EVs, an exception to Jevon’s Paradox?

Blossoms at pa harakeke – it’s always easier to be hopeful in spring

Jevons’ Paradox proposes that increased efficiency in the use of a resource leads to a rise, rather than a fall, in total consumption of that resource. William Stanley Jevons described the paradox in his 1865 book The Coal Question. He observed England’s consumption of coal soared after the Watt steam engine was introduced. The Watt steam engine greatly improved the efficiency of coal-fired steam engines, making coal a more cost-effective power source, and leading to increased use of the steam engine and thus total coal consumption.

Put simply, Jevons said, “It is a confusion of ideas to suppose the economic use of fuel is equivalent to diminished consumption. The very contrary is the truth.” In today’s world, we are seeing the limits of resources because human use of nearly every type of energy and material has increased over time (with a few exceptions like sheep wool, whale oil, mercury and asbestos). Therefore, Jevon’s Paradox is a concerning concept – how can we reduce energy use, if increased efficiency leads to more consumption.

Jevons’ Paradox has been quantified for Scottish iron production in the 1800s where the amount of coal produced per ton of iron fell by 66% and the iron production business boomed, resulting in a 10x increase in the amount of coal used. However, this is pretty old.

Air travel could be another example. Planes have become increasingly more fuel efficient, supposedly over 1% a year over the last 6 decades. The relative cost of plane travel has plummeted. In my lifetime, the cost of a ticket to the UK has stayed pretty similar at around $2000 since 1987. Average weekly NZ incomes have increased from $566 to $1,730. Unsurprisingly, the number of people travelling by air has boomed and the amount of passenger air travel globally has increased by seven times.

AI is an another area where Jevons’ Paradox appears to apply. Compute has become more efficient and consequently less expensive. In a single generation of AI models, the cost of the best available answer has decreased by about 20% and the cost of producing a fixed level of LLM performance has fallen by orders of magnitude. However, the resultant vastly increased capability of AI has driven more usage, so the total amount of compute has hugely increased and is forecast to increase even more leading to the current (insane) scale of demand for data centres.

Jevon’s Paradox is commonly applied to fossil fuels. Many government policies push for increased efficiency of machines using fossil fuels, on the basis this will result in an overall reduction of fossil fuel use. In the 1980s, economists Daniel Khazzom and Leonard Brookes (then chief economist of the UK Atomic Energy Authority) argued attempts to reduce energy consumption by increasing efficiency would raise demand for energy in the economy as a whole: the ‘Khazzom-Brookes postulate’.

In 1992, economist Harry Saunders built on the ‘Khazzom-Brookes postulate’ and suggested increased energy efficiency would increase energy consumption in two ways:

• Increased energy efficiency would reduce energy prices, encouraging increased use.
• Increased energy efficiency would increase real incomes and lead to economic growth, thereby increasing energy use across the economy.

Saunders pointed out it’s important to consider both microeconomic and macroeconomic effects. Energy efficiency improvements may reduce energy consumption in a specific market while they increase it at the macroeconomic level.

The impact of EVs on energy use is a common topic as cheap Chinese EVs spread around the globe. So how does use of EVs relate to Jevons’ Paradox? Are EVs actually reducing energy use in driving? Or are they so great and cheap to drive that people are driving more, obviating the efficiency gains?

EVs consume far less energy consumption compared to an ICE vehicle. In ICE vehicles, only 12-30% of the fuel energy gets turned into forward momentum with most of the rest lost as heat. EVs convert >77% of their electrical energy into power in the wheels. In addition, regenerative braking in EVs captures energy for reuse, leading to up to 94% efficiency of energy use in an EV.

EV’s efficiency means they are cheaper to run per kilometre, which could encourage additional travel and potentially more travel by car rather than walking or cycling. I certainly know of people who justify driving, rather than using less energy-consumptive modes of transport, based on owning an EV rather than an ICE car. Personally, I can feel I am less resistant to using our EV than our diesel Ford Everest because not only is the Kia EV6 cheaper to run, we can power it off our roof.

Norway is the best real world test of the energy effects of electrifying passenger vehicles because it has the most electrified passenger fleet and good data. According to the IEA, 88% of cars sold in Norway in 2024 were EVs and 31% of passenger car kilometres were driven in EVs. The calculated total energy use per passenger kilometre is estimated to have fallen from 0.41 kWh in 2005 to 0.32 kWh in 2024. At the same time, the IEA reports Norway’s fossil fuel consumption for passenger transport in 2024 was 12% lower than in 2021. Therefore the rebound effect – increased driving as a result of cheaper vehicle-kilometres – has not outweighed the efficiency gains from EVs.

A meta-analysis of the rebound effect in passenger transport estimates it at 10-12% in the short run and 26-29% in the long run. We can compare this to average car energy use:

	Energy/fuel use per 100km	Energy use in kWh/ kWh equivalent for fossil fuels
EV	12-17 kWh	12-17
Petrol	3-8 litres	29-57
Diesel	6-8 litres	64-86

In the worst cast scenario, an EV is nearly twice as energy efficient as a Toyota Yaris. In the best case scenario, a small EV is seven times as energy efficient as our Ford Everest. That’s why, at population level in Norway, the rebound effect has not eliminated the reduction in energy resulting from use of efficient EVs. In other words, Jevons’ Paradox does not apply to uptake of EVs if we simply consider energy use.

Of course, you have to take into account that the reduced energy use is only environmentally beneficial if the electricity consumed by the EVs is generated through efficient power generation. I found a variety of numbers for fossil fuel power plant efficiency but the general range appears to be 30-40% efficiency i.e. a power plant is no more efficient than an ICE car so there’s no net energy gain burning fossil fuels to power electric vehicles. In New Zealand, our electricity generation is around 85% from renewables. Therefore, shifting to EVs will reduce overall energy usage and EVs will mostly not be powered by fossil fuels one step removed.

I find this all heartening. I have tended to view Jevons as a gloom merchant because improvement seems futile if all you get is more resource consumption. However, Jevons Paradox doesn’t hold in the case of EVs, likely because the scale of improvement is so great and the potential for increased travel demand is sufficiently constrained (as opposed to the demand for AI compute power!). Got to take those wins…