Organizing the global value chain: A firm-level test

Abstract

In this paper we study the organization of Global Value Chains on a sample of about 4,000 manufacturing parent companies operating more than 90,000 affiliates on a global scale, which chose to integrate at least once in the period [2004][2005][2006][2007][2008][2009][2010][2011][2012]. Assuming a technological sequence of production stages, a recent property rights framework (Antràs and Chor, 2013;Alfaro et al., 2015) predicts that a choice of vertical integration is crucially based on both the position of a supplier along the chain and on the relative size of demand elasticities faced by the final producer and the supplier. We positively test whether, if final demand is sufficiently elastic (inelastic), producers of final goods integrate production stages that are more proximate to (far from) the consumers. However, this is not valid for cases of midstream parents, i.e. for producers of intermediate inputs that can integrate either backward or forward along the chain. We document that midstream are at least as common as are downstream parent companies but that existing theory neglects them. In these cases, we find that demand elasticities do not play a significant role in integration choices. Interestingly, both midstream and downstream parents tend to integrate affiliates that are more proximate in segments of a supply chain. Our findings point to a role for technological determinants that may be as important as are contracting frictions in organizing Global Value Chains.

Measuring the Input Rank in Global Supply Networks

“…For sake of comparison with previous studies, we compute the Input Rank on U.S. Input-Output tables, sourced from the Bureau of Economic Analysis (US BEA, 2002), and world Input-Output tables, sourced from both WIOD and EORA. Finally, we test the correlation of the Input Rank with choices of vertical integration, in the fashion of Antràs and Chor (2013), Alfaro et al (2017), and Del Prete and Rungi (2017). On a sample of 20,489 U.S. parent companies controlling 154,836 affiliates worldwide, we find that a higher Input Rank is positively associated to higher odds that a (direct or indirect) input is vertically integrated, relatively more when final demand is more elastic.…”

“…In this case, the main prediction is that final-good producers integrate stages that are relatively more downstream (upstream) when final demand is sufficiently elastic (inelastic). However, when it comes to firm-level empirics, Del Prete and Rungi (2017) find that vertical integration choices are not always in line with theoretical predictions, as parent companies and affiliates locate not so far from each other along upstreamness/downstreamness segments. In this contribution, we build on a similar framework and find that a higher Input Rank is always associated with higher odds that that input is vertically integrated, even after controlling for the position on the downstreamness/upstreamness segments, and that proximate inputs on the network are more likely integrated than distant ones.…”

“…See also Rungi et al (2017). Similar data structures have been used in Alviarez et al (2016), Cravino and Levchenko (2017), Del Prete and Rungi (2017). 14 The Network Distance between any input and any output in the U.S. I-O tables is the minimum number of downstream linkages that connect them through 15 For similar mappings of firm-level sourcing based on input-output tables and industry affiliations, see Charlton (2009), Acemoglu et al (2010), Alfaro et al (2016), Rungi and Del Prete (2018). with industry-level estimates of demand elasticity sourced from Broda and Weinstein (2006), and with a measure of input contractibility retrieved from Antràs and Chor (2013).…”

“…For our purpose, we will make use of a dataset of U.S. parent companies that have integrated at least one production stage over time. Our empirical strategy explicitly takes on the theoretical framework by Antràs and Chor (2013), while augmenting the estimates by Del Prete and Rungi (2017) with the inclusion of the Input Rank. To validate our sample, we compare with official 'Data on Activities of Multinational Enterprises' (BEA, 2018) and OECD Statistics on Measuring Globalization (OECD, 2018).…”

The global value chain under imperfect capital markets

“…They find that whether a firm integrates upstream or downstream suppliers depends crucially on the elasticity of demand for the final product. Del Prete and Rungi () use a detailed firm‐level data set covering about 4,000 manufacturing parent companies and more than 90,000 affiliates in 150 countries and find supporting evidence for the property‐rights model. Using city‐level Chinese processing export data, Luck () provides evidence that is consistent with the theoretical predictions of Antràs and Chor ().…”