Closure notice rules not as simple as they sound

Last week, just before the federal election date was announced, the Federal Energy Minister submitted a rule change to the Australian Energy Market Commission (AEMC) to extend the existing National Electricity Rule (NER) that obliges scheduled and semi-scheduled generators to provide 3 ½ years notice of closure to five years.

The Australian Energy Council responded with caution. Early market information is certainly beneficial, but attempting to mandate fixed dates in this way introduces complexities, is not readily enforceable and is potentially counter-productive. Importantly, the thinking behind such proposals appear to confuse forecasts and commitments.

We thought it useful to take a look at the history of notice of closure rules and the issues they raise.

Hazelwood and Finkel  

Prior to 2018 there were no obligation to provide an “official notice of closure” apart from the non-binding forecasts that the Australian Energy Market Operator (AEMO) requested from generators in order to prepare its planning outlooks, such as the Electricity Statement of Opportunities (ESOO). A number of significant plants had closed through the 2010s, including Northern, Playford B, Anglesea, Morwell, Munmorah, Wallerawang and Swanbank B. Many of these gave accurate, long-term forecasts, whilst in some cases, such as Northern, operational circumstances changed unexpectedly quickly.

Everything altered, of course, following the Hazelwood closure with  a 5 month official forecast of its closure at the end of the 2016-17 summer.  That was followed by a period of tight reliability in Victoria and South Australia and raised wholesale prices across the National Electricity Market (NEM).

Hazelwood’s plant circumstances had been challenging for some time. In early 2016 it was still hoped these issues could be economically rectified, in which case the station could have kept running. It is not clear that had an official “notice to close” obligation existed, that the outcome would have been any different.

Nevertheless this event coincided with the Finkel Review, which recommended an obligation to provide at least 3 years notice of closure be placed on large generators. The Finkel Review’s recommendation seized on the obvious advantages of providing more predictable forecasts. Although it is not clear the Review adequately worked through its practicalities.

Given the political desire to support the Finkel Review at the time, the recommendation was immediately endorsed by governments, and the Energy Security Board (ESB) put forward a successful rule change. The result is that all scheduled and semi-scheduled generators[1] have to provide an indicative closure date to AEMO, which is published. It is non-binding until the last three years, subsequently extended to 3 ½ years.

How the existing rule works

Under NER 2.10.1(c3), generators must list their proposed closure dates, and are allowed to extend them at any time, but may not bring them forward in the last 42 months unless granted an Australian Energy Regulator (AER) exemption.

There is an obvious asymmetry, in that a near-term plant closure can be deferred, but not brought forward. This was intentional: Finkel’s objective was to create an “overlap” in new capacity ahead of a closure. There was never an intention to inhibit the economic deferral of a planned closure.

Under what circumstances may the AER provide an exemption? The most obvious is a plant failure: it’s pretty hard to keep a plant running after it has blown up. But less extreme cases exist. For example a plant may receive a corrective order from a safety or environmental regulator. If money were no object then such things can be fixed, but clearly it makes no sense if the plant is scheduled to close in the near future anyway.

The AER’s guideline acknowledges that changes in circumstances will require both a technical and economic assessment of the closure plan, which recognises that contemporary market conditions will be taken into account. Whilst few would disagree with such common-sense, it raises an obvious question of what the original rule is trying to achieve.

Under Corporations Law, company directors have obligations to not operate insolvent businesses. As the AEMC acknowledged in its final determination on the rule, Corporations, Environmental and Safety laws would, if they were in conflict, necessarily over-ride the NER.

Finally, generators make their own day-to-day decisions on how they run and price their operation. If a plant is incurring losses when operating, then the owner will not run it. It will remain registered, and therefore not “closed”, but the substantive effect on the market is the same. Whilst some stronger transparency rules are being proposed in this area, there will always be ways to cease operating rather than to continue operating at a loss. Attempting to prohibit such rational behaviour seems unlikely to succeed.

As one works through hypothetical examples, it becomes apparent that many stakeholders have placed far too much faith in the ability of notice of closure rules to provide certainty. Outlawing change seems as likely to succeed as outlawing gravity.

Ultimately, the only way to make sure that a generator will do its best to keep a plant operational is for it to be profitable. If customers genuinely need its physical capacity, then the market should be paying for it. If there is a fear that such genuinely needed capacity won’t voluntarily run, then this can only be fixed through the market design itself, not by creating rules that try to hold back the tide of the market’s signals.

Alternative proposals

The Grattan Institute previously investigated these issues and:

• Took the view that closure notices should be symmetrical, in that new entrants should feel confident that when they enter the market they will not have to face unexpected competition from deferred closure; and
• Recognised the unenforceable nature of a mandatory notice of closure rule.

Their alternative proposal was fixed closure dates (i.e. no advancements nor deferrals) enforced by large bonds forfeited upon changes to plans.

Immediately there is a logical problem: if a needed generator is receiving insufficient revenue to stay operational, it is not clear how it would also manage to stump up a bond.

But more fundamentally, such proposals intentionally discard something of great value: optionality. The ability to change business plans, in response to the emergence of clearer information, is obviously beneficial to owners, but also to society. For example, Germany has just completed its staged nuclear power station shutdown which was originally lauded for the certainty it provided.

However, this is now regretted in a context of European gas crises. It would have been very useful right now to have been able to extend the operation of some of the last nuclear plants.

Eraring notice

On 17 February 2022, Origin announced that it had alerted AEMO in relation to the “potential early retirement” of Eraring in 42 months. They also noted “We will continue to assess the market over time, and this will help inform any final decisions on the timing for closure of all four units.”

Origin’s approach seems consistent with the intent of the rule, in that it is alerting the market to the possibility of closing the plant in mid-2025, which AEMO will now plan around, but leaving open an option to extend. What was surprising however is how media commentary subsequently overlooked the intentional and beneficial optionality provided by the rule and in Origin’s announcement.  Thus the date was widely misinterpreted as fixed.

Proposed 5-year rule

The minister’s rule proposal expressly responds to the Eraring notice. The rule has proposed three features:

1. Extending the 42 months to five years;

2. Creating a new concept of "longer term mothballing" which would be subject to the same notice requirements as closure; and

3. Introducing a rule for the AER to oversee and prevent "gaming" of the notice period in relation to closure deferrals.

The second and third features above unavoidably draw the AER into subjective judgement about internal business practices. Such arrangements generally lead to unproductive games of cat and mouse between the regulator and businesses..

Forecasts versus Commitments

Reflecting on this series of events, it appears that over the last five years policy makers have confused the idea of a “good faith forecast” with that of a “commitment”.

Forecasts have always been provided in the NEM and are highly beneficial to its functioning. However, it is entirely appropriate that they remain subject to change.

Any business owner knows that the real world is probabilistic not deterministic, and that forecast returns of an asset are spread across a wide distribution curve which depend on variables that cannot be known in advance. More profitable parts of the distribution curve would imply a later closure, others an earlier closure. A very sophisticated approach would be to provide such a probabilistic distribution to AEMO, however that would be excessively complex.  Instead a “most likely” single date forecast is provided, with all parties understanding that the “most likely” can or will change over time.

Such forecasts can fairly provide the full market with the best available information, without unnecessarily restricting the optionality that is so valuable to the entire market community. Two things can go wrong with such forecasting however:

• Forecasts can get out of date as parties that have altered their plans omit to update the information to AEMO. This is remarkably easy to do: plans tend to change very gradually, and, like a boiling frog, it is easy to overlook the obligation to continuously update information.
• Much more rarely, if it occurs at all, is a more nefarious behaviour: deliberately providing misleading information in order to confuse competitors. Such behaviour is arguably illegal under many laws, and the NER has a specific rule, 3.8.22A, in relation to pre-dispatch bidding.

Rules can play a part in the first of these, and, if only to maintain confidence in the information’s integrity, potentially in the second.

Commitments are however a different matter. This is where, as per the Grattan proposal, actions are locked in without an ability to update. For the reasons described previously, these are problematic.

An excellent example of the difference exists in the short-term pre-dispatch timeframe. Numerous proposals for bidding “gate closures” have been raised and rejected in the NEM’s history. Gate closures lock generators’ bids several hours ahead of time, which, prima facie, would appear to reduce uncertainty. However when these proposals are considered in detail, it is soon realised that they are counter-productive, as they inhibit the market’s ability to efficiently respond to natural changes. In doing so they actually increase rather than reduce volatility and uncertainty.

This reasoning formed the basis of the AEC’s detailed 2020 consultancy into Ahead Markets.

Conclusion

Since the Finkel Review, the NEM has implemented firm closure rules which sound initially attractive, but when thought through problems are evident. This is because they attempt something unnatural – the prohibition of change.

Their existence has led to a misplaced trust in their effectiveness. When the inevitable happens and the rule fails to operate as expected, the appropriate course should be to review and reconsider the fundamental merit in such rules.

However, this latest development instead appears to double-down on the false belief that such things can be regulated.

[1] This is effectively any generator, including solar and wind farms, over 30MW capacity.