A rule change on transmission losses for wind and solar farms could save billions for consumers

(Matthew Henry / Unsplash)

A small fraction of electricity is lost via electrical losses on the transmission network. Someone needs to pay for these losses, and it is not unreasonable that this cost is borne by the generators. However, Australia’s national electricity market has a rule that results in generators being charged for twice the actual electrical losses.

This rule is likely to add tens of billions to consumer electricity bills over the next few decades, lead to a bias against renewable generation, an inefficient build out of renewable generation with increased social license issues and even adds complications to setting a high renewable electricity target. These issues could be greatly reduced by charging generators for actual losses, or via a loss factor hedging market.

The Australian Electricity Market Operator (AEMO) is responsible for ensuring that electricity supply always meets demand, and for facilitating payments from consumers to generators for the electricity that they use. Some generators are supplying distant loads and lose a higher proportion of their generation in transmission losses. It makes sense that these generators are paid less per unit of generation, to account for these losses. What makes less sense is to charge generators for twice their electricity losses. However, this is how our electricity market has been designed.

The justification for this is the assumption that pricing electricity at the margin will lead to the most economically efficient market.

The spot electricity market has been designed so that all generators get paid the same price as the very last MW that is needed to meet demand. To transfer this concept to electrical losses, instead of charging for the average electric loss, it charges for the electrical losses that applies to the very last MW of generation.

Electrical losses increase in proportion to the square of transmitted power. Simple algebra demonstrates1 that the electrical loss that applies to the very last MW is twice the average electrical loss.

To introduce some jargon, AEMO uses the marginal loss factor (MLF) instead of average loss factor (ALF) to determine how much generators are paid. If the actual electrical losses for a generator are 3%, then the loss factor applicable to the last MW is 6%, and AEMO would assign to the generator a MLF of 0.94.

As a result, it gets paid for 94% of its generation. If AEMO were instead to charge for actual losses, then it would receive an ALF of 0.97 and get paid for 97% of its generation.

MLFs discriminate against wind and solar

The difference between the ALF and the MLF represents the amount that a generator is over-charged for transmission losses. Generally, generators that are further away from demand centres are overcharged to a much larger extent, and these tend to be wind and solar generators, as shown in Figure 1.

This figure also shows that MLFs tend to be getting worse at a rate of more than 0.01 per year for solar farms and 0.004 per year for wind, while coal and gas show a slight positive to no trend. In short, wind and solar is being overcharged more than other generators, and the degree of over-charging is getting worse each year.

Figure 1: Wind and solar tend to have worse MLFs than fossil generators, resulting in a larger difference between ALF and MLF (left). MLFs are steadily getting worse for wind and solar generation (right)

MLFs add uncertainty and increases finance costs

It is expected that Australia’s main electricity grid, the NEM, will see over 100 GW of new utility solar and wind farms costing almost $200b built over the next few decades2. This cost will ultimately be borne by consumers. The total cost that is passed on to consumers is also heavily dependent on the cost of financing. Every extra percent of financing cost will add over $30b to the cost borne by consumers.

The cost of financing is mostly determined by interest rates, but also depends on risk premiums. Uncertainty in revenue forecasts add risk and increase the cost of financing.

It is impossible to accurately forecast loss factors many years in advance. To do so requires precise knowledge of when and where new generators are going to be built, how rapidly the population will install rooftop solar, changes to transmission and how demand is going to change at all locations across the grid.

MLFs were relatively stable a decade ago when changes in supply and demand were small. But the electricity grid is currently undergoing a massive transformation from one powered mostly by a small number of large coal generators to one powered mostly by hundreds of wind and solar farms and millions of residential PV systems.

Electricity flows are changing greatly, and loss factors are changing with them. Generators located far from demand centres are most at risk of volatile loss factors, and these tend to be wind and solar farms, as shown in Figure 2.

Figure 2: Volatility in wind and solar farm MLFs is far greater than volatility in coal or non-peaking gas generation.

The variability of loss factors is greatly adding to revenue risk for new renewable projects. It is plausible that marginal loss factor uncertainty is adding 1% to the cost of capital for wind and solar developers, which would add approximately $5.50/MWh to the cost of new wind generation and $3.50/MWh to the cost of new utility solar.

Figure 3 shows that it could add over $25 billion to consumer costs over the next 27 years, and another $8 billion more in the following decade.

Figure 3: Additional annual cost to consumers due to a 1% in the cost of capital for new utility solar and wind farms.

MLFs do not reward generators for reducing transmission losses

Imagine your house is suffering from large electrical losses. You get an electrician to come and fix it. Once the electrician has finished the work, you claim that the electrical losses are now fine, so you refuse to pay them.

This is probably not legal, and at the very minimum is not a good approach to get repeat work from an electrician. But that is how our electricity market works when it comes to loss factors.

To explain, let’s look at the situation in the Broken Hill region. In 2016/17, Broken Hill solar farm was generating a little over 100 GWh/y and was assigned an MLF value of 1.12, as shown in Figure 4. A MLF greater than one shows that the generator is getting paid for more than what it generated.

This is not because transmission losses are negative, rather, it is because the generator is displacing generation that would have higher transmission losses. In other words, the addition of Broken Hill solar farm was reducing losses on the network.

Figure 4: MLFs and annual generation for Broken Hill Solar Farm and Silverton Wind Farm.

The high value of the MLF in the Broken Hill region is supposed to send a price signal that more generation should locate in the area.

The price signal worked, and a couple of years after the construction of Broken Hill solar farm, construction of Silverton wind farm also begins. First generation is in 2017/18, but it is not till 2019/20 that generation starts approaching full capacity. In the space of 2 years, Broken Hill solar farm’s MLF plumets almost 40% from a high of 1.25 to 0.76.  Silverton wind farm’s MLF drops 22%.

The large change in MLF to a value much less than one indicates that total generation from both Silverton wind farm and Broken Hill solar farm was too big for local load, and so much of the generation instead had to be exported away from Broken Hill suffering electrical losses along the way.

But even if Silverton wind farm was more appropriately sized to the local load, and reduced the MLF to 1.0, it would not have received the benefit of reducing transmission losses. Just like refusing to pay the electrician who fixed the losses in your house, so too does MLF fail to compensate generators who reduce electrical losses.

Payment is determined by the loss situation after the generator is producing, not by whether the generator increased or reduced losses.

The changes in MLF heavily penalised the Silverton Wind Farm that was too large for the remaining local load, but it penalised the Broken Hill Solar Farm even more heavily.

This is despite the solar farm dramatically reducing losses in the region, not being oversized relative to load, and having no plausible way of forecasting or influencing the decision to build the wind farm or its eventual size.

A hedging market might make MLFs acceptable

The situation described above is a natural result of marginal pricing. It is not unique to electrical loss factors. But in many cases, particularly where investments worth many billions of dollars are involved, there is the opportunity to sign contracts to hedge this marginal pricing risk.

As an example, the electricity spot market has been designed around marginal pricing. But it is virtually impossible to finance a new wind or solar farm based on spot market revenue alone, it is far too volatile and risky. Nearly all new renewable energy projects sign some form of power purchase agreement to manage spot pricing risk. Would it be possible to hedge MLF risk?

MLFs are also used to determine costs for load (consumers). Costs are given by consumption multiplied by the MLF. Returning to the Broken Hill region, in the 2-yr period as Silverton wind farm ramped up towards full capacity, the MLF for load at the 220kV sub at Broken Hill reduced by 20%, from 1.05 down to 0.83.

Before Silverton started producing consumers were charged for more than they used, due to the high electrical losses involved in meeting demand in that region. After Silverton Wind Farm ramped up to full power, consumers were charged for less than what they consume, as any extra demand then reduced transmission losses.

Before Silverton Wind farm was constructed, it is plausible that a financial contract could have been reached between the Silverton wind farm and Broken Hill consumers to fix the MLF for Silverton wind farm at some value between the what the MLF would have been with or without Silverton generation, perhaps a value around 1.0.

Consumers should benefit from the MLF reduction caused by Silverton, so they could share some of the benefit with the wind farm to encourage its construction. Such an arrangement would be highly analogous to consumers signing a power purchase agreement (PPA) with new generator.

Just as PPAs reduce exposure of both consumers and generators to volatile spot prices, a MLF contract could shield both consumers and generators from volatile MLF values.

Alas, there is no market for fixing MLFs, and it is hard to imagine there ever being one. Consumers mostly buy their power through the large retailers and it is not clear how the retailers pass on changes in local MLF values to consumers.

It also seems unlikely that the retailers would willingly and in good faith participate in a contract to fix MLFs for new generators, particularly if those generators may be competing with some of their own generation assets.

Thus, the situation remains that wind and solar farms are reducing loss factors for consumers in the regions but are getting none of the benefit for doing so.

MLFs can lead to inefficient build of generating assets and increased social licence issues

It is commonly stated that marginal pricing leads to the most economically efficient solution. But a very simple example shows that this is not always the case.

 Loss FactorLCOE
 ALFMLFpre-losspost ALFpost MLF
Table 1: Loss factors and levelized cost of energy (LCOE) for two hypothetical generators.

Consider a simple network with a load and two proposed generators to supply that load. One generator (G1) is located close to the load, and is forecast to receive an ALF of 0.98 and MLF of 0.96.

Its levelized cost of energy (LCOE) is $55/MWh before losses, $56/MWh after the average loss factor is applied, or $57/MWh after the MLF is applied. The second generator (G2) is located further from the load, but has a lower cost of energy before losses and post ALF, but higher after MLF, as shown in Table 1.

Under the current market rules where MLF is used, G1 would be built in preference to G2, as it has a lower post-MLF LCOE. But it is G2 that is would be able to supply generation to the load more cheaply, because the post-ALF LCOE reflects actual losses. The choice to use MLFs for settlement would bias the playing field away from G2 even more heavily if G2 had a higher cost of capital due to increased MLF uncertainty.

Furthermore, it is wind and solar farms that tend to be located further from demand centres. They are the ones bearing the brunt and uncertainty of being charged for twice actual losses. In the example above, it is plausible that G2 is a wind or solar farm while G1 might be a coal generator. The bias against projects with a lower MLF is usually a bias against renewable generation.

But even if G1 and G2 are both renewable projects, the use of MLF biases generation towards wind and solar farms that are located closer to consumers than would be the case if generators were charged for actual losses. Placing wind and solar farms closer to consumers leads to increased social licence issues, which is one of the key risks that could delay our transition to a mostly renewable grid.

MLFs amplify forecasting errors

MLFs are fixed for each generator for each financial year. In April of this year, AEMO released MLFs for the 2024/25 financial year using data from 2022/23. Loss factors are calculated using data that is two years older than the period for which they apply. Because of this 2-year delay, AEMO needs to do significant modifications to the data to estimate how power flows might change over that 2-year period.

In late 2020 AEMO presented analysis of how well their MLF forecasts for 2019/20 compared to what the loss factors would have been if AEMO had perfect foresight. Figure 5 shows the results. In most regions, the forecasts weren’t too bad, just out by a percent or two. But note that even a 1% error will affect revenue by 1%, which can have a large impact to the profitability of a renewable project. However, two regions, South-West NSW and North-West Victorian had huge errors of around 10%.

The main reason for the error was that curtailment in the region was much higher than forecast. Heavy curtailment of the renewable generators resulted in much less power being exported from the region, greatly reducing transmission losses.

Projects in these areas were hit doubly hard. Not only did they suffer higher than expected levels of curtailment, but they were also charged 10% more for losses than they should have been had the forecasts had been more accurate. Indeed, generators in the NW Vic region were charged for 15% losses, they should have been charged 5%, but actual (average) losses were more like 2.5%.

Figure 5: historical versus forecast MLFs. Source: AEMO Meeting Pack MLF Forum 16 December 2020

If the average loss factor (ALF) was used for determining how much generators were charged for losses, then these forecasting errors would have been half the size. A 10% error and loss of revenue using MLFs would instead be a 5% error and loss of revenue using ALFs.

These errors are not a criticism of AEMO, indeed it is likely that much of the error came from the developer. However, if AEMO struggles to forecast MLFs only 2 years in advance, what hope does a developer have in forecasting the MLF over the 25- or 30-year life of a wind or solar farm? And the main point is that the decision to use MLFs instead of ALF has the result of amplifying errors by a factor of two.

There is no mechanism for generators to be compensated for these errors in estimating loss factors. It is no surprise that investors are requiring higher financing costs to manage MLF risk.

MLFs will complicate the setting of renewable electricity targets

When a wind or solar farm generates electricity, it also creates renewable energy certificates (LGCs, to be replaced by REGOs after 2030). The number of certificates it creates is given by the amount of MWh it generates multiplied by the MLF.

If a government, company or any other entity wants to claim it is powered by 100% renewable electricity, then to prove this claim it must obtain and then surrender renewable energy certificates equivalent to its usage.

Let’s assume all users in Australia want to claim that they are powered by 100% renewables, and lets also assume all wind and solar farms have a MLF of 0.9, implying 5% of generation is lost via transmission losses. To supply 100% of demand, they will need to generate a little over 105% of demand to allow for those 5% losses.

The number of certificates created will therefore be equivalent to 95% of demand (0.9 x 1.05). We find ourselves in the absurd situation that there will not be enough certificates for everyone to claim that they are 100% powered by renewables, even though renewables were supplying 100% of demand.

The rule that means generators are charged for twice actual losses has now complicated the setting of renewable energy targets. If MLFs are used for determining renewable certificate creation, then the regulator will have to estimate a correction factor to adjust for this mismatch.

What about the 2018 rule change request to use ALF instead of MLF?

In November 2018 Adani Renewables submitted to the AEMC a rule change request seeking to change from using MLFs to ALFs. In February 2020 the AEMC made a final decision to deny the request.

The main arguments for the refusal were:

  1. The use of ALF goes against the fundamental economic principle of marginal pricing, and would reduce efficiency of dispatch and dampen locational signals for efficient new investment.
  2. Commissioned modelling indicated that the use of ALF would lower the wholesale market price, but this would be more than offset by a reduction in the intra-regional settlement residue (IRSR), so the net result for consumers would be an increase in prices.
  3. Consumers have no means to manage transmission losses

Regarding point 1, as explained in an earlier section, the use of MLFs can result in a less efficient build out of generating assets, which in turn would raise costs for consumers.

For point 2, it is worth emphasising that this analysis was focused entirely on spot market outcomes and the intra-regional settlement residue (IRSR). Because AEMO charges for twice actual losses, it usually collects more money from customers than it pays out to generators. This residual is called the Intra-regional settlement residue (IRSR). AEMO distributes this settlement residue to the transmission network service providers (TNSP) who in turn pass it on to consumers via a reduction in their transmission charges.

It is widely acknowledged that the use of ALF will reduce spot prices but also reduce the IRSR. Whether the reduction in IRSR outweighs the benefit of lower spot prices is probably in dispute. Section 11.1 of AEMO document “Treatment of Loss Factors in the National Electricity Market” shows a very simple example illustrating that the reduction in market price exactly offsets the reduction in IRSR. The AEMC commissioned research, which was more complicated, though not necessarily more accurate, found that customer payments would be $42 million higher using ALFs. It was also higher under most of other sensitivities that they modelled.

However, these spot market and IRSR outcomes are 1-2 orders of magnitude smaller than the outcomes that result from higher financing costs for new renewable development. The former is of the order of 10s of millions of dollars per year, the latter is of the order of billions of dollars per year.

The AEMC paid lip service to concerns that financing costs increase due to MLF uncertainty. The justifications were:

  1. The cost of debt had been falling, so that any relative decrease in the credit rating would have a lesser impact on the cost of debt compared to a scenario where the cost of debt was increasing.
  2. While the cost of capital would increase due to a reduction in gearing level, market analysis suggested that electricity developers and asset owners have a relatively low-risk rating.
  3. Overseas investors have committed substantial funds to invest in Australian generation assets.

Point 1 appeared short-sighted even in 2020 and clearly no longer applies. But the bigger issue is that none of the points address the impact of lowering the cost of finance, and that this is likely to deliver vastly greater benefits to consumers than any change to spot market outcomes. Despite this, the AEMC stated that “increases in their cost of capital does not outweigh the reduction in efficient investment signals and dispatch decisions that would occur across the NEM or the impact on the affordability of electricity for consumers”. It is hard to see how they came to this conclusion.

It is also worth mentioning that the NEM is rapidly moving towards a grid where most supply comes from zero marginal cost generation assets. The importance of remaining devoted to the economic dogma of marginal pricing is diminishing each year. Efficiency of dispatch is much less important when most generators have zero marginal costs. Moreover, there are already several prominent aspects of the electricity market which deviate from marginal pricing:

  • Power purchase agreements and hedging contracts
  • Distribution loss factors are calculated on an average rather than marginal basis
  • Marginal loss factors are fixed for a year, so can significantly deviate from the actual marginal loss at any given moment in time
  • The proposed Priority Access component of Transmission Access Reform, which will see existing generation favoured over newer generation facilities. Indeed, this would be highly analogous to locking in MLFs for generators for a period of time, generally advantaging them relative to newer generators

In short there are many sensible exceptions to deviating from pure marginal pricing. There doesn’t seem to be a strong reason for transmission loss factors to be calculated at the margin, and there are many good reasons for it to not.

An updated design paper for the Capacity Investment Scheme (CIS) has recently been published. The CIS is being designed to provide renewable developers with a more certain revenue floor to help projects achieve financing. The current design is to exclude both negative pricing risk and MLF risk, leaving generating assets fully exposed to these risks. These are two of the biggest risks faced by renewable developers. It is understandable that the Government does not want to distort the market too greatly with regards to negative pricing. It is important that projects are not rewarded for generation that is not valued3. But MLF risk arising from a decision to charge for twice the actual losses is a different matter. There is no great risk to supply and demand or the market in general if generators are charged for actual losses.


The decision to charge for twice actual transmission losses represents a large risk to developers of new generation assets. The use of marginal loss factors without the ability to hedge results in:

  • Increased financing costs, likely to cost consumers tens of billions of dollars over the next few decades.
  • A bias to fossil over renewable generators.
  • A less economically efficient build out of generation facilities.
  • A bias towards generators closer to demand, which will increase social license issues and possibly jeopardise the transition to a low emission grid.
  • Complicate the setting of renewable energy targets.
  • Amplify errors in forecasting loss factors.

These are all strong reasons to consider a move away from marginal calculations of transmission loss factors.

  1. See section 5.1 of https://aemo.com.au/-/media/files/electricity/nem/security_and_reliability/loss_factors_and_regional_boundaries/2016/treatment_of_loss_factors_in_the_nem.pdf?la=en ↩︎
  2. [Source: AEMO Draft 2024 ISP combined with analysis from author ↩︎
  3. A valid exception to this rule relates to delays in the retirement of a coal generator. The delay is likely to greatly increase the incidence of negative pricing, and it is reasonable that new generation that was built to replace that generator is not penalised for a delay in its retirement. ↩︎
Get up to 3 quotes from pre-vetted solar (and battery) installers.