SWIS Rooftop Solar Was Worth Under an Eighth of the Average Price in 2024

One measure, two markets

A value factor weighs each half hour's wholesale price by what solar produced in it, and divides the result by the plain average price over the same half hours. At 1.00, solar's output is worth the average. Below it, solar produces when the market is cheaper than usual; below zero, it produces, on average, when the market is paying to take energy away. The measure is a ratio, so it removes the price level and leaves the timing.

Western Australia's Wholesale Electricity Market changed underneath that measure on 1 October 2023 at 08:00 AWST. Before the reform, the Balancing Market set one SWIS-wide Final Price for each 30-minute Trading Interval. After it, the Real-Time Market clears a price every five minutes, and AEMO publishes a Reference Trading Price for each half hour as the average of its six five-minute prices. A value factor before and a value factor after are computed against different prices, so this report reports them separately, never averages them together, and never draws a line across the break. 2023 appears twice, as two part-years.

The utility series is the output of five PV facility codes, about 251 MW, which Solar Analytica identified by name in AEMO's facility registers because the registers record no fuel. It is a thin fleet, and it grew inside the record: about 51 MW reported at the start, joined by the 100 MW MERSOLAR_PV1 from April 2020 and the 100 MW SBSOLAR1_CUNDERDIN_PV1 from August 2024. The rooftop series is AEMO's estimate of rooftop output, built from sensor data and extrapolated on installed capacity, including what households use themselves. AEMO's estimate starts in November 2020, so the rooftop series starts in 2021.

Solar farms before and after the reform

Before the reform the solar farms' value factor was 0.87 in 2019, 0.71 in 2020, 0.75 in 2021 and 0.79 in 2022, and 0.76 in the part of 2023 before the reform. In 2019 three of the facilities were reporting.

After the reform it was 0.66 in the 2023 part-year, which covers spring and early summer only, 0.85 in 2024 and 0.79 in 2025. A higher figure after the reform is not evidence that solar became more valuable. Three of the five facilities are Semi-Scheduled and can receive Dispatch Caps, and output a farm does not produce in a low-priced interval would drop out of its own weighting and raise its value factor. This report does not measure curtailment and attributes no part of any figure to it.

Rooftop solar before the reform

Before the reform there is one rooftop series, AEMO's 30-minute estimate. Its value factor was 0.61 in 2021, 0.68 in 2022 and 0.61 in the 2023 part-year. These are points, each screened for defects and none corrected, and the screen's findings are set out below.

After the reform, a range rather than a point

From the reform AEMO publishes a second rooftop estimate, a reading every five minutes, beside the 30-minute series it already produced. The two do not agree. Over whole years the 5-minute series' energy divided by the 30-minute series' was 1.0611 in 2024 and 0.9235 in 2025, and neither series is method-stable. Which one is closer to the rooftop fleet's real output is not settled, so neither is chosen. The 5-minute series is also summed to half hours at three alignments, lags of 0, +5 and +10 minutes, because how its timestamps line up with the half hour is itself uncertain. All four members are priced at the Reference Trading Price, and the rooftop value factor after the reform is the range they span.

The range was 0.30 to 0.36 in the 2023 part-year, -0.000071 to 0.12 in 2024 and 0.65 to 0.67 in 2025. In each period the 30-minute series gives the low end and the 5-minute series at lag 0 the high end. No point inside a range, and not its midpoint, is the rooftop value factor.

2024: what rooftop solar was producing into

The 2024 range runs from just below zero to 0.12. On the 30-minute series the value factor was -0.000071: rooftop energy was weighted to a price of -$0.01/MWh against an average of $79.19/MWh across the year. On the 5-minute series it was 0.10 to 0.12, at weighted prices of $8.24 to $9.47/MWh.

Both estimates sit low for the same reason: the prices rooftop output met. Of the 17,272 Trading Intervals included in 2024, 3,783 carried a negative Reference Trading Price, and 50.6 per cent of the 30-minute series' rooftop energy was produced in them, 48.2 to 48.5 per cent on the 5-minute series. Split by half year, the range was 0.31 to 0.42 from January to June and -0.23 to -0.12 from July to December: in the second half of 2024 every series put rooftop solar's value factor below zero.

That floor count is checked against the regulator. Counting floor-priced Trading Intervals over every published price in the regulator's review window gives the same 44 intervals, on the same days, that the Economic Regulation Authority lists in Table 22 of its 2025 price floor review, and the three pre-reform floor-priced intervals in its Table 33, interval for interval. At the five-minute level the counts do not agree: the regulator counts 614 Dispatch Intervals at the floor and this build 617 over every published row in the same window. That difference is recorded and unresolved.

The solar farms' value factor in 2024 was 0.85, with 28.0 per cent of their output at a negative price. This report does not measure why the two fleets differ, and the curtailment boundary above applies to that difference too.

What the rooftop record carries with it

The 30-minute series is screened every year for night readings, held readings, midday zeros and, where the 5-minute series exists, days on which the 5-minute series' energy is below half or above double the 30-minute series'. The night readings are material in no period. The broken days are a different matter. After the reform they can be measured on a fixed sample: every day kept, each broken day's 30-minute values replaced by the 5-minute series, and the value factor compared over the identical intervals. That moves the 30-minute value factor by +0.0236 to +0.0242 in 2024 and +0.0082 to +0.0089 in 2025, against a materiality threshold of 0.005, so the broken days are material in both years. The patched figures measure that materiality. They are not a corrected series and replace no figure in this report, so the 30-minute member, the low end of the 2024 and 2025 ranges, carries those defects as published. Before the reform there is no 5-minute series to patch from, and the broken days' materiality there is not determined.

Two further limits apply to the range. The lag that best aligns the two series reaches past +10 minutes in 7 of the 27 post-reform months, all from November 2024, beyond the three lags the range uses; that is a known limitation, not a reason to widen it. And AEMO's own figures do not settle which series to trust in 2024: the 5-minute series sits furthest above AEMO's Quarterly Energy Dynamics figure in the second quarter, while AEMO's Quarterly Energy Dynamics Q2 2025 says its own figure for that quarter was likely underestimating rooftop output. Agreement with AEMO's quarterly figures cannot test either series, because both are AEMO estimates.

The reference price, checked half hour by half hour

AEMO describes the Reference Trading Price as the average of the half hour's six five-minute Market Clearing Prices, so every included post-reform Trading Interval was checked against that identity: 38,438 of them. Of those, 20 fail it: one in the 2023 part-year, 19 in 2024 and none in 2025. Every one equals, within half a cent per megawatt-hour, the average of the first five of its six prices, so each is classified as a price that omits the final five-minute interval, and none is left unclassified. The headline keeps AEMO's published price. Pricing those intervals at the six-price average instead, or leaving them out altogether, moves no value factor by as much as 0.005.

Why the published price omits the last interval is a hypothesis, not an established fact. AEMO's release notes for its WEM Dispatch Engine 3.0.3, issued on 12 August 2024, list an item increasing the time allowed to calculate the Reference Trading Price; when that change took effect is not known, and whether settlement used a revised price is unverified. Solar Analytica has not asked AEMO.

A second sensitivity weighs five-minute output by each five-minute Market Clearing Price rather than the half-hour reference price. It gives the solar farms 0.66, 0.85 and 0.79 for the three post-reform periods, and the 5-minute rooftop members 0.10 to 0.11 in 2024. It is a sensitivity and does not replace the headline.

What these numbers are not

• Not revenue. A value factor is a ratio of prices. It is not what any solar farm received, a contract price, a settlement amount, or what any household is paid or saves.
• Not one series across the reform. The figures before and after 1 October 2023 are computed against different prices and are not pooled, averaged, joined or trended. Each 2023 part-year is a part-year, not a full-year value.
• Not the whole utility fleet. The five listed PV facility codes are about 251 MW. Emu Downs Solar Farm has no facility code of its own and is not included.
• Not metered rooftop output. Both rooftop series are AEMO estimates from sensor data scaled to installed capacity, and both include what households use behind the meter. Neither is treated as the true output.
• Not a curtailment measure. Nothing here measures output a facility did not produce, and no change in a value factor is attributed to it.
• Not settlement metering. Utility output is AEMO's SCADA, not loss-adjusted settlement meter data. Intervals AEMO marks as replaced are left out whole, and none is treated as zero.
• The SWIS only. The NEM runs a different market with different price and time rules, and no figure here is comparable with a NEM region.

One question about late 2025 stays open: the Reference Trading Price was still published through the year, but whether five-minute settlement replaced it in the final quarter, and whether it remained the right headline price there, has not been confirmed.