Policy Research Working Paper 9156 World Development Report 2020 Background Paper icio Economic Analysis with Inter-Country Input-Output Tables in Stata Federico Belotti Alessandro Borin Michele Mancini Development Economics World Development Report 2020 Team February 2020 Policy Research Working Paper 9156 Abstract Several new statistical tools and analytical frameworks have for given empirical questions on trade in value-added and been developed recently to measure countries’ and sectors’ participation in global value chains of countries and sec- involvement in global value chains. Such wealth of meth- tors. By exploiting inter-country input-output tables, icio odologies reflects that different empirical questions call for provides decompositions of aggregate, bilateral, and sectoral distinct accounting methods, along with different levels of exports and imports according to the source and destina- aggregation of trade flows. This paper is a companion to tion of their value-added content. As different measures the conceptual framework presented in Borin and Mancini are suited to address distinct economic questions, icio is (2019). The paper describes a new Stata module, icio, that designed to be flexible also in this respect. allows the user to construct the most appropriate measure This paper is a product of the World Bank’s World Development Report 2020 Team, Development Economics. 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://www.worldbank. org/prwp. The authors may be contacted at federico.belotti@uniroma2.it, alessandro.borin@bancaditalia.it and michele.mancini@bancaditalia.it. 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 icio: Economic Analysis with Inter-Country Input-Output Tables in Stata∗ Federico Belotti1 , Alessandro Borin2 , and Michele Mancini3 1 University of Rome Tor Vergata 2,3 Bank of Italy Keywords : global value chains; input-output tables; trade in value-added; Stata. JEL classification : E16, F1, F14, F15. ∗ This paper is a product of the World Bank’s World Development Report 2020 Team, Devel- opment Economics, the International Relations and Economics Directorate of the Bank of Italy and the University of Rome Tor Vergata. A pilot version of the icio Stata module was used to compute some of the measures of GVC participation used in the analysis of the World Bank’s World Development Report 2020 “Global Value Chain: Trading for Development”. New updates and more information about icio can be found here: http://tradeconomics.com/icio. To install icio, run in Stata: ssc install icio. The views expressed in this paper are solely those of the authors and do not involve the responsibility of the Bank of Italy. The usual disclaimer applies. 1 Introduction The diffusion of global production networks has called for new statistical tools providing a representation of complex production linkages between and within economies. New types of data sources, the Inter-Country Input-Output (ICIO) tables, and new analytical frameworks have been developed to measure supply and demand contributions of countries and sectors in global value chains (GVCs).1 In this paper we describe icio, a new Stata command that computes countries’ and sectors’ participation in GVCs as well as relevant measures of trade in value- added, following the conceptual framework proposed by Borin and Mancini (2019), which – in turn – extends, refines and reconciles the other main contributions in this strand of the literature.2 The command is flexible in many aspects. It allows to choose from different accounting methodologies, called perspectives. Each of these perspectives is best suited to address specific empirical questions, such as tracking production-demand linkages, assessing countries’ participation to the global production sharing, quan- tifying value-added embedded in countries’ and sectors’ exports, evaluating the potential exposure to macroeconomic and trade policy shocks. It exploits the most famous ICIO tables - the World Input-Output Database (Timmer et al. 2015), the OECD TiVA database (OECD, 2018), and the Eora Global Supply Chain Database (Lenzen et al. 2013). Moreover, any user-provided ICIO table can be straight- forwardly loaded and used to compute value-added trade and GVC participation measures. More specifically, icio encompasses the most relevant measures of value-added in exports and imports at the aggregate, bilateral and sectoral levels. For a given trade flow, it disentangles the source country/sector and the destination coun- try/sector of value-added content. Moreover, for export flows at any level of dis- aggregation, icio computes the component related to GVC trade, i.e., the one entailing more than a single border crossing. This measure - and its backward and forward GVC participation sub-components, is featured in the World Bank World Development Report 2020 (WDR, 2020).3 In addition, the icio command can be used to retrieve, from the ICIO tables, 1 See, among others, Johnson and Noguera (2012), Wang et al. (2013), Koopman et al. (2014), Borin and Mancini (2015), Los et al. (2016), Nagengast and Stehrer (2016), Johnson (2018), Miroudot and Ye (2018), Los and Timmer (2018). 2 The essential features and the algebraic derivation of the conceptual framework are reported in Appendices A to F. 3 GVC measures are based on Borin and Mancini (2015, 2019), which consistently refine the vertical specialization index proposed by Hummels et al. (2001). 2 the GDP (i.e. value-added) produced by a given country or industry (origin), the final demand in different countries and sectors (destination), or a combination of the two (when both origin and destination are specified). We are aware that many questions can be addressed only partially with the current version of icio. Thus, our plan is to release new modules in the near future. First, we will add a much broader set of measures to assess the participation of countries and sectors in GVCs (Borin and Mancini, 2017; Wang et al., 2017) and their position (Antr`as and Chor, 2013; Fally 2012; Antr`as and Chor, 2018). Second, we will build a set of indicators to better evaluate the direct and indirect effects of trade policies, taking into account the GVC structure. The rest of the paper is organized as follows. In Section 2 we show how to load ICIO tables in icio using the command icio_load. In Section 3 we show how supply, demand and supply-demand linkages can be computed in icio. This is useful when the empirical questions are related by supply, demand or require linking the origin of the value added to its absorption in final demand, without considering explicitly international trade flows. In Section 4 we provide the tools for gathering a comprehensive overview of the production process and cross-country relationships, i.e., how to obtain value-added decompositions of trade flows by country of origin and destination. In Section 5 we focus on the measurement of GVC participation. In several instances, and for illustration purposes, we show how to replicate some of the measures of GVC participation and figures presented in the World Bank’s Word Development Report 2020. Section 6 concludes. Each section shares the same structure. At the beginning we provide a brief overview of the measures therein discussed. Then, we show how to compute those measures in icio, providing also some examples and a list of several relevant ques- tions that could be addressed, with the related icio’s syntax. In addition, in the Appendices we present the related conceptual frameworks. 2 Inter-Country Input-Output tables Input-Output (IO) models were developed by Leontief (1936) to represent and an- alyze production and consumption relationships within an economy. The related statistical tools, the IO tables, indicate the monetary amount of inputs of each sector necessary to produce the total output of a given industry and, in turn, how this output is used as final consumption (or investment) or as intermediate inputs for other productions. National IO tables distinguish only between domestic and foreign inputs; on the output side, exports represent one of the possible ‘final’ uses 3 of output, as domestic consumption and investment. ICIO tables, which have been developed combining national IO statistics with trade data, describe sale-purchase relationships between industries within and between economies as well as the uses in different final demand components (e.g. consumption, investment and government spending). In particular, an ICIO table specifies the country-sector pairs that pro- vide intermediate inputs to a given industry and the country-sector pairs to which that industry sells its output - in the case of intermediate products - or the ultimate destination markets for final goods. In Appendix A we present the basic conceptual framework of ICIO models while in Section 2.1 we show how to load ICIO tables with the command icio_load. 2.1 Implementation: Loading ICIO tables using the icio_- load command In order to use the icio command one needs to load a particular ICIO table by using the icio_load command. icio_load allows to work directly with the most popular ICIO tables - OECD TiVA database (OECD, 2018), World Input-Output Database (WIOD, Timmer et al. 2015), and Eora Global Supply Chain Database (Lenzen et al. 2013); in addition, any other user-provided ICIO table can be loaded. 2.1.1 Syntax The basic syntax for icio_load is icio_load, [options ] where the main options are: iciotable(table name [, usertable options ]), specifies the ICIO table to be used for the analysis; default is wiodn, the last WIOD release available (release 2016, see below for more details on the available tables’ versions); year(# ), sets the year to be used for the analysis; the default is the last available year: 2014 for the WIOD tables (wiodn), 2015 for TiVA tables (tivan), 2015 for the Eora Global Supply Chain Database tables (eora). Not needed for user- provided tables; info, shows the data sources and the versions of the loadable ICIO tables. 2.1.2 Examples icio_load can be used for the following purposes: 4 1. To display the list of the directly available ICIO tables and their releases:4 . icio_load, info table version from to wiodn 2016 2000 2014 tivan 2018 2005 2015 eora 199.82 1990 2015 wiodo 2013 1995 2011 tivao 2016 1995 2011 In this way the user can always recover which ICIO tables are directly available via the icio command. As can be seen from the previous output, at the time of writing, the following tables have been made available: the “2013” and the “2016” releases of the WIOD tables, the “2016” and “2018” of the TiVA tables, and the “199.82” version of the Eora Global Supply Chain Database tables. 2. To load a specific year of the ICIO table of interest; for example the following syntax allows to load the year 2014 of the WIOD tables released in 2016 (i.e. wiodn):5 . icio_load, iciot(wiodn) year(2014) 3. To load a user-provided ICIO table, by specifying user, instead of a specific ICIO table’s code, in the iciotable(table name ) option. The user-provided ICIO table and the related country list must be provided using two different comma separated files (i.e. files with extension .csv). For example: . icio_load, iciot(user, userp("path_to_the_table_folder") tablen(ADB_2011.csv) // countrylist(adb_countrylist.csv) The syntax above shows that additional information has to be provided in order to instruct icio to load the user-provided table: i) the path to the folder where the .csv file containing the user-provided table is located, via the 4 The ICIO tables directly available in icio are automatically downloaded the first time the user requests them through the icio_load command. These files are saved within the Stata system folder "../ado/plus/i" using a .mmat format (notice that the filename begins with the prefix icio_). For further details on the last release directly available through icio, run icio_load, info or visit the official websites oe.cd/tiva for the OECD TiVA database, wiod.org for the World Input-Output Database and worldmrio.com for the Eora Global Supply Chain Database. Please, also remember to cite the reference to the ICIO database you are analyzing through icio. 5 Multiple years have to be loaded sequentially and then the results should be appended. 5 sub-option userpath("string"); ii) the name of the .csv file containing the table, via the sub-option tablename(string); the name of the file containing the country list, via the sub-option countrylistname(string). Notice that the table’s csv file must contain only one matrix of dimension (G × N ) × (G × N + G × U ), where G is the number of countries, N the number of sectors and U the number of uses (i.e. consumption, investment, etc.). See the icio help file for more details. 3 Supply, demand and supply-demand linkages ICIO tables can be used in combination with long-established accounting relation- ships (Leontief, 1936) to measure the net value of production (GDP) of a coun- try/sector, the value of final demand in a given country/sector and to pin down the links between the country-sector where the value-added originates and the market where it is absorbed in final demand. In Appendix B we show how to retrieve supply, demand and supply-demand linkages, i.e. GDP by country/sector of origin and/or destination, from ICIO tables while in Section 3.1 we describe how to compute these measures in icio, providing also some examples. 3.1 Implementation: Supply and demand with icio The icio command can be used to retrieve the GDP (i.e. value-added) produced by a given country or industry (origin of value-added) by specifying the option origin(), to measure the final demand in different countries and sectors (destina- tion of the value-added) by specifying the option destination(), or a combination of the two (both origin() and destination()). 3.1.1 Syntax 1. Gross Domestic Product: icio, origin(country code [,sector code]) [standard options ] 2. Final demand: icio, destination(country code [,sector code]) [standard options ] 3. Value-added by origin and final destination: 6 icio, origin(country code [,sector code]) destination(country code [,sector code]) [standard options ] The list of available country/sector codes for the loaded ICIO table can be dis- played by running icio, info. As for the [standard options], they are: save(filename [, replace]), which saves the command output (scalar, vector or matrix) in memory to an excel file (.xls); groups(grouping rule “group name” [...]), which specifies a user-defined grouping of countries. In this way, output measures can be computed for a country group (e.g., the “Euro area”, “MERCOSUR” or “ASEAN”) as a whole while taking into account the specific supply/demand/trade structure of each member of the group. To define one or more country groups, the user has to provide a list of comma-separated country codes, which is the “grouping rule ”, followed by a user defined “group name ”. 3.1.2 Examples icio is useful to address empirical questions related to supply, demand and supply- demand relationships, without considering explicitly international trade flows. For instance, it is possible to retrieve the GDP of Germany that is finally absorbed in China (i.e., Johnson and Noguera (2012) “value-added exports”), that according to the WIOD database in 2014 is: . icio_load Loading table wiod 2014... loaded . *What is the value-added originated in Germany and absorbed in China? . icio, origin(deu) destination(chn) Value-Added by origin/destination: Origin: DEU Destination: CHN Output: Value-Added Millions of $ % of total Value-Added 101042.25 100.00 The Stata output displays the value, in millions of US dollars, and the share of total value-added produced in a specific country, when the complete list of countries or sectors of destination is selected by specifying the code all6 (or absorbed by 6 This is the case if a certain country/sector of origin is specified, origin(country - code [,sector code]), as well as the complete list of countries/sectors of destination, i.e. 7 a specific country/sector, if the complete list of countries or sectors of origin is specified with the same code).7 If the option all is not specified the share will be clearly equal to 100%. For example . *What is the value-added originated in Germany and absorbed in each country? . icio, origin(deu) destination(all) Value-Added by origin/destination: Origin: DEU Destination: ALL Output: Value-Added Millions of $ % of total AUS 12048.65 0.33 AUT 36839.13 1.02 BEL 20719.98 0.57 BGR 3004.85 0.08 BRA 15943.49 0.44 CAN 15305.40 0.42 CHE 37205.73 1.03 CHN 101042.25 2.79 CYP 768.14 0.02 CZE 16171.75 0.45 DEU 2446528.60 67.58 DNK 12205.44 0.34 ESP 33469.66 0.92 EST 1219.86 0.03 FIN 9049.83 0.25 FRA 85684.36 2.37 GBR 77158.04 2.13 GRC 6543.15 0.18 HRV 2473.71 0.07 HUN 8782.43 0.24 IDN 4689.61 0.13 IND 12267.21 0.34 IRL 5407.92 0.15 ITA 52128.41 1.44 JPN 23269.19 0.64 KOR 17377.73 0.48 LTU 1978.03 0.05 LUX 3989.54 0.11 LVA 1134.20 0.03 MEX 10619.36 0.29 MLT 321.40 0.01 NLD 34154.59 0.94 NOR 10257.55 0.28 POL 33030.97 0.91 PRT 6773.81 0.19 ROU 9065.05 0.25 ROW 235768.63 6.51 RUS 38509.83 1.06 SVK 6112.61 0.17 destination(all[,sector code ]) or destination(country code ,all). 7 This is the case if the complete list of countries or sectors of origin is specified, i.e. origin(all[,sector code ]) or origin(country code ,all), as well as a specific country/sector of destination, destination(country code[,sector code ]). 8 SVN 2618.50 0.07 SWE 20894.17 0.58 TUR 19070.35 0.53 TWN 6318.91 0.17 USA 122388.23 3.38 As can be noted, the share of the German GDP absorbed in China (on the total GDP produced by Germany), around 2.8%, is obtained either by looking for the CHN row in the previous output, or by taking the ratio of the dollar values obtained throughout the options origin(deu) destination(chn) and the value of the total German GDP, that can be obtained using: . *What is the Germany´s GDP? . icio, origin(deu) Value-Added by origin/destination: Origin: DEU Output: Value-Added Millions of $ % of total Value-Added 3620310.26 100.00 Other examples of questions that could be answered (see the reported syntax) with the analysis of supply-demand linkages through icio are: • What is the GDP (value-added) produced by each country? icio, origin(all) • How much value-added does each country produce in a given sector (e.g., sector code 19)? icio, origin(all,19) • What is the aggregate final demand of each country? icio, destination(all) • Where the value-added produced in the Italian sector 19 is absorbed? icio, origin(ita,19) destination(all) • Which final demand sectors in China are the most important for the absorption of US-made value-added? icio, origin(usa) destination(chn,all) • Where the GDP produced in each country is absorbed (and save the output as “supply demand.xls” in the current working directory)? icio, origin(all) destination(all) save("supply_demand.xls") 9 • How much USMCA (former NAFTA) countries’ final demand in sector 20 is satisfied by Chinese productions? icio, origin(chn) destination(usmca,20) groups(usa, mex, can, "usmca") 4 Value-added in trade flows The accounting relationships presented in the previous section provide a useful tool to link the origin of value-added (GDP) to its absorption in final demand. However, they provide only partial information on overall production processes and cross- country relationships. For instance, no information is provided on the production stages that take place as well as on the national borders that are crossed between the stage in which the value-added is generated and the one of its final absorption. Both the above represent critical information to understand countries interdependence in GVCs. In many empirical applications it is important to trace value-added in gross trade flows, for instance, when we want to measure the value-added produced by a country that is involved in a certain trade relationship. Depending on the empirical issue under investigation, it is also necessary to consider trade flows at different levels of disaggregation and analyze their value-added content. In fact, in some cases we may be interested just to single-out the value-added embedded in global trade flows or in the total exports or imports of a country. In other cases, also the bilateral and sectoral dimensions of trade flows may matter. For instance, when studying the implications of GVCs, it is relevant to consider the position of a country (or sector) within the production chain and identify its direct upstream and downstream trade partners. This may be relevant in order to geographically map the production networks and analyze the international propagation of macroeconomic shocks. A key issue in the value-added accounting of trade regards the definition of “double counted” components, i.e. items that are recorded several times in a given gross trade flow due to the back-and-forth shipments that occur in a cross-national production process (Koopman et al. 2014). For instance, imagine that a country is exporting cotton. After some processing stage abroad, the cotton is imported back, embedded in some fabric, to be further re-exported as apparel. The value of cotton will be counted twice in the aggregate exports of the country, i.e. “double counted” - in the first export flow and in the second one, embedded in the apparel - but once in the GDP (i.e. value-added). Now imagine that the goal is to allocate the value-added across the two export flows. A reasonable way is to consider the cotton production as value-added in the 10 first shipment, i.e. the one of cotton, and as double-counting in the second one, when cotton is embedded in the shipment of apparel. Summing the value-added terms in the two export flows, we end up with consistent aggregate figures at the country level, i.e. cotton classified once as value-added and once as double counting. Consider now a different goal, i.e. assessing the value-added exposed to a specific trade barrier imposed by a partner. To this end, suppose that a tariff impairs the exports of apparel. Now, when considering the value-added embedded in the second flow, the one exposed to the tariff, also that one of cotton needs to be taken into account. In this way, we are able to correctly assess the value-added that could be impaired by the tariff. More in general, depending on the type of trade flow and the objective of the analysis, it is necessary to define the “perimeter” according to which something is classified as “value-added” or “double counted”, i.e. a specific accounting “perspec- tive” has to be chosen. Each perspective is better suited to address specific empirical issues. When- ever the empirical application requires to retrieve the entire value-added of a coun- try/sector of origin that is embedded in a given trade flow, as in the second example above, the accounting perspective to be chosen should match the level of disaggre- gation of the trade flow considered, as reported in the first column of Table 1 (e.g., “exporting country” perspective for aggregate exports, “bilateral” perspective for an aggregate bilateral trade flow, “sectoral-bilateral” perspective for a trade flow between two countries in a given sector, etc.).8 For instance, suppose a tariff is im- posed on the imports of a given sector from a certain partner, and we are interested in evaluating what part of the exporting country GDP is exposed to the tariff. In this case we want to consider as “value-added” the entire GDP that is involved in this sectoral-bilateral relationship, even if part of that was previously exported to other countries/sectors (i.e. double counted in an “exporting country” perspective). The specific sectoral-bilateral relationship becomes the new relevant perimeter, and only the items that enter multiple times in this trade flow are considered as “double counted”. Indeed, this is what is called a “sectoral-bilateral perspective”, the one used in the second example on cotton. However, these measures cannot be summed up to get a precise assessment of value-added contents in more aggregate trade flows, e.g. value added in the total exports of a country. In other words, they are non-additive.9 Thus, if we 8 See Section E for a complete overview of the perspectives available in icio). 9 See also Johnson (2018) and Los and Timmer (2018) on this point. More specifically, when we use an accounting perspective based on a narrower perimeter to define the double counted items (e.g. the sectoral-bilateral one), by summing these indicators we obtain value-added content 11 are seeking a breakdown of the value-added measures by sectors of exports, by importing partners or by sector-partner combinations, consistent with exporters’ aggregate figures, as in the first example on cotton, we need to apply the exporting- country perspective also to the decomposition of more disaggregated trade flows (see column two of Table 1). In this way, the resulting measures are additive, i.e. measures at a more aggregate level can be obtained summing disaggregate results. This accounting perspective can be used also for other purposes such as, for instance, to single out the portion of trade in any type of exports that crosses just one international border. Indeed, Section 5 shows that this is instrumental for measuring GVC-related trade. Whenever the perspective is set at a more aggregate level as compared to the considered trade flow, it is also needed to select an approach to allocate double counting. icio implements two alternative approaches:10 the first method, so called “source-based” approach, accounts for the value-added the first time it leaves the country of origin; the second, “sink-based” approach, considers it the last time it crosses the national borders. The choice between these two approaches depends on the particular issue we want to address. The “source” approach is designed to examine the production linkages and the country/sector participation to different types of production processes. This makes it more suited to assess, for instance, the share of an export flow that crosses just one border, i.e. traditional exports, as opposed to the share that is further re-exported, i.e. GVC-exports. Conversely, the value-added in the “sink” approach is recorded as closely as possible to the moment when it is ultimately absorbed. This makes it more suited to studying the relationship between value-added in exports and final demand, as in the analysis of bilateral trade balances. As already mentioned, each perspective is more suited to address specific em- pirical questions. In Table 2 we provide a non-exhaustive overview of the most common ones, with the best suited accounting method to provide an answer. See Section 4.1 for additional examples, with the related icio syntax. In the Appendix we show the conceptual framework of the value-added account- ing in total exports, following an exporting-country perspective (Appendix C), in bilateral exports, both with a bilateral and an exporting-country perspective (Ap- pendix D), and in all the possible trade flows (Appendix E). Instead, in Section 4.1, we show the implementation in icio and provide some empirical examples with the measures which exceed the correct ones for the aggregate trade flow (i.e. those based on a broader perimeter for defining double counted items). 10 These approaches were proposed by Nagengast and Stehrer (2016) and fully derived by Borin and Mancini (2015, 2017). 12 related syntax. Table 1: A summary of the available perspectives and approaches for each trade flow Perspective Perspective in line with consistent with the trade flow more aggregate flows (i.e. additive) (1) (2) 1. Total exports 1a. Aggregate Exporting-country World (source/sink) 1b. Sectoral Sectoral-exporter Exporting-country (source/sink) 2. Bilateral exports 2a. Aggregate Bilateral Exporting-country (source/sink) 2b. Sectoral Sectoral-bilateral Exporting-country (source/sink) 3. Total imports 3a. Aggregate Importing-country n.a. 3b. Sectoral Sectoral-importer n.a. 13 Table 2: Overview of the most common empirical questions Empirical question Trade flow to select Accounting method GDP embedded in the total exports of a country total aggregate exports exporting-country perspective GDP potentially exposed to: — a shock on a bilateral trade relation (e.g. generic trade frictions bilateral aggregate exports bilateral perspective between two countries) — a shock on a specific sectoral-bilateral trade relation (e.g. a bilateral sectoral exports sectoral-bilateral perspective specific tariff imposed by a trade partner in a given sector) — a shock on the imports of a country (e.g. trade restrictions total aggregate imports importing-country perspective a-vis all partners) vis-` 14 — a shock on the imports of a country in a given sector (e.g. a total sectoral imports sectoral-importer perspective a-vis all partners) specific sectoral tariff vis-` — a shock on the sectoral exports of a country (e.g. negative total sectoral exports sectoral-exporter perspective demand shock on the exports of a given country and sector) Value-added breakdown in disaggregated export flows, consistent sectoral/bilateral/sectoral- exporting-country perspective, with total aggregate measures bilateral exports source or sink approach Value-added breakdown of bilateral trade balances bilateral exports exporting-country perspective, sink approach Traditional exports vs GVC-exports any export flow exporting-country perspective, source approach 4.1 Implementation: Accounting for value-added in gross trade Depending on the specific empirical question, the user needs to choose the appro- priate icio’s options in order to select: i) the desired trade flow; ii) the best suited accounting methodology to single out “double counted” components (see above for a definition of “double counting”); iii) the appropriate output measure(s). As to the first point, the following types of trade flows are considered: i) aggre- gate exports; ii) sectoral exports; iii) bilateral exports, iv) sectoral-bilateral exports, v) aggregate imports, vi) sectoral imports. It is also worth recalling that the option group() allows to consider value-added decompositions for country aggregates, so that the set of trade flows’ combinations is actually broader (see Section 3.1.2 for examples). For each trade flow, we consider the accounting perspectives and ap- proaches that appear to be more economically important (see Appendix E for an overview). We structured the icio’s syntax according to the different trade flows as follows 1. Value-added and GVC participation in total exports of a country: a) Value-added and GVC participation in total aggregate exports: icio, exporter(country code) [methods 1a ] [output exports] [origin destination ] [standard options] b) Value-added and GVC participation in total sectoral exports: icio, exporter(country code [, sector code]) [methods 1b ] [output exports] [origin destination ] [standard options] 2. Value-added and GVC participation in bilateral exports: a) Value-added and GVC participation in bilateral aggregate exports: icio, exporter(country code ) importer(country code) [methods 2a ] [output exports] [origin destination ] [standard options] b) Value-added and GVC participation in bilateral sectoral exports: icio, exporter(country code [, sector code]) importer(country code) [methods 2b ] [output exports] [origin destination ] [standard options] 3. Value-added in total imports of a country: a) Value-added in total aggregate imports: icio, importer(country code) [methods 3a ] [output imports] [origin destination ] [standard options] 15 b) Value-added in total sectoral imports: icio, importer(country code [, sector code]) [methods 3b ] [output imports] [origin destination ] [standard options] 4.1.1 Accounting methods’ options The options perspective() and approach() can be used to select the appropriate accounting methodology (i.e. [methods * ] in the syntax reported in Section 4.1) for answering the empirical question of interest. For the different trade flows, Table 3 reports a summary of the available perspectives. As already mentioned at the beginning of Section 4, whenever the perspective is set at a more aggregate level as compared to the considered trade flow, two alternative approaches are available. By using the icio option approach(source) the item is classified as “value-added” the first time it crosses the national border, whereas the option approach(sink) allows to consider it as “value-added” the last time it crosses the border. Table 3: A summary of the available perspectives and approaches for each trade flow: syntax Perspective Perspective in line with consistent with the trade flow more aggregate flows (i.e. additive) 1. Total exports persp(world) approach(source) 1a. Aggregate persp(exporter) persp(world) approach(sink) persp(exporter) approach(source) 1b. Sectoral persp(sectexp) persp(exporter) approach(sink) 2. Bilateral exports persp(exporter) approach(source) 2a. Aggregate persp(bilateral) persp(exporter) approach(sink) persp(exporter) approach(source) 2b. Sectoral persp(sectbil) persp(exporter) approach(sink) 3. Total imports 3a. Aggregate persp(importer) n.a. 3b. Sectoral persp(sectimp) n.a. 4.1.2 Output options For the selected trade flow, icio allows to compute the main indicators of gross trade and value-added through the output() option. 16 For export flows (i.e. [output exports] in Section 4.1) the default output option is output(detailed). It allows to get a complete value-added decomposition of the trade flows according to the conceptual scheme of Figure C.2 in Appendix C. Gross trade - gtrade - is split in the part that is originally produced by the exporting country (domestic content - dc) and the part that is produced abroad (foreign content - fc); in turn, each of these components is broken up in a part of value- added item (domestic value-added - dva - and foreign value-added - fva) and in a part of double counting.11 The methodology used to single out the value-added and double-counted components changes according to the selected perspective/approach options, while the gtrade, dc and fc measures are, by construction, the same regardless of the accounting methodology. The detailed output also includes additional indicators of trade in value-added that have been singled out in the literature (e.g. VAX by Johnson and Noguera, 2012; Reflection by Koopman et al. 2014; DAVAX and VAXIM by Borin and Mancini, 2019, see Appendix C for an overview) and also measures of GVC partici- pation12 as developed in Borin and Mancini (2015, 2019). The additional indicators that are included in the detailed output vary consistently with the selected perspec- tive/approach. The user can also ask for a specific trade indicator by specifying one of the following arguments of the output() option: gtrade, dc, dva, fc and fva.13 As an additional feature, it is also possible to single out the country/sector where the goods/services were originally produced by specifying the origin(country - code [, sector code]) option, as well as the market/sector where they are absorbed in the final demand, by specifying the destination(country code [, sector code]) option (i.e. [origin destination] in Section 4.1). Results for all countries or all sectors can be computed and displayed simultaneously, using all as argument for country code or sector code. Notice that country code and sector code cannot be both all. If the aim is to compute the value-added produced by a specific coun- try/sector of origin, the option output(va) has to be specified.14 The gross content term (i.e. value-added + double counted items) for a specific country (and sector) 11 Double-counted terms are not singled out as output options in icio, but can be easily com- puted by subtracting the value-added component, either domestic or foreign, from the correspon- dent domestic or foreign content. 12 See Section 5 for more details on these indicators and how to compute them. 13 In addition to value-added and gross trade measures, for any export flow when perspective(exporter) and approach(source) are specified - these options are actually the default - it is also possible to compute the value of trade that is related to GVCs and its back- ward and forward sub-components, specifying gvc, gvcb or gvcf in output(), respectively. These measures are discussed in detail in Section 5. 14 Note that, when the country in origin() corresponds to that specified in exporter(), icio provides the same results when selecting output(dva) or output(va). 17 of origin can be computed by specifying output(gtrade). As far as import flows are concerned (i.e. [output imports] in Section 4.1), the distinction between domestic and foreign items is less relevant, as the former would refer only to the items produced, exported and then re-imported by the importing country itself. For this reason, the default in this case is to compute the gross trade value (i.e. gtrade). The imported value-added (gross-content) can be traced back to the country of origin specifying the option origin(country code [, sector - code]) together with output(va) (output(gtrade)). Of special note is that it is possible to pin down also a specific market/sector of final demand by specifying the destination(country code [, sector code]) option. As for the standard options, (i.e. [standard options] in Section 4.1), the save() and group() options are available for both export and import flows (see Section 3.1 for a description of these options). 4.1.3 Examples: Value-added in trade flows In this section we provide some examples of the insights that icio, dealing with the break down of value-added in trade flows, can bring for the economic analysis of ICIO tables. As running example, we select a specific trade flow, the Chinese total aggregate exports. After having loaded the year 2014 of the last release of the WIOD tables using icio_load, iciot(wiodn) year(2014) the user can easily obtain a detailed breakdown of the selected trade flow, both in millions of US dollars and as a share, using . icio, exporter(chn) Decomposition of gross exports: Perspective: exporter Exporter: CHN Importer: total CHN exports Millions of $ % of export Gross exports (GEXP) 2425406.15 100.00 Domestic content (DC) 2039474.07 84.09 Domestic Value-Added (DVA) 2016712.86 83.15 VAX -> DVA absorbed abroad 1957739.47 80.72 Reflection 58973.39 2.43 Domestic double counting 22761.21 0.94 Foreign content (FC) 385932.09 15.91 Foreign Value-Added (FVA) 380473.47 15.69 18 Foreign double counting 5458.62 0.23 GVC-related trade (GVC) 781287.59 32.21 GVC-backward (GVCB) 408693.30 16.85 GVC-forward (GVCF) 372594.29 15.36 The detailed decomposition can be also computed for a particular sector of export, e.g., “Manufacture of computer, electronic and optical products” (sector code 17 for the loaded ICIO table) by using15 . icio, exporter(chn,17) Decomposition of gross exports: Perspective: exporter Exporter: CHN Importer: total CHN exports Sector of export: 17 Millions of $ % of export Gross exports (GEXP) 560552.89 100.00 Domestic content (DC) 417041.87 74.40 Domestic Value-Added (DVA) 404306.15 72.13 VAX -> DVA absorbed abroad 386215.71 68.90 DAVAX 315342.79 56.26 Reflection 18090.44 3.23 Domestic double counting 12735.72 2.27 Foreign content (FC) 143511.01 25.60 Foreign Value-Added (FVA) 140161.87 25.00 Foreign double counting 3349.14 0.60 GVC-related trade (GVC) 245210.09 43.74 GVC-backward (GVCB) 156246.74 27.87 GVC-forward (GVCF) 88963.36 15.87 DAVAX: Value-Added directly absorbed by the importer We now show how the results can change by using a different perspective on the same trade flow. We move from the default - exporting country perspective - to a sectoral-exporter perspective by adding the option perspective(sectexp) . icio, exporter(chn,17) perspective(sectexp) Decomposition of gross exports: Perspective: sectexp Exporter: CHN Importer: total CHN exports Sector of export: 17 Millions of $ % of export Gross exports (GEXP) 560552.89 100.00 15 Run icio, info after icio_load to get the complete country and sector lists. 19 Domestic content (DC) 417041.87 74.40 Domestic Value-Added (DVA) 409968.49 73.14 VAX -> DVA absorbed abroad 391624.68 69.86 Reflection 18343.80 3.27 Domestic double counting 7073.39 1.26 Foreign content (FC) 143511.01 25.60 Foreign Value-Added (FVA) 141076.94 25.17 Foreign double counting 2434.07 0.43 While the output confirms that domestic and foreign contents are not affected by changing perspective, the value-added terms are higher and, as a consequence, double counting items are smaller. This is not surprising since the sectoral-exporter perspective features a more restrictive definition of double counting (see Appendix E). Which perspective should be used? It depends on the specific empirical ques- tion. If the goal is to measure to what extent the GDP of a country could be exposed to a certain shock on the exports of a sector, a sectoral-exporter perspective might be appropriate. Indeed, in this case, the relevant border to trace value-added is the exporting country/sector’s one. Instead, the default perspective (the “exporting country” one) is the most appropriate if the goal is to compute GVC-related trade indices - since to this end the relevant border is always the exporting country’s one - and is suited to obtain measures of value-added trade traced in disaggre- gated trade flows that are consistent with the aggregate figures. Notice that this additivity property is a feature of the “exporting country” perspective only.16 It can be easily verified by showing that value-added components and GVC-related trade in the aggregate Chinese exports - as computed before using the synatx icio, exporter(chn) - equal the sum of the very same measures obtained for each sector. A possible implementation of this check is the following . qui icio, exporter(chn,all) output(gtrade) . mata : st_matrix("sum_gtrade", colsum(st_matrix("r(gtrade)"))) . di "aggregate Gross exports " %14.2f sum_gtrade[1,1] aggregate Gross exports 2425406.15 . . qui icio, exporter(chn,all) output(dva) . mata : st_matrix("sum_dva", colsum(st_matrix("r(dva)"))) . di "aggregate Domestic Value-Added" %14.2f sum_dva[1,1] aggregate Domestic Value-Added 2016712.86 . . qui icio, exporter(chn,all) output(fva) . mata : st_matrix("sum_fva", colsum(st_matrix("r(fva)"))) 16 Instead, all the other perspectives are non-additive, i.e. measures at a more aggregate level cannot be obtained summing disaggregate results. 20 . di "aggregate Foreign Value-Added" %14.2f sum_fva[1,1] aggregate Foreign Value-Added 380473.47 . . qui icio, exporter(chn,all) output(gvc) . mata : st_matrix("sum_gvc", colsum(st_matrix("r(gvc)"))) . di %11.0g "aggregate GVC-related trade " %14.2f sum_gvc[1,1] aggregate GVC-related trade 781287.59 We now select a different trade flow, moving to bilateral sectoral exports. In particular, we consider the Chinese exports of computer, electronic and optical products to the United States. The default assessment of the extent of GVC par- ticipation, as well as a breakdown of the flow in terms of value-added components consistent both with the aggregate Chinese exports and with total Chinese exports to the United States, is obtained using the default “exporting country” perspective as: . icio, exporter(chn,17) importer(usa) Decomposition of gross exports: Perspective: exporter Exporter: CHN Importer: USA Sector of export: 17 Millions of $ % of export Gross exports (GEXP) 107292.76 100.00 Domestic content (DC) 79824.00 74.40 Domestic Value-Added (DVA) 77386.31 72.13 VAX -> DVA absorbed abroad 76957.76 71.73 DAVAX 72570.84 67.64 Reflection 428.55 0.40 Domestic double counting 2437.68 2.27 Foreign content (FC) 27468.76 25.60 Foreign Value-Added (FVA) 26827.72 25.00 Foreign double counting 641.04 0.60 GVC-related trade (GVC) 34721.92 32.36 GVC-backward (GVCB) 29906.44 27.87 GVC-forward (GVCF) 4815.48 4.49 DAVAX: Value-Added directly absorbed by the importer Again, we can select a perspective in line with the level of aggregation of the chosen trade flow, i.e. a sectoral-bilateral perspective - perspective(sectbil), for instance to measure to what extent the Chinese GDP could be exposed to a tariff imposed by the United States on the imports of computer, electronic and optical products: . icio, exporter(chn,17) importer(usa) perspective(sectbil) 21 Decomposition of gross exports: Perspective: sectbil Exporter: CHN Importer: USA Sector of export: 17 Millions of $ % of export Gross exports (GEXP) 107292.76 100.00 Domestic content (DC) 79824.00 74.40 Domestic Value-Added (DVA) 79815.64 74.39 VAX -> DVA absorbed abroad 79373.64 73.98 Reflection 442.00 0.41 Domestic double counting 8.36 0.01 Foreign content (FC) 27468.76 25.60 Foreign Value-Added (FVA) 27465.88 25.60 Foreign double counting 2.88 0.00 According to WIOD 2014 data, Chinese value-added potentially exposed to this tariff turns out to be around $79.8 billion, as shown by the value reported for the Domestic Value-Added (DVA). This is higher than the $77.4 billion of Chinese value-added traced in the same export flow using an “exporting country” perspective (see the previous icio output above). Thus, if we had used the latter perspective, we would have understated the Chinese exposure. Again, each empirical question calls for the best suited perspective: the default “exporting country” is more useful if the aim is to assess GVC participation or to retrieve measures of value-added trade consistent with the figures at a more aggregate level; the perspective in line with the selected trade flow is more suited to encompass the entire value-added that might be affected by a shock hitting that particular flow. The same reasoning applies when the objective is to choose the best suited perspective for bilateral aggregate exports. Here the choice will be, again, between the default “exporting country” perspective and the bilateral one. For example, Figure 4.9 of the World Bank World Development Report (WDR) 2020, reported here as Figure 1, aims at quantifying the potential exposure of other countries to a US-China trade war. In fact, US and Chinese exports embed a non-negligible amount of other countries foreign value-added that would be indirectly exposed to new tariffs. For instance, around 2% of the value-added in the Chinese exports to the United States consists of Japan’s GDP and almost 1.8% of the Republic of Korea’s GDP. In turn, around 2.5% of the value-added in the US exports to China is Canadian GDP. These numbers can be easily obtained using icio by selecting a bilateral perspective and retrieving the value-added by country of origin in a particular bilateral flow. The entire value-added, domestic and foreign, that crosses 22 the specific bilateral border where the new tariffs could be in place, i.e. the GDP potentially affected by trade barriers, can be computed using the following syntax . *Replicate data of WDR2020 Figure 4.9 . icio_load, iciot(eora) year(2015) Loading table eora 2015... loaded . icio, exp(chn) imp(usa) persp(bilat) output(va) origin(all) save(wdr_4_9a.xls) Decomposition of gross exports: Perspective: bilateral Origin: ALL Exporter: CHN Importer: USA Output: Value-Added Millions of $ % of export AUS 2115.55 0.57 BEL 610.65 0.17 BRA 1046.77 0.28 CAN 955.66 0.26 CHE 977.48 0.27 DEU 4058.78 1.10 FRA 1919.28 0.52 GBR 1604.30 0.44 HKG 1489.18 0.40 IDN 1692.15 0.46 IND 1123.87 0.31 ITA 1564.26 0.42 JPN 7535.08 2.05 KOR 6554.82 1.78 MYS 1604.49 0.44 NLD 748.18 0.20 RUS 2208.83 0.60 SGP 1005.21 0.27 THA 964.69 0.26 USA 5126.99 1.39 Output saved as: wdr_4_9a.xls into the current working directory . icio, exp(usa) imp(chn) persp(bilat) output(va) origin(all) save(wdr_4_9b.xls) Decomposition of gross exports: Perspective: bilateral Origin: ALL Exporter: USA Importer: CHN Output: Value-Added Millions of $ % of export AUS 235.58 0.18 BRA 361.65 0.28 CAN 3326.80 2.56 CHE 295.33 0.23 CHN 2322.90 1.79 DEU 1255.74 0.97 FRA 559.10 0.43 GBR 649.05 0.50 23 IND 294.56 0.23 ITA 438.06 0.34 JPN 1293.39 1.00 KOR 541.96 0.42 MEX 1288.42 0.99 MYS 274.08 0.21 NGA 230.40 0.18 NLD 249.97 0.19 RUS 501.60 0.39 SAU 232.88 0.18 TTO 271.61 0.21 VEN 1547.39 1.19 Output saved as: wdr_4_9b.xls into the current working directory As in Figure 1, we deliberately reported in the above Stata output only the 20 countries with the highest foreign value-added in the bilateral exports between the United States and China. Actually, by running the previous syntax, icio would report also the value-added of the other countries in the EORA MRIO database. Since the complete list is very long, the user may find it useful to exploit the save() option. As can be seen, by adding this option, the complete icio output has been saved into a file called wdr_4_9b.xls within the current working directory. Lastly, we consider the analysis of value-added in the total imports of a country. To quantify the German GDP potentially exposed to US tariffs vis-` a-vis all partners, according to WIOD 2014 data, we can use the following syntax . icio_load Loading table wiod 2014... loaded . icio, origin(deu) imp(usa) output(va) Decomposition of gross imports: Perspective: importer Importer: USA Origin: DEU Exporter: total USA imports Output: Value-Added Millions of $ % of import Value-Added 133064.91 5.53 The Stata output indicates that around $133 billion of Germany’s value-added are imported, directly and indirectly, by the United States (around 5.5% of the total US imports), and thus could be exposed to US trade barriers, according to WIOD 2014 data. German GDP exposure to these trade barriers can be computed taking the ratio with respect to the total German GDP - obtained with icio, origin(deu). Thus, around 3.7% of German GDP could be affected by these 24 Value added (%) Value added (%) 0 0.5 1.0 1.5 2.0 2.5 3.0 0 0.5 1.0 1.5 2.0 2.5 C an Ja a da pa Ko n re C a, Re Ve hi ne na Un p. zu ite el d St a, RB a te s Ru G er Ja pa ss ia m n an n Fe y M de ra ex ic tio o n G Au er Un m str ite an al d y ia Ki ng Fr do an m ce In Fr d on an e sia Ko ce re M Ru ss a ,R Un al ia ay n ep . ite d sia Fe Ki de ng ra t H do 25 io m n on g Ko Ita ng Ita ly SA ly R, Br C az hi i na Sw l itz er In la di nd a Figure 4.9 The multilateral dimension of the U.S.–China trade war In Br az di Tr a il in M Si ng id al a d ys a ap ia or an Sw e d itz To er ba la N go nd et b. U.S. exports to China: Share of value added by U.S. trade partner, 2015 he Th rla ai nd la nd s Au C a. Chinese exports to United States: Share of value added by Chinese trade partner, 2015 an Figure 1: Replication of World Bank WDR 2020 Figure 4.9. str al ad Sa ud ia N a et iA he ra rla bi nd a s N Be ig lg er iu ia m US trade barriers. In a GVC world, the GDP exposure to a trade barrier could be direct - through the country’s exports to the economy that has imposed the trade restrictive measure - or indirect - through the exports of other countries. The former can be computed looking at the German GDP directly exported to the United States, running: icio, origin(deu) exporter(deu) importer(usa) perspective(bilateral) output(va). Thus, 2.8% of German GDP could be directly affected by US tariffs, while 0.9% could be affected through other countries exports to the US of German products. If the goal is to quantify the potential exposure of German value-added to a US tariff on a specific sector, e.g. motor vehicles from Germany, a sectoral-importer perspective is the right choice: . icio, origin(deu,20) imp(usa,20) output(va) Decomposition of gross imports: Perspective: sectimp Importer: USA Origin: DEU Exporter: total USA imports Output: Value-Added Sector of import: 20 Sector of origin: 20 Millions of $ % of import Value-Added 17216.14 6.59 . mat GDPsect=r(va) . icio, origin(deu,20) Value-Added by origin/destination: Origin: DEU Output: Value-Added Sector of origin: 20 Millions of $ % of total Value-Added 147493.71 100.00 . mat GDPtot=r(vby) . di "Germany exposure in sector 20: " GDPsect[1,1]/GDPtot[1,1]*100 "%" Germany exposure in sector 20: 11.672456% Again, the relative exposure can be easily obtained taking the ratio of the ab- solute exposure ($17.2 billion) with respect to the total German value-added in the motor vehicles industry ($147.5 billion). Thus, a US tariff hitting motor vehicles imports from Germany might affect around 11.7% of the value-added produced in the same sector in Germany. 26 Other examples of questions that could be answered using icio for the analysis of value-added trade are: • Which part of a country’s total exports is home produced, i.e. is domestic GDP? icio, exporter(deu) output(dva) • Which part of a country’s total exports can be traced back to other countries GDP? icio, exporter(deu) output(fva) • Where the foreign value-added in German exports is produced? icio, origin(all) exporter(deu) output(fva) • Considering the bilateral exports from Italy to Germany, where the Italian GDP (domestic VA) re-exported by Germany is absorbed? icio, exporter(ita) importer(deu) destination(all) output(dva) • How can be obtained the complete breakdown by origin and destination of the value-added (both domestic and foreign) for Chinese exports to the US? icio, origin(all) exporter(chn) importer(usa) destination(all) output(va) save(CHN_to_USA.xls) • How can the (corrected) Koopman et al. (2014) decomposition be retrieved using icio? icio, exporter(deu) perspective(world) approach(sink) • Which is the Chinese GDP that at any point in time, passes through a certain bilateral trade flow, say Chinese exports to the United States? In other terms, what is the Chinese GDP potentially exposed to US tariffs on imports from China? icio, exp(chn) imp(usa) persp(bilat) output(dva) • Which is the German GDP potentially exposed to US trade barriers on all imports? icio, origin(deu) imp(usa) persp(importer) output(va) • Which is the German GDP that could be affected by US tariffs on imports in sector 20? icio, origin(deu) imp(usa,20) persp(sectimp) output(va) 27 • Which is the exposure of US GDP to a Chinese tariff on US imports in sector 17? icio, exp(usa,17) imp(chn) persp(sectbil) output(dva) • To what extent are Italian sectors exposed to a shock on German’s exports in sector 20? icio, origin(ita,all) exp(deu,20) persp(sectexp) output(va) 5 Measuring GVC-related exports Following the original idea by Hummels et al. (2001), many contributions in the literature have shared the view that the trade flow related to GVC activity should consist in goods and services crossing more than one border along the production process. Borin and Mancini (2015) made this definition operational by proposing a way to isolate traditional trade from gross flows (i.e. the portion of trade crossing just one border) and considering the remaining part as a proxy of the GVC related trade. This GVC indicator presents three desirable features : i) it is bounded between 0 and 1, since it traces within a particular trade flow the share of it related to GVC activity, i.e., the numerator is a sub-component of the denominator; ii) it is additive at any level of aggregation/disaggregation of trade flows; thus, data can be summed at any level (total country exports/world exports/world sector exports/country groups) in order to obtain the proper GVC participation measures at the desired level of aggregation; iii) it can be broken down into two additive terms, i.e. a ‘backward’ component corresponding to import content of exports and a “forward” component, which measures the part of domestic production that is supplied to the importing country to be processed and re-exported. In Appendix F we provide its conceptual framework while in Section 5.1 we show how to compute GVC measures in icio and present some examples. 5.1 Implementation: GVC in exports To compute GVC measures with icio, the user needs to select: i) the desired trade flow and ii) the appropriate GVC measure to be computed (overall, backward or for- ward participation). The option perspective(exporter) is in this case imposed, since only this perspective allows to distinguish between the value of trade crossing just one border and the value of trade further re-exported, i.e. GVC trade. 28 5.1.1 Syntax The icio syntax for the different export flows is the following: 1. GVC participation in total exports of a country: a) GVC participation in total aggregate exports: icio, exporter(country code) [output gvc ] [origin destination ] [standard options] b) GVC participation in total sectoral exports: icio, exporter(country code [, sector code]) [output gvc] [origin destination ] [standard options] 2. GVC participation in bilateral exports: a) GVC participation in bilateral aggregate exports: icio, exporter(country code ) importer(country code) [output gvc ] [origin destination ] [standard options] b) GVC participation in bilateral sectoral exports: icio, exporter(country code [, sector code]) importer(country code) [output gvc ] [origin destination ] [standard options] The output() option, i.e. [output gvc ] in the reported syntax, allows to get different measures of GVC-related trade by specifying gvc, gvcb and gvcf as ar- guments for total, backward and forward GVC indicators, respectively. As can be noted from the icio results reported in Section 4.1.3, GVC-related indicators are routinely reported as part of the detailed output, when an export flow-at any level of aggregation-is specified. Also for GVC indicators it is possible to single out the country/sector where the goods/services were originally produced by specifying the origin(country - code[,sector code]) option, as well as the market/sector where the goods/services are absorbed in final demand by specifying the destination(country code[,sector - code]) option. 5.1.2 Examples: GVC-related exports Figure 1.13 of the World Bank WDR 2020, here reported as Figure 2, is based on EORA MRIO 1990 and 2015 data and shows the GVC-related trade in agriculture (sectors 1 and 2 in EORA) and agri-food sectors (sector 4 in EORA). For instance, in the case of Tanzania, one of the Sub-Saharan African countries that experienced 29 a significant increase in GVC participation in the agri-food sector, the data used for plotting panel b. of Figure 2 can be obtained by using the following syntax . *Replicate data of WDR2020 Figure 1.13 panel b . icio_load, iciot(eora) year(2015) Loading table eora 2015... loaded . icio, exp(tza,4) output(gvc) Decomposition of gross exports: Perspective: exporter Exporter: TZA Importer: total TZA exports Output: GVC-related trade Sector of export: 4 Millions of $ % of export GVC 93.22 52.74 . icio_load, iciot(eora) year(1990) Loading table eora 1990... loaded . icio, exp(tza,4) output(gvc) Decomposition of gross exports: Perspective: exporter Exporter: TZA Importer: total TZA exports Output: GVC-related trade Sector of export: 4 Millions of $ % of export GVC 38.80 33.53 In Figure B.2.1.1 of the WDR 2020, Vietnam’s integration in the electronics global value chain is discussed. Panel a, here reported as Figure 3, is based on the EORA MRIO database and shows the GVC-backward related trade, i.e. backward integration, in the electrical and machinery sector. Data for 2015 can be obtained running: . *Replicate data of WDR2020 Figure B.2.1.1 panel a . icio_load, iciot(eora) year(2015) Loading table eora 2015... loaded . icio, exp(vnm,9) output(gvcb) Decomposition of gross exports: Perspective: exporter Exporter: VNM Importer: total VNM exports Output: GVC-backward related trade Sector of export: 9 Millions of $ % of export 30 GVC backward 859.70 64.32 Figure 2: Replication of World Bank WDR 2020 Figure 1.13. Figure 1.13 GVCs expanded in both the agriculture and food industries from 1990 to 2015 a. Agriculture GVCs b. Agri-food GVCs 60 90 Agriculture GVC participation, 2015 (%) Agri-food GVC participation, 2015 (%) 80 50 RWA 70 VNM GHA ETH KEN 40 60 HUN TZA 50 SRB 30 LAO GMB 40 SSD BGR 20 30 ETH LBN YEM 20 10 10 0 0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 70 80 Agriculture GVC participation, 1990 (%) Agri-food GVC participation, 1990 (%) East Asia and Pacific Europe and Central Asia Latin America and the Caribbean Middle East and North Africa North America South Asia Sub-Saharan Africa Figure 3: Replication of World Bank WDR 2020 Figure B2.1.1. a. Backward integration of electronics and machinery as a share of gross exports 80 70 60 50 Percent Percent 40 30 20 10 0 2000 2005 2010 2015 Throughout the WDR 2020, several figures on GVC-related trade at the world level are reported. These measures can be obtained with icio retrieving and then summing the GVC-related trade of each country in the loaded input-output table. Since this could be computationally intensive, we have also released a data set with GVC indicators and the most relevant measures of value-added in trade flows 31 computed for any country/sector for each database available in icio. This database is available on the official WDR 2020 website in the data section.17 Other basic examples of questions on GVC participation and the related syntax are: • Which share of the German exports related to GVC is produced in Italy? icio, origin(ita) exporter(deu) output(gvc) • Which share of the German exports is related to backward and forward GVC? icio, exporter(deu) output(gvcb) icio, exporter(deu) output(gvcf) 6 Conclusions In this paper we described the new Stata command icio for value-added trade and GVC analysis. It’s most important features are the following: • It exploits the most famous Inter-Country Input-Output (ICIO) tables - the World Input-Output Database (Timmer et al. 2015), the OECD TiVA database (OECD), and the Eora Global Supply Chain Database (Lenzen et al. 2013) - but also allows to load any user-provided ICIO table. • It provides breakdowns of aggregate, bilateral and sectoral exports and im- ports according to the source and the destination of their value-added content, with a careful treatment of double counted items. These decompositions can be used to: – assess the exposure of countries/sectors to different kind of trade shocks, including tariffs. – get indicators for any level of disaggregation of trade flows that are con- sistent with more aggregate measures, i.e. disaggregated indicators can be summed up to get correct measures in more aggregate trade flows. • It can break down export flows in terms of “traditional” vs GVC-trade, at any level of aggregation, also distinguishing between backward and forward participation in GVC. 17 Go to https://www.worldbank.org/en/publication/wdr2020/brief/world-development-report- 2020-data. 32 • It is flexible and open, as we plan to release updates to include new ICIO database, as soon as the data become available, as well as other measures to assess the participation and position of countries and sectors in GVCs and trade policy analysis. It is worth noting that the measures computed with icio, as any other measure obtained from ICIO tables, suffer from some limitations (Antr` as, 2019). In fact, ICIO tables are built under the strong proportionality assumptions, i.e. all output within each country-industry is built with the same input mix (de Gortari, 2019). However, input-output datasets will soon start exploiting customs data to allow for more heterogeneity in production and trade (United Nations, 2018). Once ICIO tables become more detailed, value-added trade measures obtained with the differ- ent perspectives featured in icio will diverge more and more, making it even more important to have available the best suited accounting framework to answer each specific empirical question. 33 References Antr`as, P. 2019. ‘Conceptual aspects of Global Value Chains.’ NBER Work- ing Paper, No. 26539. as, P. and D. Chor, 2013. ‘Organizing the global value chain.’ Econo- Antr` metrica, 2013, 81(6), pp. 2127-2204. as, P. and D. Chor, 2018. ‘On the Measurement of Upstreamness and Antr` Downstreamness in Global Value Chains.’ World Trade Evolution: Growth, Productivity and Employment, 126-194. Taylor & Francis Group. Arto, I., Dietzenbacher, E. and J.M. Rueda-Cantuche, 2019. ‘Measuring bi- lateral trade in terms of value added’, JRC Technical Reports, 29751. Borin, A. and M. Mancini, 2015. ‘Follow the value added: bilateral gross export accounting’, Economic Working Papers no. 1026, Bank of Italy. Borin, A. and M. Mancini, 2019. ‘Measuring What Matters in Global Value Chains and Value-Added Trade’, Policy Research working paper. no. WPS 8804; WDR 2020 Background Paper. Washington, D.C.: World Bank Group. de Gortari, A., 2019. ‘Disentangling Global Value Chains.’, Dartmouth Col- lege, mimeo. Fally, T., 2012, ‘Production Staging: Measurement and Facts.’, mimeo UC Berkeley. Hummels, D., J. Ishii and K.M. Yi, 2001. ‘The Nature and Growth of Vertical Specialization in World Trade.’ Journal of International Economics, 54, pp. 75-96. Johnson, R. C., 2018. ‘Measuring Global Value Chains’, Annual Review of Economics, Vol. 10:207-236. Johnson, R. C. and G. Noguera, 2012. ‘Accounting for Intermediates: Produc- tion Sharing and Trade in Value Added.’ Journal of International Economics, 86, Iss. 2, pp. 224-236. Koopman, R., W. Powers, Z. Wang and S. Wei, 2010. ‘Give Credit Where Credit is Due: Tracing Value-added in Global Production Chains.’ NBER Working Paper, No. 16426. 34 Koopman, R., Z. Wang and S. Wei, 2014. ‘Tracing Value-Added and Double Counting in Gross Exports.’ American Economic Review, 104(2): 459-94. Lenzen, M., D. Moran, K. Kanemoto and A. Geschke, 2013. ‘Building EORA: a global multi-region input-output database at high country and sector reso- lution’, Economic Systems Research, 25:1, pp. 20-49. Los, B., M. P. Timmer and G. de Vries, 2014. ‘How Global Are Global Value Chains? A New Approach to Measure International Fragmentation.’ Journal of Regional Science, 55, No. 1, pp. 66-92. Los, B. and M. P. Timmer, 2018. ‘Measuring Bilateral Exports of Value Added: A Unified Framework.’ NBER Working Paper No. 24896. Miroudot, S., and M. Ye, 2018.‘A simple and accurate method to calculate domestic and foreign value-added in gross exports,’ MPRA Paper 89907, Uni- versity Library of Munich, Germany. Miroudot, S., and M. Ye, 2017.‘Decomposition of Value-Added in Gross Ex- ports: Unresolved Issues and Possible Solutions,’ MPRA Paper 83273, Uni- versity Library of Munich, Germany. Nagengast, A.J. and R. Stehrer, 2016. ‘Collateral imbalances in intra-European trade? Accounting for the differences between gross and value-added trade balances’ The World Economy. OECD, 2018. ‘Trade in Value Added database’, oe.cd/tiva. Timmer, M. P., E. Dietzenbacher, B. Los, R. Stehrer and G.J. de Vries, 2015. ‘An Illustrated User Guide to the World Input-Output Database: the Case of Global Automotive Production.’ Review of International Economics, 23(3). United Nations, 2018. ‘Handbook on Supply, Use and Input-Output Ta- bles with Extensions and Applications.’ ST/ESA/STAT/SER.F/74/Rev.1, United Nations Publications, New York. Wang, Z., S. Wei and K. Zhu, 2013. ‘Quantifying International Production Sharing at the Bilateral and Sector Levels.’ NBER Working Paper, No. 19677. Wang, Z., S. Wei, X. Yu and K. Zhu, 2017. ‘Measures of Participation in Global Value Chains and Global Business Cycles.’ NBER Working Paper, No. 23222. 35 World Bank, 2019. ‘World Development Report 2020. Trading for Develop- ment in the Age of Global Value Chains’, Washington, D.C.: World Bank Group. 36 A Conceptual framework: ICIO models A generic ICIO model with G countries and N sectors can be represented by the scheme in Figure 1, where Zij is the N ×N matrix of intermediate inputs produced in country i (rows) and used in country j (columns); Yij is the N × 1 vector of final goods and services completed in country i and absorbed in country j ; Xi is the N × 1 vector of gross output produced in country i; and VAi is the 1 × N vector of value-added generated in country i. Figure A.1: Inter-Country Input-Output scheme The specific column j, n of the ICIO table in Figure A.1 shows how the out- put of country j and sector n (xj,n ) is produced: i.e. sourcing intermediate in- puts from the same and other country/sector pairs and adding its own value- G N added (xj,n = i m zij,mn + vaj,n ). In turn, the row j, n shows how output of country j and sector n is used: i.e. as intermediate inputs for different indus- tries and countries and as final products to serve domestic and foreign demand (xj,n = G i N m zji,nm + G i yji,n ). It is worth noting that IO models hinge upon key proportionality assumptions: the input composition in sectoral productions does not change by geographical destination of output and it is identical between intermediate and final goods. 37 B Conceptual framework: Supply and demand in ICIO models Given a country s, each unit of its gross output can be either consumed as a final good or used as an intermediate good at home or abroad: G Xs = (Asr Xr + Ysr ), (B.1) r where country r can either be s itself or any given importing country, A is the GN ×GN matrix of intermediate inputs coefficients, obtained dividing Z by X (i.e. A = Z (u ⊗ X )). Then the basic relationship between gross output and final demand is given by: X = (I − A)−1 Y = BY, (B.2) where B is the GN ×GN “global” Leontief inverse that measures the total units of gross output in countries-sectors of origin necessary to produce a certain unit of final goods/services. Indeed, B accounts for all the gross output produced in all the rounds of production, as B = I + A + A2 + A3 + . . . + An = (I − A)−1 . In each production stage some value-added is generated. The value-added share in each unit of gross output produced by country s (Vs ) is equal to one minus the sum of the direct intermediate input shares of all the domestic and foreign suppliers (i.e. Vs = uN (I − G r Ars ). Then the direct domestic value-added matrix for all countries can be defined as follows:   V1 0 · · · 0  0 V2 · · · 0    V= .  . .. . . (B.3)  .. . . . . .   0 0 ··· VG Pre-multiplying the right-hand side of equation (B.2) by V, it is possible to obtain a G×G GDP matrix reporting the GDP by country pairs of source (rows) and absorption (columns). G G G   V1 r B1r Yr1 V1 r B1r Yr2 ··· V1 r B1r YrG G G G  V2 B2r Yr2 V2 B2r Yr2 ··· V2 B2r YrG    r r r GDP =  . . ... . . (B.4)   . . . . . .   G G G VG r BGr YrG VG r BGr YrG · · · VG r BGr YrG 38 More specifically, the GDP produced in country s can be computed as: G G G G G GDPs = Vs Bsk Ykl = Vs Bsk Yks + Vs Bsk Ykl . (B.5) k l k k l =s domestically GDP absorbed absorbed GDP abroad (VAXs ) where we have singled out the part that is absorbed at home and the part that is finally consumed abroad in country l, as a final good assembled in country k , which correspond to the “value-added exports” as defined by Johnson and Noguera (2012). It is also possible to decompose the final demand of country s by distinguishing between the part of value-added domestically produced and the one that originates abroad: G G G G G FDs = Vj Bjk Yks = Vs Bsk Yks + Vt Btk Yks . (B.6) j k k t=s k domestically FD produced produced FD abroad To get a decomposition of GDP by sectors of origin, it is sufficient to substitute the direct value-added Vs in equation (B.5) and (B.6) with its diagonalized form Vj (i.e. the N ×N diagonal matrix with the direct value-added coefficients along the principal diagonal and zeros elsewhere). Similarly, the decomposition by sectors of final absorption is obtained by replacing the vector of final demand with its diagonalized form. For instance, for goods completed in country k and absorbed in country l, the N ×N diagonal matrix of final demand is as follows:   ykl,1 0 · · · 0  0 ykl,2 · · · 0    Ykl ≡   . . ... . .  . . . . . .   0 0 · · · ykl,N . Then the decomposition of value-added by combinations of county-sector of origin and country-sector of final destination can be obtained from the GN ×GN matrix: VA(origin/destination) = VBY. (B.7) 39 C Conceptual framework: Value-added in total exports The problem of isolating value-added in trade flows has been addressed at length in the literature.18 To provide a useful starting point, we begin from the analysis of the aggregate exports of a country. Gross exports of country s can be broken down according to the country that initially produced each component. The part that originated in country s itself is referred to as the ‘domestic content of exports’ (DCs ), whereas the remaining part is called the ‘foreign content of exports’ (FCs , Koopman et al. 2010): G uN Es∗ = Vs Bss Es∗ + Vt Bts Es∗ . (C.1) t=s domestic foreign content (DCs ) content (FCs ) Although the above formula resembles closely those used to decompose the GDP of a country in equation (B.5) or the final demand in equation (B.6), the two components in (C.1) cannot be considered as “net” measures of production, i.e. value-added. In other terms, while they were indeed generated at home and abroad, respectively, they are not a measure of the GDP produced by the different countries. The reason is that VB pre-multiplies the vector of gross exports Es∗ which does not include only final products (i.e. Ys∗ ), as in equations (B.5) and (B.6), but also intermediate goods that later can be can re-imported and re-exported by the same country many times. Indeed, Koopman et al. (2014) point out that the same value-added may cross country s’s borders several times along the production process so that it would be counted many times in its gross exports (Es∗ ). This phenomenon, called “double counting”, can be easily figured out by considering the following example of a simple sequential production chain. Suppose that 1 USD of value-added originally produced in A is first exported to B as intermediate inputs, processed there, then shipped back to A and used to produce final goods for re- export to C. The value-added generated in the very first stage of production in A is counted twice: one in its gross bilateral exports with B and one in its exports to C. Koopman et al. (2014) isolate these double counted items in aggregate trade flows, by proposing an accounting framework which allows to single out the entire 18 See for example Wang et al. (2013), Koopman et al. (2014), Borin and Mancini (2015), Los et al. (2016); Nagengast and Stehrer (2016). 40 domestic and foreign value-added embedded in the aggregate exports of country s, as well as the double counted items originally produced at home and abroad. Figure C.2 shows a scheme of the basic breakdown of aggregate exports decomposition of total exports. Figure C.2: A scheme of value-added decomposition of total exports based on Koop- man et al. (2014), extended by Borin and Mancini (2019) Gross Exports (Es∗ ) Domestic Content Foreign Content (DCs∗ ) (FCs∗ ) Domestic Domestic Foreign Foreign Double value-added Double Counted value-added Counted (DVAs∗ ) (DDCs∗ ) (FVAs∗ ) (FDCs∗ ) Value-added Reflection Exports (REFs∗ ) (VAXs∗ ) Directly absorbed VAX (DAVAXs∗ ) Indirectly absorbed VAX Notice that VAXs is a subcomponent of the domestic value-added embedded in gross exports, the remaining part being the value-added that is finally absorbed by the exporting country itself (labeled “reflection” by Koopman et al., 2014). Borin and Mancini (2015) show how the VAXs can be further split in the part that is directly absorbed by the countries which are importing from s, called DAVAX (i.e. Directly Absorbed Value-Added in exports), and a part that is re-exported to third countries. This distinction is particularly useful for identifying the portion of exports that is involved in GVCs (see Section 5). Albeit the original Koopman et al. (2014) decomposition presents some draw- 41 backs and limitations,19 the general scheme they proposed remain a useful concep- tual framework for the value-added decomposition of trade flows at any level of disaggregation. Indeed, in most of the cases, the default output of icio replicates the basic part of the scheme depicted in Figure C.2. Different methodologies have been developed in the literature aiming to pin down the value-added embedded in gross export flows (see, among others, Wang et al., 2013; Koopman et al. 2014; Borin and Mancini, 2015, 2019; Los et al., 2016; Johnson, 2018; Miroudot and Ye, 2018).20 Here we present one of the possible methodologies for measuring value-added in aggregate exports - i.e. the one pro- posed by Borin and Mancini (2019) that, in the accounting of domestic value-added, is algebraically equivalent to Los and Timmer (2016). Double counting in the total gross exports of a given country s occurs whenever items that are first exported by s are then re-imported and used to produce goods and services to be exported again. Conceptually, one way to distinguish between “value-added” and “double counting” is to split the production chain in phases, each one delimited by an export flow of country s: what is generated within that particular production phase is accounted for as “value-added” in exports, what comes from further upstream production stages is “double counted”. This can be implemented in a general ICIO framework by modifying the matrix B in such a way that we can slice down the production process along the outward boundaries of the exporting country s. To this end, consider the representation of the global Leontief inverse as a sum of infinite series of the gross output generated in all upstream stages of the production process: B = I + A + A2 + A 3 + . . . + An n → ∞. (C.2) We can split the production process along country s’s borders by carving out its intermediate export linkages at any stage of the above series. Algebraically, it can be implemented by setting to zero the coefficients of matrix A which identify the 19 See Nagengast and Stehrer (2016), Miroudot and Ye (2017) and Borin and Mancini (2019) for a detailed discussion on this point. 20 These contributions differ in the types of trade flows they consider, in the targeted measures, in the solutions they propose for the value-added decomposition of disaggregated trade flows (i.e. at bilateral and/or sectoral level) and in their approaches to the foreign value-added accounting. Nevertheless, they reach the same results when considering the domestic value-added embedded in the total exports of a country, while exploiting different computation techniques. 42 direct requirement of intermediate inputs from country s (i.e. Asj = 0 ∀ j = s):   A11 A12 · · · A1s · · · A1G  . . .. . . .  . . . . .   . . . . . .   s A/ = 0 0 · · · Ass · · · 0 . (C.3)    . . . . . ..   . . . . . .  . . . . .   AG1 AG2 · · · AGs · · · AGG Then, the corresponding inverse Leontief matrix is: s s −1 B/ = (I − A/ ) . (C.4) s / s / Given that Bis = Bis + Bis Asj Bjs , equation (C.1) can be re-written so j =s that we can single out the “value-added” and “double counted” terms within each component: domestic foreign content (DCs∗ ) content (FCs∗ ) G G s s s / s / uN Es∗ = Vs B/ ss Es∗ + Vs B/ ss Asj Bjs Es∗ + Vt Bts Es∗ + Vt Bts Asj Bjs Es∗ . j =s t=s t=s j =s domestic value domestic double foreign value foreign double added counted added counted (DVAs∗ ) (DDCs∗ ) (FVAs∗ ) (FDCs∗ ) (C.5) Equation (C.5) reproduces the breakdown of bilateral exports into the main items identified in Koopman et al. (2014) (see Figure C.2). The double counted items are measured by isolating the portion of country s that have been already exported by s in a previous stage of the production process. As far as the domestic s / components are concerned, it is worth noting that Bss corresponds to the so-called local Leontief matrix (I − Ass )−1 . This means that the domestic value-added in exports is obtained by isolating all the domestic stages of production needed to produce the exported goods, while ignoring the domestic content of imported in- puts.21 The foreign value-added in (C.5) follows the same rationale, i.e. considering as value-added only the items crossing country s border once.22 21 Notably, this measure of domestic value-added in exports represents the complement to the “import content of exports” proposed by Hummels et al. (2001), but it is also numerically equiva- lent to the domestic value-added found in other contributions that have analyzed aggregate export flows (e.g., Koopman et al., 2014; Los et al., 2016; Johnson, 2018; Miroudot and Ye, 2018). 22 Instead, other contributions (see Koopman et al., 2014; Wang et al., 2013; Nagengast and Ster- her, 2016; Miroudot and Ye, 2018) in the literature adopt a different rationale - world perspective - for foreign value-added accounting, making this measure not commensurate to the domestic 43 In addition to the breakdown of the value-added by country of origin, it is also possible to consider the linkages with the market of final absorption. To this aim, total exports Es∗ can be split into final goods ( G r=s Ysr ) and intermediate inputs required by the production of gross output of the importing countries ( Gr=s Asr Xr ): G G Es∗ = Ysr + Asr Xr . (C.6) r =s r =s Then, intermediate inputs imported by the direct partner (Asr Xr ) can be fol- lowed through the country of final completion and the market of ultimate demand. According to one of the basic IO accounting relation (i.e. X = BY), all the remain- ing (and potentially infinite) stages of production are accounted for by the Leontief inverse matrix B. Finally, the domestic value-added (DVA) and the foreign value- added (FVA) in the total exports of s can be re-expressed as: G G G G −1 DVAs∗ = Vs (I − Ass ) Ysr + Asr Brk Ykl , (C.7) r =s r=s k l and G G G G G s / FVAs∗ = Vt Bts Ysr + Asr Brk Ykl . (C.8) t=s r =s r =s k l It is worth recalling that the two subscripts on final demand matrix Y refer to the country of final completion and the market of final absorption.23 With the icio command it is possible to single out the markets (and sectors) of final absorption using the option destination() (see Section 4.1). One can also obtain information on specific countries/sectors of origin of the value-added with the option origin(). The details on the sectors of origin/destination are obtained with the same algebraic formulation of matrices V and Y shown in equation (B.7). In addition to identifying specific countries-sectors of origin/destination from equations (C.7) and (C.8), the domestic value-added can be also broken down in two main aggregate indicators (see Figure C.2) to distinguish between the DVA ultimately absorbed in the country of origin s (i.e. the “reflection” terms in Koom- value-added at the country level. See Section 5.1 in Borin and Mancini (2019) for further details on this point. 23 For instance Ykl identifies the vector of goods finalized in k and sold in l. 44 pan et al., 2014, terminology, REFs∗ ) or in a foreign market (i.e. the “value-added exports”, or VAXs∗ , in Johnson and Noguera, 2012 nomenclature): G G G REFs∗ = Vs (I − Ass )−1 Ysr + Asr Brk Yks , (C.9) r =s r=s k G G G G −1 VAXs∗ = Vs (I − Ass ) Ysr + Asr Brk Ykl . (C.10) r=s r=s k l =s D Conceptual framework: Value-added account- ing in bilateral exports Bilateral perspective If we are interested in measuring the total value-added that crosses a specific bilateral border, for instance to assess the exposure to tariffs imposed by the bilateral partner, we need an accounting method for value-added in bilateral exports that excludes from gross trade figures only the items that are double counted in the very same bilateral flow. In other words, the specific bilateral relation represents the perimeter for defining double-counted flows in gross exports. This matters, for example, when we are interested in singling out the value-added crossing a specific border which could be exposed to trade tensions between two countries on each side of the relationship. By proceeding as for the derivation of value-added decomposition for aggregate trade flows (see Appendix C), we can modify the input coefficient matrix A to split the production process along the new perimeter and single out the “value-added” and “double counted” items. While in the exporting-country perspective we set to zero the coefficients that identify the direct requirement of intermediate inputs from country s to all the other countries, here we only set to zero the bilateral coefficient matrix Asr :   A11 · · · A1s · · · A1r · · · A1 G  . . . . . . .  . . . . . . .   . . . . . . .   sr A  =  As 1 · · · Ass · · · 0 ··· AsG  . (D.1)    . . . . . . .   . . . . . . .  . . . . . . .   AG1 · · · AGs · · · AGr · · · AGG Then, the corresponding inverse Leontief matrix can be defined as: 45 sr sr −1 B = (I − A ) . (D.2) By analogy with the derivation of the decomposition of aggregate exports in (C.5), we can express the complete decomposition of bilateral exports based on a bilateral perspective: domestic foreign content (DCsr ) content (FCsr ) G G sr sr sr sr ss Esr + Vs Bss Asr Brs Esr + uN Esr = Vs B  ts Esr + Vt B ts Asr Brs Esr . Vt B  t=s t=s bilateral bilateral bilateral bilateral perspective perspective perspective perspective DVA∗ sr DDC∗ sr FVA∗ sr FDC∗ sr (D.3) The measures of “domestic value-added” and “foreign value-added” in (D.3) correspond to those proposed by Johnson (2018) in a two-country context; the same measure of “domestic value-added” in bilateral exports is also obtained by Los et al. (2016) by using hypothetical extraction. Similarly to the derivation of equations (C.6) to (C.8), equation (D.3) can be further developed to consider all the forward production linkages, as well as the countries of completion and the markets of final absorption. Exporting-country perspective for bilateral trade flows The methodology presented above provides a correct measure of the whole value- added that crosses a specific bilateral border, but these indicators cannot be summed across bilateral destinations to get the correct aggregate measure, i.e. they are not additive. Conversely, to obtain a consistent breakdown across bilateral flows, the “exporting-country perspective” has to be applied also to the decomposition of disaggregated trade flows. However, in this case an approach to allocate value-added and double counted items across the different disaggregated trade flows is needed. In order to address this issue, we exploit two alternative approaches proposed by Nagengast and Stehrer (2016) and fully derived by Borin and Mancini (2015, 2017). The “source-based” approach, in which a given item is accounted for as “value- added” the first time it leaves the country of origin and, in the case of multiple crossing, it is considered “double counted” in subsequent shipments; this definition is in line with the logic behind the accounting procedure presented in equations 46 (C.2)-(C.5) (C.2)-(C.5).24 The “sink-based” approach, in which a given item is accounted for as “value-added” the last time it leaves the country of origin and, in the case of multiple crossing, it is considered “double counted” in prior shipments. Suppose, for instance, that along the production process a certain item is exported by country A first to country B and then to country C. With the “source-based” approach, the item is classified as “value-added” the first time it crosses the national border (i.e. in the exports toward B), whereas the “sink-based” one allows to consider it as “value-added” the last time it crosses the border (i.e. in the exports toward C). As already mentioned, the “source” approach is useful to separate traditional exports - crossing one border - from GVC-exports - crossing more than one bor- der. Instead, the “sink” approach is more suited for the analysis of bilateral trade balances in terms of value-added. The value-added decomposition of the exports form country s to country r ac- cording to a exporting country perspective/source-based approach can be obtained simply substituting the total exports of s in equation (C.5) with the considered bilateral trade flow (i.e. Esr ): domestic foreign content (DCsr ) content (FCsr ) G G s s s / s / uN Esr = Vs B/ ss Esr + Vs B/ ss Asj Bjs Esr + Vt Bts Esr + Vt Bts Asj Bjs Esr . j =s t=s t=s j =s domestic value domestic double foreign value foreign double added counted added counted (DVAsourcesr ) (DDCsourcesr ) (FVAsourcesr ) (FDCsourcesr ) (D.4) As to the value-added decomposition of the exports form country s to country r according to a exporting country perspective/sink-based approach, it is first nec- essary to isolate the portion of ultimate shipments within a certain bilateral trade s / −Y ) (→ flow. These “ultimate exports” (Esr ∗ ) are made up of final goods (Ysr ) and of intermediate goods that do not re-enter country s’s exports, before reaching the s / − Y∗ ) (→ ultimate destination (Asr Xr ). Since the latter are commensurate with final goods as concerns the exporting country s, the overall value-added can be computed 24 There are also other ways to single out value-added in aggregate exports. See Koopman et al. (2014), Borin and Mancini (2015), Los et al. (2016), Miroudot and Ye (2017), Arto et al. (2019) for other accounting procedures yielding the same results, at least for the domestic value-added in the total exports of a country. As regard the foreign value-added in total exports, the differences between the various contributions are mainly due to the specific accounting perspective used for this component (the point is discussed in detail by Borin and Mancini, 2019). 47 by pre-multiplying the vector of “ultimate exports” by the VB matrix (i.e. in the same way as how the VBY matrix is used to measure the total value-added in final demand in Appendix B). In order to single out the value-added and double-counted components in the exports of intermediates of country s according to a exporting country perspective/sink-based approach, we can use the same algebraic device pre- sented in equation (C.4) that allows to distinguish between the items re-exported by s and those that are not:25 G G s / s s Asr Xr = Asr Brk Ykl + B/ rs Yss +Asr B/ rs Es∗ . (D.5) k =s l s / Xr − Y∗ ) → ( Then, the value-added breakdown of bilateral exports in a “exporting country perspective”/“sink-based” framework can be expressed as follows: domestic content (DCsr ) G G s / s s uN Esr = Vs Bss Ysr + Asr Brk Ykl + B/ rs Yss + Vs Bss Asr B/ rs Es∗ k =s l domestic value domestic double added (DVAsinksr ) counted (DDCsinksr ) foreign content (FCsr ) G G G G s / s s + Vt Bts Ysr + Asr Brk Ykl + B/ rs Yss + Vt Bts Asr B/ rs Es∗ . t=s k =s l t=s foreign value foreign double added (FVAsinksr ) counted (FDCsinksr ) (D.6) As highlighted above, the three different value-added decompositions of bilat- eral trade flows in equations (D.3), (D.4) and (D.6) can be used to address different issues, for instance the analysis of the exposure to tariffs imposed by the bilateral partner, the analysis of GVC-exports and that of bilateral trade balances, respec- 25 A simple way to figure out how to decompose the exports of intermediates to country r is to re-express the general relationship of production and trade in our global I-O setting (see equation s B.1) by separating the export flows from country s as follows: X = A/ X + As X + Y / s + Ys , s s / s / where A = (A − A ), Y is the final demand matrix Y with the block matrix corresponding to exports of final goods from s equal to 0 (but including domestic final demand Yss ), and Ys s is simply equal to (Y − Y/ ). This expression can be simplified by taking into account that the sum of A X and Y is a GN ×N matrix with the total exports from country s (i.e. Es∗ ) in the s s corresponding block submatrix and zeros elsewhere. 48 tively. Nevertheless, it is important to highlight that: i) at the bilateral level, the domestic and foreign contents are the same in the three breakdowns, only the value-added and double-counted components differ; ii) the value-added and double- counted terms of the two decompositions based on the exporting-country perspec- tive (i.e. the source-based in equation (D.4) and the sink-based in equation D.6) differ only at the bilateral level and when summing across the destinations of a given exporter we obtain exactly the same aggregate indicators as those in equation (C.5).26 E Conceptual framework: Value-added account- ing in different types of trade flows and ac- counting perspectives Here we provide an overview of all the trade flows that can be analyzed with icio, with the corresponding perspectives that are available. 1. Total exports of a country a) Total aggregate exports - Exporting-country perspective: both the logic and the algebraic for- mulation of this accounting perspective are presented in Appendix C. - World perspective: This perspective has been considered only for the decomposition of the foreign content of exports. According to this methodology a certain item is accounted for as foreign value-added only once in all (i.e. world) trade flows, whereas in the export- ing country perspective it occurs only once in all the exports of a single country. More specifically, by using a source- (sink-) based approach, a certain item is considered as value-added only the first (the last) time it crosses a foreign border whereas, all the other times it does, it is classified as double counted. The decompositions based on a “world perspective” can be used to address interesting ques- tions regarding the breakdown of total world trade (i.e. by aggre- gating across countries the total exports’ decompositions obtained with icio). For instance, we can measure the share of world’s GDP 26 See Borin and Mancini (2019) for a formal proof. 49 entering the exports of some other country. However, these mea- sures are usually unsuited to addressing relevant issues regarding a country’s exports. 27 b) Total sectoral exports - Sectoral-exporter perspective: This method, formally derived in Borin and Mancini (2019), can be chosen when the aim of the analysis is to compute the entire value-added that is embedded in all the exports of a country in a given sector. This occurs, for instance, when an economic shock (or policy intervention) affects all the exports of a country in a given sector (across all the destinations), and the inter- est of the analyst is to measure the spillovers from this shock into different countries/sectors. The domestic and foreign value-added embedded in total exports of country s and sector n can be com- puted similarly as in the decomposition of bilateral flows according to the “bilateral perspective” (see equation D.3). The only differ- ence is that the original matrix of technical coefficients A needs to be modified such that asj,n is set to zero ∀ j = s; thus the inverse Leontief matrix is computed accordingly. - Exporting country perspective: This accounting method for the anal- ysis of sectoral trade flows follows exactly the logic of the same per- spective described above for the analysis of bilateral trade flows. It provides a breakdown of sectoral exports consistent with the value- added indicators computed for the total aggregate exports of a coun- try. Depending on whether the focus of the analysis is on the origin of the production or on the final absorption, a source- or a sink-based approach needs to be considered, respectively. The algebraic expres- sions follow closely the formulas in equations (C.5) (or eq. D.4) and (D.6), where total sectoral export flows are singled out through s proper diagonalizations of the VB and the VB/ matrices.28 2. Bilateral exports of a country 27 Borin and Mancini (2019) provide a more detailed discussion on this point, as well as the algebraic expressions for source- and sink-based breakdowns of the foreign content of exports based on a world perspective. The source-based decomposition corresponds to the one proposed by Borin and Mancini (2017) and Miroudot and Ye (2017). The sink-based one is similar to that reported in Koopman et al. (2014), however this part of their decomposition is affected by some drawbacks (see Borin and Mancini, 2019 for details). 28 See Borin and Mancini (2019) for more details. 50 a) Bilateral aggregate exports: both the logic and the algebraic formu- lation of these accounting perspectives are presented in Appendix D. - Bilateral perspective - Exporting-country perspective b) Bilateral sectoral exports: - Sectoral-bilateral perspective: This methodology, developed in Borin and Mancini (2019), is useful for empirical analysis aiming at mea- suring the whole value-added of a country entering in the exports of country s in a specific sector (say n)to an importing country r. It can be used, for instance, to evaluate the GDP exposure to a tariff imposed by a country vis-` a-vis a certain partner in a specific sec- tor. As for the previous decompositions, for which the perspective corresponds to the trade flow under investigation, the value-added indicators are derived by modifying the input requirement matrix A, setting to zero all the coefficients corresponding to the intermediate exports from s to r in (exporting) sector n. - Exporting country perspective: This can be used to obtain a break- down of total exports’ value-added indicators across sectoral-bilateral flows. The formulation is a direct extension of that used for total sectoral exports to bilateral trade flows. 3. Total imports of a country a) Total aggregate imports: - Importing country perspective: This methodology can be exploited to compute the GDP of a given country j that enters, directly or indirectly, in the total imports of a given country r. This measure can be interesting, for instance, when a certain country is going to a-vis all the exporting adopt a general protectionist stance (i.e. vis-` partners) and we want to compute the portion of the other countries’ GDP at stake. In this case we define the relevant perimeter at the level of the importing country’s borders as a whole. This can be implemented by following a procedure similar to that used to derive the (exporting) “country perspective” of Appendix C and is formally derived in Borin and Mancini (2019). b) Total sectoral imports: 51 - Sectoral-importer perspective: This perspective, derived in Borin and Mancini (2019), can be useful when the focus is on a particular sector of a given importing country, e.g. when a certain shock affects only the imports of a country in a specific sector. The derivation is similar to that of the importing country perspective, where the double-counting perimeter is defined at the level of sectoral imports of a country. F Conceptual framework: GVC participation The “traditional” exports of country s to country r can be defined as the production of s that is directly absorbed in r without any further re-export. This component, called DAVAX - i.e. Directly Absorbed Value-Added in exports - can be computed as: DAVAXsr = Vs (I − Ass )−1 Ysr + Vs (I − Ass )−1 Asr (I − Arr )−1 Yrr . Then GVC-related exports can be simply obtained by excluding the entire do- mestic value-added of country s absorbed directly by its direct importer (DAVAXsr ) from its exports to r: GVCXsr = uN Esr − DAVAXsr . Therefore, GVC-related trade share in total exports is given by: GVCXsr GVCsr = , uN Esr where uN Esr is the total exports of country s to country r. For the total exports of country s the GVC share will be easily computed as G r =sGVCXsr GVCs = , (F.1) uN Es∗ while at world level we have: G G s r=s GVCXsr GVCworld = G . (F.2) s (uN Es∗ ) As already mentioned, the overall GVC indicator of equation (F.1) can be de- composed into a ‘backward’ component, corresponding to the VS Index proposed 52 by Hummels et al. (2001) (see Borin and Mancini, 2019 for a formal proof) and a “forward” component, i.e. the part of domestic production that is supplied to the importing country to be re-exported: GVCsr = GVCbackwardsr + GVCf orwardsr (F.3) where G G Vs (I − Ass )−1 Asj Bjs Esr + Vt Bts Esr j =s t=s GVCbackwardsr = (F.4) uN Esr and G G G G Vs (I − Ass )−1 Asr (I − Arr )−1 ( Yrj + Arj Bjk Ykl ) j =r j =r k l GVCf orwardsr = . uN Esr (F.5) The GVCf orwardsr indicator differs from the VS1s index proposed by Koop- man et al. (2014). VS1s is computed by aggregating the content of a country’s production embedded in other countries’ exports and thus it is not necessarily a portion of country s’s exports (like VS). Suppose, for instance, that a certain inter- mediate component exported by country s later undergoes other processing phases in different countries; the original component will be double-counted several times in the summation of country s’s content in other countries’ exports. The discrepancy between the original value of goods exported by s and the related amount that en- ters in Koopman et al.’s (2014) indicator increases with the relative “upstreamness” of country s’s production. 53