Commodity Price Formation: Physical Constraints, Financial Flows, and Structural Drivers

Commodity prices do not form like financial asset prices. They emerge at the intersection of two fundamentally different logics: a physical logic (geological, energy, climatic, logistical) evolving over long time horizons (years, decades), and a financial logic (expectations, speculative positioning, portfolio arbitrage) evolving in real time. Confusing these two logics is the source of most analytical errors in commodity markets. On this point, see arabica’s climate fragility at altitude.

Brent crude averaged $115.6 per barrel in March 2022 before falling back to $80.9 in December 2022, without any meaningful change in global production capacity during that period. Copper nearly doubled between March 2020 and March 2022, from $5,183 to $10,231 per ton in monthly average terms, while physical inventories were low. European natural gas averaged $70.04 per million BTU in August 2022, about nine times its 2011-2020 average, before falling back to around $10 in mid-2023 (World Bank Pink Sheet). These moves cannot be explained by physical supply alone or financial speculation alone: they result from the interaction between the two, and this page formalizes that interaction.

The structural tension: physical supply vs financial demand

A commodity market operates on two simultaneous layers. The physical market (spot) connects producers, processors, and end users: those who extract, transport, refine, and physically consume the resource. The financial market (futures, options, swaps) connects actors seeking to hedge price risk with those taking speculative positions. The two markets are linked through arbitrage, but their time horizons and underlying logics are fundamentally different. The 2026 breakdown of a French litre of fuel covers an adjacent question.

In the physical market, adjustments are slow. Developing a new copper deposit takes many years from discovery to first production (S&P Global Market Intelligence). Building an LNG terminal takes years and billions of dollars in investment (IEA). Oil investment cycles (exploration → development → production) span several years (Rystad Energy). Physical supply is inherently inelastic in the short term: it cannot quickly respond to price spikes, and production continues even when prices fall because fixed costs dominate cost structures. These long-horizon physical constraints are formalized in our sub-pillar on physical growth constraints.

In financial markets, adjustments are instantaneous. Net speculative “managed money” positions in WTI crude ranged between +200,000 and +500,000 contracts from 2020 to 2024 (CFTC): each contract representing 1,000 barrels. Assets under management in commodity futures reached $3.5 trillion (BIS, 2023). Commodity ETFs have also drawn substantial net new capital since 2020. These financial flows can move prices in the short term independently of physical fundamentals, creating sometimes persistent divergences between market prices and real supply-demand conditions. The same reasoning is carried over to silver in the exposure differences between physical silver, ETFs and miners. A complementary angle: the long-run change in power-generation cost curves.

A deeper analysis of this tension is developed in our reference article: Physical supply vs financial demand, what really determines prices, and the specific mechanics of both markets in Physical vs derivatives markets.

Marginal cost: the invisible price floor

The marginal cost of production, the cost of the last unit required to meet demand, acts as the medium-term gravitational anchor for commodity prices. When market prices fall below marginal cost, marginal production becomes unprofitable and is eventually withdrawn, reducing supply and creating rebound conditions. When prices rise far above marginal cost, producers invest to expand capacity, increasing future supply and creating downward pressure. This price-as-regime-signal mechanism is detailed in our sub-pillar on commodities as macroeconomic regime signals.

This mechanism is empirically observable. Oil marginal cost, the breakeven level for the highest-cost producers required to balance the market, stands around $60–70 per barrel for US shale (Dallas Fed Energy Survey, 2024). Methodology detailed in the EIA weekly petroleum status report. When Brent fell below $20 in April 2020 (EIA), a large share of US production became unprofitable, and output fell from 13.1 to 9.7 mb/d (EIA). For copper, marginal cost is read at the 90th percentile of the industry cost curve (Wood Mackenzie). For lithium, it rose as new, higher-cost capacity came on stream (Benchmark Mineral Intelligence).

But marginal cost is not static: it evolves with technology, regulation, energy costs, and the US dollar exchange rate. US shale breakevens fell from $80 per barrel in 2014 to $45–50 in 2020 due to productivity gains (EIA), before climbing back under FX-driven cost pressure examined in the analysis of imported inflation and FX shocks, then climbed back toward $60–70 under oilfield service inflation and sweet-spot depletion (Dallas Fed). Marginal cost is a moving floor, not a fixed point. A detailed analysis is developed in Marginal production cost and equilibrium price.

Inventories: the buffer between supply and demand

Physical inventories (commercial, strategic, and in transit) serve as the adjustment mechanism between inelastic supply and fluctuating demand. Their level and direction often provide a more reliable market signal than supply and demand estimates, which are subject to significant revisions.

Inventory signals are especially clear at extremes. OECD commercial inventories of crude oil and other liquids fell from a Covid peak of about 3.2 billion barrels in July 2020 to about 2.6 billion in March 2022, the lowest level since 2014 (EIA, Short-Term Energy Outlook). LME copper inventories fell to very low levels by end-2023, far below their average of the prior decade. European gas inventories moved from 95% storage (Oct 2021) to a 26% low after the Russian invasion (Mar 2022, GIE/AGSI), triggering the August 2022 price spike. Worth reading alongside: the arabica-robusta price spread.

The forward curve structure directly reflects inventory conditions. When inventories are ample, markets are in contango: future prices exceed spot prices, reflecting storage costs. When inventories are tight, markets shift to backwardation: spot prices exceed futures, signaling immediate scarcity premiums. Brent has remained in near-continuous backwardation since mid-2021 (ICE), indicating structurally tight physical markets despite price volatility. LME copper went through episodes of steep backwardation in 2024. This dynamic is mapped in The role of inventories in price formation and in Copper inventories: a ticking time bomb.

Why the adjustment overshoots: the lags that build the cycle

The marginal-cost loop does not close smoothly, because its two sides run on different clocks. Demand moves with industrial activity, fiscal policy, credit conditions and global trade, and the price incorporates a change in outlook almost at once, even while physical volumes stay unchanged. New capacity answers the price signal only after a project has cleared its regulatory, financial and technical stages. Supply therefore tends to arrive when the demand that justified it has already begun to slow. The result is asymmetric: tight phases produce rapid and at times excessive price increases, while easing phases drag on, because excess capacity is slow to clear when output keeps running through falling prices.

Financing conditions sit inside that lag. When real interest rates stay durably positive, the higher cost of capital slows extractive projects and lengthens lead times further, stiffening supply adjustment at the moment capacity is short. Part of the consensus expects normalising global growth to smooth commodity cycles, which assumes supply adjusts quickly to past price signals. A reading weighted toward accumulated investment delays and fragmented production chains suggests that even a demand slowdown does not guarantee a quick exit, since the imbalance was built over years. A coordinated acceleration of investment under easier financing, a sharp negative demand shock or a major technological change could each shorten parts of the cycle. The risk this framework guards against is reading a cyclical adjustment as a regime change while the structural forces remain in place.

Three families, three pricing logics

Commodities fall into three major families whose pricing mechanisms, while sharing the general principles described above, exhibit important structural differences.

Energy: the most geopoliticized market

Oil (more than 100 million barrels per day of world consumption of petroleum and other liquids in 2024, EIA) and natural gas are the largest commodity markets and the most directly exposed to geopolitical constraints. Downstream, at the pump, that exposure reaches drivers with a lag, and a 21-year test of whether French pump prices rise faster than they fall puts the symmetry of that lag to the data. The 22 OPEC+ countries pumped 54.6% of world crude oil in 2025 (EIA), and the group keeps spare production capacity, a price lever unmatched in any other commodity. LNG markets are undergoing structural transformation: global liquefaction capacity is set to expand sharply by 2030 (IEA, World Energy Outlook 2024), reshaping gas trade flows. Refining margins, the spread between crude and refined product prices, often provide a more informative signal than crude prices alone. The year-by-year record of US refinery closures since 2020 sets out a related case.

Industrial metals: the market of structural transformation

Copper, aluminum, nickel, lithium, and cobalt are core inputs for electrification and the energy transition. An electric vehicle requires far more copper than an internal combustion vehicle (IEA). Lithium demand is expected to rise sharply by 2040 (IEA, Critical Minerals Market Review). Production and refining are highly concentrated: China refines 65% of global copper, 70% of cobalt, and 60% of rare earths (USGS). The market-by-market analysis is consolidated in our physical commodity markets sub-pillar. Long project lead times (many years for a copper mine, S&P Global MI) create structurally inelastic supply amid accelerating demand. A related read: cocoa’s 2024-2026 round-trip.

Agricultural commodities: the most shock-prone market

Grains (wheat, corn, rice), oilseeds (soybeans), and tropical products (coffee, cocoa) are the only commodities whose supply is directly exposed to climate shocks: lost harvests cannot be recovered. US wheat (SRW) jumped 31.7% from February to March 2022 in monthly average terms (World Bank Pink Sheet) after Russia’s invasion of Ukraine, as Russia and Ukraine account for 30% of global wheat exports (FAO). Those price shocks reach households through food, a chain set out in how wheat prices pass through to food inflation. Cocoa prices more than doubled in 2024, reaching a monthly average of $10,320 per ton in December (World Bank Pink Sheet), due to persistent drought in Côte d’Ivoire and Ghana (70% of global output, ICCO). Family-specific dynamics are analyzed in Differences between energy, metal, and agricultural commodities.

Further reading

← Back to pillar page Commodities

Last updated — 20 October 2026