A published cost curve ranks producers on modeled cash costs using assumed feedstock terms. It breaks where those terms are negotiated rather than indexed — long-term contracts, integrated supply, captive utilities. A producer's true position can sit a quartile away from where the curve puts it.
What is a cost curve for?
Judging who survives a price trough. It ranks producers by modeled cash cost so that an analyst can say which capacity comes out of the market first when prices fall, and that is a genuinely useful thing to know.
It is a modeling convention rather than an observation. Every position on it is computed from assumptions, and the quality of the ranking depends entirely on whether those assumptions hold for the producer in question.
| Published curve | Operator-adjusted | |
|---|---|---|
| Feedstock | Assumed at index | Structure established from people who negotiated it |
| Coverage | Every producer | The three or four the thesis rests on |
| Best used as | A starting hypothesis | A tested position |
What does a published curve assume?
Typically that feedstock is bought at index, that utilities are procured at market, and that logistics costs scale with distance. Those assumptions are reasonable in aggregate and wrong for specific producers in predictable ways.
They are also necessary. A curve builder cannot know what a producer negotiated, so indexing is the only defensible default — which is a limitation of the method rather than a failure of it.
Where does the curve stop working?
Where feedstock is contracted rather than indexed. A ten-year supply agreement signed in a different price environment, integrated upstream supply, or captive power can each move a producer materially without appearing anywhere in the model.
The effect is largest exactly when it matters most: in a trough, when the question is who is still profitable, contract structure is what decides it — the same way an offtake agreement decides a power asset.
Captive power deserves particular attention because it is both common and invisible in a modeled curve. A producer generating its own electricity at historical capital cost is insulated from exactly the input that moves most in a squeeze, and the curve will place it alongside producers buying at market. In a trough that single difference can decide which of them is still operating.
Who knows the real feedstock terms?
Former procurement and commercial staff at the producer or its suppliers. Terms are confidential and they will not quote you a number, but on an expert call they will describe the structure — indexed or fixed, term length, whether there is a floor — and structure is what the model needs.
Traders and logistics providers are a useful cross-check, because they see the physical flows that reveal whether an integration claim is real.
How should you use one?
As a starting hypothesis that tells you which producers to ask about. The curve identifies the three or four positions your thesis depends on; primary work tests those rather than the whole ranking — starting with what those plants actually run at.
Treating it as an answer rather than a question is where chemicals diligence most often goes wrong, and it is an easy mistake because the output looks precise.
Using it well also means being explicit about which positions have been tested and which have not. A curve annotated to show that three producers were checked against primary sources and the rest taken as modeled is far more useful than one presented uniformly, because it tells a reader where the ranking is evidence and where it is convention.