Jevons Paradox
Making something more efficient can make it cheap enough to use so much more often that total consumption rises instead of falling.
What Is It?
Economist William Stanley Jevons observed in 1865 that improvements in the efficiency of coal-burning steam engines didn’t reduce Britain’s total coal consumption, as many assumed they would. They increased it. More efficient engines made coal-powered machinery cheaper to run, which made it worth using in more places, for more purposes, more of the time, and the resulting growth in total use outpaced the efficiency gain per unit. The mechanism isn’t unique to coal. Making any resource cheaper or easier to use, per unit, lowers the effective cost of using it, and lowering the effective cost of something tends to increase demand for it. If demand responds strongly enough, total consumption rises even though each individual use is now more efficient than before.
Why Does It Matter?
Organizations often assume an efficiency gain will show up as lower cost, lower workload, or freed capacity, and are surprised when it doesn’t. Automating a step that used to take an analyst a day doesn’t necessarily reduce the analyst’s total workload, it can just as easily make that kind of analysis cheap enough to request far more often, absorbing the saved time with new volume rather than releasing it as slack. A faster approval process doesn’t reliably reduce the total number of approval-hours spent, it can make approval-seeking cheap enough that more things get routed through the process that previously weren’t worth the friction of asking. Making first drafts nearly free to produce, with AI-assisted writing or design, can dramatically increase how many drafts get requested, reviewed, and circulated, rather than simply freeing up the time that used to go into producing one. This matters most when leadership plans around an efficiency gain as though it were a straightforward capacity freed up for something else. If usage expands to absorb the gain, the anticipated slack never materializes, and the organization is left with a faster process handling more total volume, no calmer than before and often busier.
What Changes Once You See It?
You stop assuming that making something more efficient automatically frees up capacity, and start checking whether the efficiency gain is instead going to get absorbed by increased demand for the now-cheaper thing. You start asking, before committing to a plan built on freed-up capacity, what would have to be true for people not to simply use the faster or cheaper process more often once it exists. You also get more deliberate about efficiency gains you actually want to convert into slack, because the savings only persist if something, pricing, quotas, prioritization, a governance decision, actively prevents new demand from expanding to absorb them.
Common Misunderstandings
- It isn’t a claim that efficiency improvements are pointless or bad. Total output or value delivered typically still rises, the paradox is specifically about consumption or usage not falling the way a naive before-and-after comparison would predict.
- It isn’t the same as any increase in usage after an efficiency gain. A modest uptick that merely reduces the expected savings is usually called a rebound effect. Jevons Paradox is the stronger case, where the increase in demand fully offsets the efficiency gain and pushes total consumption higher than it was before.
- It isn’t the same as Parkinson’s Law, which is about work expanding to fill available time regardless of efficiency. Jevons Paradox is specifically about a resource becoming cheaper to use and that lower cost pulling in more demand for it.
- It doesn’t happen automatically every time something gets more efficient. It requires that demand for the resource is genuinely responsive to its effective cost, some efficiency gains really do just produce straightforward savings with no meaningful increase in usage.
- It isn’t limited to physical resources like fuel. Anything with a real cost of use, analyst time, approval cycles, compute, meeting time, is a candidate for the same dynamic once that cost drops.
Diagnostic Question
If we make this cheaper or faster, how much more of it will people demand, and could that added demand consume the savings we’re counting on?
Explore Further
Field Notes
None yet.
Related Field Guide
Origin
William Stanley Jevons, The Coal Question (1865), observing that more efficient steam engines increased, rather than decreased, Britain’s total coal consumption.