Global Value Chains and Economic Development Pahl, Stefan
DOI:
10.33612/diss.121326589
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Pahl, S. (2020). Global Value Chains and Economic Development. University of Groningen, SOM research school. https://doi.org/10.33612/diss.121326589
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Global Value Chains
and
Economic Development
Publisher: University of Groningen Groningen, the Netherlands
Printed by: Ipskamp Printing
ISBN: 978-94-034-2607-5 eISBN: 978-94-034-2606-8
Copyright © 2020 Stefan Pahl
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system of any nature, or transmitted in any form or by any means, electronic, mechanical, now known or hereafter invented, including photocopying or recording, without prior written permission by the publisher.
Global Value Chains and Economic
Development
PhD thesis
to obtain the degree of PhD at the University of Groningen
on the authority of the
Rector Magnificus Prof. C. Wijmenga and in accordance with
the decision by the College of Deans. This thesis will be defended in public on
Monday 4 May 2020 at 11.00 hours
by
Stefan Pahl
born on 6 April 1988 in Stuttgart, GermanyProf. R.C. Inklaar
Assessment Committee
Prof. B. Los
Prof. J. van Biesebroeck Prof. K. Sen
This PhD thesis is the result of several years of research. During such a process, many initial ideas have to be adjusted or even dismissed, and many things just turn out very differently than expected. This high level of uncertainty of research can cause a great deal of frustration. With the right guidance and support, however, this experience turns into something highly enjoyable and into something I would have never wanted to miss! Therefore, I want to thank a number of people who are invaluable to this thesis.
First of all, I want to thank my supervisors Marcel Timmer and Robert Inklaar who greatly contributed to the research of this thesis and beyond. I came to Groningen for the second year of my master’s degree, initially for only one year. During that time, however, you got me excited about research: I enjoyed teaching from both of you on growth, value chains and development and was motivated by the way and type of research you do and your didactics. Marcel, I would like to highlight that, already back then, I was most amazed about how much more motivated I would leave your office than coming in because of the challenging discussions. These bilateral discussions have continuously motivated me over the past years and significantly improved the content of the chapters. It is needless to say that your contribution to the research of the thesis as a whole, and as co-author in particular, is invaluable. Moreover, I am very thankful for all else that goes beyond the direct work on the thesis. For example, Gaaitzen de Vries and you organised my stay at IFPRI before I started the PhD, and you were also instrumental for my work with UNIDO. Both were great opportunities to see more of the research world and to get in touch with other researchers. Robert, I would like to highlight your guidance at the beginning and end of the thesis project, and in particular your feedback on the last chapter and your role in tracking progress. I would lastly like to stress that I admire in both of you that, besides the research skills, you are able to give very critical feedback in a very respectful, constructive way.
I would also like to thank the members of my reading committee Jo van Biesebroeck, Bart Los and Kunal Sen. Thank you for taking the time to read my thesis and for providing stimulating suggestions that will help working towards future publications of the last two chapters. I also want to thank many colleagues from Groningen. Bart, besides your role in the reading committee, I am happy that we also already met early on when you facilitated my first contact
also want to thank you, Gaaitzen: you introduced me to the work of the GGDC with the 10-Sector Database, you facilitated the stay at IFPRI, and you have remained approachable on many aspects of research and beyond throughout the years. Jop Woltjer and Reitze Gouma, it was fruitful, and a lot of fun, to work together with you on chapter 4 in this thesis. I have further benefitted from many other interactions with colleagues in the GEM and EEF department, among many others Tristan Kohl, Erik Dietzenbacher, and Dirk Bezemer. Further, I want to thank everyone who helped with administration, the GEMmies and SOMmers. Luckily, I only really encountered such issues as I recently engaged in more teaching activities and had to organise the finalisation of the thesis: thanks! I also want to thank David Chilosi, Bert Kramer and David Gallardo for our joint organisation of the PEGDECH seminar, which was a nice way of getting to know different fields of research (mostly economic history though). During my stay in Vienna, I also had great support from Alejandro Lavopa, Adnan Seric and Michael Windisch and I am happy that we could still keep up this relationship over the past years (e.g., with the workshop in Thailand). Lastly, thanks to all people I met at conferences throughout the years, which made exchanging ideas so interesting.
I also gratefully acknowledge financial support from the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO) [grant number 453-14-012] for the thesis as a whole. Furthermore, financial support from the World Bank Group for chapters 3 and 4 is also acknowledged [SST task code p162461].
Pursuing a PhD, however, is also about meeting many other interested and interesting people. You did indeed also contribute by exchanging ideas on the content but in particular by providing a great environment. I always felt that a day in the office without having lunch with you was a bit of a sad day. I am very happy that I joined the PhD, while there were so many other nice people around. The yearly weekend trips and the fact that this year’s trip includes 13 people is a clear sign of this! Daan and Maite, it was just great on many aspects to meet you early on in Groningen because we had many, many great and funny moments! Aobo, it was also really nice getting to know you at the very start of the PhD and in particular that you always give a positive turn to things. Thanks to Johannes for being so motivated for nights out; Nikos and Ferdinand, for your sarcastic comments; and Romina for always being fun! Juliette, thanks for being caring and for canoeing in Giethoorn - a great distraction from Duisenberg. Tobi, Jos and Gianmaria, it is great that you are always motivated to do (new) sports from tennis to beachvolleyball to bouldering to football, and for everything people do afterwards. Xianjia, Joeri and Ibrahim, you
otherwise never heard about (e.g., milk powder). Joeri thanks for bringing my Dutch to the next level and Ibrahim for sharing your football insights. Timon thanks for, perhaps involuntarily, making us all a bit more relaxed by telling us about all your small weekend trips. Mart, Daniel, Nick, Duc, Fred, Kailan, Bingquian, Femke and Eda and many others from Duisenberg – all of you contributed to a nice atmosphere and to the many nice experiences we had!
Eduard, thanks for making my arrival to Groningen so easy by introducing me to new meetups right away, such as the film club. Luzia and Marieke, it was great meeting you there and to continue meeting up from Vienna to Berlin. Marianna, Dimitrios and Wen, thanks for always nice conversations. Also thanks to many other people who supported me from outside Groningen. This includes all friends formerly from Göttingen (mostly now in Berlin) – Sascha and Mario great that you showed (as far as I know: genuine) interest in the topic. Ines, thanks a lot for enduring support during the PhD and way beyond! It helped a lot that you understood all the potential issues that come up during the process of writing a PhD thesis (much better than when the roles were reversed). Of course, thanks to my family: my sister Anne and my parents Elisabeth and Peter. Thank you for making studying possible in the first place, motivating me to study whatever I wanted to and, most importantly, encouraging me to always go for new and interesting endeavours and to see places of the world. This is a major reason why I ended up in economics in the Netherlands – ultimately leading to this PhD thesis.
1 Introduction ... 13
1.1 Background ... 13
1.2 This thesis ... 18
1.3 Outlook ... 23
2 Patterns of Vertical Specialisation: Long-run Evidence for 91 Countries ... 29
2.1 Introduction ... 29
2.2 Methodology ... 30
2.2.1 Measuring vertical specialisation ... 31
2.2.2 Identifying periods of vertical specialisation ... 33
2.3 Data sources and construction ... 35
2.4 Empirical findings ... 38
2.5 Concluding remarks ... 53
2.6 Appendix: additional tables and decomposition method ... 55
2.7 Supplementary material: data construction ... 59
3 Do Global Value Chains Enhance Economic Upgrading? A Long View. ... 69
3.1 Introduction ... 69
3.2 Methodology ... 72
3.2.1 The concept of value added and employment in exports ... 72
3.2.2 The measurement of value added and employment in exports ... 74
3.3 Data sources ... 75 3.4 Empirical results ... 79 3.4.1 Econometric model ... 80 3.4.2 Main results ... 84 3.4.3 Extensions ... 90 3.5 Conclusion ... 95
3.6 Appendix: additional table ... 99
3.7 Supplementary material: robustness and data construction ... 100
4 Jobs in Global Value Chains: A Measurement Framework ... 113
4.1 Introduction ... 113
4.2 Methodology ... 116
4.2.2 Methodology to calculate GVC jobs ... 119
4.2.3 Decomposing growth in GVC jobs ... 120
4.3 Data sources ... 124
4.3.1 World Input-Output Database ... 125
4.3.2 New countries built into WIOD ... 125
4.4 Competitiveness in GVCs of goods: empirical results ... 129
4.5 Growth of jobs in GVCs of goods: empirical results ... 137
4.5.1 Sources of GVC job growth ... 138
4.5.2 Sectoral distribution of GVC job growth ... 141
4.5.3 Extending the decomposition ... 144
4.6 Concluding remarks ... 148
4.7 Appendix: additional tables ... 152
4.8 Supplementary material: Method and sources for seven new countries added to the World Input-Output Database 2016 release ... 155
5 Value-Added Gains from Trade Facilitation: Evidence from Developing Countries in Africa and Asia ... 191
5.1 Introduction ... 191
5.2 Methodology ... 194
5.2.1 Value-added effects ... 195
5.2.2 Sectoral trade elasticities ... 198
5.3 Data ... 203
5.4 Results ... 208
5.4.1 Trade elasticities ... 208
5.4.2 Value-added effects ... 213
5.5 Discussion ... 220
5.5.1 Possible general-equilibrium effects ... 220
5.5.2 Exploration of value-chain adjustments ... 223
5.6 Conclusion ... 225
5.7 Appendix: additional tables ... 229
6 References ... 233
Table 2.1 World VAX-D ratio for exports by manufacturing industries ... 42
Table 2.2 Share of country-years in periods of vertical specialisation ... 46
Table 2.3 Number of countries by trend in VAX-D ratios ... 47
Table 2.4 Decomposition of change in VAX-D ratio for selected countries ... 50
Table 2.5 Vertical specialisation and GDP per capita. Dependent variable: VAX-D ratio... 52
Table 3.1 Difference in means: average annual growth rates in 10-year periods ... 80
Table 3.2 Summary statistics: 10-year periods ... 83
Table 3.3 GVC participation and labour productivity in exports growth ... 86
Table 3.4 GVC participation and employment in exports growth ... 87
Table 4.1 Overview of main sources used for adding seven countries to WIOD ... 128
Table 4.2 Competitiveness in GVCs of final goods ... 131
Table 4.3 GVC specialisation indices by product GVCs ... 133
Table 4.4 Final destination of a country’s GVC value added, 2014 (% shares) ... 137
Table 4.5 Sources of GVC job growth in GVCs of final goods, 2000-2014... 140
Table 4.6 GVC jobs by sector of employment ... 143
Table 4.7 Sources GVC job growth in manufacturing, 2000-2014 ... 147
Table 5.1 Summary statistics ... 208
Table 5.2 Trends in trade facilitation ... 208
Table 5.3 Baseline gravity regression ... 212
Table 5.4 Trade elasticities to trading time by exporting sector groups ... 213
Table 5.5 Net impact of trade facilitation as % of GDP ... 214
Table 5.6 Sectoral structure (%) induced by global improvements to best practices ... 219
Table 5.7 Net impacts of trade facilitation using alterative baseline structures: global changes of observed changes in trading time ... 225
Figure 2.1 VAX-D ratio for exports by all manufacturing industries ... 41 Figure 2.2 World VAX-D ratio for exports by machinery and textiles industries ... 42 Figure 2.3 World VAX-D ratio for exports by manufacturing under alternative scenarios ... 44 Figure 2.4 Examples of countries with non-monotonic trends in VAX-D ratio ... 49 Figure 3.1 Domestic value chains in export production ... 73 Figure 3.2 Marginal effects of GVC participation on labour productivity growth, by levels of labour productivity in exports ... 88 Figure 3.3 Marginal effects of GVC participation on employment growth, by levels of labour productivity in exports ... 88 Figure 4.1 Stylised example of GVC incomes ... 118 Figure 4.2 Sources of GVC job growth, low-income countries ... 141
Chapter 1
13
1 Introduction
1.1 Background
The remarkable growth of world trade is one of the defining features of the global economy of the past decades. At least partly, this is due to the emergence of a new form of integrated global production, which allows countries to specialise in specific production stages rather than entire goods and thereby to trade semi-finished parts and components (Yi, 2003). Sinn (2006) argues, for example, that exporting German firms are importing their components from firms in Central and Eastern Europe and then they only assemble the goods and subsequently re-export them. More specifically, Dudenhöffer (2005) estimates that a Porsche Cayenne finalised in Germany includes only 33% of German value added, while the rest is added abroad. The Porsche is no longer ‘Made in Germany’ but includes contributions from multiple lower-income countries in Europe. This production pattern is referred to as trade in global value chains (GVCs) and is pervasive at the global scale. Other industries with well-documented GVC trade include clothing (e.g., Gereffi, 1999; Gereffi and Frederick, 2010) and electronics (e.g., Dedrick et al., 2010), both characterised by GVCs passing through East and South-East Asia.
This highly interconnected production setup has a myriad of implications. In this thesis, I focus on the implications for developing countries and study the role for GVCs in and for economic development.
There is widespread enthusiasm about GVCs in international organisations, resulting in numerous reports praising GVCs as a possible panacea for economic development (e.g., Taglioni and Winkler, 2016; UNIDO, 2018a; World Bank, 2017; 2020). This enthusiasm essentially stems from the argument that GVC trade provides greater opportunities for development than traditional trade because it is different in two ways. Firstly, it is argued that transactions within GVCs incur a relational aspect, as opposed to pure market-based transactions. Through close cooperation in joint production setups, GVCs allow for fast technology and knowledge transfers. This should allow countries to catch up faster with the technology leaders. Secondly, GVCs potentially yield fine-grained specialisation at the level of production stages (e.g., assembly of cars instead of entire cars) and thereby generate higher efficiency and welfare gains than traditional trade. Specialisation in small parts of the value chain further allows for easier exporting and thus for easier reach of foreign markets, as countries do not need to build up whole value chains at home but can import critical components (e.g., Baldwin, 2014).
14
An often cited stylised comparison that illustrates this view is between Malaysia’s and Thailand’s automotive policies (e.g., Baldwin, 2014; Wad, 2009). Malaysia aimed to build the whole value chain at home by producing a complete national car, following strategies of earlier industrialisers like South Korea or Japan. By the early 2000s, Malaysia indeed built a fully domestically produced car including high-value components like the engine and it reached a dominant role on the domestic market. With plummeting demand after the Asian financial crisis, however, the industry suffered because it could not compensate through exporting due to lacking international competitiveness. Thailand, on the other hand, abolished its plans to produce a fully domestically produced car and aimed to integrate into GVCs of foreign car manufacturers. One element of this strategy was the abolition of its local content requirements in the early 2000s. Through integration into GVCs, Thailand combined high-technology inputs from advanced countries with its low-cost labour to be internationally competitive. Thailand has indeed become a large car exporter, and many global manufacturers have production sites in Thailand, suggesting successful development through GVCs.
Yet, this strategy might have caveats too. To name only one, Thai firms might in fact only ‘borrow’ the technology and once multinationals decide to combine their technology with cheaper labour elsewhere, there might be little production left in Thailand (e.g., Baldwin, 2014; Rodrik, 2014). In this particular case, the question arises how Thailand can ensure that multinationals stay even if wages increase. As GVCs have rapidly become a defining feature of the world economy, such questions are relevant for all developing countries. Yet, there is a relative lack of quantitative evidence on the interaction of GVCs and issues of economic development, which is the primary motivation for the four studies in this thesis.
In the academic literature, GVCs have first been made explicit in case-study literature in political economy, sociology and economic geography.1 This literature has widely studied development opportunities of firms in developing countries in GVCs, but remains relatively cautious about the overall prospects (e.g., Kaplinsky, 2000). One concern are governance and power structures. Gereffi (1994) and Gereffi et al. (2005) importantly emphasise that GVCs are typically not characterised by market transactions between two firms but that they imply the relational aspect. These relationships result in specific governance structures depending on the type of transactions and capabilities of firms involved. Such governance structures are argued
1 Note that the so-called French Filière approach is another predecessor. This literature has primarily focused on agricultural chains with a focus on former French colonies (Raikes et al., 2000). This thesis focusses primarily on manufacturing value chains.
15
to importantly affect whether and how firms in developing countries can move into new, better rewarded activities or capture larger shares of value, broadly known as upgrading (Humphrey and Schmitz, 2002). For example, GVCs can be characterised by so-called captive governance structures with strong lead firms and limited development opportunities for suppliers located in developing countries. Others emphasise that upgrading is only possible given a set of preconditions, such as a large domestic market or an innovation-friendly infrastructure (e.g., Pietrobelli and Rabellotti, 2011). These findings point to a potentially less optimistic relationship between trade in GVCs and development. This literature has produced many case studies investigating development aspects of particular sectors, firms, geographic regions, or institutional environments. However, it has remained largely conceptual with limited quantification or efforts to provide generalizable results.
The economics literature is less focussed on development questions. Ongoing discussions centre on how to model GVCs and how to measure them. The first theoretical novelty compared to traditional trade is the degree of specialisation. GVCs allow countries to exploit comparative advantages not only at the level of final goods but also at the level of production stages. The direct prediction is that larger gains from trade are possible when this is accounted for (e.g., Caliendo and Parro, 2015). The second novelty is that international trade is no longer only a transaction between two countries but that the production process involves multiple countries. This makes it much more difficult to identify clear specialisation patterns by production stages, and to predict trade and welfare outcomes. In this regard, it becomes important whether GVCs are modelled as a strict consecutive process, because theoretical models based on consecutive and non-consecutive production setups provide different predictions on bilateral trade and welfare (e.g., Baldwin and Venables, 2013; Johnson and Moxnes, 2019; Yi, 2003; 2010). Furthermore, consecutiveness in production might add another level of specialisation. For example, countries with higher trade costs might tend to specialise in more upstream production stages if trade costs accumulate along consecutive production chains (e.g., Antras and de Gortari, 2019). Moreover, as GVC trade is not only the exchange of final goods but describes a joint production setup, additional factors might determine specialisation, trade or ownership patterns, such as contractual insecurity (e.g., Antras and Chor, 2013). These considerations add complexity compared to more traditional models of final-goods trade which typically suggested a clear North-South pattern based on factor endowments. More empirical evidence is thus needed to inform theoretical modelling to obtain a better understanding of the role of developing countries in the world economy when GVCs are present.
16
The measurement of GVCs, however, is not straightforward because official statistics do not reflect GVC trade. Trade data are collected by customs authorities, only recording gross exports and imports. That is, a country records its (gross) value of exports but not how much value has been added by the exporting country and where the embodied intermediate inputs came from. It also only indicates the direct export partner but not the ultimate destination. For example, even an exported final good might be directly re-exported to another country instead of being consumed in the direct destination market. The main idea in the literature to measure GVCs and specialisation therein is to trace how much of the gross value of a good can be attributed to particular countries. In a seminal contribution, Hummels et al. (2001) measure how much foreign value added is embodied in one unit of exports (see also Chenery et al., 1986). The higher this foreign value-added content (and thereby the share of foreign intermediate inputs in production), the higher is a country’s so-called vertical specialisation. This approach is based on input-output tables, which have the useful feature of mapping supply and use relationships between producers, allowing for tracing the origin of intermediates.2 Several
recent initiatives not only exploit those supply and use relationships within countries but also attempt to track them across the world economy (e.g., Lenzen et al., 2013; OECD, 2018; Timmer et al., 2015b). Recognising the pervasiveness of GVCs, this is now slowly percolating into the agenda of official national accounting practices (e.g., Eurostat, 2008), but data availability issues are still pressing as many developing countries have not released a single official input-output table. While a lot of progress has been made on tracing trade patterns using multi-region input-output tables, there is an ongoing debate on implementation of key concepts, such as specialisation and participation in GVCs (e.g., Johnson, 2018; Koopman et al., 2014; Los et al., 2016; Wang et al., 2017). For example, it is discussed whether or for what purpose it is more suited to study exporters’ (foreign) backward linkages (in the spirit of Hummels et al., 2001) or an industry’s forward linkages into consumption or production of other countries (in the spirit of Johnson and Noguera, 2017). Another approach is suggested by Wang et al. (2017), for example, attempting to define shares of production as traditional, simple GVC and complex GVC trade by separating value added in trade flows by the number of border crossings. No consensus is found.
2 Simultaneously, the micro literature has studied GVCs at the firm level. The basic identification of GVCs is to track whether firms import and/or export. Importantly, firm-level studies face the challenge that the aspect of a value chain (domestic or global) is difficult to implement. Kee and Tang (2016), for example, combine firm-level statistics with country-level input-output data to obtain measures in the spirit of Hummels et al. (2001) at the firm level.
17
GVCs are not yet as prominent in the broader development economics literature. On trade more generally, this literature also recognises the importance of efficiency and welfare gains derived from specialisation. Yet already since Prebish (1950) and Singer (1950), it is further argued that exploitation of comparative advantages may not be beneficial over the long run and diversification from exports of primary commodities into specialisation in more advanced products is emphasised (e.g., Hausmann et al., 2007). The long-standing debate is whether exploiting the short-run opportunities also provides long-run benefits or whether it holds back capability development in sectors or activities that provide more sustained development. The generation of long-run benefits is thus a key question with respect to GVCs. As for Thailand, productivity gains and increasing competitiveness may be achieved in the short run but it is an open question whether or how this can be sustained (e.g., Baldwin, 2014, Rodrik, 2014; 2018). The development literature further highlights more generally structural change into manufacturing as an essential driver of economic development due to several beneficial characteristics, such as fast productivity growth, technology spillovers to other sectors through linkages or scope for capital accumulation (e.g., Lewis, 1954; Chenery et al., 1986; Rodrik, 2016). In particular, it is often argued that developing countries need to generate jobs in manufacturing to develop (e.g., Felipe et al., 2019, Rodrik, 2013). Yet, it is observed that several countries do not experience structural change into manufacturing or even experience declining manufacturing activities, and globalisation is considered one potential reason (e.g., Rodrik, 2016). Hence, a second key discussion with respect to GVCs is whether and how they allow countries to industrialise, and in particular whether job growth (in manufacturing) is fostered. The goal of this thesis is to contribute to the discussions on structural change and long-run development by providing systematic, cross-country evidence on the generation of value added, productivity and employment in relation to trade in GVCs. The first two chapters consider a very broad set of countries and a long time horizon since 1970. Chapter 2 studies the patterns of GVC engagement of this large set of developing countries and its relation to countries’ stages of development. Chapter 3 investigates the productivity and employment opportunities in manufacturing through participation in GVCs. Chapters 4 and 5 zoom in on a smaller set of countries since 2000. Chapter 4 investigates proximate drivers of job growth in an interconnected world economy for a set of 25 low and middle-income countries, decomposing job growth into global demand, countries’ competitiveness and technology. Chapter 5 analyses how trade policies generate value added when GVCs are pervasive. It evaluates the effect of trade facilitation on sectoral value added, focussing on three African and five East and South-East Asian countries. I elaborate on these chapters in the next section.
18
1.2 This thesis
The economics literature on GVCs has predominantly focused on a relatively small set of mainly high-income countries. Therefore, stylised facts on the engagement of developing countries in GVCs or how GVC engagement evolves over the course of development are not yet established. Evidence on the development opportunities of GVCs typically relies on only few countries, which are treated as archetypal (China in particular), and/or on data of only recent years after 1990 (e.g., World Bank, 2020). Robust patterns on those issues are further useful to provide guidance for future theoretical modelling on the engagement of developing countries in GVCs. Chapter 2 in this thesis aims to fill this gap by providing new stylised facts on GVC participation based on 91 developed and developing countries between 1970 and 2013. The two main antecedents are Hummels et al. (2001) documenting vertical specialisation of 14 economies between 1970 and 1990, and Johnson and Noguera (2017) documenting the rise of GVCs for a set of 40 mainly developed countries between 1970 and 2009. The chapter further elaborates on the different measures of GVCs. In particular, it contrasts the measure popularised by Johnson and Noguera (2017) that tracks forward linkages in the world economy as an indicator of GVCs with tracking backward linkages of exports in the spirit of Hummels et al. (2001).
We follow the latter and base our results on an inverse measure of Hummels et al.’s (2001) vertical specialisation, namely the domestic value-added content in exports (VAX-D) ratio. Through identification of structural break points following Bai and Perron (1998; 2003) in the time series of VAX-D ratios, we identify specific periods of vertical specialisation. We find that the world VAX-D ratio has continuously declined since the mid-1980s (until the crisis 2008/2009). We date the start of global vertical specialisation to the year 1986, providing quantitative support for the conventional narrative (e.g., Baldwin and Lopez-Gonzalez, 2015). Yet, these global trends hide important variation at the country level. We find that almost all countries initiate vertical specialisation, but we find large heterogeneity in timing and speed. The first region of developing countries to experience vertical specialisation is East and South Asia, which is consistent with case-studies on early offshoring (e.g., Hobday, 1995). Interestingly, we also find vertical specialisation in Sub-Saharan African countries in the 1970s, but this trend slows down later on. This early period coincides with relatively high but fluctuating growth rates of GDP per capita in that region (World Bank, 2018b), and thus might deserve further investigation. As a first exploration of this link, and more generally of the link between GVC participation and stages of economic development, we correlate the VAX-D ratio
19
to GDP per capita and find that higher vertical specialisation is associated with higher GDP per capita.
In chapter 3, we investigate this link in more detail by asking whether GVC participation enhances economic upgrading, measured by labour productivity and employment growth in formal manufacturing. Importantly, we not only focus on direct exporters, as in firm-level analysis (e.g., Del Prete et al., 2017; Okafor et al., 2017), nor only on the exporting sector, as in related macro analyses (e.g., Kummritz et al., 2017). Using input-output tables in combination with industry-level statistics on employment and value added in formal manufacturing, we track all direct and indirect exporters in the exporting sectors as well as all supplying ones. We choose the two outcome variables - productivity and employment - because they speak to the potential key effects of GVCs. It is argued that GVC participation affects productivity growth through specialisation and through learning due to the relational aspect (e.g., Criscuolo and Timmis, 2017). Employment growth is less often emphasised in the GVC literature but it is argued to play an important role in development, especially with a focus on manufacturing and discussions on structural change (e.g., Rodrik, 2018; Sen, 2019). The question arises whether increased productivity coincides with increased scale of production and thereby with employment growth. Alternatively, Rodrik (2013; 2018) hypothesises that engagement in GVCs may spur the diffusion of technologies that are ultimately labour-saving and thereby hurt employment growth in developing countries. We refer to this as the ‘mixed-blessing’ hypothesis of productivity growth but limited employment growth.
We correlate growth of the two outcome variables over 10-year periods to GVC participation at the beginning of each period based on a sample of 57 countries between 1970 and 2008. We find a strong association of GVC participation with productivity growth, robust to different cuts of the data. This association appears to be stronger for the set of Asian countries than for the remaining set of developing countries. One might argue that their location makes it particularly suitable for benefitting from GVC participation due to embeddedness in and proximity to ‘factory Asia’ (e.g., Baldwin and Lopez-Gonzalez, 2015). Strong linkages across countries at different stages of development may offer easy access to sophisticated inputs and opportunities to take over production stages as wages rise in the relatively more developed countries. In terms of employment, we do not find evidence that higher GVC participation is associated with higher employment growth. Across multiple cuts of the data, we find correlations that are close to and not statistically different from zero.We conclude that GVC participation is a mixed blessing: on average stimulating productivity growth but not
20
employment growth. We hypothesise, in the spirit of Rodrik (2018), that modern technologies are diffusing throughout GVCs, which are biased against (unskilled) labour. This might be driven by stringent requirements of lead firms.
Despite this less optimistic outlook on employment growth on average, it is evident that there are countries with fast employment growth, potentially through integration into GVCs. These countries may have also experienced falling labour requirements through technological change but they increased the scale of production to such a large degree that employment growth was still fast. In chapter 4, we explore this issue by studying 25 low and middle-income countries and delve into the proximate sources of job growth. We track three proximate sources with meaningful interpretations in the debate on job growth through GVC participation: technology, GVC competitiveness, and demand. We do so by applying a new decomposition method implemented with global input-output tables. To study low-income countries, we extend the existing World Input-Output Database (Timmer et al., 2015b) by four countries from Sub-Saharan Africa and three from South-East Asia between 2000 and 2014.
With the technology term in the decomposition, we speak to Rodrik (2018) and chapter 3, which argue that the diffusion of production technologies through GVCs moderates employment growth. Essentially, employment opportunities are expected to be limited because firms need to adopt production technologies that require less labour. We measure this by the labour requirements (number of workers) per unit of value added. If the hypothesis is true, employment generation through GVC participation must instead come from enlarged scale of production. The scale of production in a GVC depends on the size of the end market for a country’s value-added exports (global demand) combined with a country’s income share in the GVC (our measure of competitiveness). A crucial feature of GVCs is that countries are linked to consumers through forward linkages that can span multiple countries (following Johnson and Noguera, 2012; 2017). As such, demand shocks can be transmitted through these linkages to final-goods producers but also to all (foreign) upstream suppliers of intermediates (e.g., Bems et al., 2011). Being integrated into GVCs delivering to fast-growing markets thus might contribute to cross-country differences in job growth. We capture this demand effect by mapping a country’s value added to consumption of final goods in specific end markets. Importantly, a crucial difference to analysing gross exports is that one has to measure the end market and cannot rely on a country’s direct export partner. The scale of production is further moderated by the success of a country to capture income in that particular chain. We measure
21
a country’s income in a GVC as a share relative to all other countries. This so-called GVC income share is introduced as a measure of GVC competitiveness by Timmer et al. (2013). Several studies argue that the most successful countries, such as China, have managed to become major production centres supplying labour-intensive production stages to many GVCs, and therefore to concentrate jobs (e.g., Gereffi, 2014; Haraguchi et al., 2017; Kee and Tang, 2016). The flipside is that many other countries might have low or even declining income shares in GVCs, generating low job growth.
We find that the demand term is positive for all 25 low and middle-income countries. Interestingly, it seems not necessary to be integrated into GVCs that deliver to Asian countries with particularly fast expenditure growth (e.g., China). Kenya’s jobs, for example, are to a large degree dependent on demand growth in the domestic market, which has grown particularly fast. Vietnam, on the other hand, is very dependent on foreign demand, and benefitted strongly from expenditure growth in Asia. The country with the lowest contribution of demand growth is Mexico, which suffered from its dependence on North America with slow expenditure growth. The technology term contributed negatively to job growth in all countries in our sample. In many cases, the reduction of labour requirements fully offset the positive contribution of demand expansion, in line with Rodrik’s hypothesis. GVC competitiveness, however, seems to vary largely across countries. We find that some countries experience tremendous improvements, while other developing countries even experience declining competitiveness. Countries like China and Vietnam have very fast increases in GVC competitiveness, again counteracting the declining labour requirements and thereby generate fast job growth, in particular in manufacturing. Overall, diverging patterns of GVC competitiveness seem to be a driving force of cross-country differences in (manufacturing) job growth, which means a better understanding should be high on the research agenda on structural change and job growth through GVC participation.
In chapter 5, we investigate the role of one specific trade policy, namely trade facilitation, for sectoral value-added generation when GVCs are present. Trade facilitation has the broad goal of facilitating goods trade across borders through simplification and harmonisation of rules, regulations and infrastructure across countries. It is promoted by international organisations, such as through the WTO’s trade facilitation agreement, as a way to lower trade barriers, especially as many developing countries already benefit from preferential access agreements. A key issue in the evaluation of such initiatives is that traditional analyses of trade policies, and trade facilitation in particular, are based on gross exports only (e.g., Djankov et al., 2010;
22
Freund and Rocha, 2011). This may yield misleading results on the net impact and on the distribution of impacts across sectors. Gross exports overstate the expected net impact compared to value-added effects because countries typically require imports to exports. Secondly, GVCs open opportunities of third-country gains, as GVCs link firms through forward linkages to consumer markets in multiple countries, as pointed out in chapter 4. Hence, if trade costs are reduced in one country, this may affect all (foreign) upstream suppliers. Lastly, analyses based on gross exports misrepresent the sectoral distribution of the impacts because it omits indirect suppliers to exports. This sectoral dimension is particularly important in developing countries with respect to discussions on structural change.
To address this, we use a two-step approach following Vandenbussche et al. (2019). We first obtain sectoral trade elasticities with respect to trade facilitation, measured by the time it takes to import and export. Based thereupon, we predict gross-export effects and derive sectoral value-added effects using the global input-output system. Sectoral trade elasticities are important because they determine which exports are stimulated and thereby affect a country’s value added. Moreover, they affect a country’s value added through stimulated third-country exports. For example, we do not find evidence that exports of processed food and textiles are stimulated by trade facilitation. This is important for some of the studied low-income countries because they are relatively strong exporters of those sectors, and because they have strong forward linkages into precisely those sectors.
We investigate these effects for eight low and lower-middle income countries in Africa and Asia. We find large potential value-added effects from trade facilitation for global improvements to best practices. Yet, we find that in particular Ethiopia and Kenya miss out on additional benefits as they do not link into stimulated exports of other countries. Moreover, in terms of sectoral implications, we find trade facilitation not to be a driver of structural change into manufacturing per se. In particular in the set of African countries, we find that the stimulated sectors are mostly agriculture and services (in particular wholesale and retail trade; excluding business services).
23
1.3 Outlook
This thesis studies several issues on the relation of GVCs to economic development, and shows several results that can be helpful for understanding its role in and for economic development. Key discussions in the development literature centre on job growth (in manufacturing), and long-run productivity and value-added generation. GVC participation seems to be positively correlated to GDP per capita (chapter 2) and to productivity growth in formal manufacturing (chapter 3), findings which hold across relatively large sets of countries and over long time periods. These associations might be driven by efficiency gains of specialisation, by the diffusion of technology within GVCs, and by learning based on the relationships between participating firms. Yet, we find no evidence of faster employment growth in formal manufacturing on average (chapter 3), and we find that there are large differences in job growth across countries due to their performance within GVCs (chapter 4). GVCs thus appear not to be a driver of jobs and structural change per se, but a number of top performers engaging in GVCs appears to benefit greatly. One might thus hypothesise that GVC participation is necessary for structural change but not sufficient. We further find that trade facilitation, which is a popular tool to foster trade and growth of developing countries, benefits countries very heterogeneously. GVCs require to look beyond gross export effects to understand the net impact and the sectoral implications of trade policies. Yet, GVCs also open new avenues for value generation, namely through the integration through forward linkages. Whether countries potentially benefit from trade policies depends on their export specialisation and on their forward integration into GVCs, which can make a sizeable difference (chapter 5). Benefitting from integration into GVCs thus requires designing trade policies that take a country’s peculiarity in this regard into account.
The topics in thesis are clearly not exhaustive, and several future research areas are of importance. While each chapter in the remainder of the thesis lays out a set of future research questions in a specific field, I would like to highlight three broader directions here.
As pointed out, structural change plays an important role in debates on economic development. Yet, I would like to encourage the discussion on how to think about structural change when GVCs are present. Initially, structural change was thought of as a shift of labour and other production factors into relatively more productive, ‘modern’ activities, which have characteristics that make them conducive for economic development (e.g., Lewis, 1954). Typically, this is conceptualised as a shift into the manufacturing sector (e.g., Chenery et al.,
24
1986; Kuznets, 1966; 1971). According to Szirmai and Verspagen (2015), the main reasons for manufacturing to be a driver of economic development are its relatively high productivity, the scope for capital accumulation, economies of scale, embodied and disembodied technological progress, linkages and spillover effects into other domestic activities, and Engel’s law that relative demand for manufactured goods will rise as countries become richer. Rodrik (2016; 2018) also emphasises its potential to absorb (unskilled) labour. Yet, a trend is observed that many developing countries do not manage to industrialise or that they even deindustrialise at relatively low levels of GDP per capita (Rodrik, 2016). Instead, it is observed that some countries directly shift into services with the emergence of so-called consumption cities (e.g., Diao et al., 2018a; 2018b; 2019). To take account of such developments, it is sometimes argued that other sectors, such as industrial agriculture or business services, may take manufacturing’s place and perhaps it is necessary to enlarge the definition of the modern sector (e.g., Page, 2012; Newfarmer et al., 2019).
Yet, the literature still relies on a very traditional concept of the sectors in the economy, while several studies have shown that GVCs may blur the boundaries of sectors due to offshoring and outsourcing. Manufacturing firms may outsource or offshore non-core activities and thus often do not perform traditional manufacturing activities anymore but mainly services, such as branding (e.g., Bernard and Fort, 2013; Crozet and Milet, 2017; Kelle, 2013). In offshoring destinations, manufacturing might only consist of production activities (assembly, fabrication) and not include any pre- and post-production. Overall, the size of the manufacturing sector (or of any other sector definition of the modern sector) in a country may thus tell us relatively little about whether the activities in that sector have characteristics associated with the ‘modern’ sector, and to what extent those resemble what used to be bundled within manufacturing. It is an interesting future research area to investigate whether manufacturing in general or in specific countries does still fulfil those properties when GVCs are present. For example, one might argue that linkages and spillover effects have become weaker with GVCs because firms may not have domestic but foreign backward linkages. A starting point to investigate this is to measure the type of activities that are performed within a sector (e.g., by occupations or tailored surveys; see Sturgeon and Gereffi, 2009; Timmer et al., 2019) and to investigate the characteristics of those activities (e.g., linkages to other sectors, technology adoption; capital accumulation). These results would inform the debate on structural change whether using traditional sectors is still suited or whether a more activity-based view needs to be adopted.
25
Secondly, I would like to highlight the role of informal firms in GVCs, as informality is highly prevalent in many developing countries. De Vries et al. (2012) document that about 80% of Indian manufacturing employment is informal and that it was in fact increasing between 1993 and 2004. Moreno-Monroy et al. (2014) document that this trend was due to outsourcing from formal activities to informal ones. Such trends might relate to countries’ performance in GVCs. Gereffi (2014) argues that success in GVCs might be due to the exploitation of informal firms, such that formal exporters thrive due to poor labour standards of their informal suppliers. Yet, there is also a recent literature that argues that the informal sector can be a vibrant part of the economy, which in fact shares many characteristics with the formal sector and fosters economic growth (e.g., Diao et al., 2018a; 2018b). One might thus also hypothesise that informal suppliers foster export success due to their capabilities rather than their low labour standards. Strong linkages from exporters to the informal sector might further be important because of potential technology and knowledge spillovers, and employment generation through exporting. Understanding the role of informal firms in GVCs would further open avenues to investigate how GVC participation relates to characteristics of those informal jobs or to formalisation rates, which is an important issue in fostering ‘good’ jobs (e.g., Rodrik and Sabel, 2019).
Currently, however, it is difficult to investigate the role of informal firms as suppliers to exporters and their role in GVCs. Official input-output tables are based on surveys of typically large, formal firms. The supply and use relationships shown in such tables are thus representative for those larger firms and not for small or even informal ones. Following up on this research agenda would require a country-by-country analysis. One approach would be to build up an input-output table from micro data that are representative for the universe of firms, including informal ones. Most developing countries do not yet have such surveys but Tanzania produced a useful survey that is nationally representative of all micro, small and medium-sized firms (e.g., used by Diao et al., 2018a; 2018b). This could be a starting point for such an exercise. While there are several other directions to include firm-level heterogeneity (firm size, export status, ownership), the inclusion of the informal sector seems to be one of the most pressing issues with respect to studying economic development.
Lastly, I would like to highlight the lack of understanding of the relational aspect of GVCs (as also pointed out by Antras, 2019; World Bank, 2020). The relational aspect has long been emphasised in the case-study literature but it has not yet been as prominent in economics. The focus has been on theorising about and measuring specialisation in GVCs. The theoretical literature has mostly focussed on determinants of GVC trade as market transactions, and the
26
empirical literature has attempted different ways of measuring the foreign value-added content in exports. The relational aspect has mostly been implicitly included by hypothesising about the channel through which GVCs may affect the domestic economy (e.g., learning). Including the relational aspect more explicitly, however, could provide new insights. For example, the relational characteristic of GVCs could in fact make GVC trade less sensitive to trade or factor-cost changes, once the GVC is established. Antras (2019) argues that, among other reasons, this might be due to relationship-specific investments and trade of intangibles between firms within GVCs. This could have important implications for theoretical modelling of GVCs. Incorporating the relational aspect explicitly could also provide novel insights on the type of GVCs or type of relations within GVCs that provide scope for learning and potentially better long-run opportunities for firms in developing countries.
First explorations of this idea can be found in studies on foreign affiliates, which typically investigate ownership decisions (e.g., Feenstra and Hanson, 2005). Producing in-house is the extreme example of relational transactions, as this setup fully circumvents the market. Gereffi et al. (2005) refer to this as hierarchical governance with full managerial control by the lead firm, where upgrading of suppliers is almost fully determined by the lead firm’s strategy. It is important to better understand also more intermediate cases between pure market transactions and full integration. As pointed out by Antras (2019), a first empirical approach to address this is to follow Martin et al. (2019) who recently developed a measure of relationship-stickiness measured by the length of firm-to-firm trade relationships for a sample of French firms. More generally, an empirical approach to investigate the bargaining power of participating firms would prove useful. Building thereupon and incorporating such ideas into economic analyses of GVCs would add to the understanding about the role of GVCs in and for economic development.
Chapter 2
Patterns of Vertical Specialisation in Trade:
Long-run Evidence for 91 Countries
Abstract
We estimate the domestic value-added content in exports of manufacturing goods (VAX-D ratio) for 91 countries over the period from 1970 to 2013. We find a strong decline in the world VAX-D ratio since the mid-1980s mostly accounted for by the substitution of foreign for domestic intermediates. Using a break-point detection method, we identify three waves of vertical specialisation in the world economy: 1970-79, 1986-95 and 1996-08. We find that most countries (79) initiated a period of vertical specialisation at least once. We find that the VAX-D ratio correlates negatively with GVAX-DP per capita, and that the negative slope is flattening out at higher levels of income.
29
2 Patterns of Vertical Specialisation: Long-run Evidence for 91 Countries
2.1 Introduction
Countries may specialise in particular stages of production, relying on imports of intermediate goods and services to produce for exports. This process is known as vertical specialisation in trade as proposed by Hummels et al. (2001). Using input-output tables to measure the import content of exports, they found that vertical specialisation increased over the period 1970-1990 in 13 out of the 14 countries studied. Yi (2003) showed how the increased interdependence of countries can have major implications for trade policy, for example through cascading effects of import tariffs and other types of trade protection. Relatedly, Johnson and Noguera (2012, 2017) introduced a new metric that measures the value added of a country that is absorbed abroad (expressed as a ratio of gross exports). Based on a panel dataset of 42 OECD countries and major emerging markets, Johnson and Noguera (2017) documented a decline in this ratio for almost all countries over the period 1970-2008, interpreted as a widespread process of production fragmentation in the world economy. At the global level, the ratio was falling roughly three times as fast during 1990-2008 compared to 1970-1990.
This chapter contributes to the literature on vertical specialisation in two ways: methodologically and empirically. Johnson and Noguera (2017) studied developments in a set of mainly rich and middle-income countries. In this chapter, we provide new evidence on trends in vertical specialisation in trade for a large set of 91 countries at various stages of development, including many low-income countries, for the period 1970-2013. We also extend the sectoral detail in the data (19 detailed industries, up from 4 broad sectors as in Johnson and Noguera, 2017). This puts higher requirements on the data, but improves the measurement and opens the avenue for studying vertical specialisation in the production of particular manufacturing product groups. We track vertical specialisation in trade through the share of domestic value added in gross exports, which we refer to as the VAX-D ratio.3 We focus on the exports of manufactured goods which includes value added in the exporting sector, as well as value added from other domestic sectors that contribute through backward linkages. These can be other manufacturing industries, but also non-manufacturing industries delivering primary materials or business and support services. We follow Bai and Perron (1998, 2003) and estimate structural breaks in the time series of VAX-D ratios to identify periods of vertical specialisation and
3 This share is equal to (one minus) the import content of export measure of Hummels et al. (2001), see section 2.2 for discussion.
30
vertical integration. Thus, we are able to provide an overview of long-run trends in vertical specialisation in trade for a wide set of countries and explore possible correlates, in particular GDP per capita.
Our methodological contribution is in elucidating the difference between an indicator that tracks vertical specialisation, as defined by Hummels et al. (2001), and an indicator that tracks value added absorbed abroad as defined by Johnson and Noguera (2012, 2017). The latter has been developed as an alternative measure of exports that fits international trade models that are written in value added terms rather than gross flows (Johnson, 2014), which we refer to as VAX-C (following Los and Timmer, 2018). Its measurement is built upon tracing forward linkages rather than backward linkages, which are central in the concept of vertical specialisation and picked up in VAX-D. The difference between VAX-D and VAX-C measures is not only conceptually, but also empirically relevant. They are quantitatively comparable at the level of aggregate exports, but not at the sectoral level. This is further discussed in section 2 and also highlighted in our results.
The remainder of the chapter is structured as follows. In section 2.2, we discuss the calculation of our main indicator, the VAX-D ratio. We also present our approach to the estimation of structural breaks and the identification of periods of vertical specialisation. In section 2.3, we discuss the construction of our dataset. A novelty in our empirical strategy is in using untapped data of value added and gross output in manufacturing in developing countries at a high sectoral disaggregation and annual frequency (UNIDO, 2016). This is combined with detailed trade data from Feenstra et al. (2005) and benchmark input-output tables. We present our main findings in section 2.4. Section 2.5 concludes.
2.2 Methodology
In this section, we first outline our measure of vertical specialisation in trade. Next, we discuss our methodology to identify structural breaks in time series of this measure following the techniques introduced by Bai and Perron (1998, 2003).
31
2.2.1 Measuring vertical specialisation
To track vertical specialisation in trade, we measure domestic value added in exports as introduced by Koopman et al. (2012). We follow the terminology of Los and Timmer (2018) and refer to it as the VAX-D ratio. For a particular country, it is defined as
𝑉𝐴𝑋𝐷𝑟𝑡 =𝑉𝐴𝑋𝐷𝑡
𝑒𝑡𝑡𝑜𝑡 ,
(1) where 𝑒𝑡𝑡𝑜𝑡 is the sum of exports, and t a time-subscript. This ratio is bound between zero and one and a lower value indicates a higher level of vertical specialisation in trade. VAX-D is the domestic value added in exports measured as
𝑉𝐴𝑋𝐷𝑡 = 𝐯𝐭′(𝐈 − 𝐙𝐝𝐨𝐦,𝐭)−𝟏𝐞𝐭,
(2) where 𝐞 is a (column) vector of gross exports by industry. There are n industries so 𝐙𝐝𝐨𝐦 is an 𝑛𝑥𝑛 matrix of direct domestic input coefficients. Elements zij of this matrix denote the amount
of inputs from domestic industry i needed to produce one unit of output in industry j. Further, I is the identity matrix and (𝐈 − 𝐙𝐝𝐨𝐦,𝐭)−𝟏 the well-known Leontief inverse such that (𝐈 − 𝐙𝐝𝐨𝐦,𝐭)−𝟏𝐞𝐭 denotes gross output in all domestic industries that is needed for the production of 𝐞𝐭. It accounts for the fact that the production of a good needs intermediates,
which themselves are also produced making use of intermediates, etcetera. The Leontief inverse summarises all prior production steps as it can be written as a geometric series: (𝐈 − 𝐙𝐝𝐨𝐦,𝐭)
−𝟏
= 𝐈 + 𝐙𝐝𝐨𝐦,𝐭+ 𝐙𝐝𝐨𝐦,𝐭𝟐 + ⋯ + 𝐙𝐝𝐨𝐦,𝐭∞ , under the assumption that the production
technology as represented by Z is the same in all stages of production. To find the domestic value added related to the production of exports, one needs to multiply industry output by the transpose of (column) vector 𝐯𝐭, with element vi the value added over gross output ratio in
industry i. In our empirical analysis, we focus on domestic value added generated in the production of manufactured exports and exclude exports from mining and agriculture. This is because fragmentation in production of these goods is difficult as by nature they contain a large share of location-bound inputs. Thus 𝐞𝐭 contains zeros in all non-manufacturing entries.
Our measure for vertical specialisation is almost identical to the one introduced in the seminal work by Hummels et al. (2001). They proposed to track the import content of exports. Koopman
32
et al. (2012) defined domestic value added in exports and showed that it is equal to gross exports minus the import content of exports. We follow the value added terminology as it has a clearer link with other measures of trade (Johnson, 2018; Los and Timmer, 2018). VAX-D is related, but different, from the well-known VAX-C measure introduced by Johnson and Noguera (2012). C tracks the amount of value added in a country that is absorbed abroad. VAX-C was developed as a measure of trade in value added (Johnson 2014). At the aggregate level, VAX-C and VAX-D are equal when the exports of a country consist of final goods only. As most countries also export intermediates, VAX-C is typically lower than VAX-D.4 The numerical difference appears to be generally small as shown in supplementary material section 3, indicating that the share of value added exported through intermediates and returning home is minor. The difference between VAX-C and VAX-D is not necessarily small for sector-level measures however.5 VAX-D in manufacturing exports captures all domestic value-added in
products exported by the manufacturing sector. This value added is generated in the production chain that includes the manufacturing industry that exports, but also other manufacturing and non-manufacturing industries (such as agriculture, mining and services). In contrast, the manufacturing VAX-C measure of Johnson and Noguera (2017) captures how much value added is generated in the manufacturing industry that is ultimately absorbed abroad, embodied in exports by all industries.6 Put otherwise, while the measurement of VAX-C is based on tracing forward linkages in the use of manufacturing value added, VAX-D is based on tracing backward linkages in the production of manufacturing exports (see Los and Timmer, 2018 for further discussion).7 This is a major conceptual difference. Hence, the choice of indicator depends crucially on the purpose of the study. VAX-C, for example, lends itself to study demand spillovers through input-output linkages (e.g., Bems et al., 2011). The process of vertical specialisation in trade as described by Hummels et al. (2001), however, is about the
4 See Koopman et al. (2014) and Los et al. (2016) who provide a full decomposition of gross exports, encompassing the concept of VAX-D and VAX-C.
5 VAX-C and VAX-D measures also differ for bilateral flows as shown in Los and Timmer (2018). This should not come as a surprise as the two measures have different aims.
6 A sectoral VAX-C ratio can be bigger than one when the sector exports mainly value added through other sectors (see e.g. Table 1 in Johnson and Noguera, 2012). A sectoral VAX-D can never be bigger than one, as domestic value added in an export flow can never be bigger than the export flow itself.
7 To construct VAX-C for our set of countries, we would need to construct an integrated multi-region input-output table rather than a set of national input-input-output tables. Put otherwise, we would need to add information on the country-industry destination of a country’s exports. Only then one can trace where a country’s value added is ultimately absorbed. This would require additional bilateral trade data and add another layer of
complexity to the data construction process (including balancing of trade mirror flows) which we did not attempt here. Using the existing WIOD world input-output tables, we find that the VAX-D and VAX-C measures correlate highly for aggregate exports, see supplementary material section 3.
33
fragmentation of backward linkages in the production of exports, and therefore we use VAX-D as our measure of vertical specialisation in trade.
2.2.2 Identifying periods of vertical specialisation
We define a period of vertical specialisation as a period in which there is a significant trend decline in the VAX-D ratio. To this end, we follow Bai and Perron (1998, 2003) and identify structural breaks in the time series for each country. We proceed in two steps. Firstly, a given maximum number of potential structural breaks is identified in a time series, and secondly the actual number is selected by testing statistical significance of each break.
Let m be a predefined maximum number of structural breaks in a time series. The time periods in between the breakpoints are called ‘regimes’, and are indexed by i.8 We will estimate a trend in a given regime i by
∆𝑉𝐴𝑋𝐷𝑟𝑡(𝑖) = 𝛼𝑖 + 𝜀𝑡(𝑖),
(3) where 𝑡(𝑖) indicates year t in regime i, ∆𝑉𝐴𝑋𝐷𝑟 the first-differenced VAX-D ratio (annual observations), 𝛼𝑖 is a regime-specific constant and 𝜀𝑡(𝑖) the error term, which is allowed to have different distributions across regimes. This is a pure structural change model in which all parameters vary with regimes. To locate break years 𝑇𝑖 (the last year of regime i), the following
sum of squared residuals is minimised,
∑𝑚+1𝑖=1 ∑𝑡(𝑖) (∆𝑉𝐴𝑋𝐷𝑟𝑡−𝛼̂𝑖)²,
(4) with 𝛼̂𝑖 the estimated parameter obtained from equation 3 and t(i) running from Ti-1 + 1 through
Ti . The number of breaks m is set before the estimation (we start with 5). In addition one has
to choose a minimum length h of a regime. Our choice is guided by the aim to capture long-term developments rather than business cycle fluctuations. We start with h is 5 years. So 𝑡(𝑖) is endogenously pinned down (with minimum distance h) for a given set of breaks.
8 So there are m+1 regimes, including a begin period (up to first breakpoint) and an end period (from last breakpoint to end of period).
