Herfindahl-Hirschman Index in Commodities: How Supply Concentration Quantifies Mineral Volatility Risk
Why Supply Concentration Is a Volatility Input, Not Just a Geopolitical Narrative
Most commodity risk frameworks treat supply disruption as a tail event. In practice, for minerals with concentrated production, disruption risk is priced continuously. The Herfindahl-Hirschman Index (HHI) provides the quantitative backbone for measuring that concentration, and its levels correlate directly with the frequency and magnitude of volatility regime shifts in exchange-traded metals.
HHI is calculated as the sum of squared market shares across producers (countries, firms, or mines, depending on the unit of analysis). The index ranges from near zero (perfectly fragmented) to 10,000 (single-source monopoly). The U.S. Department of Justice considers markets with HHI above 2,500 to be highly concentrated. In critical minerals, several commodities blow past that threshold by a wide margin.
Cobalt's mine-stage HHI, driven by the Democratic Republic of Congo's ~70% production share, sits above 5,000. Rare earth elements, with China controlling roughly 60% of mine output and over 85% of processing, register similarly. Lithium is lower but rising as production consolidates around Australia and Chile. Compare these to copper, where the top five producing countries collectively hold roughly 55% of output but no single nation exceeds 28%, yielding an HHI closer to 1,500. The volatility profiles of these commodities reflect the difference in concentration.
HHI and Realized Volatility: The Empirical Relationship
The link between high HHI and elevated realized volatility is not theoretical. Minerals with production HHI above 3,000 exhibit materially higher annualized volatility, wider bid-ask spreads, and more frequent gap moves than their lower-HHI counterparts. The mechanism is straightforward: when a small number of jurisdictions or producers control most supply, any policy change, export restriction, labor action, or logistical disruption removes a larger fraction of global output from the market. Price discovery becomes asymmetric, with upside shocks outsizing downside corrections.
This pattern appears clearly in cobalt and lithium, where structural supply chain risk drives recurring volatility regimes. Nickel's 2022 LME short squeeze, while driven by positioning, was amplified by the concentration of Class 1 nickel refining in a handful of facilities. High HHI doesn't cause every spike, but it sets the conditions under which shocks propagate faster and further.
For risk managers running VaR models or options desks pricing vol surfaces on metals, ignoring HHI means systematically underestimating the kurtosis of return distributions for concentrated commodities. A Gaussian assumption calibrated on copper's return profile will misprice tail risk on cobalt or lithium by a wide margin.
How HHI Enters the Volterra Model
The Volterra prediction pipeline incorporates supply concentration as a standing feature set alongside 96 daily GDELT GKG news files and market context signals. Geographic HHI at both mine and refining stages is encoded per mineral and updated as production share data refreshes. This allows the XGBoost classifier to weight concentration dynamics when generating 7-day, 14-day, and 30-day volatility probability forecasts across its five risk levels (LOW through EXTREME).
In walk-forward cross-validation, supply concentration features rank consistently among the top feature importance scores for cobalt, lithium, and rare earths. For lower-HHI metals like copper and aluminum, news sentiment and exchange-specific liquidity signals carry more weight. The model adapts its feature reliance per mineral, which is one reason it maintains a mean AUC of 0.815 across a diverse coverage universe. Full detail on the feature engineering approach is available in the methodology section.
Figures from the Volterra daily pipeline. Full historical backfill available on AWS Data Exchange.
Practical Applications for Desks and Risk Teams
For systematic traders, HHI provides a static or slow-moving prior that informs position sizing. A high-HHI mineral warrants tighter stop-losses, wider confidence intervals on vol forecasts, and potentially asymmetric hedging structures (e.g., skewed collars rather than symmetric straddles).
Procurement teams can use HHI trends to identify minerals where supply chain risk is intensifying before it shows up in spot prices. If a mineral's processing-stage HHI is rising due to capacity consolidation, forward volatility should be re-priced even if spot remains stable.
Options desks should calibrate implied vol surfaces against concentration-adjusted realized vol, not raw historical vol. The Volterra dataset, available through AWS Data Exchange, provides the signal layer to do this systematically across 12 exchange-traded critical minerals, capturing the concentration dynamics that raw price data alone cannot reflect.