Why Electricity Is Different
The instantaneous balance constraint, frequency as the signal that balance is holding, and why power cannot leave the grid it was born on.
Just after two in the afternoon on an unremarkable Tuesday, somewhere in a control room with too many screens, an operator watches a number that almost never moves. In North America it reads 60.00. In most of the rest of the world it reads 50.00. It is the frequency of the alternating current on the grid, and it is the only real-time report card the entire system has.
If more power is being consumed than generated, that number falls. If more is being generated than consumed, it rises. Not eventually. Now. There is no warehouse between the two sides of the trade, no tank farm, no floating storage, no inventory of any kind sitting between the generator and the kettle. Supply and demand are the same quantity on this system, continuously, or it comes apart.
Every other market in this series has a buffer. Oil has tankage at every stage from the lease to the refinery to the terminal, and when the market ran out of onshore space in 2020 it went to sea. Gas has salt caverns and depleted reservoirs and the whole seasonal rhythm of injection and withdrawal. Those buffers are what let a price be wrong for a while without anything breaking. Electricity has no such forgiveness, and almost every strange feature of a power market traces back to that one fact.
What the buffer was doing
Storage does several distinct jobs in a commodity market, and power has to reproduce every one of them by other means.
Table 1-1: What storage does, and what power has instead
| The job | Oil and gas | Electricity |
|---|---|---|
| Absorb a supply shock | Draw down inventory | Start a faster unit, or shed load |
| Move value across time | Store it and carry it forward | Mostly cannot, so the forward curve is a forecast rather than a carry |
| Move value across space | Ship, pipe or rail it | Only along wires that already exist, and only within one synchronous grid |
| Let the price be wrong briefly | Inventory absorbs the error | Nothing absorbs it, so the price moves instead |
The second row is the one traders should sit with. In oil, the shape of the forward curve is disciplined by arbitrage: if a distant month trades far enough above the front, somebody buys the front, pays for tankage and insurance and financing, and delivers into the distant month. That trade is what makes contango and backwardation mean something physical. In power, for most of the curve, that trade does not exist. A megawatt-hour in January cannot be manufactured in October and held. So the forward price is somebody's estimate of what the marginal unit will cost on a day nobody has seen yet, with no arbitrage available to discipline it.
Frequency is the balance sheet
The reason the frequency moves at all is that the system has mechanical memory. A large thermal or hydro plant spins a heavy mass of steel in step with the grid, and that rotating mass stores kinetic energy. When demand suddenly exceeds supply, the first response is mechanical. The spinning machines physically slow down, giving up some of their kinetic energy to cover the gap, before any control system has acted. The frequency falls because the machines are decelerating.
That effect buys the operator seconds, and seconds are what the entire control hierarchy is built on. Governors respond, then automatic generation control, then the operator, then, if all of it fails, the protective relays begin disconnecting load in blocks to save the rest of the system. This is the sequence that a market design has to pay for, and Chapter 13 is about who gets paid to sit ready at each stage.
Grid frequency is the only quantity in commodities that is simultaneously a physical measurement, a real-time inventory report and a contractual obligation. When it falls, it is telling you the system is consuming its own stored motion to cover a shortfall.
This matters more than it used to because the mechanical memory is optional now. Wind and solar connect through power electronics rather than through a synchronised spinning mass, and an inverter has no inherent inertia to give up. That is a genuine engineering change, and it is the subject of Chapter 6. It is also, as that chapter will show, not the explanation for the blackout most people attribute it to.
The commodity that cannot travel
The second constraint is geographic, and it is more absolute than anything in oil or gas. Power flows only along conductors that already exist, and only within a single synchronous area where every machine turns in step. Between synchronous areas, alternating current cannot simply be handed across. It has to be converted to direct current, moved, and converted back.
North America runs on three of these areas: the Eastern Interconnection, the Western Interconnection, and Texas. They are not meaningfully one market. Texas is connected to its neighbours only through a small number of direct-current ties, which is a deliberate arrangement with regulatory consequences that Chapter 18 takes up in full. The practical effect is that a generator with surplus power in one interconnection and a customer who needs it in another are, for most purposes, in different worlds.
Readers of Commodities 101 will recognise the argument from the theme on logistics: for a whole class of markets the delivered price is created after extraction rather than at it. Electricity is the extreme case. A megawatt-hour is perfectly fungible with any other megawatt-hour on the same grid at the same instant, and completely unavailable to anyone anywhere else. There is no ship to charter, no route to reroute, no arbitrage to put on. There is only the wire, or no wire.
What follows from all this
Hold those two constraints together, instantaneous balance and captive geography, and most of what looks eccentric about power markets stops looking eccentric.
Prices go negative, which in any storable commodity would be an anomaly requiring a special explanation, and here happens thousands of hours a year as a structural feature. Prices spike to thousands of dollars in a single interval and return to normal an hour later. The same megawatt-hour is worth different amounts at two substations a few miles apart. Regulators argue endlessly about whether the market pays enough to build a plant that will run four days a decade. Every one of those is a consequence of having no inventory and no exit.
It also explains why there is no single world price for electricity, and never will be. Oil has one. Gas has been converging toward one as liquefaction turned a piped commodity into a shipped one. Power has dozens of prices that cannot be arbitraged against each other in any meaningful sense, and the differences between them come from different answers to the same engineering problem, written by different regulators in different decades.
Which is the argument of this book. Every market design in the chapters ahead, nodal or zonal, energy-only or capacity-paying, dispatched by an independent operator or by a vertically integrated utility, is an attempt to solve one problem: you cannot store the product, so you must arrange, in advance and by contract, for exactly the right amount of it to exist at exactly the right instant, in exactly the right place. Nobody has solved that cleanly. The rest of this book is about how differently people have failed to.