Assessment of real-time pricing options

Transcription

1 Assessment of real-time pricing options Information paper Submissions close: 24 May April 2016

2 Executive summary The Authority is seeking submissions to decide whether to explore real-time pricing in more detail and, if so, which option(s) it should develop further. If the Authority considers that a realtime pricing option is consistent with its statutory objective, it will prepare a formal Code amendment proposal for consultation in the 2016/17 financial year. In New Zealand, the pricing manager publishes final prices at least two business days after real time. Parties making decisions linked to spot prices do not know final prices until two days later. They rely instead on a series of indicative prices (ie, forecasts of final prices) to estimate the financial consequence of their decisions. New Zealand s two-day lag before publishing final prices is unusual. As far as the Authority is aware, all other organised electricity markets determine and publish spot prices during, or close to, real-time. Spot prices could be more actionable and resource efficient The Authority has identified two concerns with current spot pricing arrangements: 1. Indicative spot prices could be more actionable, ie, they should provide information parties can act on with confidence. Currently, the pricing manager determines and publishes final spot prices (used for settlement) two or more days after the fact. Indicative prices are available closer to real-time, but they can be unreliable predictors of final prices. Further, indicative prices may not always be available, especially during times of system stress. This means participants face uncertainty not only about future conditions (to some degree inevitable), but also about current and recent past conditions. 2. Final spot prices could be more resource efficient. That is, no consumer thinks afterwards they would have preferred to consume less/more at the final spot price. Likewise, no generator would have preferred to generate more/less at the final spot price. At present, inefficiencies can arise because final spot prices do not necessarily reflect the resources used in the dispatch process and current arrangements discourage some parties from participating. Changes in technology are making it easier for consumers to control how much electricity they use in response to spot prices. For some commercial entities, information about spot prices also affects how much electricity they generate. If spot prices were more actionable, these parties could participate more and make better decisions. More participation would ensure spot prices are resource efficient more often. Making spot prices more actionable and resource efficient has benefits Making spot prices more actionable and resource efficient would promote all three limbs of the Authority s statutory objective. It would promote: a) more efficient generating decisions and levels of demand response b) greater competition between generators and consumers c) a more efficient level of reliability in the power system. More actionable and resource-efficient spot prices would help parties make better decisions about when to invest in, and retire, generation plant. They could also promote retail innovation and confidence in the wholesale market. Better asset retirement and investment decisions would promote a more efficient level of reliability in the power system. Finally, increasing the ii 11 April p.m.

3 response to spot prices would promote greater competition in both the wholesale and retail markets. It would encourage new and innovative retail tariffs and interactions between consumers and retailers or third parties. Adopting real-time pricing would address both of the Authority s concerns identified above. It would allow final prices to be published in, or just after, real-time and for pricing to be more closely aligned with dispatch. Real-time pricing would also make prices more efficient when facing tight generation conditions or even shortages. Although participants would still face some price uncertainty about future conditions, the Authority is considering other avenues to improve spot prices, including quick wins such as improving the accuracy of demand and wind generation forecasts. The Authority is also exploring the merits of an hours-ahead market. The Authority considered four options, two of which are promising The Authority has considered the costs and benefits of four real-time pricing options (see section 3 for details): look-ahead 30-minute price (option A) look-ahead 5-minute dispatch-based price (option B) look-behind 5-minute price (option C) look-behind 30-minute price (option D). Based on its initial assessment, the Authority considers options B and C to be the most promising. Of these two options, the Authority prefers option B to option C because option B appears to deliver a greater net benefit. If the Authority decides to adopt real-time pricing, it would likely take more than two years to implement. The most actionable spot price would be one set in advance for a whole trading period (option A). This would allow people to decide how much energy to produce and consume after the price is set for the next thirty minutes. Setting prices in advance would allow both generation offers and demand bids to set prices. This includes bids for involuntary load shedding at scarcity prices. However, a lot can change in thirty minutes: option A would require arrangements to prevent manipulation, such as a balancing market for differences. In effect, option A would be an aheadmarket option, which the Authority is exploring separately. That is why the Authority has not considered it further in the quantitative assessment. Option D is the same as option C, except it would produce a single price for the trading period immediately after it finishes. Under option D, prices would be significantly less actionable than option C and it would be only slightly cheaper. Option D is not a promising option because, although it is cheaper, the net benefit is likely to be far less than option C. Therefore, the Authority has not considered it further in the quantitative assessment. Option B (dispatch-based pricing) looks preferable to option C (5-minute look-behind price) Option B would provide prices that are less actionable. Parties would know the price for only the next five minutes or so, rather than for a full 30-minute trading period. However, as for option A, demand bids could set prices (with less onerous conditions than existing dispatchable demand provisions). This includes bids for involuntary load shedding at scarcity prices. This would iii 11 April p.m.

4 encourage parties to submit their offer/bid quantities and prices so that generation and consumption decisions were optimal. The assessment of costs and benefits shows this option has the greatest net benefit, though it is much more costly to implement than option C. Option C would produce prices that are slightly less actionable than option B. This is because prices would be available every five minutes in arrears. Dispatchable demand provisions would remain as they are now. Processes to resolve reserve and generation shortages would also remain unchanged but would run automatically to produce prices every five minutes. Option C is less expensive to implement than option B but the potential benefits are less again. The table below summarises the size of the potential net benefits of real time pricing under options B and C. Table 1: Net benefits of real time pricing options B and C Benefit Option B Option C Demand response Retail competition Reliability 10 7 Total Less costs 6 4 Net benefit Source: Electricity Authority iv 11 April p.m.

5 Contents Executive summary 1 What you need to know to make a submission What this consultation paper is about How to make a submission By when you should make your submission 8 2 The Authority has concerns about spot prices Final spot prices and indicative prices Comparison with real-time pricing arrangements in other jurisdictions Spot prices could be more actionable and more resource efficient Why existing spot prices are less actionable and less resource efficient Improving spot pricing fits with the Authority s priorities Improving spot pricing is consistent with the statutory objective 14 3 The Authority considered four real-time pricing options The Authority identified four options Option A look-ahead 30-minute price Option B look-ahead 5-minute dispatch-based prices Option C look-behind 5-minute prices Option D look-behind 30-minute price Features common to all options Options B and C require a method for establishing a 30-minute price from a series of 5-minute prices Identifying quick wins 20 4 Assessment of the options Qualitative assessment suggests options B and C most promising The quantitative assessment favours option B over option C 23 Appendix A Format for submissions 26 Appendix B Examples of problems with current arrangements 27 Managing demand cool store owner 27 Retail competition and innovation - retailer 27 Demand response - major industrial user 27 Generation scheduling hydro generator 28 Managing spot price risks futures contracts 28 Appendix C TASC report from the system operator 30 Appendix D Constructing 30-minute prices from 5-minute prices for options B and C 31 Options to calculate price for 30 minute period 31 Appendix E Example of flexible user 34 A flexible user operating under the status quo 34 A flexible user operating under option B 34 A flexible user operating under option C 35 Overall comparison 35 Appendix F Benefits and costs of options B and C 36 Assessment framework 36 Categories of expected benefit 36 Categories of expected cost 41 Estimated benefits and costs 42 Glossary of abbreviations and terms 45 ii v 11 April p.m.

6 Tables Table 1: Net benefits of real time pricing options B and C iv Table 2: Estimated benefits and costs ($M) 24 Table 3: Summary of effects for a flexible user 35 Table 4: Estimated benefits and costs 43 Table 5: Performance ratio where Options B and C provide same net benefit 43 Figures Figure 1: Timeline 7 Figure 2: Examples of indicative prices and when they are available 9 Figure 3: Spot market is at the core of the electricity market 13 Figure 4: Diagram comparing real-time pricing options 17 Figure 5: Converting 5-minute prices to 30-minute prices 32 Figure 6: Effect of spot price signals on reliability 40 vi 11 April p.m.

7 1 What you need to know to make a submission 1.1 What this consultation paper is about This paper seeks views on ways to improve spot pricing arrangements in the wholesale electricity market. In particular, the Authority has considered four real-time pricing options, under which the pricing manager would determine and publish final spot prices during or immediately after the finish of each trading period. This compares with current arrangements, where the pricing manager does not determine or publish final spot prices until at least two days after real-time We have developed a range of options to a conceptual level, two of which we expect to deliver net benefits compared with current arrangements At this stage, we are considering whether to explore real-time pricing in more detail and, if so, which option(s) to develop further. As Figure 1 shows, if we consider that a real-time pricing option would be consistent with our statutory objective, in the 2016/2017 financial year we will consult on a proposal to amend the Code. Figure 1: Timeline Consult on real-time options for further investigation Propose amendment to the Code Develop design details Implement Start date Source: Electricity Authority If the Authority chooses to implement real-time pricing, it would likely take more than two years to implement At this time, the Authority is seeking views on: a) the range of real-time pricing options that it has considered b) the two specific options that it has short-listed, and the expected benefits and costs c) proposed next steps. 1.2 How to make a submission The Authority invites you to make a submission on this paper Please note the Authority wants to publish all submissions it receives. If you consider that it should not publish any part of your submission, please indicate which part, set out the reasons why you consider the Authority should not publish it, and provide a version of your submission that the Authority can publish (if it agrees not to publish your full submission) If you indicate there is part of your submission the Authority should not publish, the Authority will discuss it with you before deciding whether to publish that part of your submission April p.m.

8 1.2.4 However, please note that all submissions the Authority receives, including any parts that it may not publish, can be requested under the Official Information Act This means the Authority would be required to release them unless good reason existed under the Official Information Act to withhold them. The Authority would normally consult with you before releasing any material that you said we should not publish The Authority would prefer to receive submissions in electronic format (Microsoft Word) in the format shown in Appendix A. Submissions in electronic form should be ed to with Information Paper Assessment of real-time pricing options in the subject line If you cannot send your submission electronically, post one hard copy to either of the addresses below. Postal address Submissions Electricity Authority PO Box Wellington 6143 Physical address Submissions Electricity Authority Level 7, ASB Bank Tower 2 Hunter Street Wellington 1.3 By when you should make your submission You should deliver your submission by or otherwise so it arrives by 5pm on 24 May Please note we are unlikely to consider late submissions We will acknowledge receipt of all submissions electronically. Please contact the Submissions Administrator if you do not receive electronic acknowledgement of your submission within two business days April p.m.

9 Indicative prices 2 The Authority has concerns about spot prices 2.1 Final spot prices and indicative prices The prices at which wholesale buyers and sellers transact are final spot prices (final prices). Some parties also use final prices to calculate amounts payable under risk management contracts for electricity, such as futures contracts In New Zealand, the pricing manager publishes final prices at least two business days after real time. Parties making decisions about consumption, generation or trading that depend on spot prices do not know what final prices will be until at least two days later. They rely instead on a series of indicative prices (ie, forecasts of final prices) to estimate the financial consequence of their decisions As Figure 2 illustrates, these indicative prices include: a) ASX futures prices, which provide an indication of average spot market prices b) prices in the weekly dispatch schedule, which the system operator issues once a day for every trading period for the next week c) prices in the price-responsive schedule (PRS) and non-response schedule (NRS). The system operator produces these schedules every half hour for the next four hours, and every two hours for the next 36 hours d) 'real-time' prices (RTPs), which the system operator calculates every five minutes and which indicate prices for the previous five minutes. Figure 2: Examples of indicative prices and when they are available -30days -7days -36hrs +30min 0 +1days +2days Futures prices Weekly dispatch schedule PRS/NRS schedule RTPs Provisional prices Interim prices Source: Electricity Authority Provisional, interim, and final prices are not indicative prices because they are not available until after the trading period is over. 2.2 Comparison with real-time pricing arrangements in other jurisdictions New Zealand s spot pricing arrangements with the two-day lag in publication are unusual. As far as the Authority is aware, all other organised electricity markets 9 11 April p.m.

10 determine and publish spot prices during, or close to, real-time. For example, the PJM market calculates settlement prices for each hour-long trading period from dispatch prices that are determined and published at the start of every 5 minute interval The National Electricity Market in eastern Australia uses similar arrangements. That market calculates settlement prices for each 30-minute trading period based on the average of six dispatch interval prices. The market calculates these dispatch prices at the start of each 5-minute dispatch interval Both these markets have adopted spot-pricing arrangements that create a high degree of alignment between dispatch and settlement processes. Participants in these markets have a greater level of assurance about spot prices and the financial consequences of their decisions. Put another way, prices in these markets are more actionable. 2.3 Spot prices could be more actionable and more resource efficient The Authority has two concerns about the existing arrangements in New Zealand s wholesale electricity market: a) Spot prices could be more actionable. Final prices are available too late to inform participants' short-term decision-making. Further, indicative prices do not provide reliable predictors, for the reasons discussed below. Consequently, participants face uncertainty not only about future conditions (which is inevitable to some degree), but also about current and recent past conditions. b) Spot prices could be more resource efficient. Spot prices are resource efficient if no consumer thinks afterwards they would have preferred to consume less/more and no generator would have preferred to generate more/less. Spot prices are not always resource efficient because they are calculated independently of the dispatch process and because current arrangements discourage some parties from participating Making spot prices more actionable and resource efficient would promote all three limbs of the Authority s statutory objective. It would promote: a) more efficient generating decisions and levels of demand response b) greater competition between generators and consumers and between retailers c) a more efficient level of reliability in the power system In the short term, more actionable prices would provide more efficient demand response from existing (mostly industrial) consumers. This is because they would have more certainty about the consequences of their decisions to consume/generate electricity. More actionable prices would also result in more resource efficient outcomes because final spot prices are less likely to either: a) exceed the value at which some users would have preferred to voluntarily reduce their consumption or increase their onsite generation b) be lower than the value at which some users would have preferred to voluntarily consume more electricity or reduce their onsite generation More actionable prices would also promote more efficient generating decisions because generators would have better information for short-term planning. This is especially important for generators facing important short-term decisions such as whether to get a unit ready for operation. It is also important for smaller generators that April p.m.

11 are self-dispatched. 1 Again, making prices more actionable would result in more resource efficient outcomes. That is, final spot prices should be higher than a generator s willingness to generate more often than they currently do (and vice versa) Prices that are more actionable would encourage greater levels of demand response. This would promote greater competition between generators and consumers, particularly when spot prices are high. Greater competition should reduce the scale of price spikes and average spot prices. This would directly benefit consumers who pay prices based wholly or partially on spot prices. It should also have a similar effect on the trading of electricity risk management (hedge) contracts and retailer costs. Competition in the retail market should encourage retailers to pass on any reduction in costs to retail consumers In the long term, greater participation by consumers in the wholesale market also contributes to a more efficient level of reliability. In tight supply situations, consumers could influence outcomes in three ways: a) deciding between consuming at high prices b) generating their own electricity c) delaying or reducing their power consumption More actionable prices should also promote greater retail competition via tariff innovation. Improvements in technology (eg, batteries, and appliances with communication devices) are making it easier for consumers to participate in the electricity market. For example, consumers can program some models of refrigerators and air conditioners/heat pumps to alter their electricity usage in response to external signals, including information about spot prices. Some retailers now offer consumers time of use tariffs based on spot prices so consumers can alter their electricity consumption in response to them Finally, more actionable prices could reduce the number and scale of regrets or financial distress for some participants, which could increase confidence in the market Appendix B sets out examples that illustrate these effects. 2.4 Why existing spot prices are less actionable and less resource efficient The Authority is concerned about some important differences in the way indicative and final spot prices are calculated. There are: a) Differences in the methodologies used for calculating prices in the different timeframes. For example, if there is not enough generation to meet projected demand, real-time prices default to artificial placeholder values. The pricing manager calculates final prices using a different approach, which involves postprocessing of model outputs (eg, for scarcity pricing). b) Differences in the inputs used to calculate prices. For example, indicative real-time prices use the average level of wind generation for the previous five minutes, whereas final prices use the average level of wind generation over the 30-minute trading period. 1 Self-dispatched means generators have discretion to alter their level of output in real-time April p.m.

12 2.4.2 Differences in pricing methodologies largely reflect historic market design choices, and there is no inherent reason to seek to perpetuate those choices. At present, shortages of electricity or reserve cause the market model to produce artificially high prices. Ideally, the actual costs of non-supply provided by the affected parties would replace these artificial placeholder values. Only if there is a residual shortfall would pre-defined values in the Code represent the expected cost of non-supply In relation to inputs, indicative prices use information on expected conditions available when the pricing manager calculates the price. Final prices use inputs that reflect a mix of expected and actual conditions. For example, the quantity of generation offered into the spot market reflects expected plant availability, while demand uses actual (or estimated) load Differences between indicative and final prices can also arise when there are manifest errors of some kind, such as where incorrect data is used. Part 13 of the Code relating to pricing errors governs these types of events, which are relatively rare. If the Authority were to adopt some form of real-time pricing, it would be desirable to retain a mechanism to address manifest errors after real-time, along the lines of those used in other electricity markets Putting the rare instance of manifest errors to one side, the inputs used to determine indicative and final spot prices can differ for a variety of reasons, such as: a) generation cleared in forecast prices unexpectedly becomes unavailable during the trading period b) transmission elements unexpectedly become unavailable or are put into service during the trading period c) generation is dispatched to meet real-time demand in excess of or lower than the average demand measured by meters for the trading period d) actual demand varies from the forecast e) intermittent generation varies from the forecast At present, the highest cost generator dispatched during the trading period does not set the final price if, when viewed with the benefit of hindsight, its output is not required. Yet the generator was dispatched and a real resource cost was incurred. Because the real cost is not reflected in the spot price, generators receive constrained-on payments that compensate them for being dispatched at offer prices above the final price If a real-time final price had signalled the resource cost, consumers could have altered their consumption. This could have reduced or avoided any constrained-on payment In addition, demand response actions like this would lower the overall cost to consumers and contribute to economic efficiency by better ensuring that the least cost resources were utilised at all times This example illustrates the more general point that, under current arrangements, there can be divergence between the resource costs incurred in real-time and final prices. Alternative pricing arrangements that narrow or remove such divergences are more efficient. They also reduce the likelihood that generation resources are constrained-on, especially in situations where their incremental cost exceeds the benefit to consumers April p.m.

13 If dispatch and pricing were better aligned, it would also provide information to participants that is more actionable, because it would eliminate the two-day lag in publishing final prices The Authority considered these factors when it developed the real-time price options discussed in section Improving spot pricing fits with the Authority s priorities Improving spot pricing fits with the Authority s current priorities of improving retail competition and enhancing the efficient operation of the industry As shown in Figure 3, the spot market is central to the wider electricity market. Weaknesses in the spot market affect competition and efficiency across the sector. More specifically, current spot pricing arrangements make it harder for retailers to offer innovative demand-response products that would stimulate retail competition. Further, the absence of timely and reliable real-time price signals compromises efficiency, because parties are more likely to make sub-optimal decisions about consumption and generation. Figure 3: Spot market is at the core of the electricity market Source: Electricity Authority Improving spot market arrangements also builds on the Authority s earlier work. In mid- 2015, the Authority published a paper on a range of possible spot market refinements. 2 2 See April p.m.

14 After considering stakeholder feedback, in September 2015 the Authority released a decision paper indicating it would further explore two potential refinements: 3 a) options for settling on prices based on system conditions in real-time (real-time pricing) b) options for adding or facilitating an hours-ahead market (ahead market) In that decision paper, the Authority also indicated it would consider options to enhance the accuracy of the inputs into the pricing process, such as wind and demand forecasts. It noted these enhancements could add to, or reduce, the net benefits available from real-time pricing and/or an ahead market Recent industry trends make it especially timely to consider spot pricing arrangements. Technology changes in appliances, communications and metering are making it easier for smaller electricity users to respond to spot prices. In addition, the tightening supply/demand outlook is reinforcing the importance of efficient prices to signal investment in the right type and level of resources on the power system (both generation and demand management) Spot pricing arrangements are also very relevant to other aspects of the Authority s work programme, especially: a) Cap products the Authority wants to introduce tradeable cap products. Simplifying the way spot prices are determined during periods of system stress would make them more actionable and support greater and more competitive trading in these products. b) Gate closure the Authority recently approved a Code change to shorten the gate closure period. Enhancements to real-time pricing would complement this, by improving the quality and robustness of information available to participants. Participants would have more scope to refine their offers and bids, and better information with which to make such refinements. 2.6 Improving spot pricing is consistent with the statutory objective Although this paper is not proposing a formal Code amendment, the Authority has explicitly considered whether a future change toward real-time spot pricing would be consistent with its statutory objective As explained in subsection 2.3, the Authority expects that real-time spot pricing would promote all three limbs of the statutory objective: a) If demand-response providers have more actionable price signals, and if they can compete more directly with supply-side resources, this would enhance competition. There should also be flow-on competition benefits in the retail and hedge markets from improving spot market competition See An ahead market would allow parties to lock-in prices a number of hours ahead of real-time for their planned demand and generation volumes. Any differences between planned and actual volumes would be settled at spot prices. An ahead market is intended to help participants manage the uncertainty that exists in the lead up to real-time, whereas real-time pricing is intended to reduce uncertainty about prevailing market conditions. For this reason, real-time pricing and an ahead market are being considered in parallel rather than as one potential initiative. The Electricity Authority's objective is to promote competition in, reliable supply by, and the efficient operation of, the New Zealand electricity industry for the long-term benefit of consumers April p.m.

15 b) Reducing the extent of reliance on administrative pricing processes during periods of system stress would strengthen reliability. This should improve the level of confidence in prices, and reduce the likelihood of sub-optimal decisions about investment, retirement, or consumption. c) The availability of more reliable real-time price signals would reduce the likelihood of parties taking actions that they later regret once they know final prices, and this would improve efficiency As real time pricing promotes all three limbs of the statutory objective, the Authority does not have to make trade-offs between one limb over another. Q1. Do you agree the spot pricing issues identified by the Authority are worthy of further investigation? April p.m.

16 3 The Authority considered four real-time pricing options The Authority has identified and assessed four high-level options for real-time pricing. This section, and the system operator s report (Appendix B), describe the options in more detail. 3.2 The Authority identified four options Figure 4 compares the four real-time pricing options it has identified with the current arrangements and a (theoretically) ideal set of arrangements. Under the current arrangements, parties see indicative prices in advance of and during the trading period. However, those prices are not actionable because they do not help form the settlement price, which is not available until two or more days after the trading period In theory, the ideal approach would be to align the trading period interval with the intervals used for dispatch and pricing. Doing this would make prices both actionable and more efficient. However, shortening the current 30-minute trading period would require extensive changes to settlement systems and processes affecting everyone in the wholesale and retail market. Given the likely substantial cost to implement such changes, the scope of the investigation and options did not include changing the length of the trading period April p.m.

17 Figure 4: Diagram comparing real-time pricing options Dispatch intervals Real time dispatch is not changing Current Actionable prices None Settlement prices 2 days or more Ideal Actionable prices Settlement prices Intervals align with dispatch Key: Dispatch instructions sent Real time price published Option A 30-min lookahead Actionable prices Settlement prices Price for trading period published Look-ahead Option B Dispatchbased price Actionable prices Settlement prices Cumulative (time-weighted average) Look-behind Actionable prices Option C 5-min lookbehind price Settlement prices Cumulative (time-weighted average) Option D 30-min lookbehind price Actionable prices Settlement prices None pm 5.30pm Source: Electricity Authority Figure 4 shows that option A (look-ahead 30-minute pricing) provides up to 30-minutes notice of the price for the next trading period. Options B (look-ahead 5-minute dispatchbased pricing) and C (look-behind 5-minute pricing) provide actionable prices that cumulatively form the settlement price. Prices under option D (look-behind 30-min pricing) are no more actionable than the current arrangements. 3.3 Option A look-ahead 30-minute price Under this option, the pricing manager would calculate final prices shortly before the start of each trading period, using the information available at the time. Prices would reflect the interaction of offered generation and demand-side bids ie, a demand-side bid could set the final price if it was the marginal resource. Prices would reflect starting circumstances and expected conditions for the next 30-minute trading period Changes in system conditions after the trading period starts would not affect price calculations for that period. The pricing manager would take into account changes in system conditions only when calculating the prices for the next trading period April p.m.

18 3.3.3 A broad equivalent of this option would be settling on the prices in the last PRS published before each trading period starts. 3.4 Option B look-ahead 5-minute dispatch-based prices The system operator uses a nominal 5-minute dispatch interval, but actual dispatch intervals may be shorter or longer. For convenience, this paper uses the term lookahead 5 minute dispatch-based prices for Option B, even though actual dispatch intervals may differ from 5 minutes Under option B, the pricing manager would calculate spot prices based on the dispatch schedules used by the system operator to issue new dispatch instructions. Prices would reflect the interaction of offered generation and demand-side bids ie, a demand-side bid could set spot prices if it was the marginal resource Prices would reflect starting circumstances and expected conditions for a short lookahead period. The system operator currently uses a nominal period of five minutes in its dispatch process. However, it could be shorter or longer depending on how frequently the system operator needs to issue new dispatch instructions to reflect changing conditions. This approach would limit the effect of any forecast error because if actual conditions diverged materially from expectations, the system operator would re-dispatch and generate a new price The pricing manager would calculate final prices for each 30-minute trading period as the average of the real-time dispatch-based 5-minute prices in that period (see section 3.8 for more details about ways to do this). A broad equivalent for this option would be settling on the prices in the real-time dispatch (RTD) schedules generated by the system operator. However, all load would be priced (including scarcity prices assigned to involuntary load shedding), which would eliminate the need for provisional pricing processes. 3.5 Option C look-behind 5-minute prices Under this option, the pricing manager would calculate spot prices based on conditions in the preceding five minutes. Prices would reflect the interaction of offered generation and actual demand. This means that demand-side bids could not directly set 5-minute prices, except for users that participate in the dispatchable demand scheme The pricing manager would calculate final prices for each 30-minute trading period as the average of the 5-minute look-behind prices (see section 3.8 for more details about ways to do this) A broad equivalent for this option would be settling on the prices in the RTP schedule that the system operator publishes each five minutes for the preceding 5-minute period. However, redefining some aspects of RTPs would align them with existing postprocessing that occurs to final prices (eg, applying scarcity pricing rules, resolving provisional pricing situations). 3.6 Option D look-behind 30-minute price Under this option, the pricing manager would calculate spot prices at the end of each trading period based on the conditions in the preceding 30 minutes A broad equivalent for this option would be the existing final price process, except that prices would be calculated and published soon after the end of each trading period April p.m.

19 3.7 Features common to all options There are some features that are common to all of the options to varying extents: a) An automated process to resolve infeasibilities. Infeasible outcomes result where the pricing program cannot compute a feasible market-clearing solution without relaxing some constraints (such as allowing demand to exceed supply). At present, the pricing manager resolves infeasibilities manually. If a market-clearing solution cannot be calculated from offered resources (voluntary demand response and generation), prices are determined in accordance with procedures defined in the Code. These procedures set prices that represent the cost of the load shedding or reserves reduction required to clear the market. This manual process helps ensure spot prices settle within predefined ranges if forced load shedding has occurred (scarcity pricing), or for final prices to be based on a predefined multiple of offer prices (for reserves shortfalls). It might be necessary to automate this manual process under some real-time pricing options. b) An automated process to resolve high spring washer situations. Under current arrangements, the pricing manager uses another manual process when it detects a high spring washer price situation. 6 This process modifies final prices in some circumstances. It might be necessary to automate this manual process under some real-time pricing options. c) A process to resolve pricing errors. Current arrangements allow the pricing manager to re-process settlement data if there is a manifest error. While this arguably reduces real-time certainty for participants, it provides a protection against manifest input errors. Similar provisions exist in other electricity markets that use real-time pricing. is the Authority proposes retaining the existing or similar provision for dealing with pricing errors, perhaps with a minimum threshold for reopening prices. d) A process to resolve undesirable trading situations. Current arrangements allow the Authority to determine prices if an undesirable trading situation (UTS) occurs. 7 As with pricing errors, while this process may arguably reduce real-time certainty, there is a high threshold for invoking this process so is the Authority proposes retaining it. 3.8 Options B and C require a method for establishing a 30-minute price from a series of 5-minute prices Under all of the options, a 30-minute trading period would not change, with a single final price applying at each node for the 30-minute trading period For options B and C, the pricing manager would need to determine 30-minute final prices from the 5-minute prices it calculated within a trading period. Other markets address this issue with pricing intervals that are shorter than trading periods. Appendix D contains a discussion of alternative methods. 6 7 A high spring washer price situation is an event where the spot price at one or more locations is much higher than at other locations due to the interaction between a transmission constraint and electricity loop flows. Under Part 1 of the Code, an undesirable trading situation is a situation that threatens or may threaten confidence in, or the integrity of, the wholesale market, and that in the reasonable opinion of the Authority cannot be resolved by any other mechanism under the Code April p.m.

20 3.9 Identifying quick wins The Authority is considering if there are any quick wins as part of its investigation into spot market refinements. A quick win could be either: a) an alternative to the proposal that can be implemented quickly at little cost, that partly captures the benefits of a more comprehensive solution implemented later (eg, inter-island reserve sharing ahead of a national market for instantaneous reserve) b) one element of a comprehensive solution that can be implemented before the rest of it The Authority proposes moving to a final price that encourages participants to contribute to, and receive, spot prices in real-time. However, it is aware that the accuracy of the forecasts is also important. More accurate estimates of inputs to indicative prices, independent of spot prices, are equally important to forming efficient spot prices. That is why the Authority is considering such improvements as part of its investigation into an hours-ahead market. It is working on this investigation now and expects to be able to release its findings in a few months. Q2. Are there any options you think we missed? If so, please describe them April p.m.

21 4 Assessment of the options 4.1 Qualitative assessment suggests options B and C most promising Option A look-ahead 30-minute Of the options considered, option A provides the greatest level of price certainty to participants. At the start of each trading period, participants would have a firm price on which to base their decisions. 8 For this reason, the Authority considers it would provide the most actionable prices of the options However, one concern with this option is the integrity of forecast information, especially demand and non-scheduled/intermittent generation. Because these inputs would affect final prices, participants would have a strong incentive to bias their forecast inputs with the objective of pushing final prices up or down (depending on their wholesale market position). For example, demand-side parties could have an incentive to under-estimate their demand quantities, since the intersection of the offered generation schedule and the forecast demand schedule would determine the settlement price While compliance procedures should be able to identify any gross manipulation, it would be difficult to identify more subtle biases given the presence of genuine uncertainty in forecasting processes An alternative would be to use forecast estimates from a neutral party such as the system operator. However, that would reduce rather than eliminate the incentive issue, as the system operator would still require some information from participants. It would also limit consumers to indirect participation in the market (ie, responding to forecast prices, rather than submitting their own bids) Even discounting the incentive for bias, there are likely to be larger divergences between dispatch and final prices because of events that occur in the trading period after the final price is prepared. Existing constrained-on payments would be higher than they are now and therefore final prices would be less efficient A more efficient way of dealing with divergences would be for participants to face a spot (or balancing) price for any deviations between their quantities scheduled in an hours-ahead market and their actual quantities. The spot price would be determined in accordance with options B, C or D. Consequently, option A represents the boundary between a real-time and an hours-ahead market, the other spot market refinement the Authority is investigating. Option B Look-ahead 5-minute dispatch-based Publishing firm prices would improve the response to prices, compared with the existing arrangements and options C and D. However, it would be less actionable than option A, because the 30-minute final price would progressively crystallise through each trading period, 9 rather than at the start of each trading period Unlike option A, this option would not create an incentive for participants to bias their forecast inputs in an attempt to influence final prices. The reason such biases would not arise is that the system operator s dispatch decisions are based on actual system 8 9 Other than to address any manifest errors or a UTS. As discussed in Appendix D, the settlement price for each 30-minute trading period would be computed from the 5 minute prices for each dispatch solution April p.m.

22 conditions, providing a correction mechanism for any biased inputs. For example, if actual demand was higher than expected, this could indicate unrealistically low quantities in demand-side bids intended to (say) bias final prices downwards. This would lead the system operator to re-dispatch and determine new spot prices that reflect the actual (higher) level of demand. This key difference should remove the incentive for parties to game the forecasts when compared with option A Equally important, if demand-side participants have a genuine capacity and intent to manage their demand at various price points, they would have a strong incentive to reveal this information via price-responsive bids. The incentive arises because such bids can help them to actively manage their electricity purchase costs. It also allows demand-side bids to determine the market clearing level for a 5-minute price, and hence directly influence final prices. In this respect, option B provides a better mechanism for demand-side parties to participate and improves upon the existing dispatchable demand scheme Option B would also substantially align final prices with the arrangements used for realtime dispatch. Therefore, it should provide the sharpest marginal signal and, consequently, the greatest scope for reducing inefficient outcomes and constrained-on payments Appendix E provides further explanation of how this option would operate for an electricity user that has the flexibility to rapidly alter its level of demand. Option C look-behind 5-minute This option would provide a slightly lower level of response than option B. While the 30- minute final price would progressively crystallise during the trading period, 10 there would be a lag as the price becomes known for each preceding 5-minute period. In comparison, under option B, parties would know that a real-time price would be good until changed. Under option C, they would have to rely on the real-time price for the last five minutes being a good indicator of the price in the next five minutes (or more) This option would use actual demand and intermittent generation for the preceding five minutes, and should not create any incentive for participants to bias their forecast input information The option should provide a lower incentive than option B for demand-side parties to submit price-based bids, because bids would only be able to set final prices if they were made under the existing dispatchable demand provisions Relative to current arrangements, this option would significantly reduce (but not eliminate) the scope for divergences between dispatch and final prices because each 5-minute price is backward looking, whereas dispatch is inherently (slightly) forward looking. Option D look-behind 30-minute While this option would substantially reduce the lag in publishing final prices relative to the current position, it would not alter the information available to participants in realtime. For this reason, the Authority considers it will provide the least improvement in terms of actionable prices. It would also not improve the alignment between dispatch and settlement arrangements. 10 See Appendix D for further explanation April p.m.

23 The main benefit of this option is the potential to improve short-term planning decisions. Generators and demand-side participants would have firm information on where prices have been tracking in the last few hours, rather than needing to wait for two days for this data. Only options B and C are suitable for quantitative assessment Options B and C (look-ahead 5-minute and look-behind 5-minute respectively) each offer significant qualitative benefits over the status quo and the Authority has therefore considered them in more detail in terms of expected benefits and costs The Authority considers that neither option A nor option D provides clear and significant benefits that rival the other two options. Therefore, the Authority has not assessed them in detail as potential alternatives to current arrangements Option A (look-ahead 30-minute price) creates gaming incentives because it provides parties with a firm price for the next 30 minutes based on their declared intentions. This is an inherent feature of the option, which would require creating a separate mechanism to address the gaming incentive Further, if strong benefits exist for creating a firm look-ahead price, it is likely that the optimal ahead period would exceed 30 minutes. Given that the Authority is separately considering an ahead-market mechanism, it has excluded option A from the quantitative assessment of benefits and costs Option D (look-behind) 30-minute does not appear to offer significant benefits relative to the status quo. Using the existing calculation to calculate the final price would maintain the existing differences between dispatch and pricing that discourage efficient responses by consumers and generators Accordingly, the Authority expects option D would yield much lower benefits than option C, yet cost about the same to implement. Consequently, the Authority has excluded option D from the quantitative assessment of benefits and costs. 4.2 The quantitative assessment favours option B over option C The quantitative assessment of options B and C adopts the following approach: a) the analysis is undertaken from an economy-wide perspective based on the expected incremental benefits and costs of adopting real-time prices b) effects are assessed over a 15-year period, starting from the time that a decision is made to implement real-time pricing c) values are estimated in 2016 dollars using a 6% real discount rate; sensitivity cases with discount rates of 4% and 8% are also considered d) the counterfactual to real-time pricing assumes that existing arrangements remain in place The Authority has developed quantitative estimates for the following sources of benefit: a) more efficient levels of demand-response b) greater retail market competition and innovation c) more efficient levels of reliability The Authority has not developed quantitative estimates for the benefits of more effective risk management and improved market confidence because there is not April p.m.