trade · macro
Export destination concentration: Gini, HHI and entropy
How concentrated a country's exports are across the places they go, measured three ways that disagree. Concentration is the quantity behind every argument about trade dependency, and which measure you pick changes who looks vulnerable.
Definitions and method
Computed in app/services/network.py from the bilateral trade already on this site, for every reporting economy rather than one region. For each exporter and year, shares are taken over destinations, self-trade and aggregate counterparts removed, and countries with fewer than five destinations dropped because every index is noise on a handful of partners. HHI is the sum of squared shares on a 0-10,000 scale and is dominated by the largest partner; it also falls mechanically as partners are added, so a normalised form is given beside it. Entropy is Shannon's, in nats, weighting the tail far more generously; entropy_norm divides by ln(n). Gini measures inequality across destinations rather than dominance of the largest. Read all three: where they disagree is where the interesting structure is.
| Indicator | Code | Unit |
|---|---|---|
| Shannon entropy of destination shares | dest.entropy | nats |
| Entropy, normalised for the number of destinations | dest.entropy_norm | 0-1 |
| Gini coefficient across destinations | dest.gini | 0-1 |
| Herfindahl-Hirschman index of destination shares | dest.hhi | index, 0-10000 |
| HHI, normalised for the number of destinations | dest.hhi_normalised | 0-1 |
| Destinations with a recorded flow | dest.partners | partners |
| Largest destination's share | dest.top_share | % of exports |