Mapping intra firm trade in the automotive sector: a network approach

TL;DR

This paper introduces a multilevel network approach to map and analyze intra-firm trade within the automotive sector, combining firm, country, and trade data to uncover underlying trade patterns and determinants.

Contribution

It develops a novel multilevel network methodology with motif detection and ERGM analysis to study intra-firm trade, addressing data scarcity in global trade patterns.

Findings

Identifies potential intra-firm trade ties using motif detection.

Analyzes determinants of intra-firm trade with ERGM models.

Constructs detailed country-level intra-firm trade networks.

Abstract

Intra-firm trade describes the trade between affiliated firms and is increasingly important as global production is fragmented. However, statistics and data on global intra-firm trade patterns are widely unavailable. This study proposes a novel multilevel approach combining firm and country level data to construct a set of country intra-firm trade networks for various segments of the automotive production chain. A multilevel network is constructed with a network of international trade at the macro level, a firm ownership network at the micro level and a firm-country affiliation network linking the two, at the meso level. A motif detection approach is used to filter these networks to extract potential intra-firm trade ties between countries, where the motif (or substructure) is two countries linked by trade, each affiliated with a firm, and these two firms linked by ownership. The motif detection is used to extract potential country level intra-firm trade ties. An Exponential Random Graph Model (ERGM) is applied to the country level intra-firm trade networks, one for each segment of the automotive production chain, to inform on the determinants of intra-firm trade at the country level.