Last week saw several media reports of a new analysis of the NEM predicting five large coal plants may close by 2025 (over and above Liddell, already scheduled for closure in 2022-23). The report is by the Institute for Energy Economics and Financial Affairs (IEEFA), in partnership with Green Energy Markets.
It’s a big call from agencies with a pedigree that is more anti-coal than anchored in cold reality. But that doesn’t make the claim wrong. Since 2009 economists have predicted that the closure of large coal fired power stations would create temporary price relief for the survivors. This has been experienced first hand after the closures of Northern in 2016 and Hazelwood a year later. Is anything different now?
The first red flag is that the report is not based on market modelling, but rather on the thesis that current rates of renewables investment will continue; flat underlying demand, and a couple of price scenarios.
It reasonably assumes that renewables, with their zero marginal cost, will underbid coal plant (and gas and hydro) and so displace it by getting dispatched ahead of it. Coal plant revenues are expected to decline significantly from the reference year of 2018.
Why 2018? Well that is the year after Hazelwood closed, when wholesale prices shot up. It was also the start of the big renewables boom, as developers raced to get projects commissioned before the Renewable Energy Target plateaued in 2020. So, it allows the report to show much bigger numbers for falls in coal plant revenue and renewables growth than if they started in 2020.
Individually none of the input assumptions are especially outlandish. The report forecasts 28GW of new entrant renewables (including 15GW of rooftop solar) from 2018-2025. Given 2020 saw 6.9GW of combined utility-scale and rooftop renewables and 2019 saw 6.2GW, a rate of 4GW a year is not especially bullish. Especially since the likelihood of a slowdown due to falling wholesale electricity prices is limited – state renewables schemes will keep the pipeline going regardless of price signals, and rooftop solar is “competing” against the retail price of electricity rather than the wholesale price.
Major new sources of demand and demand growth appear unlikely before 2025: the economy needs to recover from the impact of COVID restrictions and the likely candidates for new demand: hydrogen and electric vehicles (EVs) will need a few years to really ramp up. A demand fall – if large industrial consumers like the Portland smelter close – is at least as likely.
Second red flag is that prices are assumed, (unlike market modelling, price is an input here, not an output). So wholesale electricity prices are assumed to be similar to current prices (their scenario A uses an average price of $50, the same as last year, while their scenario B is $34). This may be a reasonable assumption if the only change on the supply side is the continuing influx of new renewable generation. However the implication of their analysis is coal exit and prices are unlikely to stay as low in the face of thousands of megawatts of dispatchable generation exiting the market over a four year period. In turn this raises the issue of reliability. The report claims reliability is not under threat, but the analysis does not appear adequate to support that claim. This would require testing scenarios of cloudy, still days when renewable output is low.
This exposes the most obvious flaw in the analysis: the profitability of all coal plants is based on none of them closing, but then claiming several will lose money so will probably close early. But if one did close, that would improve the fortunes of the rest, as this analysis by Aurora Research explains.
Even if their guesstimates of wholesale prices prove correct, the revenue numbers are likely understated for a few reasons.
- They assume generators only earn the spot price, whereas most coal plant output is hedged and these contract prices are often at a (modest) premium to spot prices. This demand for guaranteed supply in the future (from retailers and large customers) doesn’t change after a coal generator closes. Supply contracts and so the value of these contracts increases, immunising other coal generators.
- More renewables and less coal generation will result in greater price volatility, with prices falling the most at times of abundant renewables supply. This won’t hurt hedged coal generators, it will hurt uncontracted renewables generators.
- They ignore ancillary services revenues, such as frequency control. While a small part of revenue today, they are likely to grow especially as AEMC is considering adding new markets to keep the system secure.
- They overestimate the displacement effect of renewables, because they don’t account for curtailment of renewables when there is more renewable output than the system can handle. This is growing and is likely to grow further, given renewables output often coincides.
On the generator cost side, they use figures from AEMO’s ISP modelling. But using these costs even when they result in losses ignores the logic of markets – when a business’s revenues diminish, they typically find ways to cut costs. This may not be easy for coal plants, but it’s not impossible.
Finally, the 2025 figures are just a snapshot. Owners of long-lived assets like coal plants will usually ride through a bad year or two if they are confident that things will change in the future. The reality of the complex interactions of demand and supply in the NEM is that there is no inexorable price trend in one direction. Historical prices look more like a random walk.
In summary, the analysis is high-level but useful. Like any forecast, it’s going to be wrong in the specifics, but could be right in the general story it tells. And the basic story that Australian coal plants are on their way out and many will close before their current scheduled date is not a new one (unless you’re Barnaby Joyce or Matt Canavan…). More authoritative sources, such as Energy Security Board chair Dr Kerry Schott, have said as much.
But will three to five coal plants close by 2025? This seems a stretch. As the report itself admits, more sophisticated modellers such as Aurora Research and Frontier Economics, are not coming up with this kind of result. These consulting houses actually get paid based on the quality of their forecasts for clients. When they start predicting such outcomes, then it is time to take these claims seriously.