Functional division of labour and value capture in global value chains: a new empirical assessment based on FDI data

Abstract

This work provides a new empirical assessment of global economic hierarchies and the associated unequal distribution of value between core and peripheral economies. This is accomplished by looking at the functional division of labour induced by the international fragmentation of production and the related value capture dynamics in global value chains (GVCs). To this aim, we introduce and compute an indicator of ‘functional specialization’ of economies based on their ability to attract Foreign Direct Investment, which allows us to detect the value adding activities in which more than 100 countries have specialized from 2003 to 2018. We show that the most intangible-intensive activities are concentrated in core capitalist economies, while production operations at the lower end of the value chain are mainly the prerogative of low- and middle-income countries. Although China and India have emerged as partial but significant outliers, a substantial persistence of this functional division of labour across world macro-regions is also observed over the period. Most notably, we find that a higher specialization in the most intangible-intensive functions allows countries to capture a greater amount of value from trade in GVCs, thus providing novel empirical support to the ‘Smile curve’ hypotheses and the underlying ‘intellectual monopoly’ perspective.

Keywords:

• Global value chains
• foreign direct investment
• functional specialization
• intangible assets
• smile curve

JEL CODES:

• F14
• F21
• O19

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 In a similar vein, a pioneering contribution by Strange (1994) introduced the concept of ‘knowledge structure’ to highlight the mechanisms by which structural power can be exercised by those who possess, can limit and decide, at least in part, the conditions of access to knowledge.

2 While classical examples of tangible assets are land, building and equipment, intangible assets are mainly constituted by patents, brands, customer data and software. More precisely, intangibles can be grouped in three broad categories, the first one being computerized information (such as software and databases); the second is made of innovative property (such as scientific and nonscientific R&D, copyrights, designs, trademarks); lastly, the third regards economic competencies, including brand equity, firm-specific human capital, networks joining people and institutions, organizational know-how that increases enterprise efficiency, and aspects of advertising and marketing (Corrado et al., 2005).

3 See Sell and May (2001) for a critical reconstruction of the key ‘moments’ in the history of intellectual property rights that led to the TRIPS Agreement.

4 This phenomenon can also be interpreted as a modern version of the Prebisch-Singer hypothesis (Prebisch, 1949; Singer, 1950), whereby increased competition in fabrication functions, mainly performed by low- and middle-income countries, leads to a deterioration in the terms of trade of manufactures, thus reducing the amount of value captured by these economies from trade in GVCs (Kaplinsky, 2000; Milberg & Winkler, 2013). From this perspective, Gimet et al. (2010) talked of ‘immiserizing specialization’, which recalls the concept of ‘immiserizing growth’ proposed by Bhagwati (761958). In fact, while technology transfer from core to peripheral economies can trigger productivity gains in the latter, strong competitive pressure in these value chain segments leads these productivity gains to be capitalized especially by leading firms acting as global buyers (Kaplinsky et al., 2002).

5 A short description of the fDi Markets database, including a brief discussion of the nature of the data, can be found in the Online Appendix. Further details on this dataset are reported in Coveri and Zanfei (2022).

6 The list of value chain functions included in the fDi Markets database is reported by Table A.1 in the Online Appendix.

7 FDI and pure market-based exchanges are conceptually at the two opposite poles of GVC governance types, the former being typical of hierarchical GVCs and the latter of GVCs in which exchanges take place primarily through inter-firm arm’s length transactions. Between the two, hybrid forms of GVC governance exist (Gereffi et al., 2005; Buckley & Strange, 2015; Strange & Humphrey, 2019). Focusing on FDI, our analysis mostly concerns GVCs of the former type.

8 The EU28 macro-region – being an aggregate of 28 formally independent high-income economies – shows a very high share of intra-region FDI, making comparisons with other world regions more difficult. Hence, for this macro-region we also compute the functional specialization net of intra-region FDI.

9 The complete list of countries by macro-region that are included in our analysis is reported by Table A.8 in the Online Appendix.

10 While we adopt the convention largely used in the GVC and smile curve literature according to which R&D is assumed to be an ‘upstream’ function, we are also aware that this is a simplistic way of articulating our discourse. Based on the seminal contribution by Landau and Rosenberg (1986) it has become apparent that in most circumstances innovative activities should be conceived a non-linear, chain-linked process wherein R&D often plays a key role in most if not all production and commercialisation stages along the whole value chain.

11 The classification of value chain functions according to the GVC stage they can be allocated to is provided by Table A.1 in the Online Appendix, while Table A.2 reports the values of the specialization in FDI of macro-regions across GVC stages.

12 Note that, as a further refinement of the dataset, we do not consider country-year observations for which the specialization indices are computed over a total number of inward FDI lower than three (this threshold is equal to the total number of inward FDI for the 25th percentile of the distribution of total inward FDI received by low- and middle-income economies, a large share of which draws very few or zero FDI per year). This is a fair adjustment in order to improve the reliability of the sample as it allows to avoid biases in the computation of the specialization indices; conversely, the latter would risk being driven by a very small number of total inward FDI for a series of country-year observations, with the result that some countries would report, in a given function, very large specialization indices in specific years and zeroes for all the remaining years. Also note that setting a higher, more ‘cautious’ threshold – as we will do in the econometric analysis (Tables 3, 4 and 5) – leaves our findings largely unchanged (results are available upon request).

13 Similarly, Timmer et al. (2019) find a very low (positive) correlation between GDP per capita and functional specialization in marketing activities.

14 According to the smile curve hypothesis, the functional specialization of economies is largely associated with their level of economic development. GDP per capita generally shows indeed a strong correlation with several aspects related to the skill intensity, technological capabilities and labour market conditions of economies, thus representing a good proxy for their overall position in the global economic hierarchy. However, GDP per capita of countries is certainly not the only aspect that explains their functional specialization and identifying all determinants of specialization of economies across GVC stages goes beyond the scope of this work. The different factors associated with value capture in GVCs will instead be considered in the empirical model described by equation (2).

15 The selected functions, together with a brief description, are reported by Table A.3 in the Online Appendix, while Table A.4 shows the values of the specialization of macro-regions in these functions. Note that we avoid focusing on activities which go under the head of ‘ICT & Internet infrastructure’ and ‘Business services’ notwithstanding the high share of FDI related to these functions. The reason is that they constitute rather broad and heterogeneous classes of activities aimed to provide a very large series of telecommunications and business support functions (Sturgeon, 2008).

16 As regards the index of specialization in logistic functions, a decreasing trend is found for China (Fig. A.2 in the Online Appendix, Panel A) while India shows fluctuating values but always well below one.

17 The RFS index differs from the production specialization index used by Stöllinger (2021) since the latter indicator does not account for the evolution of functional specialization, i.e., it does not vary over time.

18 These indicators are built by exploiting the EORA Multi-Region Input-Output tables and details on the methodology and comparisons with other value-added trade databases are reported in Casella et al. (2019).

Additional information

Notes on contributors

Andrea Coveri

Andrea Coveri holds a Ph.D. in Economics from the Marche Polytechnic University (Ancona, Italy). He is currently Assistant Professor of Applied Economics at the University of Urbino (Italy), where he teaches Economics of Globalisation and Global Political Economy. His research interests include global value chains, the economics of innovation and the theory of value, price and distribution.

Antonello Zanfei

Antonello Zanfei is Professor of Applied Economics at the University of Urbino (Italy), where he directs the Ph.D. in Global Studies. His research focuses on FDI, Trade, Global Value Chains, and innovation. He was visiting professor at: Stanford University; Université L. Pasteur, Strasbourg; ENST, Paris; Southbank University, London; Trinity College, Dublin; Universidad Complutense, Madrid; SPRU, Sussex University, and Roskilde University.