Commodities

Market drivers

Commodity forward prices are driven mainly by the fundamentals of the supply chain. The forward price results from the participants' expectations regarding the demand-supply balance and the actual price at which it is deemed to be reached.
As such, commodity markets are forward markets. What is called 'spot' means prompt or nearby delivery which ranges from in a couple of days to in a few weeks. Spot markets exist but are far less liquid because used mainly for last minute delivery adjustments. Market participants schedule production and consumption and anticipate delivery constraints, hence the most liquid markets are forward, within the planning horizon of participants in the supply chain.

Many factors affect the price, both spot and forward, of a given commodity. Some of the major ones:

• Supply-demand balance
• Consumption rate and stock availability
• Actual or anticipated scarcity and supply disruptions
• Earnings, market performance and expectations
• Investment level
• Globalization, growth and trends
• Geopolitics and governmental policies
• Risk management policies, Speculation

Commodity prices are volatile. There are many reasons for it:

• Many consumers and producers with unpredictable behaviours.
• Dependency on exogenous factors, both short and long term:
• Supply: weather, economic cycles, disruptions (political crises, strikes, piracy, bottlenecks in logistics)
• Demand: weather, economic cycles, gross domestic product and growth, consumption patterns
• Economics of storage and transport media (pipe, freight).
• Perishability, Crop or vintage effect.
• Switch to substitutes when prices spike above acceptable level.
• Often not much elasticity from both supply and demand side.
• Institutional structure of a particular market, whereby some players may dominate.
• Lack of liquidity

The forward curve typically anticipates expected tensions in the supply-demand balance:

• Curves will be convex where high demand or supply constraints are anticipated, concave in the opposite situation. Storage withdrawal or injection and transport obviously mitigate extreme shapes.
• Recurrent 'humps' exist for many commodities at various levels: yearly (seasonality due to climate or crops in different hemispheres), weekly (business days versus week-end), or intra-day (peak energy consumption in the evening).
• Release of extra availability happens rather per batch, due to production constraints, both short term (power generation stack) and long term (mining).
• Bottlenecks in storage or transport capacity may exacerbate local imbalances.
• Short term tensions can switch abruptly to 'back to normal'.

For several commodities, strong correlations exist between forward curves.

• Products whereby the entire production chain depends from a single input (e.g. crude for the oil complex).
• Products which may substitute each other, thus compete with each other (e.g. energy, oilseeds).
• Products which are a key input in the production of other ones (e.g. coke in steel production).
• Geographical spreads for the same product versus cost of transport.
• Market makers obviously try to service both buyers and sellers, hence building natural hedges within their portfolio, between exposures with clients seeking to benefit from price movement in either direction. But they also research correlations which may exist between the products for which they offer risk management contracts. Not all products are very liquid, but strong correlations with more liquid ones may exist (e.g. refined product vs crude oil).
• Correlations however cannot be hedged, experience lags, distortions and may temporarily breakdown or even vanish.

Price discovery

The price will depend on the conditions in which the delivery or title transfer will effectively happen. What, where, how and when the delivery is agreed to happen will impact the price paid for it. The combination of delivery conditions is endless. But obviously there exist active trading places (large harbours, major network hubs), where many transactions occur, becoming liquidity pools.

Specialized news agencies (Platts, Argus, Opis and others) poll physical markets on activity. They interview buyers and sellers, identify done deals, even facilitate broadcasting of those, cross check the information, perform statistical checks, some cleansing if need be and consolidate all this information. This thorough methodology allows them to release on regular interval (often daily) high/low prices representative of what effectively happened in the market. Those prices are widely accepted by market participants, both physical and financial, to fix price formulae they may have agreed upon among each other.

Those agencies report on facts, on hit transactions for delivery in a very near future. What they don't offer is a view on assessments for deliveries further ahead. Even if some quote 'forwards' the horizon doesn't reach beyond more or less 3 to 6 months.

Exchanges propose future contracts. Price discovery happens on a marketplace where bid and offer for contracts for forward delivery sets the price for a range of maturities ahead (usually from 18 to 60 months, depending on the contract). This creates a term structure for prices immediately given by the market itself. A large open interest and many participants are needed for the contracts to be liquid, thus representative of its market. To concentrate liquidity, contracts are proposed only for a couple of products and delivery terms. Those contracts can ultimately be physically settled (even if very few are) which anchors the financial contract to its underlying physical reality.

More stringent disclosure obligations and collateral requirements have given the Exchanges the opportunity to propose as well listed, standardized look-alikes of structures which are widely traded in OTC mode. They organize a trading platform and offer clearing services. Unlike futures, final fixing and settlement of those contracts are however exogenous to the platform on which they are traded. They rely on closing prices published by the exchange itself, or a competing exchange, or news agencies. A term structure is proposed albeit not very liquid yet beyond the immediate front months.

To resume some important features of the price discovery chain:

• Each index or future contract mark specific delivery conditions.
• Indices published by agencies are ‘post facts’ and are rather settlement prices than forward prices.
• Futures exist only for a limited set of deliveries.
• The cleared equivalents of OTC structures ultimately rely on index prices to be settled.

Price discovery, price fixing and dissemination of quotations happens in many ways, from a verbal agreement between counterparts to real-time electronic exchanges. Among the key participants:

• Exchanges
  Price fixing is the result of a market. Propose a term structure, but for a limited set of deliveries. Up to the trader to find out which are the spreads (product and location), between the delivery terms of his specific transaction and those benchmarks.
• Price Reporting Agencies
  The coverage proposed by those agencies is impressive. They scrutinize full range of products in all possible markets around the globe. Some agencies have a global reach, some cover in detail particular markets.
  They report on done deals and do not offer a real term structure of forward prices. It is a commercial service which has a cost.
• Brokers
  Obviously not public, quotes given by brokers when servicing their clients remain an important source of price discovery. They are very close to their markets, often ahead of what will happen and thus disclose very useful insight into the ongoing. Several electronic platforms exist, consolidating information provided by different brokers in a structured way, providing a comprehensive vision of the short term market conditions.
• Regulators
  If not immediately involved in price discovery, regulators and public agencies, release information of great importance: statistics on production, consumption, export and import, customs, observed retail prices, support level prices for assisted productions, policies, etc.

Benchmarks

For many markets availability of price benchmarks the market has confidence into, is of utmost importance.

• Price is a key driver to producers, transformers and end-users. It determines whether new sourcing capabilities are explored, developed or closed down, transformation capacity built or abandoned and consumption reoriented.
• Most commodities are sold or bought on an unknown forward average flat price, suiting all parties. This because the market players are confident that when the title transfer will effectively occur, a 'benchmark' will be available, to materialize the pricing formula 'at market level'.
• Price benchmarks are limited to some heavily traded products. As such they do the 'heavy lifting' for price discovery, enabling other grades to be traded in reference to the most liquid flat price instruments (formula pricing), providing security and liquidity to the whole market. Reference markers hence fullfill two key functions expected from financial markets, namely price discovery and risk transfer capabilities.
• Spot physical trade only represents a fraction of the total traded volume. The remainder is contract or 'term' pricing, entered into to hedge the positions taken by the different parties. This active trading crystalized around a few price benchmarks enable values to be discovered, tested and secured. Risk management can device flexible hedges for all type of products, tenors and uncertainties.
• Price benchmarks should provide robust liquidity and high price correlation to their underlying markets. They become strong signals which allow market participants to react adequately to any information they help to broadcast, hence improving market efficiency.

Cleared swaps

It is common practice to price a physical delivery based on the average of prices published by a Price Reporting Agency (PRA) for the corresponding delivery month for an index which terms are close to the effective delivery. This guarantees both buyer and seller to realize a transaction 'at the money' which is at a price which reflects market conditions at the moment product ownership is transferred. The willingness to hedge those physical transactions with paper trades gave rise to an active OTC market hooked on quotes released by Price Reporting Agencies.

There is however a valuation challenge. Except for a few, very liquid indices, those agencies do not publish forward prices. They just compute a 'spot' price for a given market, based on verified transactions for prompt, nearby, very short term, physical delivery. There is no published term structure for those indices. It is up to the trader to estimate what the forward curve looks like, which is often done, by tying the index to a related product for which a future curve exist and applying a differential. E.g. it is understood that prices for fuel oil, which is obtained by refining crude, are correlated to those of crude oil. But measuring this correlation is a far from simple: there are obvious unit conversions (BBL vs MT) and technical factors (densities, refining yield), but temporary market distortions (refining bottlenecks, surge in demand), certainly short term, are harder to anticipate. A trader who knows his market well, has to assess how each piece of information may impact the curve underlying the position he is managing.

In the aftermath of the financial crisis of 2008, regulators pushed to have this bilateral OTC market hosted by exchanges to guarantee more secure clearing mechanisms. Exchanges offered 'cleared swaps'. For a future, the link with the underlying physical market is guaranteed by an EFP mechanism, whereby a future position can ultimately trigger a physical delivery. For cleared swaps, this link is maintained by ultimately settling the future at the price which is published by the Price Reporting Agency, which gets its prices by sampling trades which effectively happened in the physical market. Two future contracts on different exchanges but hooked on the same PRA index, may have different prices as long those are forward, but once maturity has been reached, their prices will be identical, i.e. the fixing formula (usually a monthly average) of the figures released by the PRA. The two future contracts are like two different boxes for the same content, the forward prices are different bets on what this content will be. Until the PRA releases the content, which is what happened in the physical market.

It seems a paradox. The future gives a term structure to an index which does not have one, but will ultimately settle against the value of the index. It tells what the financial market anticipates the physical market will be, but settles according what the physical market effectively was. The final settlement price is not resulting from bid, ask balance reached on the exchange itself, but is exogenous to the exchange.

Curve building

To mimic pricing of other financial instruments, where the forward price is derived from spot price plus the net carry, the non-arbitrage pricing formula can be written as:

Theoretical forward

For a given asset, be: \( S_t \hspace{1.6em} \textsf{the spot price at time t} \) \( F\space_{T,t} \enspace \textsf{the forward price of maturity T at time t} \) \( r \hspace{2.1em} \textsf{the continuously compounded interest rate for one year} \)

One may write: \( F\space_{T,t} = S_t\space e^{r (T - t)} \)

Since commodities do not inherently generate income, forward prices should be higher than spot prices. However in markets with trading constraints, such as commodities markets, the cost of carry should be adjusted by at least two factors:

• Storage cost
  Can be substantial for bulky materials having low unitary value and often depreciate.


• Convenience yield
  It is a somewhat empirical measure of the fact that a user may obtain a benefit from physically holding the asset rather than a futures contract. The stock holder may profit from temporary shortages or just safeguard the ability to keep a production running. He may choose between immediate consumption versus investment for later. This choice grants a premium.
  Some argue that this yield is a mathematicians' artefact to explain a behaviour of the short term forward curve which is not witnessed in other financial products, an attempt to rationalize the fundamental different nature of this asset, whereby for most market participants, exchanging a commodity happens for much more stringent needs than holding an investment vehicle.
  

Empirical forward

Considering the above, be: \( r \enspace \textsf{the risk free rate} \) \( s \enspace \textsf{the storage rate} \) \( c \enspace \textsf{the convenience yield} \)

One can amend the forward equation as: \( F\space_{T,t} = S_t\space e^{((r + s) - c)\space (T - t)} \) or \( F\space_{T,t} = S_t \times (1 + ((r + s) - c))\space T \)

However it remains tricky to parametrize a forward curve based on spot price only as proposed in the formula above.

• Anticipated demand-supply imbalances in the near term cause price spikes often quickly followed by reversion to mean.
• The most liquid quotes are often future contracts. But each maturity can to a certain extend be considered as an independent variable.
• It is hard to predict how storage (adjustment in time) or transport (adjustment in space) will be able to smooth sudden curve jumps. Elasticity to price of storage and transport is far from a continuous process but one with thresholds and limit levels.
• Where varying qualities or markets of a given commodity are traded, markets usually only quote a limited number of very liquid deliveries. Other grades or delivery terms are then traded at a spread towards those markers or benchmark curves, based on their relative quality and delivery constraints.
• Another effect is the roll yield an investor in commodity futures captures when the futures contract converges to the spot price. It is the amount received or to be paid by rolling a future contract from a short-term maturity into a longer-term maturity.

Most commodities forward curves have a stable long end, sitting near the marginal cost of production, and a bumpy short end, governed by short term supply-demand imbalances. The industry planning well in advance, most trading volume takes place at the long end of the curve. The investors horizon is usually long term as well which rides up the curve.

Depending on the slope of the curve, markets are said to be in:

• Contango
  • Forward prices (long term) are higher than spot prices (short term). This is considered as a normal market.
  • Convenience yield is lower than cost of carry.
  • Future rolls down to spot, roll yield is negative.
  • Stock in excess, cost of carry is compensated by a positive yield. Difference should be large enough to compensate the storage cost (contango limit).


• Backwardation
  • Forward prices are lower than spot prices. This is considered as an inverted market.
  • Convenience yield is higher than the cost of carry.
  • Future rolls up to spot, roll yield is positive.
  • Shortage or disruption concerns grant a premium for spot availability. The effect is reinforced by:
    • The inability of investors and speculators to short the underlying asset (spot).
    • The extreme risk aversion of physical hedgers due to high penalties associated with delivery default.
  • Commodities with long supply lags, such as industrial metals, are most likely to experience backwardation. It happens but disappears quickly with commodities which supply is more elastic to demand.

Volatility

Historical volatility

The volatility is a statistical measure of dispersion of a set of data (returns, prices) from its mean. The more spread apart the data, the higher the deviation. The higher the volatility, the riskier the asset. It is a direct measure of the movement of the underlying's price (realized volatility) over recent history.

If one samples observable quotes for a specified period, several statistical measures can be done. Looking forward, shifting from statistical sampling to probability, those measures have well defined equations.

• The average or mean of the sampled population is the expected value of a the random variable.
• The volatility is a statistical measure of dispersion of a set of data (returns, prices) from its mean. The more spread apart the data, the higher the deviation. The higher the volatility, the riskier the asset. It is a direct measure of the movement of the underlying's price (realized volatility) over recent history.
  Volatility is the variance of the sampled data. Variance is calculated by taking the differences between each number in the set and the mean, squaring the differences (to make them positive), and averaging the squared differences, thus dividing the sum of the squared differences by the number of values in the set. An equivalent measure for the volatility is the standard deviation , which is the square root of the variance.
• Skewness is the degree of departure from symmetry of a distribution. A negatively skewed distribution has a tail which is pulled in the negative direction (down the curve). A positively skewed distribution has a tail which is pulled in the positive direction (up the curve).
  
  
  

  

  
  
  
• Kurtosis is the degree of peakedness of a distribution. Kurtosis helps gauge an asset's level of risk.
  • A normal distribution is a mesokurtic distribution.
  • A high kurtosis or leptokurtic distribution (peaked, clustered) has a higher peak than the normal distribution and has heavier tails. A leptokurtic distribution means that small changes happen less frequently because historical values have clustered by the mean. However, this also means that large fluctuations are more likely within the fat tails: risk is coming from outlier events and extreme observations are much more likely to occur.
  • A low kurtosis or platykurtic distribution (flat, dispersed) has a lower peak than a normal distribution and lighter tails. A platykurtic distribution denotes a fairly uniform lay out of data, with fewer large fluctuations, hence less risky.

Implied volatility

Out of premia paid for options and the market price of the underlying asset, one can deduce the volatility. This implied volatility is the volatility of the underlying which passed to the model, will give a theoretical fair value of the option equal to the premium paid for that option.

Implied volatility is determined by the market price of the option contract itself, and not by statistics on the prices of the underlying. Hence it is function of the supply and demand dynamics of the options market. It is a forecast of future volatility made by traders and acts as an indicator of the current market sentiment about volatility. Unlike historical volatility, implied volatility comes from the price of an option itself and represents volatility expectations for the future. Because it is implied, traders cannot use past performance as an indicator of future performance. Instead, they have to estimate the potential of the option in the market. But in options trading, it is the more prevalent metric compared to historical volatility because the latter isn't forward-looking.

According to Black–Scholes, the implied volatility of an asset should be the same for all strikes and maturities. By computing the implied volatility for traded options with different strikes and maturities, it appears that the volatility surface (the 3D graph of implied volatility against strike and maturity) is not flat. Premia paid for options are not equal to their theoretical fair values. Traders factor in that 'eccentric' derivatives (deep OTM) bear more risk than ATM or ITM options. For a given maturity, the more out of the money, the higher the premium. If symmetrical, the 2D graph plotting implied volatility against strike or moneyness, looks like a smile . If not, the implied volatility is said to be skewed (asymmetry from a normal distribution).

Caveats

As for forward curves, when analysing volatility figures, one should bear in mind the fundamentals of commodity markets which is the balance between supply and demand.

• The anticipated imbalances being more acute in the short term, volatility curves will typically be backwardated, with excessive whippiness of the front end. The mean-reverting nature of the market (over a few weeks) will ease volatility toward the long end, where the volatility moves slowly with long term demand, affected by much more predictable macro-economic factors, such as GDP or growth.
• The implied volatility surface is made up of options on future contracts, of which each maturity can behave fairly decorrelated from the adjacent ones, certainly in the short term. Liquidity is also concentrated on shorter term maturities.
• Most market participants are industrial, extremely risk averse, searching to hedge exposure. Producers want OTM puts, consumers want OTM calls. The shape of the smile, which is often skewed, will depend on which group dominates the market.

Caution should be the rule as well while pricing options. Most models still rely on Black-Scholes models, be them adjusted or corrected. Advantages are its simplicity and the fact that it is widely adopted. It became de facto the standard in pricing options such that a fair value calculated by Black-Scholes is interpreted as the premium to be paid for this option. One should however not forget that some of the assumptions which Black-Scholes itself considers as prerequisites for its model to be applicable, are not always verified in commodities markets, as detailed hereafter. This is true for all option evaluation models, even considered very sophisticated. Use the outcome as a proxy, a good indicator, but don't get fooled by a blind trust in models and never forget to check fundamentals and apply common sense.

• Efficient, frictionless markets
• Normal distribution of prices of the underlying, which is not the case in markets with spikes.
• In extreme cases, such as oversupply of generation capacity in electricity markets, prices can turn negative.
• Ability to buy or sell any amount of the stock, including short selling, which is not feasible in physical markets.