The Spark Spread

A gas-fired power station converts one commodity into another. It buys natural gas, measured in million British thermal units, and sells electricity, measured in megawatt-hours. The exchange rate between the two is its heat rate.

That makes a power plant a currency trade with physical machinery. The spark spread is the power market equivalent of the product minus crude spreads in oil, the refinery margins or crack spreads:

Take nat gas at 4 dollars per million Btu and a plant at 7. Fuel costs 28 dollars per megawatt-hour. If power sells at 40, the spark spread is 12 dollars, and that 12 dollars has to cover operations, maintenance, start costs and eventually the capital. If power sells at 25, the plant should not run, and Chapter 8 explains why it will not be dispatched.

Quoting power in units of nat gas

The inversion traders actually use flips the equation around. Divide the power price by the nat gas price and the result is the implied market heat rate.

At 40 dollar power and 4 dollar nat gas, the market heat rate is 10. That number says the market is currently willing to pay for generation from any plant efficient enough to make a megawatt-hour on 10 million Btu or less. A plant at 7 is comfortably inside that and earns the difference. A plant at 12 stays off.

Expressing power in heat rate terms strips out the fuel price and leaves the thing a generation business is really exposed to. A nat gas plant is long the ratio rather than long power or short nat gas, and it will survive a doubling of the nat gas price provided power doubles alongside it.

The same construction exists for other fuels. The coal equivalent is thedark spread. In markets where carbon carries a price, subtracting the cost of the emissions permits gives the clean spark and clean dark spreads, and in Europe the switching point between nat gas and coal generation is a comparison of those two numbers rather than of the fuels themselves.

A power plant is an option

Here the chapter connects to Oil 101’s options material.

The owner of a nat gas plant has the right, and no obligation, to convert nat gas into power at a fixed conversion ratio. That is a call option on the spark spread, struck at the plant's heat rate, exercisable in every hour of the plant's life.

Reading a generator this way explains behaviour that discounted cash flow analysis cannot. A peaker with a heat rate of 11 in a market whose average heat rate is 8 has no intrinsic value, and it is worth a great deal, because it will be exercised in the small number of hours when the market heat rate spikes far above its strike. Its value comes almost entirely from volatility, which is why a market with flat prices cannot support peaking plant however high the average price is, and why Chapter 12's argument about scarcity is really an argument about the volatility a design permits.

It also explains why more renewables can raise the value of a nat gas plant even while reducing the hours it runs. Solar and wind cut the number of hours the plant is in the money and increase the violence of the hours that remain. Fewer exercises, each worth more, and an option that is worth more overall.

Renting the spark spread option

Because generation is optionality, it can be separated from the concrete and sold on its own, which is what the structured market does.

A tolling agreement gives a counterparty the right to run the plant for a period in exchange for a fixed payment. The toller supplies the nat gas, takes the power, and makes the dispatch decisions. The owner has converted a volatile spread into rent, and the toller has acquired the option without buying the asset. The motive on each side is different. The owner turns a revenue stream that depends on hours nobody can forecast into a contracted payment a lender will accept, which is often what makes the plant financeable at all, and hands over fuel procurement and dispatch along with the risk. The toller gets physical conversion capacity without the capital, the permitting or the twenty-year commitment, which suits a retailer short of power in exactly the hours the plant would run, or a nat gas marketer that would rather sell its molecules as electricity when the spread is wide. The owner pays for that certainty by giving up the upside: in the hours the market heat rate spikes, the profit belongs to the toller.

A heat rate call option does the same thing financially. It pays the holder the amount by which the market heat rate exceeds a strike heat rate, settled in cash, with no plant involved and no obligation to generate anything. The payout is notional cash, so a party with no assets can hold the exposure and a generator can hedge without committing its own machine.

Which raises the obvious question of why anybody signs the physical version when the financial one is cleaner. A cash settlement is not electricity. A retailer short of power in those hours needs megawatt-hours delivered at a node, and a payout leaves it holding cash and still short. The toll also references a specific machine rather than a financial index, which removes the basis risk to the node and the shape risk of a benchmark heat rate no real plant has, and it can run for the life of the asset where the financial market thins out beyond a year or two. What the toller takes on in exchange is the plant itself: if the unit is on outage on the day the spread is widest, a financial option would still have paid.

Why the nat gas curve drags the power curve

Chapter 8 established that nat gas sets the marginal price in most hours in most American markets. The trading consequence is that power forwards move with nat gas forwards, and much of what looks like a power view is a nat gas view.

A reader who has been through NatGas 101 already understands storage cycles, basis between hubs, the seasonal shape of the nat gas curve and why a cold January moves Henry Hub. All of that transmits into the power curve through the heat rate, and a power trader who cannot read a nat gas curve is reading half the position.

The relationship is strong rather than fixed, and Chapter 20 showed where it weakens. In hours when solar sets the price, the spark spread relationship breaks entirely, since the marginal unit has no fuel. As that share of hours grows, the correlation between nat gas and power degrades in the middle of the day and holds in the evening, which splits a single relationship into two different ones depending on the hour. Traders now think in terms of the on-peak heat rate, covering the weekday daytime hours when demand is high, and the midday heat rate as separate quantities, and Chapter 24 covers the instruments that lets them.