Why generation offers diverge from variable cost

This post follows my last one on opportunity cost and how it shows up in the generation offer stack in the NEM. In this post, I consider a range of reasons why generation offers might diverge from variable cost and how pricing at opportunity cost is not a misuse of market power.

Please check out the published articles in the Sources section at the end of the post for published articles on the economic concepts discussed in this three-part series of posts and how they apply to electricity markets.

Wholesale market bids in the NEM diverge from variable cost because, like any other imperfect market, participants have market power sufficient to be able to profitably alter market prices. What follows is an explanation of how the following resources participating in the NEM are motivated to offer or bid according to opportunity cost:

  • fuel-limited generation resources
  • resources with relevant operational inflexibilities
  • high variable cost generation
  • price-sensitive consumption
  • batteries and pumped hydro resources.

Bidding opportunity cost is not necessarily a misuse of market power

Before we start, it is important to consider the concept of market power.

In general, many people are (rightly) distrustful of the competitiveness and outcomes from markets where there is relatively low demand elasticity and a large chunk of consumption is considered an essential good or service. It also doesn’t help when suppliers can set their own prices!

However, most people would agree that prices, as a general rule, should reflect relative scarcity. That is, prices should rise above variable cost in scarcity and fall below variable cost in certain circumstances during overabundance. It follows that generation offers that diverge from variable costs are social welfare enhancing and efficient if they signal genuine scarcity or overabundance and should not be considered a misuse of market power.

Having said that, it is beholden on participants and regulators to monitor the market and call out pricing behaviour that doesn’t appear to be in keeping with market conditions – i.e., pricing power that does not reflect genuine scarcity or abundance causes the good or service in question to be priced artificially low (which could erroneously drive out or discourage competition) or at prices higher than supply conditions warrant (creating artificial scarcity).

Fuel-limited generators price according to opportunity cost

In this examination of situations where opportunity costs affect offers, we will start with fuel-limited generating resources, a constraint naturally associated with hydro generating units. A fuel-limited generator is one that is constrained from generating electricity at full output over a relevant time horizon (could be minutes, hours, days, weeks, months) by a limited supply and stock of fuel.

Hydro generators do not offer all their water fuel at variable cost, which is very close, if not equal, to zero if it is gravity-fed from a lake. As the water stored in the lake is limited (i.e., you can run out if you operate at full output for long enough) and filled only by sporadic rainfall events, pricing your entire stock of water at variable cost is not an efficient use of scarce fuel.

With limited fuel storage and uncertain supply (i.e. lake storage and rainfall), the price a hydro generator is willing to accept to use its water to generate electricity is a function of its existing stock of fuel and expected prices and rainfall in the future. This means the value of its fuel right now depends on how much it might make instead by storing that fuel and using it to produce electricity later (the value of the next best option or opportunity cost). For more information on the economic theory behind the value of water to hydro generators, check out Diana Tam’s article in the sources below.

When water stored behind a dam is being depleted by using it to generate electricity, it is rational and profit-maximising for such a generator to allocate its scarce fuel to generate more electricity when it has the most value, over a given time horizon. As lake storage is depleted the value of a hydro generator’s remaining water tends to increase and vice versa. At one extreme, the value of a hydro generator’s water fuel is at or below variable cost when the lake is full and the generator must spill water. At the other extreme, the value of its remaining water can climb very high as lake storage is depleted toward minimum thresholds of lake storage.

Pricing offers to reflect potential fuel scarcity is not only profit-maximising but societally efficient due to the high value consumers place on electricity supply and an upward sloping supply curve. Conserving fuel for future use increases the price of electricity now, encouraging other, higher-cost, resources and demand response to take its place.

In return, this conservation displaces at least some of the output required of resources and demand response/management at those high prices in a later period. That is, conserving limited stocks of fuel now should mean there are more supply available later when the value of electricity is higher. Put another way, conserving fuel lowers supply now and increases prices now, but this conservation increases supply later and lowers (even higher) prices in this future period. Ideally, this means less higher value demand is shed in the later period.

In the sources below, I’ve included a link to a submission by Snowy Hydro Limited to the AEMC that reveals details about its hydro bidding practices and its thoughts around opportunity cost.

You should note three implications from fuel-limited generators pricing according to opportunity cost:

  1. Prices right now become a function of how market conditions and prices are expected to turn out in the future.
  2. Reallocating scarce fuel from intervals of lower prices to intervals with higher prices tends to make prices converge in the time horizon over which limited fuel is being allocated, which could be hours, days, weeks or even months. When offers and prices are equal to variable cost, the price in every trading interval is equal to the short-run marginal cost. This is not the case when opportunity costs affect prices over multiple trading intervals.
  3. Because generation offers and prices depend on expected future market conditions and rely on competition to increase or decrease generation, a fuel-limited generator’s offer is implicitly a function of the offers of every other generator. This insight has policy implications I’ll talk about in my next post.

Thermal power stations can also experience fuel scarcity

Fuel scarcity is a natural consideration for the owners of hydropower stations. However, from time to time, fuel supply problems at coal and gas power stations can also force traders to switch to the same fuel-limited trading behaviour as hydro generators. There were reports of coal supply problems in June 2022 (see slide 24 in Origin Energy’s FY2022 results presentation, for example) just as there were several years ago in the months that followed the surprise announcement of the imminent closure of the Hazelwood power station.

It is also possible that large jumps and falls in the price of thermal fuel in the future (another sign of changing fuel scarcity) can affect offer prices now – if they are known, these may be signalled over weeks or days rather than changing from one trading interval to the next on the day they occur. Advance signalling of the increasing likelihood of higher cost fuel is similar to hydro generators’ efficient trading behaviour so it should not be automatically considered to be a misuse of market power.

Operational inflexibilities force opportunity cost considerations

Opportunity cost is also a relevant consideration for any generating unit not flexible enough to turn on and ramp up to full output within a single trading interval or ramp down from full output to zero within a single trading interval. This inflexibility forces generators to consider and price their offers according to the potential revenue and associated shutdown/startup operating costs over multiple trading intervals according to the limits of their flexibility, causing prices to diverge from variable cost in each trading interval. Wind and solar PV generation may also have start up and shut down inflexibilities. However, on its own, being output-constrained by the availability of wind or sun just makes the quantity of electricity they can produce and sell more variable than gas- or coal-fired turbines.

A profit-maximising generator considering whether to ramp up (or down) or turn on (or off) estimates the cost/value of the alternative over the number of intervals affected by its inflexibility. The decision is further affected if the generating unit incurs extra costs when it starts up or shuts down or if it must maintain a minimum level of output to operate safely or stably. Generating units might also need to remain off for a minimum period of time after they are turned off or they may prefer to remain running for a minimum period before they are turned off.

These operational inflexibilities and costs are different from the recovery of other fixed costs because they are related to output. However, they have a binary (i.e., operating/not operating) over a specific period (shutdown/startup time), rather than a linear, relationship with output and often have a significant effect on a generator’s decision to turn on/off or remain on/off. For example, the threshold offer price for turning off is signalled in offer prices below the generator’s variable cost while the threshold price to turn on is above it to reflect a desire to cover any significant startup/shutdown costs.

For instance, some thermal generators incur costs or a heightened risk of tripping the unit if it is dispatched below a minimum safe operating threshold so they usually offer this minimum quantity at prices in zone 1 (i.e. at or below zero). This ensures that prices must usually be very low for long enough to warrant switching of the generating unit(s). In this way, consideration of opportunity cost can cause wholesale market prices to be lower than variable cost, not just above it. Understandably, if prices are expected to be below variable cost for some time, then eventually switching off becomes the lower cost alternative (ie cost of absorbing prices below variable cost is higher than costs associated with switching off) and the generator amends its offer, either bidding unavailable or raising its offer prices well above the prevailing price.

Note that operational inflexibilities affect renewable resources, too, and not only due to engineering considerations such as ramping and minimum operating limits. For instance, hydro generators may operate according to complicated instantaneous and time-based restrictions on the use of water. For example, water use must be below a specified flow rate per hour or higher/lower than a specified throughput per day under specific conditions.

Generators with high variable costs may also price at opportunity cost

Generators with high variable costs have a problem recovering their fixed costs. Often referred to as peaking generators, they operate only occasionally so they have to recover their fixed costs (including economic profit) in short periods to remain profitable.

Because competition is imperfect, market conditions can become tight and generators have market power when they do. At those times, merchant (uncontracted) peakers are unlikely to offer all their generation at variable cost or they would not cover their fixed costs even if they were flexible enough to turn on and reach full output in a single dispatch interval and switch off in the next.

However, having the NEM the widest range of wholesale electricity prices in the world helps mitigate against free-for-all merchant pricing by peaking generators. The potential high and low prices create large financial risks that encourage retailers and generators in the NEM to agree forward contracts that help cover their fixed costs and incentivise them to offer the contracted quantities at variable cost.

Price-sensitive consumers consider opportunity cost

Electrification throws up more activities that consume electricity that may be price sensitive and constrained by inflexibilities. Much like a peaking generator, the traditional demand response is to switch off devices using electricity when prices reach very high levels, which occur only occasionally. Similar to peaking generation, the total cost associated with an upcoming period of high prices (or benefit of an incentive payment) is compared with the utility or profit from the use of electricity by some device (i.e. opportunity cost).

There are also some discretionary processes, such as the production of ammonia, methanol or hydrogen, that could be bid so as to consume more electricity when it is of low value and less (or none) in times when it is high cost, over a given time horizon, referred to as “normally-off” scheduled load. Some of these discretionary processes might also be limited by production orders, meaning they could behave much like the opposite of hydro generator – i.e. producing only what they need to fulfil orders and choosing to do so at the lowest prices over a given time horizon. A similar situation might occur in the future for hydrogen production.

Batteries and pumped hydro are special

Batteries and pumped hydro resources are special because both their discharging/generation and charging/consumption are limited by their storage capacity. When they aren’t operating to provide network support or FCAS, their wholesale market trading is sometimes referred to as temporal arbitrage.

Arbitrage is the practice of buying something in a market at a lower price and simultaneously selling it in another market at a higher price. Temporal arbitrage is similar but it involves taking advantage of price differences in a single market at different times – buying low now and selling high later (the same market becomes two when split between time periods).

When they are engaged in temporal arbitrage, both the costs and revenues of pumped hydro and batteries are closely linked to wholesale prices and they seek to maximise their revenues and minimise their costs by taking advantage of the price spread in the future, not the absolute prices themselves. A swing in the price from $10,000/MWh to $10,450/MWh is the same opportunity to them as a price swing from $100/MWh to $550/MWh. Check out the paper by David Andrés-Cerezo and Natalia Fabra in the sources section for more information how temporal arbitrage works to reduce price volatility.

Prices diverge from variable costs over different time horizons

Note that the effect on wholesale prices from decisions based on opportunity cost varies depending on the time horizon relevant to the traders making those decisions. The effect of inflexible start-up and shut-down times on wholesale prices could extend from one or two five-minute trading intervals up to many hours ahead, in keeping with the fuel resource and their limitations in a particular market.

For example, the storage capacity of batteries and many pumped-hydro schemes are typically measured in hours. However, hydro lake storage can range from fractions of days to years, for example:

  • The Scandanavian and some South American electricity markets includes hydro lakes storing multiple years of demand for electricity. This makes them slow to fill and drain and causes prices to move equally slowly.
  • Hydro lakes in Tasmania can store over a year’s consumption of electricity in the State.
  • Hydro lakes in New Zealand hold several weeks’ worth of nationwide consumption.
  • Storage available in states of the NEM, other than Tasmania, likely sums to only several hours of regional consumption, at most.

Note that, in general, if there are no constraints on generating capacity to meet demand, then materially increasing the level of fuel/charging storage will slow the rate of change in wholesale prices in response to changing market conditions and vice versa.

What changes are implied by this and the transition?

So now we understand some of the ways in which resource limitations affect offers, my third and final post will explore how things change as the NEM transitions to a low-emission generation mix and what policy changes we might usefully make.



Short run marginal cost – Technical paper, Adam McHugh, ERAWA, 11 January 2008

Opportunity cost – the road not taken, EconClips on Youtube, 14 June 2018

Bidding in energy-only wholesale electricity markets – Professor George Yarrow,

assisted by Dr Chris Decker, written for AEMC, Nov 2014

Water is valuable: the allocation of water and other resources in the New Zealand electricity market, Diana Tam with Prof. Lew Evans, June 2013

Structure matters – storage in electricity markets, David Andrés-Cerezo and Natalia Fabra, December 2020

Submission on opportunity cost methodology from Snowy Hydro for AEMC consultation on Snowy Hydro Limited compensation claim


This post was originally published on LinkedIn. Reproduced here with permission.


About our Guest Author

Greg Williams is a Principal Policy Advisor at the Australian Renewable Energy Agency (ARENA).

You can view Greg’s LinkedIn profile here.

Be the first to comment on "Why generation offers diverge from variable cost"

Leave a comment

Your email address will not be published.