Prema-chandra Athukorala, Australian National University: This paper undertakes a comparative analysis of production networks in Europe and East Asia in the global context, with emphasis on determinants of export patterns and intra-regional linkages. The findings suggest that: exports of intermediate goods (used as a proxy for parts and components or “middle products” exchanged within production networks) are more sensitive to trade barriers than total goods exports; regionalization of production networks is primarily determined by greater sensitivity of intermediate goods trade cost associated with distance that overwhelms cost advantages arising from advances in transportation and communication technology; and the production network in these two regions are being guided by distinct factors with geographic distance and remoteness playing a more significant role in East Asia than in the EU. Overall, this is an interesting paper on an important subject. However, I would like to suggest that the authors pay attention to the following methodological imitations in their further research.Trade within global production networks (“network trade”) is based on “global production sharing”—breakup of the production process of a given product into vertically separated stages/tasks carried out in several countries. Therefore, network trade essentially has two components: trade in parts and components (involving the “middle layer” of the production process) and trade in final assembly (Jones and Kierzkowski 2001; Obashi and Kimura 2018). The empirical analysis of this paper uses “intermediate goods” (delineated using the UN Broad Economic Classification Broad Economic Categories [BEC] at the 3-digit level of commodity classification) rather than parts and components and completely ignore final assembly.BEC-based intermediate goods include both parts and components and the standard intermediate goods such as iron and steel, industrial chemicals, and metal products. These two categories of intermediate goods have different characteristics as traded goods. Unlike standard intermediate goods, which are normally traded based on market-determined prices, trade in parts and components are mostly characterized by “relationship specificity” among the user firms (Athukorala and Khan 2016). Therefore, the former is not a good proxy for latter. In addition, data relating to intermediate goods (based on the BEC classification) significantly overstate the magnitude of bilateral trade links within global production networks of resource-rich countries, such as Malaysia and Indonesia.According to my calculations, final assembly accounts for 53.2 percent of total world network exports in 2006–07 (two-year average) (Athukorala 2014). More importantly, this share varies notably among countries: For instance, final assembly bulks large in network exports from China because it is the world's finally assembly center, where as parts and components (mostly exported to China) account for much of networks exports from the other countries in the region. Given this fundamental structural feature of production sharing, ignoring final assembly significantly underestimate the relative importance of network trade in global manufacturing but also the underestimation bias varies significantly among countries, giving rise to a bid problem in econometric analyzing of trade patterns using multi-country data, as has been done in this paper.I believe that the finding of this paper relating to regional rather than global integration of East Asian production networks is a statistical artefact resulting from the omission of final assembly from the commodity coverage of network trade. China accounts for more than half of final assembly exports from the region and more than 70 percent of these exports are to markets in North America, Europe, and the rest of the world. If network trade is measured appropriately, capturing both parts and components and final assembly, you will find a China-centered global (rather than regional) production network in Asia. The other Asian countries export parts and components to China, and China exports final assembly largely to the rest of the world.The empirical analysis of the paper is a direct application of the gravity equation specified by Anderson and van Wincoop (2003) to analyze determinants of “total” trade flows. No attempt has been made here to modify it to suit analyzing trade flows within global production networks informed by the theory of global production sharing (Jones 2000; Jones and Kirzkowski 2001; and many other papers). The “missing prince from Hamlet” is the “lead firm” (multinational enterprise [MNE]). There is no variable in the model to capture the pivotal role of MNEs within production networks. The standard variable used for this purpose is foreign direct investment. The production location decision of the lead firm is determined by the “relative cost” (mostly relative wages) among countries and the “service link cost” associated with undertaking different tasks of the production process and smoothly linking/coordinating these tasks. There is no relative cost variable in the model. I think per capital gross national income is a reasonably good proxy, if data on manufacturing wages are not available. Inclusion of most of the other variables (distance, remoteness, colonial links, RTAs, common language) can be theoretically justified as capturing “service link cost.”
Read full abstract