Markets committee | January 10-12, 2023 | - PowerPoint Presentation

1. Markets committee | January 10-12, 2023 | Westborough, MAParviz AlivandDay-Ahead Ancillary Services Initiative (DASI)1Principal EconomistMitigation Enhancements for the Co-Optimized Day-Ahead Market

2. 2Day-Ahead Ancillary Services Initiative: Mitigation Enhancements WMPP ID:162Proposed Effective Date: Q4 2024/Q1 2025With DASI, the ISO seeks to procure and transparently price the ancillary service capabilities needed to meet real-time energy demand and deliver a reliable, next-day operating plan with an evolving generation fleetTo safeguard competitiveness, the ISO is concurrently proposing mitigation-related enhancements for the co-optimized day-ahead marketToday’s discussion: Market power assessment and simulations resultsConduct test reference and threshold price approach

3. Market Power Assessment and simulations3

4. 4ISO and IMM staff have collaborated in performing detailed market power assessment simulations for the co-optimized Day-Ahead Market (DAM)This section summarizes:Data on prevalence of pivotal suppliers in the DA marketScope of the market power simulationsMain conclusions from this market power assessmentStakeholder feedback and questions welcomeModule Overview

5. 5Pivotal Supplier EvaluationIt is infrequent to have a pivotal supplier in the co-optimized DAM, even with the FER and all FRS requirementsMarket Power EvaluationAbility of a supplier to materially raise DA energy and A/S prices by withholding DA A/S capability is limited, due to substitutionUnder certain conditions, absent mitigation, a large supplier could profitably raise DA A/S prices, by withholding A/S capabilityThese conditions are quite infrequent historicallyHowever, the hours in which withholding is profitable are hard for suppliers to predict, and withholding A/S can lower A/S and energy clearing pricesThis is because of both generators’ lumpiness, and the co-optimized market’s clearing engine (MCE) “pre-positions” resources to minimize the impact of withholding(continues next slide)Market Power Assessment: Summary of Key Findings

6. 6Market Power Evaluation (continued)Under certain (highly infrequent) conditions, absent mitigation, withholding a portion of a supplier’s total capability could raise DAM energy prices in the co-optimized DAMHowever, the price impacts of withholding DA A/S are primarily manifest in the DA A/S clearing prices, and not in the DAM’s energy prices (i.e., not in DA LMP or the FERP)A/S cleared quantities directly enter the A/S pricing constraints, but the energy constraints are only limitedly impacted by A/S withholdingIn the rest of this module, we discuss each of these six key findings in greater detailMarket Power Assessment: Summary of Key Findings (cont’d)

7. 7The ISO evaluated the total resources available in the DAM, and compared it to the co-optimized DAM’s requirements and the largest suppliers’ resource portfolios, each day 2016-2021Total resources available is the sum of offered EcoMax for all units that are not on outage at the close of the bidding window in the DA market (adjusted for imports and exports)Total requirements is the sum of the DA load forecast and total FRS requirements for that hourTotal system resources available, net of total requirements and net of the largest portfolio, is referred to as the “net supply margin”The DAM has a pivotal supplier if the net supply margin is negative in any hour of that dayPivotal Supplier Evaluation – Methodology

8. 8Over the last seven years, a co-optimized DAM would have had a pivotal supplier in only 144 hours (out of ~52,600 between 2016-2021)This amounts to about 0.22% of hours, and occurred on 31 daysThe net supply margin in these 144 hours ranges from ~-6 MWh to ~-2,900 MWhDays without a DA pivotal supplier, but very tight RT conditions (e.g. Labor Day 2018), would not appear with a negative net supply margin in the DA data (next slide)Even within a tight DA day that has a pivotal supplier in some hours, the majority of hours on the same day typically do notThis allows the DA MCE to schedule (“pre-position”) resources in a way that lessens the impact of any withholding in the tight hoursFinding #1: Pivotal Suppliers are Infrequent in the Co-Optimized DAM

9. 9Finding #1: Pivotal Supplier Frequency and Net Supply Margin

10. 10It is infrequent to have a pivotal supplier in the co-optimized DAM, even with the FER and all FRS requirementsThough the potential for successful market power in the DAM is infrequent, the DAM is potentially vulnerable to the exercise of market power on certain tight daysThis reinforces the necessity of mitigation in the co-optimized DAMGenerally, in most (but not all) hours, we expect competition and substitution to yield competitively-determined DAM pricesThe data shows that, generally, New England’s DAM has abundant resources available (for commitment) relative to total requirementsFinding #1: Summary and Implications

11. 11The withholding strategies the ISO studied cover two main actions:Withholding DA A/S capability (but not energy): these simulations cover the incentives to exercise market power in the new DA A/S products, and the impact of withholding A/S capability on the DA A/S and energy pricesWithholding total resource capability DA (energy and A/S): these simulations contemplate withholding of resources’ total capability from the DAM, not just A/S capacityAdding new A/S requirements and the FER could create greater incentives for withholding total capability from the DAMTerminology note: For purposes of this presentation, we use “withholding DA A/S” synonymously with “not offering DA A/S at the resource’s (ISO-modeled) competitive offer price”Market Power Simulation Methodology

12. 12“Form” of withholding (physical or economic) in simulations:From a market power simulation standpoint, successful economic and physical withholding have identical impacts on market outcomes (see ISO’s December MC mitigation presentation, slide 21)Logic: Either makes the withheld capability 'extra-marginal' in the DAM clearing process, raising DA price(s) Thus, in the simulation studies, we do not distinguish the form of withholding (economic v. physical)Their distinction is not necessary to draw (quantitative) conclusions about the potential impacts of an exercise of market powerIn the data and results summarized next, the withheld capability can be assumed to be either physically or economically withheldMethodology and Physical vs. Economic Withholding

13. 13The simulation-based market power assessments quantify withholding strategies’ impacts with a full-scale co-optimized DAM platform, including all generation assets and resources’ parametersSimulations employed the DAM module of ISO’s (in-house) Integrated Market Simulator platformThis is the platform being used for the DASI Impact Analysis simulations, (see ISO’s November MC presentation, slides 68-79)The ISO constructed resource-specific competitive DA A/S offers for each hour, using the methodology discussed with the CommitteeSee ISO’s December MC presentation, slides 15-52)ISO simulated DAM withholding strategies (per slide 12), over a large range of potential withheld MW for a number of suppliers, for the market power study days (next slide)As per the Information Policy, the suppliers’ identities (or identifying data) will not be disclosedMarket Power Simulation Methodology: Practicalities

14. 14Market power simulation study days are based on the pivotal supplier analysis reviewed above, over the 2016-2021 periodWe studied (many) withholding strategies on 46 historical days when market power may be a potential DAM concernDays with pivotal supplier (as defined earlier): 31 daysDays in which oil prices are less than gas prices: 15 daysThis includes the winter 2017-18 “holiday cold spell” period, when system conditions were generally tight (on many days)These are among the DAM’s highest price days historicallyThe simulations assume actual historical DAM conditions, but for the (hypothetical) withholding actions studiedUnits on outage in the DAM on the historical day are modeled as on outage, as actually occurred;Units that self-committed/self-scheduled in the DAM are modeled as actually occurred (unless modeled as withheld); and so onGoal is realistic quantification of the impacts of withholding in a co-optimized DAM, at the times/days it would be expected to matterMarket Power Simulation Methodology: Study days

15. 15Simulation scenarios: The assessment simulated many withholding strategies, including:Withholding of various individual A/S productsWithholding of total A/S capabilityWithholding of resource’s total capacity (i.e., from A/S and energy)In general, the simulations were performed not only for (many of) the system’s largest suppliers each study day, but also for a number of smaller A/S and energy providersSome large suppliers have little A/S capability in their fleet, and results can differ significantly depending on a suppliers’ resource portfolioMarket Power Simulations Methodology: Scope

16. 16The market power information the ISO and IMM/EMM evaluate requires caution when presenting resultsTo avoid giving competitors detailed information about supply conditions conducive to market power, and to not inform which participants may be able to do so, the ISO plans to present only certain aggregated simulation study resultsResults presented must also not inadvertently disclose or identify individual resource data, consistent with the Information PolicyISO is still evaluating the potential limits on data and results that can be reportedWe are open to stakeholder feedback on additional useful information from the simulation analyses, consistent with these limitationsThe full results (and detailed study methods) are discussed with the IMM and the EMM

17. 17We present next DA LMP and (DA LMP + FERP) impact results when A/S capability is withheld as follows:Withholding simulated: The DAM’s largest A/S provider (in each day’s competitive case) withholds DA A/S capabilityWithholding is simulated for the duration of the dayThe largest A/S provider varies across the 46 study daysAmount withheld: We present impacts of a ‘low’ withholding of 10% case (figure A) and ‘high’ withholding of 40% case (figure B)These percentages are of the supplier’s A/S capability that cleared in the competitive case for the same dayFinding #2: There is limited impact of withholding DA A/S capability on DA energy prices, even by large suppliers

18. 18Interpreting Price Impact GraphsHow to read the graphs ahead: Competitive price scenario results for both the (DA Hub LMP) and (DA Hub LMP + FERP) are shown on the horizontal axisThe withholding scenario price outcomes for both the (DA Hub LMP) and (DA Hub LMP + FERP) are shown in the vertical axisEach data point is an hour of the DAMKeys to interpreting the graphs:Data points above the 45-degree line (toward the north-west) are hours when prices are higher as a result of withholdingData points that are (approx.) on the 45-degree line are hours when the prices are (approx.) unchanged as a result of withholdingData points that are below the 45 degree line (to the south-east) are hours when prices are lower when withholding occurs

19. 19Figure A: Withholding A/S capability impacts on DA energy prices

20. 20Figure B: Withholding A/S capability impacts on DA energy prices

21. 21On average, LMP and (LMP + FERP) move very slightly with high levels of A/S capability withholdingThe average change in (LMP + FERP) is essentially $0, even on the tightest DAM days over these six years (those with a pivotal supplier in the DAM)We also observe that (on some days) the co-optimized market’s clearing engine (MCE) “pre-positions” resources to create additional A/S capabilityThat substitution by the MCE tends to minimize the impact of A/S withholdingEnergy prices are lower as a result of withholding in many simulated hoursThis occurs because of generators’ lumpiness and ‘replacement commitments’ by the DAM’s co-optimized MCE, particularly when larger amounts are withheldThe price impact of larger amounts of DA A/S capability withholding is slightly greater than smaller amounts (compare figure A vs. figure B)We have performed similar analyses over a wide range of withholding amounts; the largest mean effect on (LMP + FERP) is $0.68/MWh. Finding #2: There is limited impact of withholding DA A/S capability on DA energy prices, even by large suppliers

22. 22We present next DA A/S price impact results when A/S capability is withheld as follows:Withholding simulated: The DAM’s largest A/S provider (in each day’s competitive case) withholds DA A/S capability. Withholding is simulated for the duration of the dayThe largest A/S provider varies across the 46 study daysAmount withheld: We present impacts of a ‘low’ withholding of 10% case (figure C) and ‘high’ withholding of 40% case (figure D)These percentages are of the supplier’s A/S capability that cleared in the competitive case for the same dayFinding #3: Under certain conditions, large suppliers could raise DA A/S prices by withholding A/S capability

23. 23Figure C: Withholding A/S capability impacts on DA A/S prices

24. 24Figure D: Withholding A/S capability impacts on DA A/S prices

25. 25Overall, even in the higher (40%) withholding simulations, the average impacts on FRS clearing prices are not large (~$1/MWh)These results are for the DAM’s tightest historical daysThis occurs, in part, because we observe that DA A/S prices are lower as a result of A/S withholding in many simulated hoursHowever, in a limited number of hours, the impacts can be largeThe largest DA A/S price increases (compared to the competitive case) from withholding 40% of A/S capability is ~$80/MWh (more on this below)The largest FRS price changes are in the highest quality product, TMSRThis is consistent throughout simulation scenarios, in which we have simulated (a range of) other DA A/S withholding amountsFinding #3 Summary: Under certain conditions, large suppliers could raise DA A/S prices by withholding A/S capability

26. 26Based on the analyses, we can identify several reasons why A/S withholding is generally not an effective means to exercise market power in the co-optimized DAM:The DAM clearing engine makes substitutions (between A/S products, and between and energy and A/S products) to minimize the impact of DA A/S capability withholding For example:A slow moving resource might be scheduled to start (or start at a different time) to create additional (spinning) A/S on it or on other committed unitsThe DAM can clear more energy, freeing A/S capability otherwise used for EIR to meet other A/S (the FRS) requirementsA lumpy fast-start resource that clears for energy in the competitive case can be not committed (in the DAM) and increase its (offline) A/S capabilityThe DAM may commit (one or more) additional units if lower-cost supply is withheld, which can lower clearing prices due to generators’ lumpinessThe A/S requirement is small compared to the system’s total DA capabilityLimitations on suppliers’ ability to raise prices: Insights and explanations on Findings #2 and #3

27. 27As noted, under certain conditions, a large supplier could raise DA A/S prices by withholding A/S (Finding #3)However, the hours in which withholding DA A/S is profitable are quite challenging to predictAttempting to do so can result in lower A/S prices, even within the same day (and for adjacent hours)In the next graphs, we present data that highlight these pointsFigure E (next) is constructed similarly to those above, but shows data for two high-price days: January 6 and 7, 2018These were tight system days, but not days with a DA pivotal supplierBoth days had similar loads, temperatures, and fuel pricesFinding #4: The hours in which DA A/S withholding is profitable are challenging to predict

28. 28Figure E: Certain large resources can raise A/S prices by withholding A/S capability, but doing so can result in lower A/S prices and, as a result, hours when A/S withholding would be profitable are challenging to predict

29. 29Figure E highlights two of our conclusions:Large resources can raise prices in certain conditions (Finding #3) The largest price increases (compared to the competitive case) from withholding 40% of A/S capability is ~$80/MWh on January 6, 2018 HE 20. But hours in which withholding A/S would be profitable (versus incurring losses) are challenging to predict, even on tight DAM days (Finding #4)The largest price decrease (compared to the competitive case) from withholding 40% of A/S capability is –$60/MWh on January 7, 2018 HE 17This uncertainty is a (partial) deterrent with DA A/S withholding: The simulations reveal attempts to withhold DA A/S can be privately costly, reducing the prices paid to all MW the supplier clears DAWe conclude it would be difficult for a supplier to knowingly avoid the hours this would occur, if it attempts to withholdFinding #4: The hours in which withholding DA A/S capability is profitable are challenging to predict

30. 30We present next energy price (DA LMP + FERP) impact results when total resource capability is withheldWithholding simulated: Between 200-400 MW of capacity (that cleared in each day’s competitive case) withheld entirely from the DAM, by a large supplier The exact MW amount varies with the withheld resource’s size Withholding is simulated for the duration of the dayThe hypothetically-withholding large supplier and specific resource varies across the 46 study daysConcern. In concept, adding new FER and FRS requirements could create greater incentives for strategic withholding total capability from the DAMFinding #5: Impacts of supplier withholding of total capability from the co-optimized DAM

31. 31Figure F: In certain (highly infrequent) conditions, absent mitigation, withholding a portion of a large supplier’s total capability could raise DAM prices

32. 32The average impacts of the simulated DA withholding on DA energy prices are not large ($3.59/MWh) These results are for the DAM’s tightest historical daysAs evident in earlier results, this occurs (in part) because in many simulated hours DA prices are lower as a result of withholding, and DAM substitution/competitionBecause the withheld MWs do not receive any revenue DA, the strategy is generally a money-losing one, even when the system appears tightHowever, in a limited number of hours, the impacts can be largeThe largest DA energy hourly price increases (compared to the competitive case) from withholding in this simulation exceeds $100/MWhFinding #5: Under certain (highly infrequent) conditions, absent mitigation, withholding a portion of a large supplier’s total capability may raise DAM energy prices in the co-optimized DAM

33. 33These simulation results do not inform the impacts or incentives for total capacity withholding in the DAM relative to today’s DAM (i.e., without FER or FRS requirements)That is, the simulation results supporting Finding #5 cannot be interpreted to imply DASI creates new/greater concerns with respect to total capacity withholding (absent mitigation)Rather, they confirm the need for (continuing) mitigation to address this potential (though the simulations indicate it is infrequent in the DAM)The largest impacts of total capacity withholding on DA energy prices occur on (a subset of infrequent) days with exceptionally high gas prices, high DAM market prices, and tight DA system conditionsThese are not days with DAM pivotal suppliers (by available capacity)This highlights the limitations of performing only pivotal supplier analyses of potential market power in the proposed co-optimized DAM, or mitigation based on pivotal supplier tests alone in that marketFinding #5: Observations and caveats

34. 34In theory: When DA A/S is withheld from the market, most of the response is in the DA A/S pricesA/S cleared quantities directly enter the A/S pricing constraints, but the energy constraints are only limitedly (an indirectly) impacted by A/S withholding In practice: This is what we observe in the simulations as well (and mostly for TMSR)Implications. This observation critically limits the overall profitability to large suppliers of DA A/S withholdingWithholding DA A/S has relatively little potential benefit in increasing the energy revenues to a large supplier’s portfolio of inframarginal resourcesIn effect, the co-optimized market inherently limits the ‘portfolio benefits’ of DAM withholding of A/S, when DASI’s new requirements are in effectFinding #6: The impacts of withholding DA A/S primarily impact DA A/S prices, not DA energy prices

35. 35We next present results on the relative frequency of DA TMSR price impacts, and DA energy price impacts, when a large supplier withholds DA A/SWithholding scenario here is the 40% DA A/S capability case summarized previously in Finding #3 (see slides 24)Figure G (next) graph shows the cumulative distribution function (CDF) of the changes in DA LMPs and TMSR prices in this scenarioIn the graph: The frequency of a price impact less than the horizontal axis values is shown on the vertical axisFor example, 90% percent of TMSR price impacts are below $6/MWh, and 90% of the DA Hub LMP impacts are below $1.32/MWh.Finding #6: The impacts of withholding DA A/S primarily impact DA A/S prices, not DA energy prices

36. 36Figure G: The impacts of withholding DA A/S primarily impact DA A/S prices, not DA energy prices

37. 37In general, for any particular impact level, the frequency of a DA energy price impact (at that level) is much less than a TMSR price impact (at that level)Graphically, this means that the CDF of the response in TMSR prices is typically to the right of the CDF of the response in LMPsAdditional simulations indicate this conclusion remains unchanged across different levels of DA A/S withholding and for different DA A/S productsThis does not mean it would be appropriate to ignore the potential for changes in the DA energy prices when measuring the impact of (potential) withholding of DA A/SRather, it indicates that withholding DA A/S will tend to have relatively little private benefit to a supplier’s portfolio of inframarginal energy resourcesFinding #6: The impacts of withholding DA A/S primarily impact DA A/S prices, not DA energy prices

38. 38ISO performed a pivotal supplier analysis and market power simulations to obtain realistic quantification of the impacts of withholding in a co-optimized DAMIn nearly (but not) all hours, competition and substitution can be expected to yield competitively-determined prices in the co-optimized DAMIn certain (historically infrequent) conditions, however, absent mitigation, a large supplier could profitably raise prices by withholding A/S capability from the DAMThe findings imply the need for a DA A/S mitigation framework with DASI, though suppliers’ incentives and conduct should not be expected to necessitate (trigger) mitigation frequentlyMarket power simulations: Key takeaways

39. DA A/S Conduct Test ParametersISO’s proposed approach to setting DA A/S offerconduct test reference and threshold prices39

40. 40Conduct Test: Identifies offers potentially inconsistent with competitive conduct  (“non-competitive offers”) A resource-specific estimate of a competitive offer price for DA A/S ISO will calculate these based on competitive DA A/S offer formulas and market conditions each day (more on this further below)Reference Prices are (ISO-estimated) competitive offer price levels; can be adjusted during consultative process (§ III.A.3 of the Tariff)Threshold Prices are offer price levels above which offers are flagged as potentially  non-competitiveConduct test reference and threshold prices

41. 41Reference prices for DA A/S offers are composed of one or more of the following components:Expected closeout cost: the resource’s expected RT settlement (charge) when the Hub RT LMP exceeds the strike priceFuel component for gas resources: the resource’s (avoidable) gas-related costs incurred as a result of a DA A/S award“Fuel” components for storage resources: the resource’s (avoidable) storage-related costs incurred as a result of a DA A/S awardAll DA A/S sellers face the expected closeout charge costThe “fuel”-related components are only applicable to a subset of resources (those with avoidable input-energy costs)For additional details, see ISO’s December MC presentation (slides 15-52) and memorandum on DASI Competitive Offer Price FormulationsResource-level reference prices: A recap

42. 42ISO will calculate resource-specific DA A/S reference prices (for the mitigation conduct test)Expected closeouts will be based on the statistical model and methodology discussed today and the ISO’s memorandum Statistical Model for Real-Time Energy PricesInput-energy (fuel-related) components will be based on the avoidable-cost logic for competitive A/S discussed in December (see prior slide link for offer formulation memorandum)The resource-specific reference price values will be calculated daily (prior to the DA market), specific to each hour, and shared (on a confidential basis, consistent with current practice) with the resourceRisk premiums will be addressed in threshold prices (next)Calculating resource-specific reference prices

43. 43As discussed in December, ISO proposes to account for risk premiums in the conduct test threshold price, not resource-specific reference pricesCurrent energy market practice, and EMM’s recommendation for DASI, is to account for risk premiums in the Conduct Test threshold (not in the reference price level. See: EMM ESI Memo, pp. 7-11)ISO’s numerical offer calculations illustrate a range risk premiums that may be consistent with competitive offer behavior (see ISO’s December MC presentation, slides 34-46)Risk preferences may vary widely among participants and over timeIMM’s pre-market consultation process provides a participant the opportunity to discuss and support a risk premium that could place resource’s offer outside the test thresholdConduct Test Threshold Prices and Risk Premiums

44. 44Conduct test threshold price is intended to set a reasonable “upper bound” on the range of competitive offer prices, accounting forPossible error in estimating resources specific offer cost components; A reasonable range (variation) of participant expectations of RT outcomes affecting competitive offers; A reasonable range of risk premiums The conduct test threshold’s ‘reasonable range’ must balance an inherent tension between under- and over-mitigation risksIf it is too low, it could result in over mitigation (mitigating competitive offers), inhibiting supply and efficient market participationIf it is too high, it could result in under mitigation (noncompetitive offers may go unmitigated), impeding competitive outcomesConduct Test Threshold Prices: Rationale and Objectives

45. 45One consideration in setting the DA A/S conduct test threshold price is that it should not be less than the expected closeout cost (or expected closeout and fuel-related components, where applicable)The basic logic is that a participant that tenders an offer price using a model/method similar to ISO’s competitive offer component estimates (slides 41) should not be flagged for potential withholdingThis minimum threshold price level does not (yet) account for risk premiums (further below)Key considerations setting the conduct test threshold price: It should be greater than competitive A/S costs

46. 46Figure H (next) considers resources whose only DA A/S cost is their expected closeout, and depicts their (hypothetical) hourly competitive offers for one dayConduct test threshold prices should be greater than competitive A/S costs: Illustrative exampleWe assume these A/S sellers’ competitive offers reflect a range of different close-out cost expectations (they may use different models, etc.)Offer prices are on the vertical axis, and the distribution of the A/S offers each hour is in a ‘box-and-whiskers’ format:Blue boxes represent the inter-quartile ranges (25th to 75th percentile) of offersThe blue horizontal line in each box is the average of the offersWhiskers represent the 5th and 95th percentiles of the offersBlue dots represent outlying offersISO’s expected closeouts is in solid orange line in the next figure

47. 47Figure H: Conduct test threshold prices should be greater than competitive A/S costs (cont’d)

48. 48Offers in the yellow shaded region in the previous figure are at or below ISO’s competitive cost estimateSetting the conduct test threshold below that (hourly-varying) level means that offers the ISO deems as competitive could be flagged as noncompetitiveThis is not consistent with a sensible balance between the mitigation design objectivesHowever, this level does not yet account for additional factors, including resource-specific additional costs (e.g. fuel-related costs) and risk premiums that may comprise a competitive DA A/S offerConduct test threshold prices should be greater than competitive cost: Interpretation

49. 49As noted, the conduct test threshold should not be set below the (hourly-varying) level that the ISO deems as reflecting a competitive DA A/S offerThis means that some resource types will need to have resource-specific conduct test threshold pricesSpecifically, e.g., gas-fired and certain storage-based resources that incur (avoidable) input energy costs as a result of a DA A/S award Conceptually, this means the ‘orange line’ for certain resource types (or specific resources) would be higher than shown in the prior figureOtherwise, the same logic continues to applyConduct test threshold prices for resources with DA A/S avoidable fuel/input energy costs

50. 50As noted previously, the conduct test threshold price should allow for reasonable risk premiumsConsistent with ISO’s current practice and EMM’s recommendation (Dec. MC Presentation, slide 16)Heterogeneous nature of risk preferences leads to potentially very wide range for risk premiums, making it impossible to set a conduct test threshold level that allows all possible risk premiums ISO’s estimates of some resource-specific historical risk premiums can be quite high, potentially reflecting factors other than risk (e.g., a unique opportunity cost in a resource-specific offer)ISO proposes to use a data-based approach to setting the conduct test threshold price that will accommodate a reasonable range of risk premiumsParticipants can use the existing consultation process (under § III.A.3) to review their risk premiums with the IMM if their offer (including risk premium) exceeds the thresholdConduct test threshold prices and “reasonable” risk premiums

51. 51Figure I (next), we incorporate in the previous figure a range of resource-specific (hypothetical) risk premiums in the distributions of resource’s A/S offersThe distribution of A/S offer prices each hour is the sum of the expected closeout and the risk premium, in a similar box-whisker formatWe again ignore fuel-related offer costs here (for simplicity)The unshaded region between purple and yellow shaded regions covers the vast majority of these A/S offersIt highlights that the conduct test threshold price:Should vary over the course of the day, andMay not accommodate all possible (resource-level) risk premiums that the ISO has estimated using its historical dataThe illustrative example, continued: Conduct test threshold prices to allow for a reasonable range of risk premiums

52. 52Figure I: Conduct test threshold prices will need to reflect a range of reasonable risk premiums, which may vary over the day

53. 53The unshaded range between the purple and yellow regions covers the vast majority of competitive offers (including risk premiums)It is wider when system conditions are more uncertainSetting conduct test threshold price outside this range may fail to balance under- and over-mitigation risksSome offers will inevitably fall outside the reasonable range (blue dots in the purple area in the graph)As noted previously, we encourage participants to utilize the consultation process as neededConduct test threshold prices will need to reflect a range of reasonable risk premiums, which may vary over the day (cont’d)

54. 54Resource-specific (default) A/S offer reference prices will be set at the ISO’s competitive offer estimate, excluding risk premiumsISO will use models and formulae discussed in December and January MC to determine these (dynamic) reference pricesConduct threshold price must balance over and under mitigation riskISO’s current thinking is a dynamic, data-based threshold price derived from market conditions and competitive offer (incl. risk premium) estimatesWhile a reasonable conduct test threshold should accommodate vast majority of competitive risk premiums, no conduct test threshold price can accommodate all individual circumstances (including risk premiums)Participants can seek consultation outlined in § III.A.3 to change their reference prices as neededConduct test considerations: Key takeaways

55. Recapping and Next Steps55

56. 56Key TakeawaysISO performed pivotal supplier analysis and market power simulations to inform its DASI mitigation designDay with a pivotal supplier in the DAM are infrequentAbility to raise DA prices through withholding DA A/S exists, but is limitedThe price impact of withholding DA A/S is primarily on the DA A/S prices, not DA LMPsWithholding capability from DA energy and A/S is more impactful, but also infrequent (and commonly not profitable)The ability of a seller to predict the impact of withholding is challenging

57. 57Key Takeaways (cont’d)ISO’s current thinking is to employ a dynamic, data-based conduct test threshold price, derived from market conditions and competitive offer (incl. risk premium) estimatesStakeholders feedback are welcome on all of today’s discussions

58. 58Next StepsISO will discuss additional market power assessment results in FebruaryMore detailed discussion on the approach and the detailed design elements February – April

59. 59Parviz Alivand(413) 535-4096 | palivand@iso-ne.com