The Innovation and Structural Transformation Database: A Guide

Foster-McGregor, Neil; Nomaler, Önder; Verspagen, Bart Working Paper The Innovation and Structural Transformation Database: A Guide UNU-MERIT Working Papers, No. 2024-027 Provided in Cooperation with: Maastricht Economic and Social Research Institute on Innovation and Technology (UNU-MERIT), United Nations University (UNU) Suggested Citation: Foster-McGregor, Neil; Nomaler, Önder; Verspagen, Bart (2024) : The Innovation and Structural Transformation Database: A Guide, UNU-MERIT Working Papers, No. 2024-027, United Nations University (UNU), Maastricht Economic and Social Research Institute on Innovation and Technology (UNU-MERIT), Maastricht This Version is available at: https://hdl.handle.net/10419/326923 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-sa/4.0/ #2024-027 The Innovation and Structural Transformation Database: A Guide Neil Foster-McGregor, Önder Nomaler & Bart Verspagen Published 22 October 2024 Maastricht Economic and social Research institute on Innovation and Technology (UNU-MERIT) email: [email protected] | website: http://www.merit.unu.edu Boschstraat 24, 6211 AX Maastricht, The Netherlands Tel: (31) (43) 388 44 00 UNU-MERIT Working Papers ISSN 1871-9872 Maastricht Economic and social Research Institute on Innovation and Technology UNU-MERIT UNU-MERIT Working Papers intend to disseminate preliminary results of research carried out at UNU-MERIT to stimulate discussion on the issues raised. 1 The Innovation and Structural Transformation Database: A Guide October 2024 Neil Foster-McGregor, Önder Nomaler & Bart Verspagen Abstract This document summarizes the content of the Innovation and Structural Transformation database provides a collection of indicators for the global economy, emphasizing the dynamic nature of development and economic growth. The database is publicly available from the web (see link in the main text). This document summarizes the content of the database, further providing information on the data sources and the construction of the variables included in the database. Keywords: Structural change, innovation, economic development, economic growth. JEL Codes: O14, O30, O40, O50, O53 The Innovation and Structural Transformation Database and this guide are provided under a Creative Commons BY 4.0 license. This license allows users to distribute, remix, adapt, and build upon the material in any medium or format, so long as attribution is given to the creator. The license allows for commercial use. 2 1. Introduction The Innovation and Structural Transformation database provides a collection of indicators for the global economy, emphasizing the dynamic nature of development and economic growth. This document summarizes the content of the database, further providing information on the data sources and the construction of the variables included in the database. The database is publicly available from the web at https://dataverse.nl/dataverse/innovation_and_structural_transformation_database/ Any future updates of the database will be available at the same link. The database and this document are provided under a CC BY 4.0 license. This means that the database can be used freely, as long as appropriate credits are given.1 This credit can be given by citing the forthcoming book that illustrates the use of the database: Felipe, J., Foster-McGregor, N., Nomaler, Ö and B. Verspagen, 2025, Innovation and Structural Transformation in Asia, Abingdon (UK): Taylor & Francis. The aim of the database is to provide a resource for researchers, policymakers and others that provides information on a set of variables and indicators that are fundamental to understand longrun growth and development. These indicators revolve around the topic of structural transformation, development, and international competitiveness. Underlying the database is the premise that development is a process of structural transformation, i.e., that increasing the standard of living in an economy involves changing what the economy produces and consumes and involves changes in the way actors in the economy interact. Traditionally, this process of structural transformation was considered to involve a movement of production (and consumption) across broad sectors (e.g., from agriculture to industry to services). With the availability of new data, however, a more nuanced perspective has emerged. One dimension of this is the observation that much structural transformation is occurring within sectors, through for example the changing structure of occupations and functions within sectors, as well as changes in the composition of production and activities within sectors. Related to these changes have been developments of the global structure of production, with structural transformation depending to a great extent on the way the economy interacts with foreign markets in today’s global economy. The position of an economy in the international structure of production depends on the competitiveness of firms and other actors in the economy. In this context, competitiveness, or comparative advantage, determines where economies, or rather 1 For details of this license see https://creativecommons.org/licenses/by/4.0/. 3 the firms located in economies, can contribute value added in global production networks, which in turn will help determine overall levels of well-being in the economy. The competitiveness, or comparative advantage, of an economy is further a function of the capabilities that exist in the economy. Development can further be seen as a process of developing competitiveness in a broader range of products (i.e., diversification), including a move into more complex products within and across sectors (i.e., upgrading). An important dimension of an economy’s capabilities relates to its performance in innovation. In this context, there has been a huge amount of interest in recent years in the development of digital technologies and how innovation in these technologies is changing the nature of production and work. Based upon these arguments, the database is formed around four main pillars. The first pillar covers data on structural change and productivity, presenting indicators that document how fast, and into which direction the structure of the economy changes and the resulting impact of these changes on productivity growth. The second pillar focuses on competitiveness and comparative advantage, which is approached through the notion of product complexity, and which in turn can be seen as a measure of development potential. This pillar focuses on developing a detailed picture of the socalled product space and how the economies in the database occupy this space. The approach involves a large data set of exported products and shows how economies are specialized in these different products. The third pillar of the database is concerned with innovation, with the database making use of patent data to identify the extent of innovation as well as the direction of that innovation, further providing an overall picture of patented inventions and how they relate to the economy, as well as an overview of trends in technologies related to the so-called fourth industrial revolution. The final pillar considers the global economy and the role of individual economies in the global structure of production through an examination of so-called Global Value Chains (GVCs), which highlight that production is fragmented geographically and firms may contribute (small) parts of total value added to a variety of GVCs. The data for this final pillar is heavily based on global input-output tables. The remainder of this document describes in turn the indicators that fall under each of the four pillars, along with the data sources used to construct them. 2. Pillar 1: Structural Change 2.1. Basic structural change indicators The basic indicators for structural change consider the sectoral composition of the economy, as well as the structure of international trade (exports and imports). For the sectoral structure of the economy, the database considers the structure of value added, household consumption demand and total demand, all in constant 2010 US$. These data are drawn from the ADB input output tables. The following variables are defined: 4 𝜎𝜎𝑡𝑡𝑡𝑡𝑡𝑡 𝑉𝑉𝑉𝑉=𝑉𝑉𝐴𝐴𝑡𝑡𝑡𝑡𝑡𝑡 ∑𝑉𝑉𝐴𝐴𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝜎𝜎𝑡𝑡𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻=𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡𝑡𝑡 ∑𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝜎𝜎𝑡𝑡𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇=𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡𝑡𝑡 ∑𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 where the indices 𝑖𝑖, 𝑗𝑗, and 𝑡𝑡 denote economy, sector and time period, respectively, 𝑉𝑉𝐴𝐴 is value added produced in the sector/economy, 𝐻𝐻𝐻𝐻 is final consumption by households in economy 𝑖𝑖 of products produced by sector 𝑗𝑗 (either imported or produced domestically), and TD is the sum of household consumption, government consumption and gross fixed capital formation by residents of economy 𝑖𝑖 aimed at products produced by sector j (again domestically or imported). These shares are used to produce an indicator of structural change, which is termed the 𝑁𝑁𝐴𝐴𝑉𝑉 indicator (Norm of Absolute Value changes). This is defined as follows: 𝑁𝑁𝐴𝐴𝑉𝑉𝑡𝑡𝐾𝐾=∑�𝜎𝜎𝑇𝑇𝑡𝑡𝑡𝑡 𝐾𝐾−𝜎𝜎0𝑡𝑡𝑡𝑡 𝐾𝐾� 𝑡𝑡2𝑇𝑇 where 𝐾𝐾∈{𝑉𝑉𝐴𝐴,𝐻𝐻𝐻𝐻,𝑇𝑇𝑇𝑇} is an index that refers to the particular indicator of structure being considered and T is the number of years considered. The greater the degree of structural change between period 0 and 𝑇𝑇, the higher the value of the 𝑁𝑁𝐴𝐴𝑉𝑉 indicator. Figure 1 shows the sectoral shares of total demand (𝑇𝑇𝑇𝑇) and the associated 𝑁𝑁𝐴𝐴𝑉𝑉 indicator for Nepal.2 Note that the database does not have data for the years 2001-2006, so the period between 2000 and 2007 is interpolated. We see that the recent period 2015-19 is one of relatively rapid structural change in total demand in Nepal. 2 This figure, as those below, is copied from the graphical interface on the website. In this interface, users can select specific (combinations of) indicators, economies and sectors to be displayed. These figures are often linked to each other, e.g., clicking/selecting an observation in one figure provides more detailed information for this observation in a related figure. All figures can be saved and used in proprietary documents. 5 Figure 1. Structural change in total demand – the case of Nepal The sectors that are available in this pillar of the database are identical to the sectors used in the GVC pillar, with a selection of these sectors also used in other sections of the database. These sectors are documented in Table 1. 6 Table 1. Sectors used in basic structural change and GVC sections Sec num ISIC Rev. 4 Description 1 A Agriculture, hunting, forestry, and fishing 2 B Mining and quarrying 3 C10-C12 Food, beverages, and tobacco 4 C13, C14 Textiles and textile products 5 C15 Leather, leather products, and footwear 6 C16 Wood and products of wood and cork 7 C17, C18, J58 Pulp, paper, paper products, printing, and publishing 8 C19 Coke, refined petroleum, and nuclear fuel 9 C20, C21 Chemicals and chemical products 10 C22 Rubber and plastics 11 C23 Other nonmetallic minerals 12 C24, C25 Basic metals and fabricated metal 13 C28, C33 Machinery, nec 14 C26, C27 Electrical and optical equipment 15 C29, C30 Transport equipment 16 C31, C32 Manufacturing, nec; recycling 17 D, E36 Electricity, gas, and water supply 18 F Construction 19 G45 Sale, maintenance, and repair of motor vehicles and motorcycles; retail sale of fuel 20 G46 Wholesale trade and commission trade, except of motor vehicles and motorcycles 21 G47 Retail trade, except of motor vehicles and motorcycles; repair of household goods 22 I Hotels and restaurants 23 H49 Inland transport 24 H50 Water transport 25 H51 Air transport 26 H52, N79 Other supporting and auxiliary transport activities; activities of travel agencies 27 H53, J61 Post and telecommunications 28 K Financial intermediation 29 L Real estate activities 30 J58-J60, J62-J63, M, N77N78, N80-N82 Renting of M&Eq and other business activities 31 O Public administration and defense; compulsory social security 32 P Education 33 Q Health and social work 34 E37-E39, R, S Other community, social, and personal services 35 T Private households with employed persons Turning to indicators of the structure of international trade (both exports and imports), the database includes two groups of indicators. The first is related to global value chains (GVCs), while the other refers to innovation, and in particular the group of technologies that is referred to as 4th industrial revolution (4IR) technologies. The indicators on GVCs distinguish between different types of products, namely consumption goods, intermediate goods and capital goods. According to this distinction, consumption goods and capital goods are considered as the outputs of a value 13 In the case of employment data, it is very difficult to obtain actual data for many (most) of the economies in the database. To overcome this constraint, therefore, estimated data from the International Labor Organization was used. This data is presented in their online ILOSTAT database6, specifically the section entitled ILOEST (ILO modelled estimates and projections). The data used are from the table called Employment by sex and economic activity, which presents employment in 1,000 persons at the ISIC Rev4 sectoral breakdown. Value added at the sectoral level is drawn from a database on national accounts held at the UNDESA Statistical Division.7 Data on value added by economic activity provided by this database is used, with the data measured in constant 2015 prices and US dollars. The database uses market exchange rates, i.e., it makes no attempt to take into account real exchange rates that “correct” for price differentials between economies. Although using real exchange rates (or purchasing power parity exchange rates, PPP) is customary for comparing living standards based on aggregate GDP, correcting for price differentials at the sectoral level is impossible due to a lack of data. Value added data are expressed in billions of US$. The data on value added are collated according to the ISIC Rev. 3 classification, whereas the employment data are reported according to the ISIC Rev. 4 classification. At the aggregation level used in the construction of the database (i.e., the seven sectors), these sectors are comparable across the two revisions however. Table 4 documents the seven sectors. Table 4. Sectoral breakdown in the structural change and productivity growth section of the database Sector number Description ISIC codes Rev. 4 Rev. 3 1 Agriculture; forestry and fishing A A, B 2 Mining, Utilities B, D, E C, E 3 Manufacturing C D 4 Construction F F 5 Trade; Hotels and restaurants G, I G, H 6 Transport, Storage, Communication H, J I 7 All other activities K – U J – P These data are used to construct several indicators. The first is the NAV indicator that was already defined above for the case of value-added. In this case it is calculated using employment shares and defined as: 6 https://ilostat.ilo.org/data/ 7 https://unstats.un.org/unsd/snaama/Downloads 14 𝑁𝑁𝐴𝐴𝑉𝑉𝑡𝑡=∑�𝑆𝑆𝑇𝑇𝑡𝑡𝑡𝑡−𝑆𝑆0𝑡𝑡𝑡𝑡� 𝑡𝑡2𝑇𝑇 where 𝑖𝑖 refers to economies and 𝑆𝑆𝑡𝑡𝑡𝑡𝑡𝑡 is the share of sector 𝑗𝑗 in total employment in period 𝑡𝑡 in economy 𝑖𝑖. Labor productivity is the main focus of the other indicators in this second set of indicators. Labor productivity is defined at the sectoral as well as the aggregate level and is calculated as value added (GDP) divided by employment. We provide the trend for both overall labor productivity, as well as the two components, value added (GDP) and employment, with the trend defined as a series of index numbers with base year 2015 (2015 = 100). This index is constructed simply as 100 times the value in a particular year divided by the 2015 value. The indicators for the impact of structural change on labor productivity growth are based on a decomposition of labor productivity growth. This can be represented as follows (we drop the economy subscript i): 𝑃𝑃�𝑇𝑇=𝑃𝑃𝑇𝑇−𝑃𝑃0 𝑇𝑇𝑃𝑃0=∑𝑆𝑆𝑇𝑇𝑡𝑡𝑃𝑃𝑇𝑇𝑡𝑡𝑡𝑡−∑𝑆𝑆0𝑡𝑡𝑃𝑃0𝑡𝑡𝑡𝑡 𝑇𝑇∑𝑆𝑆0𝑡𝑡𝑃𝑃0𝑡𝑡𝑡𝑡 = ∑𝑃𝑃0𝑡𝑡�𝑆𝑆𝑇𝑇𝑡𝑡−𝑆𝑆0𝑡𝑡� 𝑡𝑡𝑇𝑇𝑃𝑃0+∑�𝑃𝑃𝑇𝑇𝑡𝑡−𝑃𝑃0𝑡𝑡��𝑆𝑆𝑇𝑇𝑡𝑡−𝑆𝑆0𝑡𝑡� 𝑡𝑡𝑇𝑇𝑃𝑃0+∑𝑆𝑆0𝑡𝑡�𝑃𝑃𝑇𝑇𝑡𝑡−𝑃𝑃0𝑡𝑡� 𝑡𝑡𝑇𝑇𝑃𝑃0 SSRE DSRE WS where 𝑃𝑃𝑇𝑇 and 𝑃𝑃0 are aggregate (labor) productivity in year T and 0, respectively, and 𝑃𝑃𝑡𝑡𝑡𝑡 is labor productivity in period 𝑡𝑡 and sector 𝑗𝑗. The three terms into which total productivity growth (between years T and 0) is decomposed are on the righthand side of this expression, and are labelled SSRE for Static Structural Reallocation Effect, DSRE for Dynamic Structural Reallocation Effect and WS for Within Sector. SSRE measures the effect on productivity growth of reallocating resources (i.e., labor) to sectors with higher productivity levels at the start of the period (year 0), DSRE measures the effect of reallocating resources to sectors with a faster growth rate of productivity between years T and 0, and WS measures the contribution of productivity growth that occurs within sectors. Figure 4 shows a set of diagrams that illustrate the structural change indicators for Indonesia. The top part of the diagram shows the value of the NAV indicator for various periods, which are displayed on the horizontal axis. We see that structural change in Indonesia was strongest during the 1991-1995 period, and weakest during the 2001-05 period. Note that not all periods on the horizontal axis are equal in length, but the NAV indicator is scaled to yearly changes, i.e., it is comparable between periods of unequal length. 15 The bottom part of the figure shows the productivity decompositions, again for different periods on the horizontal axis (where again the effects are scaled for differences in period length). Here the black line indicates total labor productivity growth over the different time periods. This growth rate is equal to the sum of the stacked bars, which represent the three effects introduced above. The figure reveals that labor productivity growth in Indonesia was negative during the period 19962000, which encompasses the “Asian crisis”. The WS effect dominates in most periods, while the DSRE is always negative. The SSRE peaks in the period just before the Asian crisis (1991-95). Figure 4. Structural change and labor productivity growth in Indonesia Figure 5 shows a further feature of the website, with the figure reporting a breakdown of the three effects in the decomposition (WS, SSRE and DSRE) into their components at the sectoral level. This uses the following definitions (we omit the time scaling variable 𝑇𝑇): 𝑀𝑀ℎ𝑖𝑖𝑐𝑐𝑀𝑀𝑖𝑖 𝑖𝑖𝑐𝑐 𝑐𝑐𝑖𝑖𝑅𝑅𝑐𝑐𝑖𝑖 𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐𝑀𝑀𝑡𝑡𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑝𝑝𝑡𝑡= 𝑃𝑃𝑇𝑇𝑡𝑡−𝑃𝑃0𝑡𝑡 𝑃𝑃0 𝑖𝑖𝑐𝑐𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖𝑐𝑐 𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑝𝑝𝑐𝑐𝑖𝑖𝑐𝑐𝑡𝑡 𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑡𝑡= 𝑆𝑆0𝑡𝑡 𝑀𝑀ℎ𝑖𝑖𝑐𝑐𝑀𝑀𝑖𝑖 𝑖𝑖𝑐𝑐 𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑝𝑝𝑐𝑐𝑖𝑖𝑐𝑐𝑡𝑡 𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑡𝑡=𝑆𝑆𝑇𝑇𝑡𝑡−𝑆𝑆0𝑡𝑡 16 𝑖𝑖𝑐𝑐𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖𝑐𝑐 𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑡𝑡𝑖𝑖𝑀𝑀𝑖𝑖 𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐𝑀𝑀𝑡𝑡𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑝𝑝𝑡𝑡=𝑃𝑃0𝑡𝑡 𝑃𝑃0 It is easily seen that these four components can be used to construct the three effects in the decomposition: • SSRE is the sum across sectors of 𝑖𝑖𝑐𝑐𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖𝑐𝑐 𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑡𝑡𝑖𝑖𝑀𝑀𝑖𝑖 𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐𝑀𝑀𝑡𝑡𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑝𝑝𝑡𝑡× 𝑀𝑀ℎ𝑖𝑖𝑐𝑐𝑀𝑀𝑖𝑖 𝑐𝑐𝑠𝑠 𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑝𝑝𝑐𝑐𝑖𝑖𝑐𝑐𝑡𝑡 𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑡𝑡 • DSRE is the sum across sectors of 𝑀𝑀ℎ𝑖𝑖𝑐𝑐𝑀𝑀𝑖𝑖 𝑐𝑐𝑠𝑠 𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐𝑀𝑀𝑡𝑡𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑝𝑝𝑡𝑡× 𝑀𝑀ℎ𝑖𝑖𝑐𝑐𝑀𝑀𝑖𝑖 𝑐𝑐𝑠𝑠 𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑝𝑝𝑐𝑐𝑖𝑖𝑐𝑐𝑡𝑡 𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑡𝑡 • WS effect is the sum across sectors of 𝑖𝑖𝑐𝑐𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖𝑐𝑐 𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑝𝑝𝑐𝑐𝑖𝑖𝑐𝑐𝑡𝑡 𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑡𝑡× 𝑀𝑀ℎ𝑖𝑖𝑐𝑐𝑀𝑀𝑖𝑖 𝑐𝑐𝑠𝑠 𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐𝑀𝑀𝑡𝑡𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑝𝑝𝑡𝑡 Using the example of Indonesia, Figure 5 plots for each of the three decomposition terms the two constituent parts of the decomposition term for each sector against each other. Effects that are positive are plotted in blue and negative contributions are plotted in red, while the size of the bubbles indicates the magnitude of the absolute value of the effect. In the case of Indonesia, and focusing on the period 1991-2000, it can be seen that the Mining & utilities sector makes a strong contribution to the WS effect, which arises largely due to rapid productivity change, despite the relatively small employment share. Similar conclusions can be drawn for the other two effects, DSRE and SSRE. 17 Figure 5. Sector labor productivity decompositions view in the database 18 The database further reports relative productivity levels for pairs of economies. An example of this is provided in Figure 6 for Japan and India. These relative productivity indicators per sector are calculated as follows: 𝑅𝑅𝑖𝑖𝑐𝑐𝑃𝑃𝑡𝑡𝑘𝑘𝑡𝑡𝑡𝑡=𝑃𝑃𝑡𝑡𝑡𝑡𝑡𝑡𝑋𝑋𝑅𝑅𝑡𝑡 𝑃𝑃𝑘𝑘𝑡𝑡𝑡𝑡𝑋𝑋𝑅𝑅𝑘𝑘 where 𝑃𝑃 is labor productivity as before (measured in 2015 US$ per worker), 𝑋𝑋𝑅𝑅 is a real exchange rate factor, and the subscripts 𝑘𝑘, 𝑐𝑐, and 𝑗𝑗 denote two countries and a sector respectively (𝑗𝑗 can also denote the total economy). The real exchange rate factor XR is the ratio between the World Bank’s PPP exchange rate for international $ and the market exchange rate for US$, both in 2015 (the base year for the price index used in the productivity calculations). Note that XR converts the productivity data into units that reflect domestic purchasing power. In the graph we see that relative productivity in Japan compared to India is gradually falling over the period 1991-2019. Figure 6. relative labor productivity view in the database Table 5 documents the variables that are available in the structural change part of the database. 19 Table 5. Indicators in structural change and productivity section Index (2015=100) of employment, by sector and for total economy Index (2015=100) of value added (GDP), by sector and for total economy Index (2015=100) of labor productivity, by sector and for total economy Sectoral shares of employment Sectoral shares of value added (GDP) Aggregate labor productivity growth for various (sub)periods Within Sector (WS) effect in aggregate labor productivity growth for various (sub)periods Static Structural Reallocation Effect (SSRE) in aggregate labor productivity growth for various (sub)periods Dynamic Structural Reallocation Effect (DSRE) in aggregate labor productivity growth for various (sub)periods 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄 𝒐𝒐𝒐𝒐 𝒑𝒑𝒑𝒑𝒐𝒐𝒑𝒑𝒑𝒑𝒄𝒄𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑 𝒋𝒋 contribution to the decomposition effect in the sector 𝒑𝒑𝒄𝒄𝒑𝒑𝒑𝒑𝒑𝒑𝒄𝒄𝒊𝒊 𝒄𝒄𝒆𝒆𝒑𝒑𝒊𝒊𝒐𝒐𝒑𝒑𝒆𝒆𝒄𝒄𝒄𝒄𝒑𝒑 𝒔𝒔𝒄𝒄𝒄𝒄𝒑𝒑𝒄𝒄 𝒋𝒋 contribution to the decomposition effect in the sector 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄 𝒐𝒐𝒐𝒐 𝒄𝒄𝒆𝒆𝒑𝒑𝒊𝒊𝒐𝒐𝒑𝒑𝒆𝒆𝒄𝒄𝒄𝒄𝒑𝒑 𝒔𝒔𝒄𝒄𝒄𝒄𝒑𝒑𝒄𝒄 𝒋𝒋 contribution to the decomposition effect in the sector 𝒑𝒑𝒄𝒄𝒑𝒑𝒑𝒑𝒑𝒑𝒄𝒄𝒊𝒊 𝒑𝒑𝒄𝒄𝒊𝒊𝒄𝒄𝒑𝒑𝒑𝒑𝒑𝒑𝒄𝒄 𝒑𝒑𝒑𝒑𝒐𝒐𝒑𝒑𝒑𝒑𝒄𝒄𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑 𝒋𝒋 contribution to the decomposition effect in the sector 3. Pillar 2: Product Complexity The pillar on product complexity builds on the idea that the performance of economies in international trade is determined by productive capabilities, and that detailed data on exports can therefore provide an indication of the capabilities that are present in the productive structure of an economy. The starting point for the construction of most of the indicators in this pillar is information on whether two (exported) products are jointly produced with comparative advantage by a single economy, information that is taken as an indication that the production of these two products share certain capabilities. The indicators in this group are based on detailed product level export data (with information on more than 5,000 products). Using this data an established method from the existing literature, known as the ECI (Economic Complexity Index) method, is adopted to calculate an indicator of product quality from the product-level export data (Hidalgo and Hausmann, 2009; Hidalgo, 2021). This indicator of product quality is referred to as product complexity in the database. Conceptually, product complexity is an indicator of the production capabilities that are needed to sell the product with comparative advantage in international markets. Thus, each of the 5,000+ products in the database has its own value of product complexity, with the complexity of the products that an economy exports with comparative advantage further being an indicator of the capabilities of firms located in a particular economy. The formal definition of the product complexity indicator is provided in the annex. 20 3.1. Basic product complexity indicators and upgrading In this part of the database, a number of basic indicators of product complexity are presented, along with information on the way in which complexity is distributed over economies and how this distribution affects the upgrading potential of economies. The basic idea behind the upgrading potential is that it is generally advantageous to move into products with higher complexity because these markets are generally more exclusive and require higher capabilities, and hence offer a higher potential for creating value added. Product complexity indicators are based on the notion of comparative advantage, which is calculated as follows 𝑀𝑀𝑡𝑡𝑡𝑡= 1 𝑖𝑖𝑠𝑠 𝑚𝑚𝑡𝑡𝑡𝑡∑𝑚𝑚𝑞𝑞𝑡𝑡𝑞𝑞 � ∑𝑚𝑚𝑡𝑡𝑖𝑖𝑖𝑖∑∑𝑚𝑚𝑞𝑞𝑖𝑖𝑖𝑖𝑞𝑞 �> 1 𝑖𝑖𝑐𝑐𝑖𝑖 0 𝑐𝑐𝑡𝑡ℎ𝑖𝑖𝑖𝑖𝑤𝑤𝑖𝑖𝑐𝑐𝑖𝑖 where 𝑀𝑀𝑡𝑡𝑡𝑡 is the (binary) comparative advantage of economy 𝑖𝑖 in product 𝑗𝑗, and 𝑚𝑚𝑡𝑡𝑡𝑡 is the value of exports of economy 𝑖𝑖 in product 𝑗𝑗. In words, the indicator of comparative advantage is one (i.e., the economy has a comparative advantage in product 𝑗𝑗) if the economy’s share in total exports of the product is larger than its share in exports of all products together, and zero otherwise.8 The values of comparative advantage 𝑀𝑀𝑡𝑡𝑡𝑡 can be arranged in a matrix, R, with each row representing a product, and each column an economy. The sum over rows in a column is the number of products in which the economy has comparative advantage, and we refer to this as the diversification level of the economy: 𝑖𝑖𝑖𝑖𝑀𝑀𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑠𝑠𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡=1 𝑐𝑐�𝑀𝑀𝑡𝑡𝑞𝑞 𝑞𝑞 where 𝑐𝑐 is the number of products in the database, such that the diversification index represents the share of all products in which economy 𝑖𝑖 has comparative advantage. Similarly, the ubiquity of a product is defined as: 𝑐𝑐𝑅𝑅𝑖𝑖𝑢𝑢𝑐𝑐𝑖𝑖𝑡𝑡𝑝𝑝𝑡𝑡=�𝑀𝑀𝑖𝑖𝑡𝑡 𝑖𝑖 Which is simply the number of economies that export a particular good, 𝑗𝑗, with comparative advantage. Using the diversification and ubiquity indicators, we follow Hidalgo and Hausmann (2009) in defining a measure called standardness: 8 An equivalent definition of comparative advantage would be that the export share of product 𝑗𝑗 in country 𝑖𝑖 is greater than the export share of product 𝑗𝑗 across all countries (i.e., the world). 21 𝑐𝑐𝑡𝑡𝑖𝑖𝑐𝑐𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡=∑𝑀𝑀𝑡𝑡𝑞𝑞𝑐𝑐𝑅𝑅𝑖𝑖𝑢𝑢𝑐𝑐𝑖𝑖𝑡𝑡𝑝𝑝𝑞𝑞𝑞𝑞 𝑐𝑐×𝑖𝑖𝑖𝑖𝑀𝑀𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑠𝑠𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡𝑀𝑀𝑖𝑖𝑚𝑚𝑞𝑞�𝑐𝑐𝑅𝑅𝑖𝑖𝑢𝑢𝑐𝑐𝑖𝑖𝑡𝑡𝑝𝑝𝑞𝑞� � Note that the expression in the numerator is the average ubiquity of the products in which the economy has comparative advantage. The denominator is the maximum value of ubiquity over all products, which we use (contrary to the definition in Hidalgo and Hausmann, 2009) to scale the standardness measure (which is thus bound by ⟨0. .1]). The database also contains sectoral versions of standardness and diversification. For this, we assign each of the products in the export database to an economic sector.9 This implies that this part of the database is limited to sectors that produce tradeable goods, which rules out, for example, many of the services sectors.10 The sectors for which we present indicators are listed in Table 6 below. The indicators are calculated by summing over the products that belong to a sector: 𝑖𝑖𝑖𝑖𝑀𝑀𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑠𝑠𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡𝑡𝑡=1 𝑐𝑐𝑡𝑡� 𝑀𝑀𝑡𝑡𝑞𝑞 𝑞𝑞∈𝑡𝑡 and 𝑐𝑐𝑡𝑡𝑖𝑖𝑐𝑐𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡𝑡𝑡=∑𝑀𝑀𝑡𝑡𝑞𝑞𝑐𝑐𝑅𝑅𝑖𝑖𝑢𝑢𝑐𝑐𝑖𝑖𝑡𝑡𝑝𝑝𝑞𝑞𝑞𝑞∈𝑡𝑡 𝑐𝑐𝑡𝑡×𝑖𝑖𝑖𝑖𝑀𝑀𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑠𝑠𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡𝑡𝑡𝑀𝑀𝑖𝑖𝑚𝑚𝑞𝑞∈𝑡𝑡�𝑐𝑐𝑅𝑅𝑖𝑖𝑢𝑢𝑐𝑐𝑖𝑖𝑡𝑡𝑝𝑝𝑞𝑞� � where k indicates a sector, and 𝑐𝑐𝑡𝑡 is the number of products in the sector. The calculation of product complexity follows the ECI method of Hidalgo (2021). For this, we first create a row-normalized and a column-normalized version of the comparative advantage matrix R. The row-normalized version is created by dividing each element of the matrix by diversity of the column, and the column-normalized version is created by dividing each element by ubiquity of the row. We then create a new matrix by pre-multiplying the column-normalized version of the matrix by the transpose of the row-normalized version. Product complexity is equal to the eigenvector that belongs to the second-dominant eigenvalue of this matrix.11 Based upon the constructed indicator of product complexity, economy fitness levels are calculated as the average value of product complexity of the products that the economy has comparative advantage in: 9 This assignment is done using concordances between ISIC (for the input-output sectors) and the Harmonised System (HS) for the export database, or, as introduced below, concordances between HS and BEC. These concordances are available on the World Bank’s WITS server https://wits.worldbank.org/. 10 The data used to construct the various indicators in this pillar is the UN Comtrade database, which reports information on goods trade only. Data on services trade does exist – and is also available from the UN – but consistent data across economies is only available at a much higher level of aggregation. 11 The eigenvectors have ambiguous sign, and we choose the sign such that products with low ubiquity tend to have high complexity. 22 𝑠𝑠𝑖𝑖𝑡𝑡𝑐𝑐𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡=∑𝑀𝑀𝑡𝑡𝑞𝑞𝐻𝐻𝑞𝑞𝑞𝑞 𝑐𝑐×𝑖𝑖𝑖𝑖𝑀𝑀𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑠𝑠𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡 The sectoral version of fitness is 𝑠𝑠𝑖𝑖𝑡𝑡𝑐𝑐𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡𝑡𝑡=∑𝑀𝑀𝑡𝑡𝑞𝑞𝐻𝐻𝑞𝑞𝑞𝑞∈𝑡𝑡 𝑐𝑐𝑡𝑡×𝑖𝑖𝑖𝑖𝑀𝑀𝑖𝑖𝑖𝑖𝑐𝑐𝑖𝑖𝑠𝑠𝑖𝑖𝑀𝑀𝑖𝑖𝑡𝑡𝑖𝑖𝑐𝑐𝑐𝑐𝑡𝑡𝑡𝑡 Figure 7 provides an illustration of the indicators described above, with data reported on the Basic metals and fabricated metal sector in 2019. Economies’ fitness in the sector is on the horizontal axis, while diversification is reported on the vertical axis. The size of the bubbles in this figure is proportional to the level of standardness. We observe a generally positive association between the two indicators, but this is not a linear relation. At first, diversification increases along with fitness, but at a certain threshold value, fitness does not increase any further. We also see that economies with high diversification and high fitness tend to be the ones with low standardness, while economies with low fitness and low diversification also have high standardness. Figure 7. Diversification and fitness in basic metals and fabricated metal, 2019 29 𝑊𝑊𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑=�𝑀𝑀𝑡𝑡𝑡𝑡 ∑𝑀𝑀𝑡𝑡𝑡𝑡𝑡𝑡∈𝑑𝑑𝑑𝑑𝑑𝑑 𝐻𝐻𝑡𝑡𝐹𝐹 𝑡𝑡∈𝑑𝑑𝑑𝑑𝑑𝑑 𝑊𝑊𝑡𝑡𝑓𝑓𝑑𝑑𝑖𝑖=�𝑀𝑀𝑡𝑡𝑡𝑡 ∑𝑀𝑀𝑡𝑡𝑡𝑡𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 𝐻𝐻𝑡𝑡𝐹𝐹 𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 Where the superscripts 𝑖𝑖𝑐𝑐𝑐𝑐 and 𝑠𝑠𝑐𝑐𝑖𝑖 denotes domestic and foreign, or the set of domestic or foreign input sectors to the chain, respectively. In addition to input complexity of the value chain, output complexity of the value chain can simply be defined as 𝐻𝐻𝑡𝑡𝐹𝐹. Figure 11 documents input complexity versus output complexity in the Food and beverages sector, for 2019. We see a clear positive relationship: economies with higher input complexity also tend to have higher output complexity. However, there are some economies that are clearly above (or below) the line that summarizes this relationship (i.e., the line of best fit). Those economies that are significantly above (below) the line can be considered to add comparatively much (little) complexity to the output of the value chain. Table 8 provides the list of indicators that are available in the section of the database on product complexity in global value chains. The list of sectors that is included in the calculations, and for which data on inputand output complexity are available is documented in Table 6 above.13 13 Though the sector “all other sectors” is excluded from the analysis here. 30 Figure 11. Input and output complexity of intermediates in the Food, beverages and tobacco sector in 2019 Table 8. Indicators in the product complexity in global value chains section Output complexity of the chain (final goods) Complexity of intermediate goods of the sector (output) Complexity of all goods (final and intermediate) of the sector (output) Input complexity all domestic and foreign value (total) into the chain Input complexity of domestic value into the chain Input complexity of foreign value into the chain 4. Pillar 3: Innovation Innovation is a multi-faceted process, in which many different kinds of inputs and outputs can be found. Technology is an important input into the innovation process, but innovation can also be organizational, or be related to marketing and design, among other things. In our database, we 31 consider only indicators related to technological innovation, and within this restriction we cover only patented inventions. There are many well-known drawbacks of patents as indicators of (technological) innovation, such as the fact that not all patented inventions lead to actual innovations that appear in the market. Also, the propensity to patents inventions differs greatly by industry. The value of patents also differs widely, with a majority of patents being of very little value, and a few patents accounting for the large majority of total value of patents. In spite of all these drawbacks, patent indicators are developed and used in the database since they are available for a large number of economies and can be used to construct indicators that have a high degree of comparability between economies. The patent indicators used in the database are constructed using the PATSTAT database. This is a database with information on individual patent applications from a variety of patent offices around the world. PATSTAT is provided by the European Patent Office (EPO), and contains, among other things, information on so-called patent families. A patent family is defined as a set of related patent documents (applications and grants) that originate from different patent offices around the world but cover a single invention. Each of the different documents, or family members, provide protection in a specific geographical jurisdiction, which is often an economy, but can also be multiple economies. The patent family is the basic unit that is used to construct the patent indicators in the database, particularly the use of the so-called DOCDB definition of families. The basic approach for constructing the indicators involves counting patent families, where only families that have members in at least two patent jurisdictions (areas where protection is sought) or at a regional patent office such as the EPO are considered. By eliminating patent families with just one jurisdiction, the intention is to include only patents above a certain threshold of originality and expected economic value. When constructing the indicators, all members of patent families – whether they be patent applications or grants – are included. PATSTAT also provides an indication of the most likely industry of origin of the patent application. This is based on a concordance between the patent class (a classification of patents into technological domains) and the NACE industry classification. This information is used to link some the patent indicators to the input-output tables that are used in the global value chain section of the database. Location of patent families (i.e., the origin of the invention covered by the patent family) is determined in a fractional way on the basis of the location of inventors. 4.1. Patent indicators for 4th industrial revolution technologies The term 4th industrial revolution has been used to describe a set of emerging technologies in which digital connectivity, automation and artificial intelligence play a central role. These technologies are applied to a range of technological domains and productive sector (in other words, they are 32 pervasive). It is generally believed that these technologies have great economic potential but will also transform the workplace and society more generally. Using the PATSTAT database 4th industrial revolution (4IR) patents are identified, with the database providing counts of the patent families that are considered to be 4IR, further distinguishing between a number of technological fields in which these patent families are classified. In order to identify 4IR patents, a methodology proposed by the European Patent office (EPO, 2020) is applied. The approach adopted does not follow the exact EPO methodology, but this approach is approximated through the PATSTAT database query that was designed to extract information on 4IR patents.14 Because there are relatively few 4IR patent families defined in this way (in total, there are about 200,000 4IR patent families), the database uses 10-year cumulative numbers. Thus, the data documented in the database for year 𝑇𝑇 always include the numbers of patents families from year 𝑇𝑇−9 to 𝑇𝑇. Figure 12 shows 4IR patenting trends for Hong Kong, China. In the upper panel we see a steady rise of the share of 4IR patenting in total global patenting, against which we can compare the share of 4IR patenting in Hong Kong, China. We see that this share is higher than the global share (for the entire period) and that it rises more steeply after 2010, suggesting that Hong Kong, China is a relatively intensive producer of 4IR technologies. The lower panel of the figure reports information on the performance of Hong Kong, China in total global patenting, with the figure showing that the share of Hong Kong, China in total global patenting is increasing over time, albeit at a slower pace after 2010. 14 The exact EPO query could not be replicated because the Query language used by EPO is not generally available outside patent offices, and because PATSTAT does not contain the full text of the patents. The query used for the database is fully documented in Menéndez et al. (2023). 33 Figure 12. Totaland 4IR-patenting view in the database, Hong Kong, China Figure 13 shows the total number of patents in 4IR in Hong Kong, China, split by subfield. Here the steep rise after 2010 is also visible. We also see that by the end of the period, connectivity, consumer goods, data security, (smart) home, IT hardware, services and software are the largest 4IR subfields in Hong Kong, China patenting. The 4IR subfields that are available in the database are documented in Table 9, while Table 10 lists the indicators in this part of the database. 34 Figure 13. 4IR patenting by subfields in Hong Kong, China 35 Table 9. Subfields of 4IR technologies Subfield # Description 1 3D support systems 2 Agriculture 3 Connectivity 4 Consumer goods 5 Core artificial intelligence 6 Data management 7 Data security 8 Geo positioning 9 Healthcare 10 Home 11 Industrial 12 Infrastructure 13 IT hardware 14 Power supply 15 Safety 16 Services 17 Software 18 User interfaces 19 Vehicles 20 Total Table 10. Variables in the 4IR patent indicators section Patent families in all 4IR fields Patent families by 4IR subfields Patent families in all technology fields Share of 4IR patent families in all patent families, per economy and globally Share of economy in all patent families 4.2. Patent indicators in the context of global value chains In order to relate patent counts to global value chains, a similar approach to that used to relate product complexity to global value chains (section 3.2 of this document) is adopted. Initially, an indicator called the patent content of value added is define, with this indicators being defined at the level of sectors in economies: 𝑄𝑄𝑡𝑡=𝑃𝑃𝑖𝑖𝑡𝑡𝑡𝑡 𝑉𝑉𝐴𝐴𝑡𝑡 where 𝑄𝑄𝑡𝑡 is the patent content of value added in sector 𝑗𝑗 (remember that this is a sector in a particular economy, e.g., the Japanese basic metals sector, or the German chemicals sector), 𝑉𝑉𝐴𝐴𝑡𝑡 is value added in the sector, and 𝑃𝑃𝑖𝑖𝑡𝑡𝑡𝑡 is the number of patent families assigned to the sector, again 36 cumulated over 10 years. Two versions for the patent content indicator are defined: patents in all technology fields and patents in 4IR technology fields only, as defined in the previous section. The indicators 𝑄𝑄𝑡𝑡 can thus be considered analogous to 𝐻𝐻𝑡𝑡𝐹𝐹 in Section 3.1. An analogue of the 𝑊𝑊𝑡𝑡 indicator can further be defined as: 𝑂𝑂𝑡𝑡=�𝑀𝑀𝑡𝑡𝑡𝑡 𝑠𝑠𝑡𝑡𝑄𝑄𝑡𝑡 𝑡𝑡 Where 𝑂𝑂𝑡𝑡 is the patent content of the value chain of sector j (where again this indicator is constructed for either all technology fields or 4IR fields only) As in section 3.1, this indicator can further be separated into two parts: 𝑂𝑂𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑=�𝑀𝑀𝑡𝑡𝑡𝑡 ∑𝑀𝑀𝑡𝑡𝑡𝑡𝑡𝑡∈𝑑𝑑𝑑𝑑𝑑𝑑 𝑄𝑄𝑡𝑡 𝑡𝑡 𝑂𝑂𝑡𝑡𝑓𝑓𝑑𝑑𝑖𝑖=�𝑀𝑀𝑡𝑡𝑡𝑡 ∑𝑀𝑀𝑡𝑡𝑡𝑡𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 𝑄𝑄𝑡𝑡 𝑡𝑡 Where the superscripts 𝑖𝑖𝑐𝑐𝑐𝑐 and 𝑠𝑠𝑐𝑐𝑖𝑖 again refer to domestic and foreign, respectively, or the set of domestic or foreign input sectors to the chain. In other words, 𝑂𝑂𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑 captures the patent content of the value chain from domestic sources, while 𝑂𝑂𝑡𝑡𝑓𝑓𝑑𝑑𝑖𝑖 captures the patent content from foreign sources. The indicators 𝑂𝑂𝑡𝑡,𝑂𝑂𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑 and 𝑂𝑂𝑡𝑡𝑓𝑓𝑑𝑑𝑖𝑖 focus on value chains, i.e., they take a production point of view. We can also use the patent content of value added to consider final demand (i.e., consumption and investment). In this way, we can construct the indicator 𝐽𝐽𝑡𝑡=�𝑤𝑤𝑡𝑡𝑡𝑡 𝜓𝜓𝑡𝑡𝑄𝑄𝑡𝑡 𝑡𝑡 where 𝑤𝑤𝑡𝑡𝑡𝑡 is the value-added contribution of sector 𝑖𝑖 to the final demand of sector 𝑗𝑗 products in the economy to which 𝑗𝑗 belongs and 𝜓𝜓𝑡𝑡 is total final demand of sector j products in the economy to which 𝑗𝑗 belongs. An example of the 𝐽𝐽𝑡𝑡 indicator is that it could reflect the average patent content of value added delivered to final demand (i.e., consumption and investment) of transport equipment in Viet Nam (in this case, 𝑗𝑗 is the Vietnamese transport equipment sector). Figure 14 plots for all economies in the database the patent intensity of own value added against patent intensity of the value chain (“all value added used”) for 2015, for the total economy. We see a strong correlation between these indicators: economies that have patent-intensive production also tend to use more patent-intensive inputs in their value chains. The Asian economies Tapei,China, Republic of Korea and Japan top the list on both indicators. 37 Figure 14. Patent intensity of GVCs in 2015 Table 11 lists the indicators in this part of the database. 38 Table 11. Indicators available in innovation and global value chains section Patent content of value added Patent content of the value chain Domestic patent content of the value chain Foreign patent content of the value chain Patent content of final demand Patent content of value added, only 4IR patents Patent content of the value chain, only 4IR patents Domestic patent content of the value chain, only 4IR patents Foreign patent content of the value chain, only 4IR patents Patent content of final demand, only 4IR patents 5. Pillar 4: Indicators for Global Value Chain participation Indicators in this pillar are heavily based on input-output tables, which requires a framework that allows for the definition and conceptualization of GVCs using such data. This involves making a distinction between so-called intermediate goods and final goods. Final goods are goods that are either used for consumption or investment (e.g., a car), while intermediate goods are goods that are used in the construction of final goods, i.e., steel that is used to build a car. The distinction between final and intermediate also holds for services. For example, a haircut is a service delivered to final demand, while a courier service delivering business documents is an intermediate delivery. Some goods or services can either be delivered as final or as intermediary (i.e., electricity, insurance), and they may appear in two places in the input-output tables that we use. The distinction between intermediate and final is used to conceptualize the value chain. Value chains are initially defined by sectors or industries, e.g., the steel industry or the food industry. These sectors are further defined by the economy in which they are located, i.e., we have the German car industry as well as the Italian car industry, and the Japanese food sector as well as the Brazilian food sector. The value chains in each of those economy-sector combinations are defined as the entire (so-called vertically integrated) chain that is needed to produce the final output of the sector. In the example of a car sector, it includes the cars themselves as final deliveries, and all the intermediate deliveries (goods and services) needed to produce these cars. In this way, trade, and hence the global dimension of the value chains appears in two forms. First, there are exports (and the associated imports) of the final good itself. This is fully in line with the traditional idea of trade in the economic literature, e.g., exchanging cloth for wine. Second, intermediate goods are traded, and hence there will be an international component to the value chain in terms of the inputs that are used in the production of final goods. 45 Figure 17. RCA in gross and value-added exports in Chemicals in 2019 46 Table 12. Indicators used in GVC positioning indicators section Backward: share of own value added in the value chain Backward: share of domestic other sectors in value added of the chain Backward: share of foreign sectors in value added of the chain Forward: share of own value chain in value added of the sector Forward: share of value delivered to other domestic value chains as share of value added of the sector Forward: share of value delivered to foreign value chains as share of value added of the sector Domestic backward integration Domestic forward integration Value added exports of the economy to a specific global sector, RCA Value added imports of the economy from a specific global sector, CII Value added exports, RCA (of the exporting sector) Value added imports, CII (of the imported sector) Gross intermediate exports of economy to a specific global sector, RCA Gross intermediate imports of economy from a specific global sector, CII Gross intermediate exports of economy-sector, RCA Gross intermediate imports from economy-sector, CII Gross final demand exports of economy-sector, RCA Gross final demand imports from economy-sector, CII Gross total exports of economy-sector, RCA Gross total imports of economy-sector, CII 5.2. Regional and global value chains This section of the database aims to provide an impression of how “global” global value chains are. It uses the notion of geographical distance to operationalize this. This starts from a distance matrix that was obtained from CEPII15, and which contains the geographical distance in km between the capitals of the economies in the input-output database. This distance is taken as representative of the general distance between economy-pairs. With this distance matrix and the matrix 𝑉𝑉 that was introduced above the following two basic measures are constructed: 𝑅𝑅𝑡𝑡𝑏𝑏𝑏𝑏=∑𝑖𝑖𝑡𝑡𝑡𝑡𝑀𝑀𝑡𝑡𝑡𝑡𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 ∑𝑀𝑀𝑡𝑡𝑡𝑡𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 ∑𝑖𝑖𝑡𝑡𝑡𝑡𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 � 𝑅𝑅𝑡𝑡𝑓𝑓𝑏𝑏=∑𝑖𝑖𝑡𝑡𝑡𝑡𝑀𝑀𝑡𝑡𝑡𝑡𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 ∑𝑀𝑀𝑡𝑡𝑡𝑡𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 ∑𝑖𝑖𝑡𝑡𝑡𝑡𝑡𝑡∈𝑓𝑓𝑑𝑑𝑖𝑖 � where 𝑅𝑅𝑡𝑡𝑏𝑏𝑏𝑏 is the backward GVC Geo Radius of the sector 𝑗𝑗, 𝑅𝑅𝑡𝑡𝑓𝑓𝑏𝑏 is the forward GVC Geo Radius of the sector 𝑗𝑗, and 𝑖𝑖𝑡𝑡𝑡𝑡=𝑖𝑖𝑡𝑡𝑡𝑡 is the distance between the economy to which sector 𝑖𝑖 belongs and the economy to which sector 𝑗𝑗 belongs. The numerator in both expressions is the weighted average of the distance to GVC trading partners for sector 𝑗𝑗, using either backward or forward foreign linkages as weights (domestic linkages are disregarded in the construction of these indicators). The denominator in both expressions is the unweighted distance to trading partners. Hence the 15 http://www.cepii.fr/CEPII/en/bdd_modele/bdd_modele_item.asp?id=6 47 indicator is a relative distance, where the GVC-weighted distance is expressed relative to the unweighted distance. A value smaller (larger) than 1 indicates that GVC trading partners tend to be relatively close-by (far away). As it turns out the values that we find in the database are almost exclusively less than one, which is consistent with the general finding in so-called gravity trade models that trade is more intensive between economies that are close to each other. Figure 18. Backward and forward geo radius in Rubber and plastics in 2019 Figure 18 shows a plot of the backward geo radius against the forward geo radius for the Rubber and plastics sector in 2019. A positive correlation between the two dimensions exists. Generally, 48 the geo radiuses are observed to be below unity, which indicates that trade (within global value chains) occurs over relatively short distances. There are exceptions to this general pattern, however, with the forward radius of Indonesia and India being above one, for example. Figure 19 shows the time series for the forward and backward geo radius of Indonesia in the Rubber and plastics sector. Here it can be observed that both of these indicators, but especially the forward radius, are rising over time, with a jump around 2014-15. Since 2017 the forward radius increases to a value above one. Figure 19. Sectoral backward and forward geo radius in Rubber and plastics in Indonesia, 2007-2019 The list of indicators provided in this section are listed in Table 13, while the sectors for which these indicators are available are the same as in previous sections (Table 1). Table 13. Indicators in GVC positioning section Backward Geo Radius of GVC integration Forward Geo Radius of GVC integration 49 References Acemoglu, D. and P. Restrepo, 2018, Demographics and automation, NBER Working Paper no. 24421, National Bureau of Economic Research. EPO (European Patent Office), 2020, Patents and the fourth industrial revolution. The global technology trends enabling the data-driven economy, Munich. Foster-McGregor, N., Nomaler, Ö. and B. Verspagen, 2019, Measuring the creation and adoption of new technologies using trade and patent data, UNIDO Working Paper Hidalgo, C.A., 2021, Economic complexity theory and applications. Nat Rev Phys 3, 92–113. https://doi.org/10.1038/s42254-020-00275-1 Hidalgo, C., and R. Hausmann, 2009, The Building Blocks of Economic Complexity, Proceedings of the National Academy of Sciences 106(26):10570–75. Menéndez De Medina , Mercedes, Önder Nomaler & Bart Verspagen, 2023, Identification of Fourth Industrial Revolution technologies using PATSTAT data, UNU-MERIT Working Paper 2023-023. https://unu-merit.nl/publications/wppdf/2023/wp2023-023.pdf Nomaler, Ö and B. Verspagen, 2022, Some new views on product space and related diversification, UNU-MERIT Working Paper no. 2022/11, UNU-MERIT. 50 Annex A1. The use of the BEC classification The approach to construct the data on global value chains in the structural change pillar (section 2 of the main text) is based upon the latest version of the Broad Economic Categories (BEC) classification, BEC Rev. 5.16 This classification allows for a split of HS products into three broad end use categories: intermediate consumptions (intermediates); gross fixed capital formation (capital); and final consumption (consumer goods). A correspondence between the 2012 revision of the HS classification and BEC Rev. 5 has been provided by the United Nations Statistics Division.17 In addition to these three dimensions, the BEC Rev. 5 classification has a number of mixed categories (e.g., intermediate/capital, consumer/capital, etc.). These mixed categories are allocated to one of the three aggregates, with intermediate/capital and intermediate/consumption being allocated to intermediates, consumer/capital and consumer/intermediates to consumer goods, and capital/consumer and capital/intermediates to capital goods. In addition to the split between the three end use categories, the BEC classification further splits intermediate and consumer goods into primary and processed goods. Primary goods are those with characteristics of primary sectors of the economy, such as farming, forestry, fishing, and extractive industries, as well as goods that undertake only minor modifications and whose value is thus still largely from primary sectors (e.g., ginned cotton). Processed goods, conversely, are those that involve extensive processing. This distinction between primary and processed has been used to help identify global value chain trade, with producers of primary goods considered higher upstream (i.e., strong forward linkages), while producers of processed goods tend to be further downstream (i.e., strong backward linkages). A novelty of BEC Rev. 5 over the earlier BEC Rev. 4 relates to the conclusion that the definition of intermediates – including the processed and primary split – is too broad to be useful in the context of global value chains. Processed intermediate goods, for example, have been found to comprise many generic products with published reference prices or commonly sold at auction as well as intermediates with a more specific use in a particular industry. BEC Rev. 5, therefore, introduces an additional split between generic and specific intermediate goods (as well as capital goods) to better identify trade within global value chains. 16 https://unstats.un.org/unsd/trade/classifications/SeriesM_53_Rev.5_17-01722-E-Classification-by-BroadEconomic-Categories_PRINT.pdf 17 http://unstats.un.org/unsd/statcom/47th-session/documents/BG-2016-11-Manual-of-the-Fifth-Revision-of-theBEC-E.pdf 51 A2. Product complexity and upgrading probability The probability measure for upgrading is based on the idea that a country’s current specialization structure (partly) determines which products are likely targets for developing new comparative advantages. The calculations start by defining 𝑋𝑋 as the familiar binary matrix of revealed comparative advantage (RCA), with elements: xij = 1 if 𝐸𝐸𝑖𝑖𝑖𝑖𝐸𝐸𝑖𝑖 � 𝐸𝐸𝑖𝑖𝐸𝐸 �≥1 and xij = 0 otherwise, where 𝐸𝐸𝑡𝑡𝑡𝑡 denotes the value of exports of product 𝑖𝑖 by country 𝑗𝑗 and the absence of a subscript indicates a summation over the relevant dimension. The matrix 𝑋𝑋 has dimensions 𝑐𝑐×𝑐𝑐, where 𝑐𝑐 is the number of products and 𝑐𝑐 is the number of countries. Typically, 𝑐𝑐 >> 𝑐𝑐. We assume that each country exports at least one product, and each product is exported by at least one country. A given 𝑋𝑋 matrix contains a total of 𝑖𝑖=∑ ∑ 𝑚𝑚𝑡𝑡𝑡𝑡∀𝑡𝑡∈{1,2,..𝑛𝑛}∀𝑡𝑡∈{1,2,..𝑑𝑑} revealed comparative advantages. The so-called density metric for related variety (Klinger and Hausman, 2006) draws on the notion of conditional probability, computed on the basis of observed co-occurrences. The conditional probability captures the idea that having a comparative advantage in one product provides information about the likelihood that a country has a comparative advantage in another product. If the conditional probability is high (low), the two products are likely to require similar capabilities to export them with comparative advantage. Formally, let 𝑘𝑘𝑖𝑖𝑞𝑞 denote the number of countries which have comparative advantage both in product 𝑢𝑢 and in product 𝑐𝑐, and 𝑐𝑐𝑖𝑖 denote the number of countries which have comparative advantage in product 𝑐𝑐 (what is usually called the ‘ubiquity’ of a product). Then 𝑀𝑀𝑞𝑞𝑖𝑖=𝑘𝑘𝑞𝑞𝑖𝑖 𝑐𝑐𝑖𝑖 ⁄ denotes the probability that a country has comparative advantage in product 𝑢𝑢, conditional on the country having comparative advantage in product 𝑐𝑐. In matrix notation, these conditional probabilities are given by: 𝐻𝐻=𝑆𝑆−1𝑋𝑋𝑋𝑋𝑇𝑇, where the superscript 𝑇𝑇 indicates a transposed matrix and 𝑆𝑆 is the matrix with the corresponding row-sums of 𝑋𝑋 on the main diagonal and zeros elsewhere. Note that the diagonal of 𝑆𝑆 thus contains the ubiquity of respective products, i.e., 𝑐𝑐𝑡𝑡𝑡𝑡=�∑𝑚𝑚𝑡𝑡𝑡𝑡∀𝑡𝑡∈{1,2,..𝑛𝑛}𝑖𝑖𝑠𝑠 𝑖𝑖=𝑗𝑗 0𝑐𝑐𝑡𝑡ℎ𝑖𝑖𝑖𝑖𝑤𝑤𝑖𝑖𝑐𝑐𝑖𝑖 52 The database involves the use of a new measure that is also based on the idea of conditional probability, but which differs from the density measure. For this new measure, a matrix with a complementary conditional probability measure is used, which is defined as: 𝐵𝐵=𝑈𝑈−1𝑍𝑍𝑋𝑋𝑇𝑇, where 𝑍𝑍=𝑂𝑂–𝑋𝑋 and 𝑈𝑈 is a matrix with the row-sums of 𝑍𝑍 on the main diagonal. Thus, 𝑍𝑍 is a matrix of “anti-RCA”, in which the elements are 𝑧𝑧𝑡𝑡𝑡𝑡= 1 if 𝑚𝑚𝑡𝑡𝑡𝑡= 0 and 𝑧𝑧𝑡𝑡𝑡𝑡= 0 if 𝑚𝑚𝑡𝑡𝑡𝑡= 1. The diagonal elements of 𝑈𝑈 are then defined as 𝑐𝑐𝑡𝑡𝑡𝑡=𝑐𝑐−𝑐𝑐𝑡𝑡𝑡𝑡, which is the number of countries that have no RCA in product 𝑗𝑗. The elements 𝑅𝑅𝑖𝑖𝑞𝑞 of matrix 𝐵𝐵 reflect the probability that a country has RCA in product 𝑢𝑢 conditional on not having RCA in product 𝑐𝑐. We interpret these conditional probabilities as follows: if 𝑅𝑅𝑖𝑖𝑞𝑞 is low (high), products 𝑐𝑐 and 𝑢𝑢 are likely to share a high (low) degree of capabilities needed to export them with comparative advantage. Thus, the conditional probabilities 𝑅𝑅𝑖𝑖𝑞𝑞 are complementary to 𝑀𝑀𝑖𝑖𝑞𝑞. The proposed measures are then defined, in matrix notation as: 𝐸𝐸=𝐻𝐻𝑇𝑇𝑋𝑋+𝐵𝐵𝑇𝑇𝑍𝑍 𝑑𝑑, which is an 𝑐𝑐×𝑐𝑐 matrix. Note that for a country 𝑗𝑗 that has actual comparative advantage in 𝑠𝑠𝑡𝑡 products, each entry in the 𝑗𝑗th column of 𝐻𝐻𝑋𝑋 is the sum of 𝑠𝑠𝑡𝑡 alternative conditional probabilities, while each entry in the 𝑗𝑗th column of 𝐵𝐵𝑍𝑍 is the sum of 𝑐𝑐−𝑠𝑠𝑡𝑡 other conditional probabilities. Thus, through the division by 𝑐𝑐, each entry in (the 𝑗𝑗th column of) 𝐸𝐸 is the sum of exactly 𝑐𝑐 alternative conditional probability estimations, which means that the elements of 𝐸𝐸 are average conditional probabilities. Matrix 𝐸𝐸 can be seen as a probabilistic estimation of 𝑋𝑋. It can be rearranged to obtain separate measures of related and unrelated variety. For this, we proceed as follows: 𝐸𝐸=𝐻𝐻𝑇𝑇𝑋𝑋+𝐵𝐵𝑇𝑇𝑍𝑍 𝑑𝑑=𝐻𝐻𝑇𝑇𝑋𝑋−𝐵𝐵𝑇𝑇(𝑂𝑂−𝑋𝑋) 𝑑𝑑=𝐵𝐵𝑇𝑇𝑂𝑂+𝐾𝐾𝑇𝑇𝑋𝑋 𝑑𝑑, where 𝐾𝐾 ≡𝐻𝐻–𝐵𝐵 is a matrix of marginal conditional probabilities. Written this way, the matrix 𝐸𝐸 has two constituent (additive) elements. The first of these, 𝐸𝐸1≡𝐵𝐵𝑇𝑇𝑂𝑂/𝑐𝑐 is an 𝑐𝑐×𝑐𝑐 matrix in which all rows are equal to each other, i.e., where there is no country-variation. In each of the country-columns of this matrix, the 𝑖𝑖th element indicates an autonomous probability of being specialized in product 𝑖𝑖, which we define as the probability for a hypothetical country without any prior comparative advantages (a country that does not trade and begins exporting just product 𝑖𝑖). 53 We characterize this component of matrix 𝐸𝐸 as the part that corresponds to unrelated diversity. The component 𝐸𝐸2≡𝐾𝐾𝑇𝑇𝑋𝑋/𝑐𝑐 (also 𝑐𝑐×𝑐𝑐) results from the particular specialization profile of the country in terms of both product relatedness (reflected in 𝐾𝐾) and the matrix 𝑋𝑋. 𝐸𝐸2 is the induced part of 𝐸𝐸 that corresponds to related diversity. In the database, we use 𝐸𝐸2 as the measure for upgrading probabilities. In order to calculate product complexity, we also use the matrix 𝑋𝑋 that reports comparative advantages. This follows the method of ECI (Hidalgo, 2021). For this, we need the “raw” diversification measure that has already been defined in the main text, and which can be re-defined here as the diagonal matrix 𝑇𝑇 with elements: 𝑖𝑖𝑡𝑡=�𝑋𝑋𝑡𝑡𝑖𝑖 𝑖𝑖 on the main diagonal. We then calculate a new matrix as follows: (𝑈𝑈−1𝑋𝑋)𝑇𝑇𝑋𝑋𝑇𝑇−1 Product complexity is calculated as the eigenvector that corresponds to the second largest eigenvalue of this matrix. The UNU-MERIT WORKING Paper Series 2024-01 The green transformation as a new direction for techno-economic development by Rasmus Lema and Carlota Perez 2024-02 Evidence on an endogenous growth model with public R&D by Thomas H.W. Ziesemer 2024-03 Higher Educational Systems and E-resilience by Mindel van de Laar, Julieta Marotta and Lisa de Graaf 2024-04 Estimating the wage premia of refugee immigrants: Lessons from Sweden by Christopher F. Baum, Hans Lööf, Andreas Stephan and Klaus F. Zimmermann 2024-05 Green window of opportunity through global value chains of critical minerals: An empirical test for refining copper and lithium industries By Jorge ValverdeCarbonell and Alejandro Micco A. 2024-06 Migration and Consumption by Roberta Misuraca and Klaus F. Zimmermann 2024-07 The power of narratives: How framing refugee migration impacts attitudes and political support towards refugees by Tobias Hillenbrand, Bruno Martorano, Laura Metzger and Melissa Siegel 2024-08 Strategies of search and patenting under different IPR regimes by Robin Cowan, Nicolas Jonard and Ruth Samson 2024-09 From drains to bridges: The role of internationally mobile PhD students in linking non-mobile with foreign scientists by Rodrigo Ito, Diego Chavarro, Tommaso Ciarli, Robin Cowan and Fabiana Visentin 2024-10 Expanding horizons: The impact of post-doctoral mobility grants on research diversification by Yağmur Yıldız, Diego Chavarro, Fabiana Visentin and Tommaso Ciarli 2024-11 Scorching heat and shrinking horizons: The impact of rising temperatures on marriages and migration in rural India by Manisha Mukherjee, Bruno Martorano and Melissa Siegel 2024-12 Labour market integration for refugees: A social justice framework by Tamara Kool and Zina Nimeh 2024-13 The evolution of development with trade in global value chains by Önder Nomaler and Bart Verspagen 2024-14 The impacts of the Microfinance Multiplied approach on seasonal food insecurity: Evidence from a high-frequency panel survey in Uganda by Ricardo Morel, Franziska Gassmann, Bruno Martorano, Nyasha Tirivayi and John Kamau 2024-15 Building CoPS Capability for Catching Up During Transitions: The Case of ACWA Power in Saudi Arabia by Kaori Mita and Michiko Iizuka 2024-16 The KSTE+I approach and the AI technologies: evidence from the European regions by Francesco D’Allesandro, Enrico Santarelli and Marco Vivarelli 2024-17 Theory and measurement in SFC models: the role of the financial sector by Huub Meijers and Joan Muysken 2024-18 Is artificial intelligence generating a new paradigm? Evidence from the emerging phase by Giacomo Damioli, Vincent Van Roy, Daniel Vertesy and Marco Vivarelli 2024-19 Industrializing Africa - a science, technology, and innovation policy perspective by Bertha Vallejo 2024-20 Innovation in the wind energy sector by Rasmus Lema, Davide Bonaglia and Ulrich Elmer Hansen