Negative prices and revenues in the NEM over the past decade

As I look at my NEOmobile screen each working day, I have noticed a very regular spot price outcome. Generally, it is negative prices in both South Australia and Victoria during daylight hours. Figure 1 illustrates an example where both NSW and Queensland also have negative prices. Here we explore negative spot prices through the lens of negative spot revenues as well as their ratio to positive revenues. This paper will also assess the possible outlook for negative prices/revenues and assess other supply and demand metrics that may influence future negative prices/revenues.

Figure 1: NEM spot prices 2 April 2024, 9:15am eastern standard time dispatch interval

Source: NEOmobile

Background

Firstly, what are the drivers of negative prices? Primarily they have been driven by the rapid uptake of rooftop PV and utility scale solar (see Figure 2). Furthermore, solar generation’s output profile is compressed into daylight hours (see Figure 3) which magnifies its impact whereas the other major renewable generation source wind, can generate electricity any time the wind is blowing.

Figure 2: Rooftop PV and Utility-scale solar capacity and output[i]

Source: NEOExpress and AEC analysis

The growth of solar has created the infamous ‘duck curve’ where operational demand during daylight hours has been declining steadily with the expansion of solar capacity. Hence, during daylight hours there is often a surplus of electricity generation output and many plant operators are offering electricity at negative prices.

Figure 3 illustrates the impact of solar generation on average operational demand in Victoria during the day. As can be seen, operational demand during daylight hours averaged almost 6,500 MW in 2010 but by 2023 this had reduced to a range between 4,000 – 5,500 MW. This ‘gap’ in operational demand is now being supplied by rooftop PV and utility scale solar. The right-hand chart shows the average summer generation profile of rooftop PV in Victoria.

Figure 3

Source: AEMO and NEOExpress

Flow Power has published some useful charts that illustrate how negative prices are occurring from a purely economics perspective (see Figure 4). Operational demand contracts and the supply curve shifts to the right due to increasing variable renewable energy resources (VRE) that are more inclined to offer negative bids. Noting that the NEM allows for bids down to negative $1,000 MWh.

Figure 4: Supply and demand (operational) traditional generation mix and high rooftop PV and VRE

Source: Flow Power

There are logical reasons for generators to offer electricity at negative prices and keep generating when the price falls below zero:

• Renewable generators that qualify for renewable energy certificates (RECs) for every MWh produced may earn enough revenue from the RECs they create to offset the negative spot price. This is regularly demonstrated in the market where the negative prices are very close to the value of RECs at that time.
• Coal-fired power stations have what is called minimum generation level (‘mingen’) which is the minimum level of generation they must produce before they start experiencing plant related costs and if a unit is actually turned off it is a slow, expensive process to restart. Hence, to avoid being turned off, they will at least offer their ‘mingen’ of output at very negative prices to ensure they are at least dispatched at their ‘mingen’.
• Generators (especially coal-fired and VRE) are long generation and need to mitigate their spot price risk. To achieve this, they contract or trade some or all of their output at a (generally) fixed price that is independent of spot price outcomes. Hence, these generators are not exposed to negative spot prices and earn their revenue through their contracts.

Negative spot market revenues

Figure 5 reveals how negative revenues have been increasing sharply in recent years and in 2023 represented just over four per cent of net spot revenue. Net spot revenue is defined as positive spot revenue less negative spot revenue. As noted in a previous Energy Insider, 2022 was an exceptionally unusual year with high prices generating net spot revenue that was more than double the 10-year average in nominal terms and roughly double in real terms. This is the most likely explanation for the drop in negative revenues and the very large decrease as a percentage of net revenues from 3.7 per cent in 2021 to 1.1 per cent in 2022. However, 2023 saw the return of the trend towards increasing negative spot revenues.

Figure 5: Negative spot revenues by state ($ million) and per cent of net spot revenue

Source: NEOExpress and AEC analysis

Figure 5 also shows how Victoria is the most affected region in terms of dollars and while Queensland is the second most affected in dollar terms when each regions’ net revenue is taken into account the picture is a bit different. Queensland’s net revenue is nearly 10 times SA’s and twice that of Victoria and NSW has revenues 26 per cent higher than Queensland. The table below takes a closer look at the regions on a monthly basis for 2023.

Table 1: Negative revenues as a per cent of net revenues by month and region 2023

Source: NEOExpress and AEC analysis

As Table 1 shows, the most affected regions are Victoria and SA with 11.3 per cent and 7.5 per cent respectively of their net revenues being negative. In October nearly half of Victoria’s net revenue was negative and 38.9 per cent of SA’s was. Figure 6 illustrates the seasonal pattern of negative revenue percentages for both 2021 and 2023 which broadly approximates an annual solar output profile by month. Generally, the percentages of negative net revenues are lower in the winter months and seem to peak in October and November and remain significantly higher than winter in the other months.

Figure 6: Negative revenues as a per cent of net revenues by month and region 2021 and 2023

Source: NEOExpress and AEC analysis

Short term outlook for negative revenues

Clearly there are a multitude of factors apart from solar generation output that can influence the amount of negative revenues (as illustrated by the 2022 results). While the data series is very short there does appear to be a trend and a non-linear function has been fitted to project  negative revenues if the trend continues.  

Figure 7 displays negative revenues with a curve fitted to the historical data. This curve has been extrapolated out for two years. The left-hand chart includes 2022 and the projection implies negative revenues will increase to around $800 million in 2024 and just over $1.1 billion in 2025. The right-hand chart excludes 2022 for the purposes of fitting the curve and not surprisingly the goodness of fit measure increases from 0.9 to 0.95. If these projections eventuate then negative revenues will be around $1 billion in 2024 and $1.2 billion in 2025.

Figure 7: Negative revenues and trendline

Source: NEOExpress and AEC analysis

Longer term outlook for negative revenues

In this section we will look at the physical factors of demand and generation to see if any inferences can be made about the longer-term outlook. As can be seen from Figure 8, the difference between operational and total electricity consumption (operational plus rooftop solar and on-site generation) is forecast to steadily increase and the bulk of this spread is caused by rooftop solar.[ii] On the surface this indicates further increases in surplus energy during daylight hours and potentially more negative revenues.

Figure 8: Operational and total electricity consumption 2014 to 2049 (TWh)[iii]

Source: AEMO Draft 2024 ISP

Figure 9 digs a little bit deeper into the data and compares storage generation and solar generation.  If a ‘big’ assumption is made, that all storage charging is drawn from solar generation, a net solar output can be inferred. AEMO’s Draft 2024 ISP is forecasting total solar generation of just over 220 TWh by 2049. Deducting forecast storage (utility-scale and Consumer Energy Resources (CER)) output from this results in net solar output of just under 160 TWh.

From this the ratio of net solar output to total electricity consumption can be derived. As can be seen this has been and is expected to continue increasing rapidly until 2026 and then it flattens out at between 23-25 per cent until 2031. Then it stabilises to between 25-27 per cent to 2041. Based purely on these results it would appear likely that we will continue to see negative prices and revenues increasing until 2026 where they may stabilise for the next five years to 2031.

A storage to solar output ratio has also been calculated and this illustrates rapid growth to 2030 then steady growth to be 34 per cent by 2037 and then declining to settle at 29 per cent by then of the period.

Figure 9: Solar net of storage generation and storage generation

Source: AEMO Draft 2024 ISP and AEC analysis

AEMO seems to be supportive of the rationale set out above as shown in Figure 10. While it only explicitly shows CER storage’s impact on operational demand the impact of utility scale storage (and other factors) can be inferred by the flattening of the ‘duck curve’

Figure 10: Impact of coordinated CER on average operational demand by time of day

Source: AEMO Draft 2024 ISP

The data-based analysis set out above does suggest some promising factors with respect to the outlook for negative prices and revenues. However, there are other factors that are likely to resolve the issue, namely competitive market forces. The energy only market is providing the price signals and where possible businesses and industry are likely to respond and change their consumption patterns to take advantage of the surplus electricity. Furthermore, new business models may be developed to utilise this resource. To assist with this new innovative retail and derivative products are likely to need to be developed.  To some extent this is already occurring with the ASX proposing changes to peak futures contracts.   The ultimate aim is to optimally utilise our energy resources into the future with a market that is delivering efficient prices.

 

 

 

---

[i] The author would like to thank Aaron Martinez of the AEC for his assistance.

[ii] Total consumption has been calculated from the Figure 5 data in Draft 2024 ISP. Energy efficiency has been excluded and the data converted to calendar years by averaging across two financial years.

[iii] Certain non-scheduled generators and supply to large market scheduled loads (such as pumped hydro and batteries) are excluded from the operational and total consumption. For further details refer to https://www.aer.gov.au/system/files/AEMO%20-%20Forecasting%20Approach%20Electricity%20Demand%20Forecasting%20Methodology_0.pdf

pp 10-12.