Commodity dependence is an exposure in the budget, and the conversion that ends it runs slower than the price cycle

Commodity dependence is usually described as a fact about what a country produces. It is better described as a fact about what its budget is a claim on. Those are different conditions and they have different exits. A country can produce a great deal of oil and not depend on it. A country can produce very little else and depend on nothing but the price of the one thing it sells. Confusing the two is how a government comes to believe that building a second industry is the same as reducing its exposure, when the exposure sits in the revenue line and not in the industrial mix.

The definition worth using is narrow. Dependence is the share of a government's own revenue whose size is set by a price formed in a market that government does not participate in. That makes it a measure of exposure rather than of endowment, and it makes the question of escape tractable, because a share has a numerator and a denominator and a state can act on either. What follows applies that definition to the published record, takes the conditions of escape in order, and ends on the one condition the record cannot test.

Where this is written from

I am an analyst writing inside The Cefa Brief, working only from published statistics and public documents: annual series redistributed by FRED and retrieved for this site on 19 September 2026, and the Census Bureau's survey of state tax collections for fiscal year 2025. I hold no brief for any producing government and no information that is not published. That position is good for one thing. It can check every figure it prints, show the subtraction, and say plainly when a claim has no series behind it. What it cannot see is the part of this subject that decides the outcome. Not one file here holds another country's figures, so every economy that has escaped commodity dependence, and every one that has not, sits outside this evidence. Nothing here records a budget negotiation, the rules of a stabilisation fund, or which official lost the argument in a year when the price was high. The condition this piece ends on is institutional, and it is the condition I cannot measure. Naming it is the most the position allows, and I would rather name it than write past it.

The exposure is in the claim, not in the ground

The clearest instance of the definition in this record is sub-national. In fiscal year 2025 North Dakota collected 2.60 billion dollars of severance tax against 5.52 billion in total state tax collections, or 47.2 percent. Alaska collected 741 million of 1.69 billion, 43.9 percent. New Mexico stands at 31.2 percent and Wyoming at 27.7. Texas raises the largest severance sum of any state, 8.00 billion dollars, and that is 8.9 percent of a base of 89.94 billion; Oklahoma is at 7.8 percent. The same industry yields a dominant revenue line in one jurisdiction and a modest one in another, and the difference between those positions is not what lies under the ground. It is the size of everything else the state taxes. That is the first condition of escape and the least useful to state, because it is a condition about the denominator: a government lowers its exposure by having a larger base to tax, which is close to saying it lowers its exposure by having already escaped.

One price, not a class of them

West Texas Intermediate crude averaged 39 dollars a barrel in 2020, 68 in 2021, 95 in 2022, 78 in 2023, 77 in 2024 and 65 in 2025. Each point is an annual average of a daily price. From the low to the high the series gained 56 dollars; from the high to the latest reading it gave back 30. Six annual observations are what the record holds, so the highest price any budget actually faced inside a year is not in this file, and nothing here should be read as though it were. The shape is still plain enough to govern against. A revenue line tied to that series is re-priced every year by a decision taken by nobody in the jurisdiction that depends on it.

Depending on a commodity is not the same as depending on commodities as a class. Producer prices in chemical manufacturing, an index on a December 1984 base of 100, ran 289, 322.8, 359.6, 356.5, 352.2 and 356.5 across the same six years. Crude in 2022 was 2.4 times its 2020 average; the chemical index in 2022 was 1.2 times its 2020 reading. Crude in 2025 stood at 68 percent of its 2022 average, the chemical index at 99 percent of its own. Two series over one window are not a law, and a price in dollars a barrel and an index on a 1984 base cannot be added or netted. They are enough to establish that a raw price and a price for processed output did not move by anything close to the same amount over the same years, and that where an economy sits between those two kinds of price is an exposure decision as much as an industrial one.

The two clocks do not match

The second condition is that revenue be converted into a base whose price is formed differently, and the record shows how long conversion takes in the one economy it covers. Construction spending on manufacturing plant ran 75, 82, 125, 202, 244 and 215 billion dollars across 2020 to 2025. That series is stated at a seasonally adjusted annual rate, so each point is a yearly rate and approximates the year. The rate rose for four years, by 169 billion dollars between 2020 and 2024, then fell by 29 in 2025. Research and development inside GDP, stated the same way, ran 754, 839, 942, 991, 1029 and 1081 billion, with no down year and annual additions of 85, 103, 49, 38 and 52.

Set those beside the price. Crude completed a rise of 56 dollars and a fall of 30 inside the same six readings in which the plant line made a single rise and a single turn down. The revenue side of a commodity budget moves faster than the asset side it is meant to buy, and patience alone does not reconcile them, because the years in which conversion is affordable are the years in which it feels least necessary and the years in which it is obviously necessary are the years with no money in them. A stabilisation or permanent fund is the standard answer to that timing problem and a real one. It does not end the exposure. It exchanges a claim on a commodity price for a claim on an asset price, and an asset price is also formed somewhere the government does not sit.

The condition this record cannot test

Whether the second base actually grows is the third condition, and here the American record is a caution rather than a model. Manufacturing output, indexed to 2017 equals 100, ran 93, 97.6, 98, 97.3, 96.4 and 97.3 across 2020 to 2025. Six years, and the index sits below its own base year in every one of them, over a period in which the plant line's annual rate rose by 169 billion dollars and research spending rose in every observation. A base that has not regained its own 2017 level in six years is not an escape route, whatever has been committed to it, and the two facts in that sentence are held by different series measuring different things, which is the honest reason they can sit together without one explaining the other.

What remains is the condition that decides the outcome and that nothing here measures. Two governments can carry the same exposure, receive the same windfall and arrive in opposite places, and the difference lies in rules that bind a legislature in a good year: what a fund may be drawn on for, who scores the drawing, which official is able to refuse. No source in this record holds another country's figures, so the cases that would test that claim sit outside the evidence and are not going to be borrowed for it. The most this position supports is a definition and a sequence. Dependence is a share of a claim, not a share of output. The share falls when the denominator grows or when the numerator moves to a part of the chain where the price moves less. Both take longer than a price cycle, and whether a government does either in the years it can afford to is a question about its institutions, which is the part that cannot be seen from here.