How an obscure Victorian economic observation might be one of the most important ideas in climate policy ๐ – and what it would take to overcome it. ๐ฉ๐ปโ๐ฌ๐ฉ๐ปโ๐ง๐ฉ๐ปโ๐ป๐๐งฉ
We tend to assume that doing something more efficiently is, by definition, a good thing. Use less energy per mile driven. Extract more crop per acre farmed. Capture more carbon per kilowatt-hour spent. Efficiency is progress. Efficiency is the goal.
But there is a paradox lurking at the heart of this assumption โ one identified not by a climate scientist or a systems theorist, but by a Victorian-era economist writing about coal in 1865. His name was William Stanley Jevons, and what he noticed then has never been more relevant than it is today, as the world begins to deploy one of its most ambitious technological bets against the climate crisis: direct air capture of greenhouse gases.
Understanding Jevons paradox โ what it is, why it happens, and crucially, how it might be overcome โ is essential to understanding whether the technologies we’re placing so much hope in will actually save us, or quietly make things worse.
Part One: The Paradox That Bears His Name
William Stanley Jevons was watching the Industrial Revolution unfold around him when he noticed something that didn’t quite make sense. Engineers were getting dramatically better at building steam engines. Each new generation of engine extracted more work from the same amount of coal. By any intuitive measure, this should have meant that Britain’s appetite for coal would slow – or at least stop growing so fast. Instead, the opposite was happening. Coal consumption was exploding.
Jevons realised why. When steam engines became more fuel-efficient, they became cheaper to run. And when they became cheaper to run, they became economical to deploy in more places, at greater scale, for more purposes. The efficiency gains didn’t reduce demand for coal โ they *expanded* the universe of things it was worth using coal for. More mills. More ships. More railways. More factories. Each one burning coal that, without the efficiency improvement, would never have been burned at all.
He published this observation in his 1865 book *The Coal Question*, and it has carried his name ever since.
The mechanism at the heart of Jevons paradox is what economists call the **rebound effect**. It works at multiple levels simultaneously. At the most direct level, if your car becomes more fuel-efficient and costs less per mile to run, you might simply drive more โ longer commutes, more weekend trips, perhaps a house farther from work than you would otherwise have chosen. That’s the direct rebound: the efficiency gain is partly consumed by increased use.
At a second level, the money you save on fuel doesn’t vanish โ you spend it on something else, and that something else has its own resource footprint. This is the indirect rebound. And at the broadest level, efficiency improvements ripple through the entire economy, enabling new industries, new behaviours, new patterns of consumption that collectively dwarf whatever savings the original efficiency gain was supposed to deliver. This is the economy-wide rebound, and it’s the most powerful of the three.
The paradox has appeared throughout economic history. Airline fuel efficiency has improved dramatically over the past fifty years โ and global aviation has grown by orders of magnitude, with total emissions rising steadily. LED lighting uses a fraction of the energy of incandescent bulbs โ and buildings now contain far more light fittings than they once did, often running longer hours, with total electricity consumption for lighting barely changed in many countries. More efficient data centres helped power an explosion in data consumption that now makes the internet one of the world’s largest energy consumers.
The pattern is remarkably consistent: efficiency lowers the cost of something, lower cost drives greater use, and greater use consumes more of the resource than the efficiency gain saved. The improvement in *intensity* is overwhelmed by growth in *scale*.
Part Two: Enter Direct Air Capture
Direct air capture โ DAC โ is one of the more audacious technologies humanity has ever attempted to scale. The basic idea is straightforward: giant machines that pull carbon dioxide directly from the ambient air, then either store it underground in geological formations or convert it into synthetic fuels or materials. Unlike carbon capture at the point of emission (a smokestack, say), DAC works on the atmosphere itself. In principle, it can undo historical emissions, not just prevent future ones.
This matters enormously because the climate problem we now face isn’t just about stopping future emissions. We have already loaded the atmosphere with more COโ than is compatible with a stable climate. Even if every country met its current pledges โ which most are not on track to do โ we would still overshoot the warming targets set at Paris. The IPCC’s pathways to limiting warming to 1.5ยฐC or 2ยฐC almost all rely on removing billions of tonnes of COโ from the atmosphere in the second half of this century. DAC, alongside other approaches like enhanced rock weathering, soil carbon sequestration, and reforestation, is one of the tools expected to do that work.
The technology works. Facilities already operate in Iceland, the United States, and elsewhere. The company Climeworks has built a plant in Iceland called Mammoth that can capture tens of thousands of tonnes of COโ per year and store it in basaltic rock, where it mineralises into stone within a couple of years. Costs have been falling.
But today’s capacity is almost laughably small relative to the task. We need to reach **gigaton scale** โ billions of tonnes of removal per year โ by the middle of this century to meaningfully affect atmospheric concentrations. Current global DAC capacity is in the tens of thousands of tonnes annually. The gap between where we are and where we need to be is roughly five orders of magnitude. It is an engineering, economic, and political challenge of extraordinary proportions.
And into this challenge walks Jevons, paradox in hand.
Part Three: Five Ways the Paradox Threatens to Undermine DAC
The relationship between Jevons paradox and direct air capture isn’t straightforward โ it doesn’t map onto the classical template of fuel efficiency and consumption. But the underlying dynamic, efficiency enabling and encouraging greater resource use, appears in several distinct and troubling forms.
The Moral Licensing Problem
The first and perhaps most insidious risk is moral licensing. When a credible technological solution to a problem exists, people’s sense of urgency about that problem tends to diminish. We’ve already seen a version of this play out with carbon offsets. Corporations buy credits from tree-planting projects or methane capture schemes and use them to declare themselves “carbon neutral” โ while continuing to operate fossil-fuel-intensive businesses more or less unchanged. The offset doesn’t reduce emissions; it *licenses* them.
DAC, at scale, could trigger the same dynamic at a far greater magnitude. If governments, industries, and citizens come to believe that the carbon will be cleaned up later by machines, the political and social pressure to restructure economies away from fossil fuels will weaken. Why accept the disruption and cost of decarbonising heavy industry, aviation, or agriculture if the atmosphere can be remediated technologically? The efficiency of the cure becomes an argument against the urgency of prevention.
Extending the Fossil Fuel Era
A closely related risk is that cheap, scalable DAC could remove one of the central arguments for leaving fossil fuels in the ground. Today, climate advocates argue that the carbon budget is finite and shrinking โ that every tonne burned now is a tonne that cannot be burned later. DAC complicates that arithmetic. If carbon can be removed from the atmosphere at reasonable cost, the fossil fuel industry gains a powerful counter-argument: burn now, capture later.
This is not a hypothetical concern. Oil and gas companies have already begun investing in carbon capture technologies, in part because it offers them a credible narrative of continued operation alongside climate action. A more efficient DAC sector doesn’t just make capture cheaper โ it makes the *case* for continued extraction stronger.
The Energy Hunger of the Technology Itself
DAC is extraordinarily energy-intensive. Current systems require somewhere between 1,500 and 2,000 kilowatt-hours of energy per tonne of COโ captured. To put that in perspective, capturing a single tonne of COโ requires roughly the same energy as the average European household consumes in three to four months. Scaling to gigatons annually would require energy inputs comparable to significant fractions of today’s entire global electricity supply.
If that energy comes from fossil fuels โ even partially โ DAC generates its own substantial emissions, potentially capturing one tonne of COโ while emitting nearly as much in the process. And here Jevons reasserts himself: as DAC becomes more energy-efficient, it becomes cheaper to operate at scale, which drives deployment, which drives total energy demand higher. The efficiency improvement in the capture process could, paradoxically, increase total energy consumption โ and with it, total emissions โ if the energy system hasn’t fully decarbonised.
The ‘Technofix’ Displacement Effect
There is a broader version of the rebound that operates at the level of political imagination. When a technological fix is available, it crowds out systemic solutions. The existence of DAC as a viable-seeming option makes it easier for politicians to avoid the harder, more disruptive, more politically costly work of restructuring economies. Why redesign cities around public transport when you can just capture the emissions from cars? Why transform agricultural systems when industrial carbon removal can offset the methane from livestock?
This isn’t irrationality. It’s a predictable response to the availability of a less disruptive option. But it means that DAC’s efficiency as a removal technology could, paradoxically, slow the rate of change in the systems that generate emissions in the first place.
Cheapening the Cost of Carbon:
Finally, if DAC scales and generates a large supply of carbon credits, it risks driving down the price of carbon in trading markets. And a lower carbon price means it’s cheaper to emit. Cheaper emissions stimulate more activity in carbon-intensive sectors โ more flights, more cement, more industrial production. The supply of removal credits becomes a subsidy for continued pollution, and total emissions may rise even as the capture industry grows.
Part Four: The Stakes Are Different This Time
Jevons paradox has played out many times throughout industrial history, and the consequences have generally been economic โ more consumption, higher costs, depleted resources. Serious, but recoverable. Countries have adapted, innovated, found substitutes.
With climate, the stakes are categorically different. Several of the tipping points that climate scientists have long warned about โ the thresholds beyond which self-reinforcing feedbacks take over regardless of what humans do โ appear to have already been crossed, or are being crossed now.
The West Antarctic Ice Sheet’s long-term destabilisation is now considered effectively locked in at current warming levels. Even if atmospheric COโ were drawn back down, the dynamics already set in motion in that ice sheet are likely to play out over centuries. Greenland is losing ice at accelerating rates, contributing to sea level rise that will eventually reshape coastlines and displace hundreds of millions of people.
Coral reef systems are collapsing at scale. The Great Barrier Reef has experienced repeated mass bleaching events that have killed large portions of the reef structure. At 1.5ยฐC of global warming, which we are approaching, models suggest that 70โ90% of the world’s coral reefs will be severely degraded. Above 2ยฐC, the figure approaches 99%.
In Siberia and northern Canada, permafrost – ground that has been frozen for thousands of years – is thawing. As it does, it releases methane and COโ that were locked inside, creating a feedback loop: warming thaws permafrost, which releases greenhouse gases, which cause further warming, which thaws more permafrost. This feedback was not fully captured in earlier IPCC models, and it represents a significant source of additional warming that operates largely independently of human emissions choices.
This context is critical. It means that the goal of climate action is no longer simply to reach net-zero and stabilise the climate at current temperatures. It means we need to **draw atmospheric COโ down below current levels** – to achieve what scientists call net-negative emissions – to slow or partially reverse these dynamics. Many researchers argue that the target we should be aiming for is a return to roughly 350 parts per million of atmospheric COโ, a level we passed in the late 1980s. We are currently above 420 ppm and rising.
The asymmetry of timescales makes Jevons paradox particularly dangerous in this context. With coal or electricity, a rebound in consumption can be corrected over years or decades as policy catches up. With climate, a rebound in emissions driven by DAC complacency could push the system further past tipping points in ways that are irreversible on any human timescale. There is no policy correction available for a collapsed ice sheet or an extinct coral ecosystem. The margin for error is essentially zero.
Part Five: The Ideal Scenario – What Good Looks Like
Against this backdrop, it’s worth asking: what does the best credible version of this future look like? Not the utopian version; the version where everything goes right by magic – but the scenario where all the serious counter-arguments to Jevons paradox are actually applied, where the policy architecture is right, and where the renewable energy transition continues at something like its current extraordinary pace.
It turns out that such a scenario is technically coherent and physically possible. Here’s what it looks like, piece by piece.
Renewables Provide the Energy Foundation:
Solar energy has followed a learning curve that has beaten virtually every mainstream projection made over the past two decades. Costs have fallen by around 90% since 2010. Wind energy has followed a similar trajectory. Both technologies are now the cheapest source of new electricity generation in most of the world, and deployment is accelerating.
In the ideal scenario, this trajectory continues and even steepens. By the mid-2030s, many regions of the world are generating surplus clean electricity during peak production periods โ more power than the grid can immediately use. This surplus is currently wasted through a process called curtailment, where generating capacity is deliberately idled because the grid can’t absorb the output.
DAC facilities, in this scenario, are designed and sited specifically to consume this surplus clean power. They run hardest when electricity is abundant and cheap, and throttle back when the grid is stressed. Rather than creating new demand for energy โ and the emissions that might accompany it โ DAC becomes a productive use of power that would otherwise be wasted. This essentially sidesteps the energy problem at the heart of Jevons paradox. The carbon intensity of each tonne of COโ captured falls toward zero, because the energy powering the capture comes from generators that would have been running anyway.
This isn’t purely speculative. Regions including Texas, parts of Europe, and Chile are already experiencing significant curtailment as renewable capacity outpaces grid and storage development. The infrastructure challenge is real, but so is the opportunity.
Emissions Caps Remain Binding and Are Tightened:
The single most important policy mechanism for containing the Jevons rebound is a hard cap on emissions โ one that does not move because DAC exists. In the ideal scenario, governments maintain legally binding emissions reduction schedules that decline regardless of how much carbon is being captured.
DAC credits, in this framework, cannot be used by oil companies or airlines or steelmakers to offset emissions they could eliminate through structural change. They are reserved exclusively for genuinely hard-to-abate sectors: the small residual emissions from agriculture, from certain chemical processes, from aviation routes where electric aircraft aren’t yet viable. The cap on the rest of the economy remains fixed.
This is the governance equivalent of building flood defences while simultaneously managing the river better. You need both, but the flood defences don’t give you permission to stop managing the river.
A global or near-global carbon price, set high enough to make fossil fuels genuinely uncompetitive, reinforces this framework. Not a nudge โ a structural shift. When carbon is priced at the level of its true social cost, the economics of the entire energy system change, and the market does much of the work of decarbonisation without requiring every decision to be made by regulators.
DAC Is Governed as Remediation, Not Absolution:
International governance frameworks โ ideally through a strengthened and better-resourced UNFCCC or a dedicated new body โ establish clear accounting rules that keep removal and reduction in separate columns.
Carbon removed by DAC is tracked in transparent public registries, audited independently, and reported separately from emissions reductions. A country cannot count tonnes of DAC removal against its obligations to reduce emissions from power, transport, or industry. The two activities are parallel tracks, not substitutes for each other. This preserves the political and social pressure to decarbonise at source. Companies and governments that are cleaning up their own emissions receive the credit for doing so. Companies and governments that are using DAC as a fig leaf receive no such credit.
This framing matters enormously for public trust. One of the risks of carbon markets is that they become opaque and gameable, generating cynicism that undermines the entire framework. Clear, simple, honest accounting โ removal is removal, reduction is reduction, and neither substitutes for the other โ is essential to maintaining legitimacy over the decades this will require.
The Fossil Fuel Economy Unravels Structurally:
In parallel with DAC deployment and renewable expansion, the fossil fuel economy reaches a point of structural decline, not just policy-induced suppression. Electric vehicles approach dominance in new car sales across major markets. Heat pumps largely replace gas boilers in the building stock of the developed world, with parallel transitions in the developing world supported by international finance. Green hydrogen and direct electrification penetrate heavy industry.
At some point in the late 2030s or 2040s, the economics of new fossil fuel investment collapse not because carbon prices make it unprofitable, but because the demand trajectory is so clearly downward that the business case evaporates. Fields that would once have been worth developing are stranded assets before a barrel is pumped. The industry contracts not because it is beaten by regulation, but because it is displaced by a superior and cheaper alternative.
In this context, DAC isn’t propping up fossil fuels by providing them with a cleanup narrative. The fuels are declining under their own economic momentum. DAC is instead cleaning up the accumulated legacy of two centuries of industrial emissions โ a remediation project for a problem that is no longer being actively worsened.
The Trajectory of Drawdown:
If these conditions cohere, the broad shape of the future looks something like this.
Through the 2020s and into the 2030s, global emissions peak and then fall sharply, driven by the renewable energy transition, the electrification of transport and heating, and the combination of policy pressure and market dynamics. DAC begins scaling during this period, initially as a niche technology powered by surplus renewable electricity, then as a growing industry as costs fall along a learning curve analogous to solar.
By the late 2030s or early 2040s, the world approaches net-zero emissions. Atmospheric COโ concentrations stabilise. The tipping point dynamics that are already in motion continue to play out โ ice continues to melt, permafrost continues to thaw โ but the feedbacks that depend on continued warming begin to slow.
Through the 2040s and 2050s, DAC at gigaton scale begins achieving genuinely net-negative outcomes. More carbon is being removed from the atmosphere each year than is being added to it. Atmospheric COโ concentrations begin, slowly, to fall.
Over the following decades, sustained net-negative emissions bring COโ levels down from their peak โ currently above 420 ppm โ toward the 350 ppm that many scientists consider a safer long-term target. This process takes generations. But it is underway, and it is working.
Part Six: What Remains Genuinely Hard
Even in the best case, intellectual honesty requires acknowledging what doesn’t resolve cleanly.
Tipping points that have already been triggered will continue to play out. There are lag times and feedback loops now in motion that no policy can immediately halt. Sea levels will continue to rise for centuries regardless of what happens to atmospheric COโ in the near term. Some ecosystems will not recover on any human timescale. The ideal scenario doesn’t undo the past; it limits how bad the future becomes.
Political continuity over the 30โ50 year timeframe required is historically very difficult to sustain. Every election cycle is a potential reversal. The institutions that need to maintain binding emissions caps and stable carbon prices need to do so across governments of radically different political complexions, across economic crises and geopolitical upheavals, for decades. That is a test that few human institutions have passed.
Justice and equity raise questions that technology alone cannot answer. DAC is expensive, and the costs and benefits of its deployment will not fall evenly across the world. The countries most vulnerable to climate impacts โ low-lying nations, tropical regions, communities already under stress โ are often least able to fund or benefit from expensive carbon removal infrastructure. If the burden of paying for DAC falls on those least responsible for the problem, it will generate conflict, resentment, and political instability that could undermine the entire framework.
And at true gigaton scale, DAC creates its own resource pressures. The sorbents and chemical processes involved require materials. Some designs consume significant quantities of water. The land and infrastructure required is substantial. Solving one resource problem at scale tends to create others, and careful accounting will be needed to ensure that the cure doesn’t generate hidden costs.
Conclusion: A Question of Institutional Will
The most striking thing about the ideal scenario described here is that none of it requires technologies that don’t exist, or physics that isn’t real. The renewable energy transition is already underway at remarkable speed. DAC technology works and is improving. The policy frameworks โ carbon pricing, emissions caps, international accounting rules โ are understood and in many cases partially implemented.
What the ideal scenario requires, more than anything else, is **governance that is smarter than our historical average**. It requires maintaining the discipline to treat DAC as a remediation tool rather than a licence to emit. It requires the political courage to keep caps binding even when the costs of doing so are high. It requires the international cooperation to sustain a shared framework across decades of changing governments, shifting interests, and unforeseen crises.
Jevons paradox is not a law of physics. It is a description of what happens in the *absence* of adequate governance โ when efficiency improvements are allowed to run free in unregulated markets without countervailing constraints. The rebound is not inevitable; it is a policy failure. And policy failures are, at least in principle, correctable.
The honest summary is this: we are in a race between the speed of technological progress and the adequacy of our institutions to govern that progress wisely. The renewable energy transition is giving us the energy foundation we need. DAC is giving us tools to address the overshoot we’ve already committed to. Whether those tools help us or become another entry in the long list of efficiency gains that made things worse is not a question of engineering. It is a question of whether we can build institutions capable of constraining our own worst tendencies over the timescale that the planet requires.
The paradox Jevons identified a hundred and sixty years ago, watching coal burn in Victorian England, turns out to be one of the central challenges of the twenty-first century. We know what it is. We know how it works. We even know, in broad terms, how to overcome it.
The question is whether we will, and the monumental global effort that it will surely require.
For the good of all on planet Earth, and the continuity of viable human civilisation into the 22nd century, and beyond. ๐๐งฉ
*Further reading: Jevons, W.S. (1865), The Coal Question; IPCC Sixth Assessment Report (2021โ2022); Fajardy, M. & Mac Dowell, N. (2017), “Can BECCS deliver sustainable and resource efficient negative emissions?”, Energy & Environmental Science.*