‘Governed’ trade: global value chains, firms, and the heterogeneity of trade in an era of fragmented production

ABSTRACT

Over the past several decades, firms have de-verticalized and internationalized increasingly complex manufacturing and service functions, a phenomenon studied across the social sciences. However, the disciplines disagree over whether the fragmentation of production is substantively novel, requiring amendments to trade theory, or is simply a secular deepening of the international division of labor. Some economists view it as ‘just trade,’ driven by well-known actor-less determinants, such as factor endowments, technology, and returns to scale, while more recent firm heterogeneity trade theories consider firm behavior. By contrast, other heterodox social science approaches differ by focusing on the strategic actions of firms and sector-specific governance as independent drivers which ‘govern’ trade and determine the division of value between countries. This paper develops novel measurements by utilizing unique transactional trade data – the raw firm-level trade transactions that comprise standard inter-country trade statistics – on 439 of China's largest exporters in 18 subsectors of the electronics and light industries, to examine whether trade is heterogeneously governed in ways theorized by the global value chain (GVC) literature. It finds substantial empirical support for GVC-governed trade, and advances both GVC and firm-centric trade theory along several fronts.

KEYWORDS:

• governance
• international trade
• global value chains
• firm heterogeneity
• vertical specialization
• China
• transactional trade data
• multinational enterprises (MNE)

ACKNOWLEDGEMENTS

Participants and commentators at Harvard Business School and the Fairbank Center for Chinese Studies at Harvard have helped shape these ideas. Union College provided research funding, and able research assistance was provided by Justin Bogardus, Justin Dempsey, and MarcAnthony Parrino.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Under the broader GVC rubric, I include literatures on global commodity chains (GCC) (Gereffi and Korzeniewicz, 1994), global value chains (GVC) (Gereffi, Humphrey, and Sturgeon, 2005), and global production networks (GPN) of the Manchester school (Coe, Dicken, and Hess, 2008), as well as in technology and knowledge diffusion (Ernst and Kim, 2002), four distinct approaches in active dialogue with each other.

2. There are many commercial sources to acquire transactional trade data, though they differ widely in content, quality, and cost. For US data, PIERS is the gold standard, but other companies, such as Panjiva, Datamyne, Import Genius, and others, also sell data (see Dallas, 2014a). This paper uses Panjiva data. However, for comprehensive and raw transactional trade data of many other countries – meaning they are not mediated through a third-party commercial entity, see www.impexp.com, and for China only, see www.e-to-china.com.

3. However, a class of ‘global suppliers,’ which follow lead firms wherever they go, can also lower barriers to entry for countries, but they pose a new competitive dimension to local suppliers (Appelbaum, 2008; Sturgeon and Lester, 2004).

4. Another firm-centric trade literature which is relevant but not directly examined in this paper considers the effects on trade of the relationship between suppliers and head quarter (HQ) firms, especially when ‘relationship-specific investments’ (RSI) are required between them. Using various measures of RSI, empirical tests show that an increase in RSI is associated with increased outsourcing; in other words, better contractibility between firms leads to more HQ internalization – a result contrary to standard transaction cost predictions (Antràs, 2014; Nunn and Trefler, 2013).

5. Furthermore, the property rights approach described in endnote 4 differs because it is largely concerned with defining the boundaries of firms, meaning that their dependent variables are dichotomous – either internalization or arm's-length outsourcing.

6. For another perspective on “emergence” in GVC, see Dallas (2014b).

7. Countries record transactional trade in very different ways, by including different firm-level and transactional variables, which could potentially differentiate relational from captive linkages.

8. Of course, each HS code contains multiple products, thus product diversification is possible within HS codes also.

9. EU-reported data are used because China does not report trade transactions to the EU using disaggregated HS codes. China's producer price index is used as the closest approximation for ex-factory inflation.

10. China uses a more conservative threshold of 25% equity for a firm to be considered an FIE. International standards are 10%.

11. Although no predictions were made for modularity, specialization is very high, averaging between 65% and nearly 90% for six of the subsectors, much higher than the averages of light industries. The only two exceptions are the computer parts and audio–video parts sectors.

12. Video game consoles (HS Code 9504 was equally volatile with a −25.8% average. Unfortunately, the transactional data were not complete enough to be included.

Additional information

Notes on contributors

Mark P. Dallas

Mark Dallas is an assistant professor of Political Science and Asian Studies at Union College in New York, and was an An Wang postdoctoral fellow at the Fairbank Center for Chinese Studies at Harvard University in 2013–2014, where this research took place. For more information about transactional trade data and the research in this article, please see his personal website www.markpeterdallas.com.