am Econ Rev (2017) 26:9 Page 7 of 35 9 123 considerably, substituted mainly by imports from East Asia. In fact, Latin America was the region where the share of East Asian parts and components increased the most, a growth of more than 20 percentage points. From the perspective of production fragmentation logic, a country purchases more parts and components from a given region if these products have some comparative advantage. The existing literature highlights two channels through which access to inputs can benefit a country: an efficiency gain in the production process by the acquisition of cheaper and/or higher quality inputs (Amiti and Konings 2007; Goldberg et al. 2010) and the possibility of having access to inputs that previously could not be produced domestically or obtained from a third country (Goldberg et al. 2009). In both cases, a gain in productivity and changes in production pattern are expected. Based on this fact, we consider the hypothesis that the increase in import of parts and components, especially from East Asia, should be beneficial to Latin American machinery production networks. In the next subsection, we analyze the trade margins to identify possible trade pattern modifications. 2.2 Descriptive analysis: the trade margins approach The trade flow can be decomposed in extensive and intensive margins, revealing how the intensification of trade in existing relations and the beginning or ending of trade relations contribute to change this flow. Although the intensive margin is expected to be the main factor responsible for the changes, authors such as Hummels and Klenow (2005) and Kehoe and Ruhl (2013) identified that in situations of considerable trade growth the extensive margin contribution is also relevant. Our main interest in observing the extensive margins is to identify evidence of changes in Latin American countries’ import and export baskets. A country’s trade relation is understood as a product–destination pair in the case of exports and a product–supplier pair in the case of imports. Given an initial and a final period, if a pair is active in both periods it is classified as a continuing pair. If country A imported (exported) a given product from (to) country B in the first period and then does not in the second period, but it still imports (exports) the product in question from (to) a third country, we have an exit of supplier (destination). If a similar situation occurs, but in the second period the product in question is not imported (exported) from (to) any other country, then it is classified as an exit of product. If in the second period, country A imports (exports) a product that was not imported (exported) from (to) any other country in the first period, then this new relation is classified as an enter of product. If in the second period, country A starts to import (export) from (to) country B a product that was already imported (exported) in the first period from (to) a third country, then this new relation is classified as an entry of supplier (destination). Considering only the number of relations, Fig. 4illustrates the margins of each Latin American country import of machinery parts and components. Mexico and Brazil had the highest number of product–supplier active pairs in 2011, while El Salvador, Honduras, Nicaragua and Paraguay had the lowest number of active pairs. The data reflect the diversification of the industrial park in each country. The entry 9Page 8 of 35 Lat Am Econ Rev (2017) 26:9 123 years, although it declines in 2011. General machinery had the second biggest share, followed by transport equipment. Figure 12 plots the share of each sector in regional exports of final products and parts and components. Despite the observed increase in the share of general machinery imports, regional exports were focused mainly in Fig. 11 Latin America machinery parts and components import composition by sector in 1996 and 2011. Source Author’s calculation, using data available from the UN Comtrade Fig. 12 Latin America machinery export composition by sector in 1996 and 2011. Source Author’s calculation, using data available from the UN Comtrade Lat Am Econ Rev (2017) 26:9 Page 15 of 35 9 123 transport equipment. From 1996 to 2011 we identify a change in the final products export pattern with electric machinery achieving the second biggest share. A similar movement occurred for the case of parts and components exports. Observing the total exports share, we identify that transport equipment products were the most exported ones inside Latin America, while electric machinery became the second most traded in 2011. Based on the importance of these two sectors and that the study of their heterogeneity be a common practice in the literature, we also investigate them independently. In view of the facts presented in this section, the change in Latin America’s machinery trade pattern, the increase in trade flows and the modification in the average sophistication level of the traded products, in the next sections, we present the data and methodology employed to study the effects that changes in the structural composition of machinery parts and components suppliers had in the expansion of Latin American production networks and in the sophistication of the intra-bloc export basket. 3 Data In the economics literature, many studies have been conducted on the fragmentation of production with different ways of defining the object of study. Some scholars employ a more comprehensive definition of production networks, including in their analysis all inputs used, from the raw materials to the final product. To capture all production steps they use international input–output tables. 7 On the other hand, some scholars do not consider the raw materials in their analysis, understanding that just the trade of parts and components used in a given industry should be analyzed. This second group of researchers adopt a more refined classification to isolate parts and components from final products. 8 We embrace the second view for two reasons: no international input–output data are available for the majority of the Latin American countries, and the second definition permits the use of more disaggregated and specific data. Considering that the machinery industry presents a high level of complexity and uses a large number of parts and components, we adopted this industry as our object of study. The analysis of machinery production networks was based on the classification of the machinery trade in parts and components and final products. 9 The data used was collected from the United Nations Commodity Trade Statistics Database (UN Comtrade), classified according to the HS disaggregated to the six-digit level. The machinery industry comprises all the goods categorized as general machinery sector 7 For example, Timmer et al. (2015) used the World Input–Output Database (WIOD), while Baldwin and Lopez-Gonzalez (2015) employed WIOD and OECD-WTO TiVA data to analyze production networks. 8 For example, Athukorala (2005) separates manufacturing parts and components from final products using the SITC data, while Ando and Kimura (2005) do the same for the machinery industry using HS data. 9 This process was performed in accordance with the classification presented in Ando and Kimura (2005). 9Page 16 of 35 Lat Am Econ Rev (2017) 26:9 123 (HS84), electric machinery sector (HS85), transport equipment sector (HS86–89), and precision machinery sector (HS90–92). We consider the import of parts and components from countries that were responsible for at least 0.01% of the international machinery trade in 2011. The selected 89 countries are grouped in six regions. 10 We define Latin America as the group of 17 countries consisting of Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela. The PRODY measure was employed as the qualitative measure for the export and import basket of Latin American countries. 11 We also used tariff data and the depth of the Preferential Trade Agreements (PTA) 12 to account for the level of integration of the Latin American economies. The tariff data were collected from the World Integrated Trade Solution (WITS) 13 and the PTAs depth measures were calculated based on Mulabdic et al. (2017) using the contents of trade agreements from the World Bank database (Hofmann et al. 2016). Given the availability of tariff and trade data, this study is restricted to the period from 1996 to 2011. 4 Methodology Because the core of production networks is regionally concentrated, in this work, we focus on the impacts that changes in the structural composition of parts and components suppliers have on Latin America’s intra-bloc exports. For this empirical exercise, we use a model known as the workhorse of empirical international trade analysis: the gravity model (Baier and Bergstrand 2007). The gravity model is distinguished by its good fit and its parsimonious and tractable representation of economic interactions among many countries, also allowing for disaggregation of the trade in different levels of geographical organization and product classification (Anderson 2011). To quantify the impact that changes in the structural composition of parts and components suppliers had on the development of the intra-bloc machinery trade, we follow the methodology proposed by Florensa et al. (2015), an augmented gravity model that accounts for the effect of the import of intermediate products. The adoption of such a framework is justified by the fact that, different from the standard gravity framework, this version accounts for the effect that the import of parts and 10 The list containing the 89 countries divided by regions is available in the Table 8in ‘‘Appendix’’. 11 We use the PRODY measures calculated by Hausmann et al. (2007). According to the authors they ‘‘constructed the PRODY measure for a consistent sample of countries that reported trade data in each of the years 1999–2001. These indexes are the result of an average of 3 years’’ (Hausmann et al. 2007). Because the chosen years are previous to the Chinese accession to the World Trade Organization, the possibility of a downward bias in the ranking of the machinery goods (in particular, final goods, given the increase in multinationals assembling their final products China) is minimized. 12 We thank an anonymous referee for suggesting the use of PTAs depth to implicitly control non-tariff barriers. 13 Tariff data were classified to HS six-digit level and when necessary was converted to the HS1992 version following the classification in Kimura and Obashi (2010). Lat Am Econ Rev (2017) 26:9 Page 17 of 35 9 123 components of a given sector from a given supplier have on the Latin American intra-bloc exports. The proposed model assumes that in the first period, Latin American countries can import parts and components from any region of the world. These parts and components are employed to produce other parts and components that in a second period will be used domestically or traded with another country. Alternatively, they can be used to manufacture a final product that will be consumed domestically or traded with a third country. We assume that Latin American production networks are created when Latin American countries use parts and components imported from any region of the world to produce a final product or other parts and components that are exported to another Latin American country. The Latin American intra-bloc trade of final products and parts and components in a given year is explained by the tariffs imposed by the importer country in the same year and the exporter country imports of parts and components in the previous year. We augment the Florensa et al. (2015) model to capture the possible effects of the non-tariff barriers by adding a PTA depth measure. The PTA depth index calculation follows Mulabdic et al. (2017): based on the content of the trade agreements database, we count the number of legally enforceable provisions covered by each agreement and normalize it between 0 and 1. Fixed-effect dummies are used to capture all sources of unobserved heterogeneity that are constant for each period of time, individual, and sector. Following Anderson and van Wincoop (2003,2004), exporter-time and importer-time dummies are employed to control for the multilateral resistance, sector dummies control for unobserved heterogeneities of each machinery sector, while exporter–importer dummies are added to guard against unobserved heterogeneities on the relation of each country pair (Baier and Bergstrand 2007). The proposed models were defined as follows: ln XFijkt  ¼a0þX 7 r¼1 arln Mpcrikt1þa8ln 1 þtariffjikt  þa9ln 1þdepthijt  þcit þujt þsij þxkþeijkt;ð2Þ ln XPCijkt  ¼b0þX 7 r¼1 brln Mpcrikt1þb8ln 1 þtariffjikt  þb9ln 1 þdepthijt  þnit þsjt þzij þlkþdijkt;ð3Þ where XFijkt and XPCijkt denote the traded value of final products and parts and components from country ito country j(these countries are limited to Latin American countries only) of sector kin year t. The traded values are explained by country i’s imports of parts and components of sector kfrom a given region rin year t-1(Mpcrikt1), the tariff imposed by the importing country jover the product of sector kprovided by country iin time ttariffjikt, the depth measure of the PTA between country iand jin year tdepthijt, exporter-time and importer-time dummies (time is defined as a 5-year period), the sector and the importer–exporter dummies. 9Page 18 of 35 Lat Am Econ Rev (2017) 26:9 123 One difference between this work and that of Florensa et al. (2015) is the definition of the object of study. As already mentioned, we selected a more specific object of study, focusing on the machinery industry alone. This allows us to use more disaggregated and detailed data, increasing the refinement of the parts and components and final products correspondence. Additionally, we can classify machinery industry products into four different sectors, decreasing the bias that can result from the use of aggregated data. Based on previous works about production fragmentation and use of imported inputs, it is expected that the purchase of parts and components from another country should provide some efficiency gain or advantage to Latin American countries. Based on this fact, we expect that the increase in imports of parts and components from the East Asian region should guarantee a production gain for Latin American countries, increasing the intra-bloc machinery trade. An important contribution of this paper is that the analysis is not limited to the quantity impact of the import of parts and components on the intra-bloc trade; we also propose a way of verifying quality changes. Since we know that there was a change in the shares of machinery parts and components providers, we attempt to evaluate if this variation also produced a modification in the Latin American intrabloc trade pattern. Once again, it is expected that if the East Asian region is more efficient in the production of machinery parts and components, providing inputs with higher quality and/or cheaper prices, Latin American countries should be able to diversify and improve the quality of their intra-bloc trade basket. To check this hypothesis, we propose a substitution of the traded values by a trade basket sophistication index that is calculated based on the PRODY index developed in Hausmann et al. (2007). The PRODY index can be used as a proxy for the sophistication of the product, and according to Hausmann et al. (2007), countries that have an export basket more similar to the developed countries tend to register economic growth in the subsequent periods. Based on this concept and the importance of economic growth for the development of Latin American countries, we use the PRODY index to calculate the composition of the import basket of parts and components (IMPY index) and the intra-bloc export basket (EXPY index). The objective is to estimate if the imports of parts and components from specific regions contributed or not to bring the Latin American intra-bloc export basket closer to the developed countries. The proposed models are defined as follows: ln EXPYFijkt  ¼a0þX 7 r¼1 arln IMPY pcrikt1þa8ln 1 þtariffjikt  þa9ln 1 þdepthijt  þcit þujt þsij þxkþeijkt;ð4Þ ln EXPYPCijkt  ¼b0þX 7 r¼1 brln IMPY pcrikt1þb8ln 1 þtariffjikt  þb9ln 1 þdepthijt  þnit þsjt þzij þlkþdijkt;ð5Þ where EXPYFijkt and EXPYPCijkt denote the EXPY index attributed to the basket of the final products and parts and components exported from country ito country j Lat Am Econ Rev (2017) 26:9 Page 19 of 35 9 123 (these countries are limited to Latin American countries only) of sector kin year t. The EXPY index is calculated as the weighted average of the PRODY index of each component of the basket of a given sector. The EXPY index attributed to the baskets of final products and parts and components traded inside Latin America is explained by the IMPY index attributed to the basket of parts and components of sector kthat country iimported from a given region rin year t-1(IMPY pcrikt1), the depth measure of the PTA between country iand jin year tdepthijt, the tariff imposed by the importing country jover the product of sector kprovided by country iin time t tariffjikt, the exporter-time and importer-time dummies (time is defined as a 5-year period), and the sector and the importer–exporter dummies. 5 Results: quantity analysis A common characteristic of trade data is the existence of missing and zero trade values. As the natural logarithm of zero does not exist, the estimated regressions do not consider the zero trade values that are important information about the trade pattern. The dropped data are information not used in the estimation, possibly leading to a bias in the regression results. 14 To avoid this problem, we estimate the regressions using the pseudo-Poisson maximum likelihood (PPML) method (Santos Silva and Tenreyro 2006). The PPML is the most accepted technique in the gravity model literature, allowing us to account for the observations with zero trade values. 15 In addition, trade data are plagued by heteroscedasticity, consequently the use of OLS can lead to the estimation of biased elasticities (Santos Silva and Tenreyro 2006). We first estimate Eqs. (2) and (3) for the values of the pooled machinery intrabloc exports. We also consider separate estimations for the main machinery sectors: electric machinery and transport equipment. The first half of Table 3contains the results for the intra-bloc exports of machinery final products. The tariff coefficient for the pooled data regression is negative and statistically significant, indicating that reductions in import tariffs of machinery final products are associated with increases in their intra-bloc exports. Consequently, the advance of the regional integration in Latin America through the decrease in the import tariffs imposed over the intra-bloc machinery trade have a positive impact on the development of Latin American machinery production networks. On the opposite side, the coefficient for PTA depth is statistically insignificant, showing that changes in non-tariff barriers do not affect production networks. The variables related to the origin of the imported parts and components reveal that an intensification of imports from the rest of the world (ROW), East Asia (EA), and Latin American (LA) countries, increase the intra-bloc trade of final products in the subsequent period. The results reveal that a 1% increase in imports 14 The share of dropped values when the OLS method is employed ranges from around 17–29% of the total values estimated with the PPML technique. 15 In our database, we also have missing tariff and trade data for a given group of countries and products that affect the independent variables. Unfortunately, the PPML model cannot address this problem. 9Page 20 of 35 Lat Am Econ Rev (2017) 26:9 123 of parts and components from LA leads to an increase of 0.53% in the intra-bloc exports of final products. The coefficient for imports from EA is also high, approximately 0.45%, while imports from the ROW have a smaller effect, approximately 0.15%. On the other hand, the coefficient of imports from the North America 16 and China are negative, indicating a decrease on the intra-bloc trade of final machinery products of 0.32 and 0.27%, respectively. Although it is not possible to identify the exact reasons why imports from these suppliers do not stimulate intra-bloc trade, we consider a few hypotheses. The first is that an important share of imported parts and components are used in domestic production networks being consumed in the domestic market. Given the lack of data we cannot verify this hypothesis. A second possibility is that a share of these parts and Table 3 The effect of imports of machinery parts and components on Latin American intra-bloc machinery exports Final products Parts and components Pooled machinery Electric machinery Transport equipment Pooled machinery Electric machinery Transport equipment Tariff 20.17*** (0.05) 20.12** (0.06) -0.05 (0.08) -0.03 (0.04) -0.03 (0.04) -0.05 (0.04) Lagged imports of parts and components from ROW 0.15*** (0.05) 20.24** (0.09) 0.18** (0.07) 0.19*** (0.05) 0.08 (0.07) 0.04 (0.06) Lagged imports of parts and components from EU -0.03 (0.08) 0.18 (0.23) -0.03 (0.21) -0.20 (0.13) 20.42*** (0.12) -0.08 (0.12) Lagged imports of parts and components from EA 0.45*** (0.06) 0.00 (0.15) 0.15 (0.19) 0.05 (0.07) 0.38*** (0.10) 0.31 (0.20) Lagged imports of parts and components from LA 0.53*** (0.08) 0.07 (0.09) -0.12 (0.18) 0.64*** (0.08) 0.16 (0.11) 0.39*** (0.14) Lagged imports of parts and components from US and Canada 20.32*** (0.08) 20.52*** (0.15) -0.32 (0.23) 0.31*** (0.08) 20.17* (0.09) 20.49*** (0.18) Lagged imports of parts and components from China and HK 20.27*** (0.05) 0.22** (0.09) 0.06 (0.10) -0.02 (0.06) 0.02 (0.10) 0.02 (0.07) PTA depth -0.24 (0.26) 20.66*** (0.22) -0.20 (0.65) -0.33 (0.29) -0.09 (0.20) 0.03 (0.49) Observations 16,943 4320 4032 16,667 4256 3794 R 2 0.87 0.90 0.94 0.87 0.94 0.97 Results that are statistically significant and different from zero are highlighted using bold text Given a restriction of space, the coefficients of secondary variables are omitted. Robust standard errors in parentheses: *p\0.10, **p\0.05, ***p\0.01 16 The term North America refers to US and Canada. Lat Am Econ Rev (2017) 26:9 Page 21 of 35 9 123 components are used to produce goods exported to countries outside Latin America. This case is beyond the scope of this paper. It could also be the case that imports from these regions are used to produce parts and components that are exported to other Latin American countries. If that is the case, coefficients will be positive in the second half of the table, when dependent variables are parts and components exports. Results in the second column refer to the electric machinery trade. The import tariff coefficient is negative, indicating that a decrease of 1% in the tariffs imposed on electric machinery parts and components imports, all ceteris paribus, stimulates an increase of 0.12% in intra-bloc exports of electric machinery final products. Once again, the coefficient for imports from North America is negative, indicating that a 1% increase in imports from this region causes a decrease in the intra-bloc exports of final electric machinery of 0.52%. Imports from the ROW also have a negative impact of 0.24%. On the opposite side, a 1% increase in imports of electric machinery parts and components from China promotes an intensification of 0.22% in the intra-bloc trade of final products. The PTA depth coefficient is statistically significant and indicates that deeper agreements lead to a decrease in the intraregional trade of electric machinery final products. According to Hofmann et al. (2016), agreements between developing countries are in general less deep and focus more on the decrease of import tariffs, since they are still high. 17 This is in accordance with our findings that show that decreases in import tariffs are more efficient than deeper agreements in the promotion of intra-bloc electric machinery final products exports. Additionally, in Latin America shallow agreements embrace more members, while deeper ones are in general bilateral agreements involving Mexico and Central American countries. Consequently, members of shallower agreements, such as the Mercosur, promote higher trade flows among themselves than members of deeper agreements. Lastly, according to Baier and Bergstrand (2007), all PTAs are ‘‘phased-in’’ over time, approximately 5–10 years. Therefore, older and shallower agreements, implemented in the 1990s, should have bigger impacts than the deeper ones, the majority of which were implemented in 2009, only 2 years before the end of the period studied. 18 With regard to transport equipment, the results are almost all statistically insignificant. The exception is the coefficient for imports from the ROW, indicating that a 1% increase in imports of transport equipment parts and components from this region leads to an increase of 0.18% in intra-bloc exports of final transport equipment. The second half of Table 3displays the intra-bloc parts and components export coefficients. In this case, import tariff and PTA depth coefficients are statistically insignificant, signaling an increase in regional integration through the decrease of the import tariffs or non-tariff barriers would not affect the production fragmentation and relocation inside Latin America. Imports of parts and components from 17 The mean of the import tariff between Latin American countries in the studied period was 6.89% for machinery final products and 5.64% for machinery parts and components. For the case of electric machinery, the average import tariffs are 9.92% for final products and 6.62% for parts and components. 18 Table 9in the ‘‘Appendix’’ contains the list of PTAs in force during the studied period. 9Page 22 of 35 Lat Am Econ Rev (2017) 26:9 123 LA, North America, and the ROW have positive and statistically significant coefficients. As observed in Florensa et al. (2015), imports of parts and components from these regions generate what they called a ‘‘complementary effect’’. In other words, they stimulate production fragmentation and its relocation inside Latin America, since parts and components imported by a Latin American country in the first period produce other parts and components that will be used in a third Latin American country, promoting these countries engagement in machinery productions networks. A 1% increase in parts and components imports from LA results in an increase of 0.64% of the intra-bloc export of parts and components. Imports from North America promote an increase of 0.31%, while imports from the ROW have a smaller effect of 0.19%. In the specific case of electric machinery, imports from EA stimulate the intrabloc trade of parts and components with a coefficient of 0.39%, while imports from European Union (EU) and North America have a negative effect. In the case of transport equipment, the LA coefficient is positive, while the North American one is negative. Florensa et al. (2015) refer to the situation when coefficients are negative as the ‘‘substitution effect’’, because instead of enhancing the development of production networks among Latin American countries, it promotes trade inside domestic markets or with countries in third regions. Evidence indicates that increases in the import of parts and components from LA have the biggest positive impact in the creation of a Latin American machinery production network. This result aligns with Florensa et al. (2015) and the production fragmentation theory that states that the core of production networks is regionally organized. Imports from North America, a region that is geographically close to Latin America and known for engaging in back-and-forth intra-firm production network transactions with Mexico and Central American countries, presented mixed results. It stimulates the intra-bloc trade of machinery parts and components in general, but decreases the trade of final products. Conversely, the results indicate that in the specific case of electric machineries, imports from EA and China foment production fragmentation. This result also aligns with Florensa et al. (2015) and Fung et al. (2015b) who verified an increase of parts and components supplies from Asian countries supporting Latin America’s engagement in production networks. As mentioned in Sect. 2, Latin America is an area where economic integration still lags behind other regions of the globe. Although there are 21 different PTAs signed by LA countries, none of them integrates the whole region. Given this fact, we perform one extra exercise to explore the heterogeneity inside the region. Based on economic proximity and negotiated PTAs, we classified Latin America into three subregions: one composed of Argentina, Brazil, Paraguay, and Uruguay, who are members of Mercosur; a second composed of Bolivia, Colombia, Ecuador, Peru, and Venezuela who are members of the Andean Community (CAN); and a third with the members of Central America (Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, and Panama), Mexico and Chile. 19 19 Venezuela was classified as part of CAN, because during the majority of the studied period it was a member of it. Chile and Mexico were aggregated with Central American economies given the existence of many PTAs among them and because both countries are not members of either Mercosur or CAN. Lat Am Econ Rev (2017) 26:9 Page 23 of 35 9 123 Table 4contains the results of machinery exports from each subregion. 20 Given the reduction in the number of observations, we consider only the pooled machinery data results. The majority of the coefficients are statistically insignificant for the case of CAN exports. Indeed, among the three subregions, CAN is the one with the smallest machinery trade. The results reveal that imports from LA and EA countries enhance the intra-regional trade of final machinery by 0.5 and 0.36%, while imports Table 4 The effect of imports of machinery parts and components on Latin American intra-bloc machinery exports by subregion Final products Parts and components CE, Mexico and Chile Andean Community Mercosur CE, Mexico and Chile Andean Community Mercosur Tariff 20.28*** (0.07) 0.06 (0.09) 20.23*** (0.07) 0.05 (0.08) -0.05 (0.10) 20.10*** (0.03) Lagged imports of parts and components from ROW 0.04 (0.06) -0.13 (0.09) 0.15** (0.07) 0.23*** (0.06) -0.14 (0.12) 0.10** (0.05) Lagged imports of parts and components from EU -0.18 (0.13) -0.18 (0.25) 0.12 (0.19) 20.45*** (0.12) -0.17 (0.20) -0.17 (0.13) Lagged imports of parts and components from EA 0.15*** (0.06) 0.36** (0.17) 0.66*** (0.13) 0.25*** (0.08) -0.04 (0.19) 0.03 (0.09) Lagged imports of parts and components from LA 0.32*** (0.09) 0.49*** (0.13) 0.10 (0.11) 0.45*** (0.07) 0.16 (0.12) 0.05 (0.09) Lagged imports of parts and components from US and Canada 0.42*** (0.10) -0.02 (0.25) 20.92*** (0.15) 0.58*** (0.10) 0.40* (0.21) 20.25*** (0.09) Lagged imports of parts and components from China and HK 20.22*** (0.05) 20.38*** (0.13) -0.12 (0.09) 0.06 (0.07) 0.13 (0.11) 0.26*** (0.05) PTA depth 20.60*** (0.22) 0.48 (0.56) 0.66 (1.02) 21.29*** (0.35) -0.06 (0.31) 1.03 (0.85) Observations 0.71 0.69 0.91 0.90 0.73 0.92 R 2 7807 5104 4032 7635 5104 3928 Results that are statistically significant and different from zero are highlighted using bold text Given a restriction of space, the coefficients of secondary variables are omitted. Robust standard errors in parentheses: *p\0.10, **p\0.05, ***p\0.01 20 Once the regions were selected based on PTAs, the PTA depth was dropped from the regressions. 9Page 24 of 35 Lat Am Econ Rev (2017) 26:9 123 am thankful for the support from Professor Laura Ma ´rquez-Ramos and Juan Nelson Martı ´nez Dahbura. I gratefully acknowledge the Keio University Doctorate Student Grant-in-Aid Program, the MEXTsupported Program for the Strategic Research Foundation at Private Universities, and Keio Economic Society for the financial support in part for this study. The analysis and results presented in this research are only the responsibility of the author. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Appendix See Tables 8and 9. Table 8 Country list by regions Region Name North America Canada North America USA China China China China, Hong Kong EA Australia EA India EA Indonesia EA Japan EA Malaysia EA New Zealand EA Philippines EA Republic of Korea EA Singapore EA Thailand EU Austria EU Bulgaria EU Czech Republic EU Cyprus EU Denmark EU Estonia EU Finland EU France EU Germany EU Greece EU Hungary EU Ireland EU Italy Lat Am Econ Rev (2017) 26:9 Page 31 of 35 9 123 Table 8 continued Region Name EU Latvia EU Lithuania EU Malta EU The Netherlands EU Poland EU Portugal EU Romania EU Slovakia EU Slovenia EU Spain EU Sweden EU UK LA Argentina LA Bolivia LA Brazil LA Chile LA Colombia LA Costa Rica LA Ecuador LA El Salvador LA Guatemala LA Honduras LA Mexico LA Nicaragua LA Panama LA Paraguay LA Peru LA Uruguay LA Venezuela ROW Albania ROW Algeria ROW Azerbaijan ROW Cameroon ROW Cote d’Ivoire ROW Croatia ROW Egypt ROW Georgia ROW Ghana ROW Iceland ROW Israel ROW Jamaica ROW Kyrgyzstan ROW Mali 9Page 32 of 35 Lat Am Econ Rev (2017) 26:9 123 Table 8 continued Region Name ROW Mauritius ROW Morocco ROW Niger ROW Nigeria ROW Norway ROW Oman ROW Republic of Moldova ROW Russian ROW Saudi Arabia ROW Senegal ROW Sudan ROW Switzerland ROW Republic of Macedonia ROW Tunisia ROW Turkey ROW Uganda ROW Ukraine ROW Tanzania ROW Zambia Table 9 List of Latin American preferential trade agreements Agreement Year that entered in force Type Number of enforceable provisions Panama–Chile 2008 FTA and EIA 12 CAN 1988 CU 15 Chile–Costa Rica 2002 FTA and EIA 16 MERCOSUR 1991 CU and EIA 17 Central American Common Market (CACM) 1961 CU 18 Chile–Honduras 2008 FTA and EIA 18 Chile–Guatemala 2010 FTA and EIA 18 Chile–El Salvador 2002 FTA and EIA 19 CAFTA-DR 2006 FTA and EIA 19 Mexico–Uruguay 2004 FTA and EIA 21 Panama–El Salvador 2003 FTA and EIA 22 Panama–Costa Rica 2008 FTA and EIA 22 Colombia–Northern Triangle 2009 FTA and EIA 22 Panama–Guatemala 2009 FTA and EIA 22 Panama–Honduras 2009 FTA and EIA 22 Peru–Chile 2009 FTA and EIA 22 Lat Am Econ Rev (2017) 26:9 Page 33 of 35 9 123 References Aminian N, Fung KC, Ng F (2009) A comparative analysis of trade and economic integration in East Asia and Latin America. Econ Change Restruct 42:105–137 Amiti M, Konings J (2007) Trade liberalization, intermediate inputs and productivity. Am Econ Rev 97:1611–1638 Anderson JE (2011) The gravity model. Ann Rev Econ 3:133–160 Anderson JE, van Wincoop E (2003) Gravity with gravitas: a solution to the border puzzle. Am Econ Rev 93(1):170–192 Anderson JE, van Wincoop E (2004) Trade Costs. J Econ Lit 42(3):691–751 Ando M, Kimura F (2005) The formation of international production and distribution networks in East Asia. In: Ito T, Rose AK (eds) International trade in East Asia. University of Chicago Press, Chicago Ando M, Kimura F (2013a) Evolution of machinery production networks: linkage of North America with East Asia. ERIA Discussion Paper Series DP-2013-32, Economics Research Institute for ASEAN and East Asia (ERIA) Ando M, Kimura F (2013b) Production linkage of Asia and Europe via Central and Eastern Europe. J Econ Integr 28:204–240 Athukorala P (2005) Product fragmentation and trade patterns in East Asia. Asian Econ Papers 4(3):1–27 Athukorala P, Yamashita N (2006) Product fragmentation and trade integration: East Asia in a global context. N Am J Econ Finance 17(3):233–256 Baier SL, Bergstrand JH (2007) Do Free Trade Agreements Actually Increase Members’ International Trade? J Intern Econ 71(1):72–95 Baldwin R, Lopez-Gonzalez J (2015) Supply-chain trade: a portrait of global patterns and several testable hypothesis. World Econ 38(11):1682–1721 Calfat G, Cassimon D, Flo ˆres RG Jr, Rivas A (2011) Far from champions, close to midgets—international production sharing in Central and South America. Rev de Econ y Estad 49:101–137 Curran L, Zignago S (2013) Regional integration of trade in South America: how far has it progressed and in which sectors? Int Trade J 27(1):3–35 Florensa LM, Ma ´rquez-Ramos L, Martı ´nez-Zarzoso I, Recalde ML (2015) Regional versus global production networks: where does Latin America stand? Appl Econ 47(37):3938–3956 Fung KC, Garcia-Herrero A, Siu A (2015a) Production sharing in Latin America and East Asia. Working Papers 0901, BBVA Bank, Economic Research Department Fung KC, Hwang HC, Ng F, Seade J (2015b) Production networks and international trade: China, Brazil, and Mexico. N Am J Econ Finance 34:421–429 Goldberg P, Khandelwal A, Pavcnik N, Topalova P (2009) Trade liberalization and new imported inputs. Am Econ Rev Papers Proc 99(2):494–500 Goldberg P, Khandelwal A, Pavcnik N, Topalova P (2010) Trade imported intermediate inputs and domestic product growth: evidence from India. Q J Econ 125(4):1727–1767 Grossman GM, Rossi-Hansberg E (2008) Trading tasks: a simple theory of offshoring. Am Econ Rev 98(5):1978–1997 Harrigan J, Venables AJ (2004) Timeliness, trade and agglomeration. CEPR Discussion Paper, No, p 4294 Hausmann R, Hwang J, Rodrik D (2007) What you export matters. J Econ Growth 12(1):1–15 Table 9 continued Agreement Year that entered in force Type Number of enforceable provisions Chile–Mexico 1999 FTA and EIA 24 El Salvador–Honduras 2008 FTA and EIA 24 Panama–Nicaragua 2009 FTA and EIA 24 Chile–Colombia 2009 FTA and EIA 25 Colombia–Mexico 1995 FTA and EIA 27 9Page 34 of 35 Lat Am Econ Rev (2017) 26:9 123 Hofmann C, Osnago A, Ruta M (2016) Horizontal depth: a new database on the content of deep agreements. Mimeo, The World Bank Hummels D, Klenow PJ (2005) The variety and quality of a nation‘s exports. Am Econ Rev 95(3):704–723 Jones R, Kierzkowski H (2005) International fragmentation and the new economic geography. N Am J Econ Finance 16:1–10 Kehoe TJ, Ruhl KJ (2013) How important is the new goods margin in international trade? J Political Econ 121(2):358–392 Kimura F, Obashi A (2010) International production networks in machinery industries: structure and its evolution. Working Papers DP-201009, Economics Research Institute for ASEAN and East Asia (ERIA) Kimura F, Takahashi Y, Hayakawa K (2007) Fragmentation and parts and components trade: comparison between East Asia and Europe. N Am J Econ Finance 18(1):23–40 Moreira MM, Blyde JS, Martincus CV, Molina D (2013) Too far to export: domestic transport costs and regional export disparities in Latin America and the Caribbean. IDB, Washington, DC Mulabdic A, Osnago A, Ruta M (2017) Deep integration and UK-EU trade relations. World Bank Policy Research Working Paper No. 7947 Ng F, Yeats A (2003) Major Trade Trends in East Asia : What are their Implications for Regional Cooperation and Growth? World Bank Policy Research Working Paper No. 3084. https://doi.org/10. 1596/1813-9450-3084 Obashi A, Kimura F (2016) Deepening and Widening of Production Networks in ASEAN. ERIA Discussion Paper Series DP-2016-09, Economics Research Institute for ASEAN and East Asia (ERIA) Organization for Economic Co-operation and Development (OECD) (2013) Interconnected economies: benefiting from global value chains. OECD Publishing, Paris Santos Silva JMC, Tenreyro S (2006) The Log of Gravity. Rev Econ Stat 88(4):641–658 Timmer M, Dietzenbacher E, Los B, Stehrer R, de Vries G (2015) An illustrated user guide to the world input-output database: the case of global automotive production. Rev Int Econ 23(3):575–605 Yamashita N, Kohpaiboon A (2011). Trade in Supply Chains between ASEAN and China: Development and Implications. In: Findlay C (ed) ASEAN?1 FTAs and Global Value Chains in East Asia. ERIA Research Project Report 2010-29. ERIA, Jakarta, pp 25–51 Yokota K (2008) Parts and Components trade and Production Networks in East Asia - A Panel Gravity Approach. In: Hiratsuka D, Uchida Y (eds) Vertical Specialization and Economic Integration in East Asia. Institute of Developing Economies JETRO Lat Am Econ Rev (2017) 26:9 Page 35 of 35 9 123