Herfindahl-Hirschman Index in Commodities: How Supply Concentration Drives Mineral Volatility

HHI as a Structural Volatility Input

The Herfindahl-Hirschman Index sums the squared market shares of all producers in a given market. For a mineral with n producing countries each holding share sᵢ, the index is simply Σsᵢ². The resulting value ranges from near zero (perfectly fragmented supply) to 10,000 (single-producer monopoly). The U.S. Department of Justice classifies markets above 2,500 as highly concentrated.

In commodity markets, HHI operates differently than in antitrust analysis. The concern is not pricing power per se but supply fragility. A mineral with an HHI above 4,000 concentrates enough production in one or two jurisdictions that a single policy shift, export ban, or logistical disruption can remove a material fraction of global supply within days. This is not a tail risk for minerals like cobalt (DRC ~70% of mined supply, HHI routinely above 5,000) or rare earths (China ~60% of mining, ~90% of processing). It is a baseline condition.

The relationship between HHI and forward-looking volatility is nonlinear. Below roughly 2,000, the marginal effect of concentration on vol is modest because substitutable supply paths exist. Above 3,500, the vol sensitivity to disruption events accelerates sharply. A 10% supply disruption in a fragmented market (copper, with production spread across Chile, Peru, DRC, China, and others) reprices slowly as alternative flows adjust. The same disruption in a concentrated market (gallium, with China controlling ~98% of production) reprices within a single session.

Where HHI Appears in the Volterra Feature Set

The Volterra model ingests geographic concentration metrics as persistent features alongside the 96 daily GDELT GKG news files and market context signals. HHI enters the XGBoost framework not as a single static number but as a composite: mining-stage HHI, refining-stage HHI, and a processing-adjusted HHI that weights the most concentrated bottleneck in each mineral's supply chain.

This distinction matters. Nickel's mining HHI is moderate (Indonesia dominates at ~50% of mined supply, but the Philippines, Russia, New Caledonia, and Australia contribute meaningfully). Nickel's Class 1 refining HHI is considerably higher, with Chinese capacity controlling a disproportionate share of battery-grade output. The Volterra pipeline captures this divergence, allowing the model to weight refining bottlenecks differently from mining bottlenecks when scoring disruption probability.

In walk-forward cross-validation, concentration features rank consistently among the top feature importance clusters for 30-day volatility forecasts. Their contribution is most pronounced when interacting with GDELT event spikes originating from high-HHI jurisdictions. A protest event in a low-HHI producer country generates a muted signal. The same event category in a jurisdiction holding 40%+ of a mineral's supply chain triggers a measurably stronger probability shift in the model output.

Practical Implications for Vol Desks and Risk Managers

For options positioning, HHI provides a structural prior on the skew dynamics of a given mineral. High-HHI minerals (cobalt, lithium, rare earths, gallium) exhibit fatter left tails in supply and fatter right tails in price. The vol surface for these minerals should carry a persistent risk premium relative to diversified-supply metals like copper or aluminum. When Volterra signals shift from MODERATE to ELEVATED on a high-HHI mineral, the implied move in realized vol tends to be larger than the equivalent signal shift on a low-HHI mineral.

For systematic risk managers running VaR on physical or derivative commodity books, HHI concentration should inform correlation assumptions. High-HHI minerals sourced from the same jurisdiction (e.g., Chinese-dominated gallium and germanium) exhibit correlated disruption risk that standard return-based correlation matrices understate. The Volterra dataset captures these co-movement patterns through shared geographic and supply chain features.

Using Concentration Data in Portfolio Construction

Procurement teams and commodity-exposed portfolio managers can use HHI as a screening metric for stress testing. If your portfolio's weighted-average supply HHI exceeds 3,500, a single jurisdiction event can simultaneously stress multiple positions. The Volterra daily pipeline flags exactly these scenarios: when GDELT event intensity in a high-concentration jurisdiction spikes, probability forecasts across all minerals dependent on that jurisdiction shift in tandem.

Figures from the Volterra daily pipeline. Full historical backfill available on AWS Data Exchange. The concentration features, disruption signals, and resulting probability forecasts for all 12 tracked minerals are accessible for integration into existing risk infrastructure and backtesting frameworks.