[Paper Review] Bridging the short-term and long-term dynamics of economic structural change

James McNerney, Yang Li+2 more|arXiv (Cornell University)|Oct 18, 2021

Economic and Technological Innovation4 citations

TL;DR

This paper proposes a unified dynamical model based on the Principle of Relatedness—where economies diversify into related activities—to bridge short-term economic diversification and long-term structural change. It shows that diversity and compositional shifts in export baskets, captured by two derived coordinates, naturally give rise to classic structural transformation patterns and align closely with established complexity metrics like ECI and PCI, reinterpreting these metrics as summaries of long-term development dynamics rather than static complexity measures.

ABSTRACT

Economic transformation -- change in what an economy produces -- is foundational to development and rising standards of living. Our understanding of this process has been propelled recently by two branches of work in the field of economic complexity, one studying how economies diversify, the other how the complexity of an economy is expressed in the makeup of its output. However, the connection between these branches is not well understood, nor how they relate to a classic understanding of structural transformation. Here, we present a simple dynamical modeling framework that unifies these areas of work, based on the widespread observation that economies diversify preferentially into activities that are related to ones they do already. We show how stylized facts of long-run structural change, as well as complexity metrics, can both emerge naturally from this one observation. However, complexity metrics take on new meanings, as descriptions of the long-term changes an economy experiences rather than measures of complexity per se. This suggests relatedness and complexity metrics are connected, in a hitherto overlooked way: Both describe structural change, on different time scales. Whereas relatedness probes transformation on short time scales, complexity metrics capture long-term change.

Motivation & Objective

• To unify two branches of economic complexity research: diversification dynamics (short-term) and complexity metrics (long-term).
• To investigate how the Principle of Relatedness—diversification into related activities—gives rise to long-term structural change.
• To reinterpret complexity metrics (e.g., ECI, PCI) not as static measures of complexity, but as summaries of long-term developmental trajectories.
• To validate the model using 56 years of global export data across ~250 countries and regions.

Proposed method

• Formalizes the Principle of Relatedness as a stochastic dynamical model of economic diversification using a proximity matrix between economic activities.
• Applies eigenmode decomposition to the model's transition matrix to extract dominant modes of change, yielding two key coordinates: one for diversity and one for compositional shift.
• Derives a modified Economic Complexity Index (ECI*) and Product Complexity Index (PCI*) based on the model’s principal components, using a non-orthogonal basis to reflect real-world network structure.
• Uses data on global trade (RCA-based proximity matrices) to compute ECI* and PCI* across multiple proximity definitions (probability-based, minimum conditional probability, correlation-based).
• Employs variance decomposition to assess how much of the data variability is explained by the derived coordinates, validating their empirical relevance.
• Compares ECI* and PCI* to conventional ECI and PCI across different proximity matrices using Pearson and Spearman correlations.

Experimental results

Research questions

• RQ1 How can the Principle of Relatedness—diversification into related activities—generate long-term structural change in economies?
• RQ2 To what extent do complexity metrics like ECI and PCI emerge naturally from a dynamical model of related diversification?
• RQ3 Do the two dominant modes of change identified in the model (diversity and composition) correspond to well-known stylized facts of economic development?
• RQ4 How do ECI* and PCI* compare numerically to conventional ECI and PCI across different proximity matrix constructions?
• RQ5 Can the model’s derived coordinates serve as interpretable summary measures of long-term structural transformation?

Key findings

• The model’s two dominant coordinates—diversity and compositional shift—capture the well-documented pattern of countries diversifying into more products while shifting from agriculture to manufacturing.
• The variance explained by the ECI* coordinate peaks at R₀ = 0.115, indicating an optimal scale for capturing long-term structural dynamics.
• Pearson and Spearman correlations between ECI* and conventional ECI exceed 0.90 across all proximity matrices, with the highest correlation (0.919) when using the probability-based proximity matrix Φᴾ.
• PCI* and PCI show near-identical values when using Φᴾ, confirming theoretical equivalence up to a constant factor.
• The model demonstrates that complexity metrics are not static measures but dynamic summaries of long-term developmental change.
• The framework reinterprets complexity metrics as descriptors of long-term structural evolution, not intrinsic complexity, revealing a hitherto overlooked link between relatedness and complexity.