Where Power Is Cheap
Why the same commodity costs a few cents a kilowatt-hour in Quebec and several times that in Dublin, what actually sits inside a delivered price, and the industries that move to the electricity rather than the other way round.
Words the industry uses are marked like this at the sentence that defines them, so you can tell a term you will hear on a desk from a phrase this book happens to be using.
A tonne of copper costs roughly the same in Rotterdam and in Shanghai, because if it did not, somebody would put it on a ship. A megawatt-hour does not work that way. Industrial electricity runs from around one US cent per kilowatt-hour in parts of the Gulf to the high twenties and beyond in the most expensive European markets, and the gap persists year after year because nobody can put a megawatt-hour on a ship.
Chapter 3 gave the physical reason: power stays inside the synchronous area it was made in, and even inside one it moves only as far as the wires allow. Chapters 17 to 20 gave the institutional reason, which is that each region made its own design choices. This chapter is about the consequence. When a commodity cannot arbitrage, its price becomes a property of a place, and anything that can move to that place eventually does.
What is actually in the price
The first thing to establish is that the wholesale price this book has spent twenty chapters explaining is frequently the smaller half of what a customer pays. A delivered bill has five parts.
Table 21-1: The five components of a delivered electricity price
| Component | What it pays for | What moves it |
|---|---|---|
| Wholesale energy | The generation itself, set by merit order | Fuel prices, weather, the fleet mix |
| Network charges | Transmission and distribution, mostly recovering money already spent | How much wire there is per customer, and how recently it was built |
| Policy costs | Renewables support, capacity payments, social tariffs, stranded costs | Political decisions taken years earlier, usually hard to unwind |
| Carbon | Emissions permits, arriving inside the wholesale price rather than as a separate line | The permit market, and the fleet it is charged against |
| Taxes | General taxation, including value added tax charged on all of the above | Fiscal policy, and whether power is treated as a necessity |
Two consequences follow, and both surprise people. A large fall in the wholesale price is barely felt by a household, because it applies to one line of five. And building generation that is free to run can raise the retail price while lowering the wholesale one, because the support that financed it sits in the policy line rather than the energy line. Chapter 8 explains why the wholesale price falls. This is where the money that made it fall reappears.
Three reasons a country is cheap
Cheap electricity has three sources, and they differ in kind rather than in degree.
The first is an endowment that has already been paid for. Norway, Iceland, Quebec, Manitoba, British Columbia and much of Brazil run on hydro built decades ago by public bodies that no longer need to earn a return on it. The marginal cost of water arriving in a reservoir is zero and the concrete is sunk, so the price reflects a decision taken two generations ago. This is the most durable form of cheap power and the least reproducible, since the good sites are taken.
The second is fuel with no alternative buyer. Gulf producers burn associated nat gas at an administratively set domestic price rather than at what the same molecule would fetch as a cargo, which is an implicit subsidy even where no money changes hands. Qatar, Saudi Arabia, Kuwait and Iran sit at the bottom of every international price table for this reason. Whether it lasts depends on whether the exporting country keeps choosing to forgo the export revenue.
The third is policy, meaning a government that treats the industrial power price as an instrument rather than an outcome. China runs a large and relatively new coal fleet under regulated tariffs designed to keep manufacturing competitive, with the market reforms of Chapter 20 arriving on top of that rather than replacing it. The United States is a partial case: the shale nat gas price is a genuine market outcome, but it reaches the power price through the merit order of Chapter 8 rather than through any decision about electricity at all.
Why Europe is expensive
Since this is the question readers actually ask, it deserves a direct answer rather than a gesture at taxes. Prices for energy-intensive industry in the European Union have been running at roughly double the United States and more than half again above China and India. Four things account for it, in descending order of size.
Nat gas sets the marginal price in a large share of hours, and Europe buys nat gas at an import price while the United States pays close to a wellhead price. That single spread, described from the nat gas side in NatGas 101, does most of the work. Chapter 19 traces what happened when it widened to a factor of twenty in 2022.
Network charges are high because the distribution network is dense, old in places and being rebuilt in others, and because a system with a great deal of distributed generation needs more of it rather than less. Policy costs are high because the renewables build was financed through the electricity bill rather than through general taxation, which was a deliberate choice and is now difficult to reverse. Taxes are high because most European countries treat electricity as a taxable good rather than a protected necessity.
The uncomfortable observation, stated in Chapter 19 and repeated here, is that the wholesale part of the European price has fallen a long way since 2022 without the delivered price following it down, because the other four components did not move.
Congealed electricity
Some products are close to being electricity in solid form, and those products move to where electricity is cheap.
Aluminium is the pure case and the one to hold in mind. Smelting by the Hall-Heroult process, the electrolytic route that has dominated for more than a century, takes roughly 13 to 15 megawatt-hours per tonne of metal. Electricity is typically 30 to 40% of the cash cost of a smelter, more than alumina feedstock, carbon anodes and labour combined. A tonne of aluminium is, in a real sense, fourteen megawatt-hours that somebody found a way to ship.
That is why the smelter map looks the way it does. Iceland, Norway and Quebec smelt on hydro. Bahrain and the United Arab Emirates smelt on Gulf nat gas. Mozambique smelts on the Cahora Bassa dam and Tajikistan on the Nurek. It is also why European smelters shut through 2022: their power cost went to a multiple of the world median and no version of the business survived it. Commodities 101 covers the metal itself. The point here is that a smelter siting decision is an electricity decision.
The test for whether an industry relocates has two conditions and needs both. Electricity must be a large share of cash cost, and the product must be cheap to ship relative to its value. Aluminium passes both. Cement fails the second, which is why cement plants sit near their customers on expensive power instead. Silicon and polysilicon, chlor-alkali, industrial gases and electrolytic hydrogen all pass, to varying degrees.
The newest entrants pass more completely than any of them. A data centre and a bitcoin mine ship their output at the speed of light and carry no logistics cost at all, which makes them the first industries whose location is decided almost entirely by the price of power and the availability of a connection. That is the demand-side story of Chapter 25, arriving from the direction of this chapter.
Cheap power attracts the load that ends it
The arbitrage has a self-limiting property that both sides tend to underestimate.
A hydro system with more energy than its domestic economy can absorb has a genuine problem, and a smelter is an excellent answer to it: enormous, flat, creditworthy and willing to sign for twenty years. But every megawatt-hour sold that way is one that no longer sits in surplus, and once the surplus is gone the price the smelter came for stops being available to anybody else. Norway and Iceland are both having that argument in public, as are several US states about data centres, and it is a distributional argument rather than a technical one: the cheap power was there, somebody else now has it, and households can see their own bills.
The instrument that makes arbitrage happen is the long-term power purchase agreement of Chapter 24, which is how a decision about where to build is converted into an acceptable price that can hold for two decades. Another way to make new demand tolerable is that a smelter’s power demand can be interrupted: potlines tolerate short reductions, so the smelter sells that flexibility back as the reserve product of Chapter 13, earning on the electricity it agrees not to consume.
What this does to policy
European electricity is expensive relative to most of the world, and the composition of that gap has changed. The wholesale part has fallen a long way from its 2022 peak while the delivered price has not followed it down, because the network charges, policy costs, carbon and taxes stacked above it did not move. Those four are self-imposed in a way the nat gas import spread is not, and the consequence is an industrial base migrating to places where electricity is cheaper.
Setting political policy via electricity prices results in market and trade flow distortions. A carbon price that relocates a smelter has moved the emissions rather than removed them. It explains why almost every jurisdiction with a carbon price also has an exemption scheme for energy-intensive trade-exposed industry. It explains the European carbon border adjustment mechanism, which charges the embedded carbon in imported aluminium and steel so that the exemption does not have to last forever. It explains why industrial electricity tariffs are negotiated politically across most of the world rather than simply quoted. And it explains why the phrase “energy-intensive industry” turns up in election manifestos, which is unusual language for what is, at bottom, a question about the cost of a kilowatt-hour.
The argument of this chapter in one line: because electricity cannot travel, its price attaches to a place, and over long periods it is the industry that moves instead.