Economic Complexity Index
Also known as: ECI
Hausmann and Hidalgo's measure of national productive capability from export structure.
The Economic Complexity Index, developed by Ricardo Hausmann and César Hidalgo with collaborators, measures the productive knowledge embedded in a country's economy by analyzing the structure of its exports — specifically the diversity of products it exports competitively and the ubiquity of those products (how many other countries also export them competitively). The intuition is that diverse exports of relatively non-ubiquitous products signal the presence of broad productive capabilities that are hard to acquire and hard to displace. The original 'method of reflections' algorithm iteratively refines country and product complexity scores until convergence; subsequent variants use eigenvalue decomposition for the same logic. ECI has demonstrated empirical predictive power for long-run GDP per-capita growth — countries whose complexity exceeds what their current income predicts tend to grow faster — and underpins the related Product Space and Atlas of Economic Complexity tools used in development-policy diagnostics.
Core components
- Revealed comparative advantage (RCA) per product
- Diversity (count of products exported with RCA above 1)
- Ubiquity (count of countries exporting a product with RCA)
- Method of reflections / eigenvector approach
- Country complexity (ECI) and product complexity (PCI)
- Product Space network
- Relatedness and density measures
Primary use case
Long-run growth diagnostics at the country level; industrial policy and export-diversification strategy; identifying nearby products a country could feasibly add; tool for development institutions (World Bank, IDB) and national planning ministries.
Common criticisms
- Reliance on goods-export data excludes services (now a large and growing share of trade) and the informal economy
- sensitivity to product classification (HS codes) and revealed-comparative-advantage threshold
- algorithmic choices (method of reflections vs eigenvector formulation) yield slightly different rankings
- some critics (e.g., Kemp-Benedict) argue ECI is largely a relabeling of diversity and information already in standard trade indicators
- predictive performance varies across time periods and country samples.