A refining margin is what is left after you subtract the cost of crude, plus shipping and operating expenses, from the money a refiner earns selling petroleum products. Korean refiners typically treat $4-5 per barrel as the break-even line. Yet on 18 August 2026, the US diesel crack spread cleared $100 a barrel. Both figures get reported under the same label.
The gap comes from what each number subtracts. A crack spread is the raw price difference between products and crude. A composite refining margin takes that difference and then strips out the cost of running the plant. Miss that distinction and you cannot explain why "record refining margins" headlines sit alongside mediocre quarterly results at Korean refiners.
What Actually Gets Subtracted
The basic form is simple: product revenue minus crude cost minus shipping and operating expenses. That last term is the one that does the work. Fuel burned in the process, catalysts, labour and depreciation on the units all land there.
That is where the $4-5 break-even comes from. Turning a barrel of crude into sellable product carries roughly that much fixed cost, so the product-crude gap has to be at least that wide before the refiner is whole. Below the line, every barrel processed adds to the loss.
The key point is that this is not "product price minus crude price" but "product price minus crude price minus cost." The overseas crack spreads quoted in the press almost always sit at the earlier stage, before costs. Plotting them on the same axis tells you nothing.
A crack spread is the price gap between products and crude. A composite refining margin is what survives after you pay to run the plant. Same name in the headlines, different numbers entirely.
How the 3-2-1 Crack Is Calculated
The most widely quoted US benchmark is the 3-2-1 crack spread. It assumes three barrels of crude are refined into two barrels of gasoline and one of distillate, subtracts the cost of the three crude barrels from that revenue, then divides by three to express the result per barrel of crude.
As a formula: (gasoline × 2 + distillate × 1 − crude × 3) ÷ 3. That result equals "average product basket price minus crude price," which means knowing the crack lets you back out the implied product price.
At the 18 August 2026 close, WTI settled at $84.94 a barrel and Brent at $91.02. Holding WTI at $84.94 as the crude input, here is what the product basket would have to fetch at each crack level. Since one barrel equals 42 US gallons, the per-gallon conversion is included.
| 3-2-1 crack ($/bbl) | Implied product basket ($/bbl) | Per gallon ($) |
|---|---|---|
| 10 | 94.94 | 2.26 |
| 30 | 114.94 | 2.74 |
| 65 | 149.94 | 3.57 |
| 70 | 154.94 | 3.69 |
| 100 | 184.94 | 4.40 |
The third column is where this connects to something tangible. A $70 crack implies a wholesale product basket around $3.69 a gallon. For reference, on 20 July 2026 the US average pump price was $4.00 a gallon for gasoline and $5.11 for diesel. Weighted at the 3-2-1 ratio, that is $4.37 a gallon, or $183.5 a barrel.
The space between the backed-out wholesale figure and the pump price holds federal and state fuel taxes plus station-level margin, and the two observations are a month apart. So do not subtract one from the other and call the difference refiner profit. What the conversion does explain is why pump prices climb when cracks move in double digits.
Why $100 Cracks and $5 Margins Coexist
Collecting the summer 2026 readings in one table exposes the mechanics. The measured product, the reference date and whether costs are deducted all differ.
| Indicator | Reference date | Level ($/bbl) | What it measures |
|---|---|---|---|
| US 3-2-1 crack | 2026-07-20 | ~70 | 2 gasoline + 1 distillate, before costs |
| European diesel margin | 2026-07-20 | ~65 | Single product |
| Northwest Europe composite | 2026-07-20 | ~30 | Product basket |
| US diesel crack | 2026-08-18 | >100 | Single product |
| Asian composite break-even | Industry convention | 4-5 | After operating costs |
| Korean composite (H2 forecast) | H2 2026 | 26 | After operating costs |
Read down the column and three things stand out. Single-product cracks always print higher than basket measures because they isolate the strongest product. Even within Europe, the composite margin at roughly $30 is less than half the diesel crack. And the $4-5 break-even sits on an entirely different layer, since costs have already been removed.
Supply conditions explain the unusual summer prints. Middle Eastern product exports ran near one million barrels per day in June, about a quarter of pre-war volumes, and global refined output fell by roughly five million barrels per day in the second quarter. When products are scarcer than crude, cracks widen regardless of where oil trades.
Do Margins Rise When Oil Rises?
Counterintuitively, no — not reliably. A refining margin is a spread, so if crude climbs faster than products, the spread compresses. That is why refiner profitability often deteriorates precisely during oil price spikes.
The current episode is different because the binding constraint is refining capacity, not crude supply. When there is crude but not enough capacity to convert it, product prices outrun crude prices. That configuration also tends to coincide with futures curves flipping into backwardation, the subject covered in the piece on crude ETF roll costs.
Sell-side forecasts assume that structure holds. Shinhan Securities projects H2 2026 WTI at $80-90 a barrel with composite refining margins near $26, above the $10 seen in Q4 2025 before the conflict. That scenario rests on production normalisation taking nine to twelve months. If Middle Eastern capacity comes back faster, the premise breaks.
Reading an indicator as a gap between two series rather than an absolute level transfers to other markets too — the same approach applies to the spread between DRAM contract and spot prices.
What to Watch
- Check the label first — crack spread or composite margin. If cost treatment differs, the levels are not comparable.
- Single product or basket — a standalone diesel crack structurally prints above any composite.
- Position versus the $4-5 break-even — Korean refiner earnings track the composite margin against that line, nothing else.
- Whether crude and cracks move together — both rising signals product scarcity; crude alone rising signals margin compression.
- Middle Eastern capacity restart pace — if the nine-to-twelve-month normalisation assumption shortens, the H2 margin forecast loses its footing.
- Product inventories — once stocks start building, cracks roll over before crude does.