Richard Lyon makes a thoughtful case for preparing early for the end of cheap oil. But every obstacle he names is a government failure, not a market failure – and his remedy is more government.
Richard Lyon opens his reply to my review of The Energy Trap with the old joke about the senior economist who frowns at a proposal and asks whether something that works in practice also works in theory. It is a good joke, and a generous reply. Lyon concedes that his claim that a $120 barrel is viable if the average is $60 was wrong. He withdraws “not real oil” as interview shorthand. He grants that the book’s line that “the arithmetic is already done” is stronger than he needs. Few authors are so candid in what they are willing to reconsider, a standard that all authors would do well to emulate.
What remains is the heart of Lyon’s case. Cheap oil is running down; the price signal will arrive too late to act on; so governments must start building the replacement now, with public money, as “insurance”. Let me turn his joke around. Planning ahead of the market works beautifully in theory. Does it work in practice?
Physics With Degrees of Freedom
Lyon rests his case on physics. Oil is finite, so once the size of the resource is fixed, the rest is arithmetic. But the arithmetic is only as good as its inputs, and the input that matters most – how much oil will ultimately be recovered – is not a constant of nature. It depends on price, technology and access. A Hubbert curve has degrees of freedom: the assumed ultimate recovery, the shape of the curve, the number of peaks. Change any of them and the date moves, by decades even. Lyon himself concedes that forecasters who counted all oil got their dates wrong.
That uncertainty is where the real decisions are made. When do high prices signal that it is time to invest, and invest in what – deepwater, shale, enhanced recovery, coal conversion, nuclear? If nuclear, which emerging technology – SMRs, molten salt, etc.? These questions fall on those who commit capital: owners betting their own money, or managers with a fiduciary duty to the owners. Being wrong costs them or their shareholders. So they are, rightly, sceptical of anyone who confidently predicts the future from a model and asks them to act on it. Their caution is not myopia. It is the discipline that comes from bearing the cost of error – a discipline no planner faces.
The Far-Sighted Bureaucrat
Lyon’s argument is not new. In the 1980s it was made on behalf of Japan’s Ministry of International Trade and Industry (MITI), whose patient, far-sighted planning was held up against myopic American investors fixated on quarterly results. Chalmers Johnson’s MITI and the Japanese Miracle (1982) hailed Japan as the best example of a state-guided market system.
MITI’s actual record of picking winners is less flattering. In the early 1960s it sought to consolidate the car industry and restrict new entrants; Honda rushed into car-making before the door closed. Its Fifth-Generation Computer project, launched in 1982 to leapfrog the world in artificial intelligence, ended as a commercial failure. Beason and Weinstein found that Japanese industrial targeting between 1955 and 1990 flowed disproportionately to slow-growing sectors.
The problem was not that MITI’s officials were stupid. It was Hayek’s knowledge problem: no ministry can gather the dispersed information that prices convey, and no official bears the losses when the bet goes wrong. Lyon concedes the knowledge problem is real, but says his test is physical, not a planner’s preference: does an investment permanently reduce dependence on hydrocarbons? What does “permanently reduce dependence” actually mean in practice, in decisions made by investors in competing and divergent technologies? Yet somebody must apply that test – decide which projects pass, at what cost, how fast and with whose money. That somebody is a ministry. Public money to build reactors “early” is MITI’s far-sightedness with uranium in place of semiconductors.
Where Is the Market Failure?
A case for public money needs a market failure: a reason why private investors systematically misjudge the future. Lyon’s candidate is timing – a nuclear fleet takes 15 years, a price spike takes a month. But the people with the most to gain from anticipating scarcity are the owners of the oil. As Harold Hotelling showed in 1931, an owner who expects prices to rise faster than the rate of interest leaves oil in the ground. Expected scarcity is priced today, in reserves held back, in futures curves and in asset values. Owners of oil reserves are – or have to be — among the most far-sighted investors in the economy.
Nor is the 15 years a law of physics. Lyon himself agrees that regulation, not physics, explains much of the gap between Korea’s $2,200 per kilowatt and Vogtle’s roughly $10,000 or higher cost. When Washington changed the rules, things moved. Under the Department of Energy’s Reactor Pilot Program, set up by executive order in May 2025, four privately built test reactors had reached criticality by early July 2026, one of them about 150 days after its project began. These are small test units, not gigawatt plants, but the lesson stands: the timetable is set by the regulator. Lyon has identified a government failure, and the remedy for government failure is not more government.
We can all agree that nuclear power is among the most promising, energy-dense technologies we have. Agreeing on that does not settle how the investment should be made. Will owners and shareholders, weighing their own money against their own judgement of risk, decide the pace and character of nuclear investment – which designs by which companies, where, how fast? Or will planners, who allegedly know when oil and gas will run out, decide for them? A ministry that gets the date wrong loses nothing. Taxpayers and private investors do.
The Hospital and the Diamond
Lyon defends energy return on energy invested (EROI) through what he calls “biophysical economics”: energy contributes far more to output than its share of costs, just as fuel is a small line in a hospital’s budget, yet the hospital stops without it. But this is just another way stating that energy demand is inelastic and the consumer surplus large. The amount we spend on energy – its share of costs – measures energy at the margin. It does not measure how much we depend on it in toto. The gap between what we would pay for the next glass of water rather than go without water altogether – in other words, the diamond-water paradox again – is what economists call the consumer surplus. For an input with no ready substitute – like certain forms of energy – it is very large. A small cost share and a large dependence are entirely consistent. Mainstream economics has no need of a new school to see that. A large consumer surplus does not make energy the only scarce factor of production, which is what EROI implicitly assumes; an energy theory of value makes no more sense than Marx’s labour theory of value. As EROI falls, the price needed to justify a barrel rises, as Lyon notes. Precisely – and that price is what calls forth new supply, efficiency and substitutes.
Prices Damp, Quantities Adjust
Lyon corrects my use of Samuelson’s Le Chatelier principle, citing Milgrom and Roberts: long-run responses exceed short-run ones. That holds for quantities – long-run demand is more elastic than short-run demand. But it implies the opposite of what he needs. Because quantities adjust more over time, prices need adjust less. A supply shock produces a sharp price spike in the short run, which is then damped as producers and consumers respond over the medium and long term. That is dampening feedback, not a runaway spiral.
The hog cycle rests on farmers forecasting tomorrow’s price from today’s. It was in the hog market that John Muth, introducing rational expectations in 1961, showed that producers forecast far better than that. Oil has deep futures markets, and James Hamilton has said shale’s quick response should make future oil shocks smaller. His ten recessions out of eleven followed largely geopolitical disruptions, and other economists blame much of the damage on the Federal Reserve’s reaction. That is politics and monetary policy, not geology.
New Provinces, Old Policies
Lyon says the 1970s cures worked only once: the North Sea and Alaska are long past their peaks. How much of that is policy? Britain imposes a 78 per cent headline tax rate on North Sea producers, and the government has committed to issue no licences to explore new fields. Alaska’s Arctic coastal plain, which the US Geological Survey thinks may hold 4.25 to 11.8 billion barrels of recoverable oil, was only reopened to leasing in October 2025 in President Trump’s second term focused on “energy dominance”. A province shut by politicians tells us nothing about geology.
Meanwhile new reserves keep appearing. Guyana went from no oil in 2019 to over 900,000 barrels a day in 2026. Brazil’s pre-salt fields, hidden beneath thick salt until seismic imaging improved, now supply 82 per cent of Petrobras’s record output. The Permian Basin, once written off, surpassed its 1973 peak in 2017. Hubbert curves can have second peaks.
Consider, too, how little of the shale resource is being tapped. Shale drillers recover only an estimated 4 to 6 per cent of the oil in place, and the US Department of Energy’s National Energy Technology Laboratory puts recovery from multi-fractured horizontal wells at under 10 per cent, typically nearer five. More than nine barrels in ten stay in the rock. That is roughly where conventional oil once stood. Primary production from a conventional reservoir typically recovers about 10 per cent of the oil in place; water and gas injection lift that to 20–40 per cent, and tertiary methods to 30–60 per cent or more.
Shale has barely begun that journey. Operators are already experimenting with cyclic gas injection – “huff and puff” – and one Eagle Ford lease showed a 35 per cent improvement in recovery over five cycles. Even modest gains on so large a base would add vast volumes. None of this shows up in a Hubbert curve drawn with today’s technology, because tomorrow’s technology is precisely what such curves leave out.
This puts Höök, Hirsch and Aleklett’s study of giant field declines in perspective. Their study covers individual giant fields, not world supply, and their own data show that decline rates diverged after 1985 because new technology revived giant fields in the Middle East and Russia. Their 2009 warning of a mounting supply challenge was followed by world supply rising to some 106 million barrels a day in 2025.
Nor are oil and gas the only hydrocarbons. It is true that oil cannot again be pushed out of power generation; it has already gone. But the world’s proved coal reserves would last well over a century at current output, and America’s recoverable reserves some four centuries. Coal can be liquefied and gasified to make much of what oil and gas now supply. Sasol’s Secunda complex makes up to 150,000 barrels a day of fuel from coal, and China converted some 276 million tons of coal into oil, gas and chemicals in 2024. Whether cheap nuclear heat or coal conversion or something not yet invented may one day make synthetic hydrocarbons competitive for ships, planes and heavy transport nuclear-made fuels wins is precisely the bet that should be made by those who carry its risk. Not planners with no skin in the game.
Hayek’s Cycle Cuts the Other Way
Lyon enlists Hayek’s theory of the business cycle against shale. But artificially cheap credit distorts investment towards long-gestation, capital-intensive projects. This refers to Hayek’s “roundabout” methods of production – whose value is most sensitive to the rate of interest. Shale is the opposite. A well is drilled and producing within months and, because output falls steeply in the first year, as Lyon himself notes, it yields most of its oil within a few years. A conventional field trapped in a reservoir structure is a different animal: years of exploration, appraisal and construction before first oil, then decades of production. Hamilton notes the very long lag between investment and output in deepwater. Interest rates affected conventional oil far more than unconventional oil.
The bankruptcies of 2020 transferred assets to new owners at a loss to creditors; the oil remained “real” despite the haircuts received by the original owners. Lyon now accepts that shale pays at around $65 a barrel. What best fits Hayek’s description of malinvestment? A 15-year, publicly financed reactor programme – the most interest-sensitive investment of all – justified as insurance against a date nobody can name.
Simon’s Master Resource
Lyon notes that 1980, the base year of the Simon Abundance Index, was a peak year for oil prices, and cites Kiel and colleagues, who found Ehrlich would have won most ten-year bets over the past century. Pooley and Tupy, using time prices and excluding war years, found Simon would have won about 70 per cent of the time. By Lyon’s own figures a barrel cost about five and a half hours of American work in 1980 and about two in 2025. If this is an “energy trap”, it is a comfortable one – as I argued some years ago about the brave new world of ample oil.
Julian Simon called energy the master resource, and Lyon says the condition – cheap energy – carries the whole argument. Agreed. Which is why the real threat to cheap energy is policy that makes it dear: punitive North Sea taxes, drilling bans and nuclear licensing that turns five-year projects into fifteen-year ones. Lyon invokes the Cambridge economic historian Sir Tony Wrigley who showed how coal offered England an escape from the limits of an organic economy of wood, muscle and harvest. But that escape was made by private miners, ironmasters and engineers responding to prices, not by a ministry planning a managed descent.
The Finitude
Lyon and I agree that oil is finite in some ultimate sense (taking technology and human ingenuity into account), that nuclear power deserves a far larger role, and that regulation, not physics, stands in its way. We part company on the remedy. His case for planning rests on a timetable that physics does not fix with certainty and on obstacles that governments created. His cure is to hand the same governments (or their appointed experts) the job of choosing the winners. The history of governments or their experts picking winners is a rather slim one, and Japan’s MITI showed how that tends to work in practice.
Lyon closes by hoping that “preparing early” works in theory as well as in practice. The question is who prepares and on whose behalf? If it means anyone apart from those putting their own or their shareholders’ capital at risk, then beware! Clear the regulatory path, let market prices do their work, let untrammelled investors invest — limited by the usual limits of common law like fraud or externalities — and the “energy trap” may prove, like peak oil before it, a date that keeps receding.
