Patterns of Sectoral Structural Change – Empirical Evidence from Similarity Networks

Abstract

Decomposing growth processes has been identified as a key requirement for an improved understanding of mechanisms of structural change. This paper conducts a sectorally disaggregated, empirical analysis of structural change. We develop a network of inter-sector similarities on the basis of value-added data derived from multi-regional input-output tables. The similarity of sectors across countries serves as an expression of near-term sectoral composition and related changes in overall GDP. After filtering insignificant inter-sectoral links by applying a Monte-Carlo-based boot-strap approach the resulting sectoral network shows to be highly modular. Next to clearly identifiable communities, which largely resemble agricultural, industry and resource extraction sectors, we can identify certain directed "bridging sectors" – mainly light manufacturing sectors. We also show that the probability of sectoral shifts and their relation to overall GDP per capita are negatively related, suggesting that the bridges effectively are bottlenecks in the transition from the agricultural to the industrialized stage.

Keywords: structural change, sectoral composition, similarity network, product space, multi-regional input-output data

JEL Classification: O11, O14, C38, C55