In short: A hedge ratio applied to an exposure nobody has measured is a number about a number, so building the exposure inventory comes before any decision about coverage. Most unintended exposure is created by ordinary commercial terms, where a cargo is bought on one pricing basis and sold on another, leaving a position in tanks and on water that was never classified as an exposure. Representing every contract by its pricing formula rather than its volume, then netting by index and pricing period, is what turns physical activity into a position anyone can hedge. Hedge accounting rules decide what is practical, because an instrument that fails the effectiveness test moves reported earnings in exactly the way the treasury team was trying to avoid.
The hedging discussion usually opens with a percentage. How much of next year's production, or next year's purchase requirement, should be hedged, and the answer comes back as a share of volume with a rationale attached.
Meanwhile the position that will actually move this quarter's result is a few hundred thousand barrels bought on one pricing basis and sold on another, sitting in tanks and on water, and it is nowhere in the treasury system because nobody ever classified it as an exposure. It was created by ordinary commercial terms, negotiated separately, by people who were not thinking about price risk when they agreed them, and a hedge ratio applied on top of an exposure nobody has measured is a number about a number.
Building the exposure inventory first
Before any instrument is chosen, somebody has to write down what the organisation is actually exposed to. Four categories cover most of it, and the last one is the one that gets missed.
Physical inventory. Product in tank, crude in pipeline linefill, cargo on water, and material sitting at a third party under a processing or storage arrangement where title has not passed. Linefill and minimum tank heels behave like a permanent long position. They never turn over, they were acquired at whatever the price was when the line was commissioned, and no instrument with a maturity date matches that exposure.
Purchase and sale commitments. Contracts signed and not yet priced, contracts priced and not yet delivered, and the volume flexibility inside term contracts. That flexibility is an option granted or held, it has a delta, and it is almost never in anybody's position report.
Pricing formula mismatches on volumes that already net. Covered in the next section, because this is where the largest unintended positions live.
Exposures embedded in contracts that are not called commodity contracts. A freight contract with a bunker adjustment clause. A processing agreement with a fee indexed to a product price. A power supply contract with a fuel pass-through. A long-term sales contract with a lagged index that reprices six months after the market moves. Each of these is a commodity position wearing a different label, and they sit in legal files rather than in a trading system.
The output is a position report in physical units, broken down by index, pricing month, grade and location, showing the net open position in each cell. It generally looks nothing like what the trading system reports, because the trading system knows about trades and this report knows about the business.
Where the unintended exposure actually lives
Take a 500,000 barrel cargo bought at the average of a marker over five days around bill of lading in March, and sold as product on the average of a published product assessment across the whole of May.
Volumes match. Grades correspond. A volume report shows a flat book and nobody has taken a position on anything. The actual exposure is the difference between a five-day March average of one index and a full-month May average of another, on 500,000 barrels, and it has a distribution, a mean and a tail like any other position.
Nobody chose it. The buyer wanted a monthly average because that is how their downstream contracts work, the seller's standard is five days around bill of lading, and the voyage took three weeks. Three reasonable decisions produced a position larger than most of what treasury is deliberately managing.
The fix is mechanical rather than clever. Represent every physical contract by its pricing formula instead of by its volume, then net at the level of index and pricing period. A book that nets barrels bought against barrels sold will report flat whenever the volumes match, which is exactly when it is least informative.
One consequence worth checking while you are in there. If inventory is carried at cost and sold at market, the length is exposed on the way through and shows up in margin. If it is carried at market, the movement lands in the result every period whether or not anyone decided to hold it. The accounting treatment determines where the exposure appears and never whether it exists.
Three kinds of exposure that want different answers
Transaction exposure. A committed cash flow of a known size at a known date. This is what derivatives were built for, and matching instrument, volume and timing is largely an execution problem once the exposure is identified.
Translation exposure. The reported value of a balance sheet item that moves with price or currency. Inventory carried at market, or the net assets of a subsidiary reporting in another currency. Hedging it with cash instruments converts an accounting movement into a cash movement, which is a real decision that deserves to be made deliberately rather than by treating every exposure as the same species.
Economic exposure. The sensitivity of the value of the business to the price level over a long horizon. It has no maturity, no fixed volume and no clean measurement, and instruments manage very little of it. What manages it is balance sheet structure, contract design, cost position and the pace of capital commitment.
The recurring error is reaching for the instrument built for the first category and applying it to the second, then describing the result as if it had addressed the third.
Hedge accounting decides what is practical
Under IFRS 9, effective for annual periods beginning on or after 1 January 2018, and under ASC 815 as amended by ASU 2017-12 (FASB, 2017), a derivative receives hedge accounting treatment only where it has been designated and documented at inception against an identified hedged item, with an economic relationship between the two.
Without designation, the derivative marks to market through profit or loss while the physical item it offsets sits at cost. A hedge that is economically correct then produces reported earnings volatility pointing the opposite way from the thing it was protecting, and somebody has to explain that variance every quarter. Desks respond by putting on fewer hedges that would have worked, an accounting outcome driving an economic decision.
Two features of the current standards change what is workable.
Risk component designation. IFRS 9 permits designating a separately identifiable and reliably measurable risk component of a non-financial item, and ASU 2017-12 introduced comparable treatment for contractually specified components under US GAAP. In practice this lets you designate the marker component of a physically priced barrel and leave the differential undesignated, which is what makes hedging a differential-priced grade workable at all.
The effectiveness test. The quantitative 80 to 125 percent bright line inherited from IAS 39 was replaced under IFRS 9 by a qualitative economic relationship requirement, and ASU 2017-12 removed the requirement to separately measure and report ineffectiveness for qualifying hedges under US GAAP. Both changes widened what can be designated, and organisations that built their hedging policy before 2018 are frequently still operating under the older constraint out of habit.
There is a trap alongside this in the own use exemption, which keeps physical contracts entered into for the entity's own purchase, sale or usage requirements outside derivative accounting, and it is easier to lose than it looks. A practice of net settling, or of taking delivery and immediately reselling, can affect the classification for a whole class of similar contracts, and the consequence is a book of physical supply agreements arriving on the balance sheet at fair value. Get the accounting policy people into the room while the strategy is being designed rather than after the first trade.
Basis risk, and how much of it you keep
The instrument settles against something that is not your barrel. Decompose the gap into three parts, because they behave differently and they move at different times.
Grade basis. Your physical grade against the marker underlying the instrument. This is the differential, and setting it is a commercial decision covered separately in this series. Here it matters because a flat price hedge leaves the entire differential unhedged, and differentials run on their own schedule.
Location basis. Where your barrel or product physically sits against where the contract settles. Pipeline constraints, a regional refinery outage, and freight can move a location basis sharply while flat price does nothing at all.
Calendar basis. Your pricing period against the contract months you hedged in. Rolling a hedge forward exposes you to the shape of the curve, and in a market that flips between contango and backwardation the roll is a real cash item that can dominate the hedge result over a year.
Sizing what remains is a regression rather than an assertion. Regress your realised physical price on the settlement price of the hedging instrument over as much history as you have, using your actual pricing convention rather than a spot series. The slope is the variance-minimising hedge ratio and the residual standard deviation is the risk you keep after hedging. That residual is the honest description of the programme, and it is usually larger than anyone expects.
Two cautions attach to it. The relationship is estimated on history and it is not stable, and the events that move flat price hard tend to move differentials at the same time, so the correlation measured in a calm period is the wrong one for the period you wanted protection in. And the variance-minimising ratio will not be one, so a book described internally as eighty percent hedged is a different number in variance terms.
The governance layer
The mandate. Which exposures may be hedged, with which instruments, out to what tenor, and an explicit statement that positions unrelated to an identified physical exposure fall outside it. That last clause is what separates the function from a trading desk, and it needs to be written down, because it will be tested by someone with a good argument during a volatile month.
Limits. Volumetric limits by index and by tenor, a maximum open position, counterparty and credit limits, and at least one limit expressed in liquidity rather than in profit and loss.
Liquidity capacity. This is where hedging programmes die. A hedge that is working generates margin calls and collateral demands now, while the offsetting physical gain arrives on delivery, weeks or months later. Committed facilities have to be sized against a price move large enough to be uncomfortable and tested against that scenario rather than an average one. A programme unwound at the worst moment because the cash was not there has delivered the loss without the protection.
Independent reporting. Position and valuation reporting produced by a function that does not report to the person putting on the trades. Daily marks, and a monthly reconciliation of hedge results against the physical results they were meant to offset, presented together in one view. That combined view is the most useful item on this list to build, because it is the only report in which the programme can be judged honestly.
The limit
Hedging converts an uncertain outcome into a certain cost. The certain cost includes premium or carry, transaction costs, collateral, and the upside you gave away, and it gets paid every period whether or not the protection was needed. In exchange, the distribution of outcomes narrows. Stating the trade in those terms to a board is worth doing early, because a board expecting the hedging programme to make money has been told something else.
The measurement problem follows directly from the trade. A programme doing its job reports losses in exactly the periods when the physical business does well, and gains when it is struggling. Give it its own line in the management accounts and judge the desk on that line, and the rational response from the person running it is to hedge less when they expect prices to move favourably and more when they do not, which is a view on price with a hedging label attached. Run that for a few years while the size grows and you have a trading desk operating under a hedging mandate, usually discovered in the quarter it loses money.
The alternative is to measure variance reduction against the stated objective, plus adherence to mandate: how much of the realised price variance in the period was removed relative to an unhedged benchmark, with physical and derivative results always reported as one number. It is far less satisfying than a profit line and it measures what the desk was actually set up to do.
One more limit is worth stating plainly. How much variance the organisation should remove is a balance sheet and strategy question, about how far prices can fall before capital spending has to be cut or a covenant is breached, and no hedging model answers it. The analysis can tell you what a given hedge costs and what it removes; deciding how much certainty is worth buying stays with the people who own the balance sheet.
Take one month of physical activity, represent every contract by its pricing formula rather than its volume, net it by index and pricing period, and compare what falls out with what the treasury system currently shows as the exposure.