MISO Competitive Retail Solution: Analysis of Reliability Implications

Transcription

1 MISO Competitive Retail Solution: Analysis of Reliability Implications September RASC Meeting P R E P A R E D F O R Midcontinent Independent Transmission System Operator P R E P A R E D B Y Samuel Newell Kathleen Spees David Luke Oates Lily Mwalenga John Imon Pedtke S e p t e m b e r 1 9, Copyright 2016 The Brattle Group, Inc.

2 Contents Executive Summary Modeling Approach Results Findings and Next Steps 1 brattle.com

3 Executive Summary Overview and Key Questions Scope MISO asked us to evaluate the likely performance of its Competitive Retail Solution (CRS), consisting of both the Forward Resource Auction (FRA ) and the Planning Resource Auction (PRA) MISO also asked us to recommend a sloped demand curve to achieve market design objectives Key Questions Will the CRS meet local and system resource adequacy objectives? How much price volatility and merchant investment can be expected? How might utility self-supply and merchant entry interact? 2 brattle.com

4 Executive Summary Methods Used modeling structure similar to that used in PJM and ISO-NE Model long-run equilibrium conditions (after current supply excess has exited) Monte Carlo simulation model, translates variability in supply and demand into variation in cleared prices, quantities, and reliability Locational supply and demand curves for competitive retail zones and system with additional features Sequential forward and prompt auctions Individual utilities planning to meet their own needs Utilities assumed to build or buy bilaterally when short on a forward basis Utilities sell forward when long by more than a buffer above their own requirements, informed by research and interviews 3 brattle.com

5 Executive Summary Preliminary Findings Status Quo: Status quo design results in poor reliability (LOLE 0.19) as structural volatility, low price cap, and prices below Net CONE do not support sufficient merchant supply to meet needs of competitive retail customers MISO Proposal with Candidate Demand Curve: Reliability meeting or exceeding 1-in-10 standard in system and zones; procuring 1,800 MW more merchant supply; price volatility improved Key Sensitivities: Performance of the Candidate Demand Curve is sensitive to how much utility excess will be offered in the forward auction, and sizes of shocks to supply, demand, and transmission 4 brattle.com

6 Executive Summary Candidate Demand Curve Candidate demand curve is tuned to achieve 1-in-10 LOLE on behalf of competitive retail customers, though other demand curves achieving 1-in-10 could also be developed Cap: 1-in-5 LOLE, 140% Net CONE Cap is slightly above the 1x CONE requested by MISO, but still below the Brattle recommendation at Max(1x CONE, 1.5-2x Net CONE) Quantity at 1-in-5 LOLE, or approximately consistent with a minimum acceptable reliability level in any one year Foot : $0, 115% of PRMR Curve is wider than other markets, reflecting need to absorb more structural volatility than other markets and smaller size, and need to right-shift to accommodate the lower price cap 5 brattle.com

7 Contents Executive Summary Modeling Approach Results Findings and Next Steps 6 brattle.com

8 Modeling Approach Basic Monte Carlo Model Structure Purpose Estimate average, range, and distribution of capacity market outcomes: Price, quantity, and reliability System-wide, by location, and in forward/prompt Compare results realized with status quo and MISO proposals and under sensitivity assumptions Approach Input locational supply curves, demand curves, and transmission parameters for forward and prompt auctions Use a locational capacity clearing model to calculate prices and quantities Simulate a distribution of outcomes using a Monte Carlo analysis of realistic shocks to supply, demand, and transmission Simulate utility offers in forward auction under a range of assumed offer behaviors Merchant generators enter the auction if prices exceed Net CONE (exit if prices are lower) consistent with long-run equilibrium conditions Supply and Demand Shocks (Illustrative) Note: Illustrative shocks are not intended to reflect exact shock magnitudes or locational clearing results. 7 brattle.com

9 Modeling Approach Local Clearing and Transmission Constraints We model a simplified MISO locational auction clearing topology: Model only MISO North portion of system Only three zones and system (only reporting results for Zones 4 and 7 due to materiality threshold ) Only import limits (no export limits) System Demand Curves Inclusive of Zonal PRMR Demand Zone 2: WI/MI Zone 7: MI Only competitive retail share of import limits modeled in forward auction MISO South not modeled Including or excluding MISO South does not impact primary metrics of interest (system LOLE and price volatility) Import Limits Modeled: Forward: Competitive Retail share of import limit Prompt: 100% Zone 4: IL Data history is limited 8 brattle.com

10 Modeling Approach Equilibrium Conditions for Merchant and Utility Entry Utility Supply Utilities assumed to hold sufficient supply to meet their own requirements on an individual basis (if there were no merchant supplies) Individual utilities will build supply or buy bilaterally if expected to be short Utilities will offer some, but not all, of their supply in forward auction if expected to be long Merchant Entry Merchants will exit (enter) until clearing prices reach Net CONE on average Long-run equilibrium conditions may have substantially lower merchant supply if utilities hold excess supply If utilities are assumed to hold large quantities of excess supply, prices may be below Net CONE in an equilibrium where merchant supply drops to zero in a particular zone (although we do not observe this in our modeling) Long-Term Equilibrium Conditions Utility Demand Outside the Forward Auction Shocks to Supply Utility Self-Supply Sufficient to Meet Own Reliability Requirement Competitive Retail Demand in the Forward Auction Forward Price at Net CONE on Average Merchant Entry (Exit) Until Forward Prices Equal Net CONE on Average Shape of Supply Curve 9 brattle.com

11 Modeling Approach Time Sequential Forward and Prompt Auctions Utility Supply & Demand Competitive Retail Demand Utilities Offer into the Auction When Long Forward Auction Expected Supply Buffer Requirement Utilities Build or Bilaterally Purchase for Self-Supply When Short Forward Price at Net CONE on Average Merchant Supply Adjust quantity until long-run price is Net CONE in equilibrium Utility Supply Offers at Expected Prompt Price Utility Self-Supply Committed Supply to Competitive Retail A fraction of uncleared merchant drops out Net Position in Prompt Auction Utility Demand Competitive Retail Demand Prompt Auction Realized Supply Utility Self-Supply Sufficient to Meet Requirement for Own Customers Prompt Price Can Converge Below Net CONE Utility Supply Merchant Entry (Exit) 10 brattle.com

12 Modeling Approach Forward and Prompt Supply Curve Shapes Forward and prompt supply curve shapes based on PJM and MISO auctions respectively Model cycles through the eight shapes for the forward auction and the four shapes for the prompt auction Lumpiness reflected in locational supply curves, based on historical MISO offer block sizes Utilities offering into the forward auction offer at the expected prompt auction clearing price Smoothed Supply Curves Sources and Notes: Smoothed supply offer curves as published from PJM forward auctions, prompt supply curves based on actual MISO auctions. 11 brattle.com

13 Modeling Approach Parameters Based on 2016/17 Auction 2016/2017 Zone 4: Illinois Zone 7: South MI MISO North System Total Zonal Parameters PRMR (MW) 10,375 22, ,702 LRR (MW) 11,799 24,372 n/a LCR (MW) 3,788 20,851 n/a CIL¹ (MW) 8,011 3,521 n/a Gross CONE² ($/MW-day) $260 $260 $247 E&AS Offset³ ($/MW-day) $65 $65 $62 Net CONE ($/MW-day) $195 $195 $185 Retail Choice Load Parameters Retail Choice Share of Zonal Load (%) 78% 10% 10.2% Share of PRMR (MW) 8,093 2,241 10,401 Share of LRR (MW) 9,203 2,437 n/a Share of LCR (MW) 2,955 2,085 n/a Share of Import Limit (MW) 6, n/a 2016/17 Capacity Resource Supply FRAP (UCAP MW) 910 1,393 33,459 Offered (Including FRAP) (UCAP MW) 11,371 21, ,781 Cleared (Including FRAP) (UCAP MW) 9,152 21, ,826 Notes: 1. Zone 4 CIL value of 8,011 MW was used in modeling, reflected expected future import limit 2. Gross CONE for Zones 2, 4, and 7 set at the average of 2016/17 values. System value set 5% below zonal value to model system that is import-constrained on a long-run average basis. Source: 2016/17 PRA parameters. 3. Average E&AS margin for a new CT in the North region. System value set 5% below zonal values. Source: 2014 State of the Market Report. 12 brattle.com

14 Modeling Approach Variability in Supply, Demand, and Transmission Local Supply 1 Total Requirement (LRR or PRMR) Import Limit 2 Local Requirement (LCR or PRMR) Net Supply (w/ CIL) 3 Net Supply (w/o CIL) 3 Average Parameter Value Zone 4: Illinois (MW) 3,788 11,799 8,011 3,788 n/a n/a Zone 7: Michigan (MW) 20,851 24,372 3,521 20,851 n/a n/a MISO North System (MW) 101, ,702 n/a 101,702 n/a n/a Standard Deviation of Shocks Zone 4: Illinois (MW) ,522 1,541 1, Zone 7: Michigan (MW) 1, ,416 1,154 MISO North System (MW) 3,092 1,031 n/a n/a 3,249 3,249 Standard Deviation of Shocks Zone 4: Illinois (%) 5.3% 1.9% 19.0% 40.7% 41.2% 7.7% Zone 7: Michigan (%) 5.3% 1.5% 24.0% 4.4% 6.8% 5.5% MISO North System (%) 3.0% 1.0% n/a n/a 3.2% 3.2% Notes: 1. LCR in the zones and PRMR in the MISO North System are used as proxies to reflect the local supply in long-run equilibrium. 2. Standard deviation of shock to Zone 4 Import Limit based on historical Zone 4 Import Limit of 6,323 MW. 3. Standard deviation is reported as a percentage of LCR. 13 brattle.com

15 Modeling Approach Utility Net Supply Variability and Offer Behavior Modeling Collection of Individual utilities Forward Planning Meeting Utilities observe their net position If short, they build or buy bilaterally If long they can sell some excess bilaterally Forward Auction If long, utilities can sell down to a buffer above their requirement Base Case buffer is assumed to be 100 MW, or 3% of average supply (i.e. utilities offer 50% of their average excess); Sensitivity cases examined 0 MW and 200 MW buffer Prompt Auction All utility supply and demand included in the market Net Supply-Demand Balance for a 3,000 MW Utility 14 brattle.com

16 Modeling Approach Reliability as a Function of Reserve Margin We calculate realized reliability in each location as a consequence of the cleared quantity MISO staff provided estimates of system and local loss of load expectation (LOLE) for system and each zone consistent with reliability requirements study System LOLE vs. Reserve Margin Zones 4 and 7 LOLE vs. Reserve Margin 15 brattle.com

17 Contents Executive Summary Modeling Approach Results Findings and Next Steps 16 brattle.com

18 Results Status Quo Status quo design results in poor reliability in the long run as structural volatility, low price cap, and prices below Net CONE do not attract merchant supply until reliability is below acceptable levels Net CONE Average Standard Deviation Price Quantity Reliability Frequency at Cap Frequency of Price Separation Total Demand (LRR or PRMR) Local Merchant Demand Supply (LCR) Cleared Utility Supply Cleared Total Supply (Merchant Cleared plus Utility Total) Excess (Deficit) above LCR or PRMR Standard Deviation of Excess (Deficit) LOLE (Zones Incremental to System) Frequency Frequency Below LCR Below or PRMR 1-in-5 ($/MW-d) ($/MW-d) ($/MW-d) (% of years) (% of years) (MW) (MW) (MW) (MW) (MW) (MW) (MW) (events/yr) (% of years) (% of years) Prompt Auction Zone 4: Illinois $195 $195 $85 67% 20% 11,789 3,781 4, ,351 1,565 1, % 10% Zone 7: Michigan $195 $195 $81 65% 25% 24,347 20,831 2,243 19,220 21, % 0% MISO North System $185 $185 $88 62% n/a 101,651 n/a 7,510 93, ,346 (891) % 41% Long-run average prices converge to Net CONE as needed to attract merchant entry Low price cap and structural volatility result in most years at the price cap Persistent shortage on system-wide average basis Poor reliability at system level (1-in-5.2), with frequent severe shortage events 17 brattle.com

19 Results Competitive Retail Solution - Candidate Curve Forward Resource Auction ( FRA) supports reliability by attracting 1,800 MW of additional merchant supply. Sloped demand curve slightly reduces price volatility. Long run average forward prices converge to Net CONE to attract merchant entry. Long-run average prompt prices are below Net CONE and below Status Quo due to utility buffer. Forward price converges to Net CONE to attract merchant entry. Price volatility drops by 6-15%. FRA reduces clearing at cap by 60-70%. Some clearing at cap still required to keep average prices at Net CONE. FRA attracts 1,800 MW of additional merchant supply, split between zones and Rest of North. System clears slightly long on average. Zones clear long only relative to local requirement. Reliability objectives exceeded. RCS loads procure their target, but utility planning to avoid shortages and assumed buffer further improve reliability. 18 brattle.com

20 Results Forward Clearing Price and Quantity Competitive Retail Needs Are Met FRA clears 120 MW long on average Sufficient for competitive retail customers to meet their own reliability requirements Price Volatility Shows Some Improvement Small market size and large variability in utility offers mean that volatility is still high in the FRA FRA clears at the cap 40% of the time (much less frequently than Status Quo), but some clearing at the cap is required to offset low prices when market clears long. Higher price cap would reduce frequency at the cap and risk of going short, but at the expense of greater price volatility Cleared Quantity Distribution Clearing Price Distribution 19 brattle.com

21 Results Candidate Curve Design Rationale Cap Quantity at 1-in-5 Ensures price reaches cap before reliability deteriorates unacceptably Allows some shortage at high prices Price Cap at 1.4x Net CONE Price cap slightly above Gross CONE helps avoid low-reliability events Cap at a multiple of Net CONE (rather than Gross CONE) avoids excessively low cap if energy prices fall Higher cap mitigates reliability risk of administrative error in estimating Net CONE Foot at 115% of PRMR Wide curve ensures enough merchant supply to achieve 1-in-10 LOLE on average in the forward auction and to ensure that competitive retail loads don t lean on utility excess supply 20 brattle.com

22 Results Range of Demand Curves Considered 1. Candidate Curve (Cap: 1.4x Net CONE, 1-in-5; Foot: 115% PRMR) Tuned price cap to achieve forward LOLE at 1-in-10 with foot anchored at 115% of requirement 2. Cap at 1x CONE (Cap: Gross CONE, 1-in-5; Kink: 24% Gross CONE, 115% PRMR; Foot: 115% PRMR) Tuned foot point to forward LOLE at 1-in-10 (subject to maximum quantity of 115% of PRMR) 3. MISO-Wide Forward Market (Cap: Gross CONE, 1-in-5, Foot: 111% PRMR) Tuned foot to 1-in-10 forward LOLE for equivalent system where all MISO supply and demand is in the FRA, then kept same width on a proportional basis for competitive retail 4. Prompt LOLE Tuning (Cap: Gross CONE, 94% PRMR; Foot: 103% PRMR) Tuned foot and cap to meet prompt LOLE. Curve is shifted to the left because it counts on utility buffer that not offered forward to become available in prompt to support reliability 5. Prompt LOLE Tuning, 100% Excess Offered (Cap: Gross CONE, 1-in-5, Foot: 136% PRMR) Tuned foot to prompt LOLE (rather than forward like curves 1-3) assuming utilities offer 100% of their excess supply. When utilities offer all their excess, curve must be wide enough on a MW basis to absorb all net supply variation in the MISO North footprint 21 brattle.com

23 Results Demand Curve Performance All demand curves considered achieved at least 1-in-10 prompt reliability under Base Case assumptions. Candidate Demand Curve also achieves 1-in-10 on a forward basis, assuming utilities will always hold their target reserve margins. Net CONE Average Standard Deviation Price Quantity Reliability Frequency at Cap Frequency of Price Separation Total Demand (LRR or PRMR) Local Merchant Demand Supply (LCR) Cleared Utility Supply Cleared Total Supply (Merchant Cleared plus Utility Total) Excess (Deficit) above LCR or PRMR Standard Deviation of Excess (Deficit) LOLE (Zones Incremental to System) Frequency Frequency Below LCR Below or PRMR 1-in-5 ($/MW-d) ($/MW-d) ($/MW-d) (% of years) (% of years) (MW) (MW) (MW) (MW) (MW) (MW) (MW) (events/yr) (% of years) (% of years) Forward Auction Candidate Demand Curve $185 $185 $83 39% n/a 101,651 n/a 9,294 1, , % 9% Alternative Curve $185 $185 $74 40% n/a 101,651 n/a 9,308 1, , % 11% MISO-Wide Forward Market $185 $185 $80 50% n/a 101,651 n/a 8,974 1, ,698 (203) % 17% Utilities Offer 100% Excess $185 $185 $61 21% n/a 101,651 n/a 9,850 1, , , % 2% Utilities Offer 50% Excess $185 $185 $82 52% n/a 101,651 n/a 8,413 1, ,138 (809) % 40% Prompt Auction Candidate Demand Curve $185 $106 $79 13% n/a 101,651 n/a 9,360 92, ,078 1,426 1, % 2% Alternative Curve $185 $107 $80 14% n/a 101,651 n/a 9,353 92, ,106 1,454 1, % 2% MISO-Wide Forward Market $185 $116 $86 20% n/a 101,651 n/a 9,053 92, ,777 1,125 1, % 5% Utilities Offer 100% Excess $185 $90 $67 5% n/a 101,651 n/a 9,872 92, ,653 2,001 1, % 0% Utilities Offer 50% Excess $185 $121 $87 23% n/a 101,651 n/a 8,495 92, , , % 14% 22 brattle.com

24 Results Alternative Assumptions for Sensitivity Analysis We tested the robustness of the CRS with our Candidate Curve to variations in shock sizes, supply shock correlations, and utility forward offers Status Quo Scenario Shocks to Supply, Demand, and Import Limit Utility Buffer Base Based on historical shock sizes 100 MW (3% of average utility) Utility Average Supply Requirement + 3% Base Based on historical shock sizes 100 MW (3%) Requirement + 3% CRS 1/2x Shocks Half historical size 100 MW (3%) Requirement + 1% 2x Shocks Double historical size 100 MW (3%) Requirement + 5% No Buffer Most Challenging to Reliability 2x Buffer Least Challenging to Reliability Higher Supply Correlations Based on historical shock sizes 0 MW Requirement + 3% Based on historical shock sizes 200 MW (6%) Requirement + 3% Correlations among individual utility and merchant supply higher than historical 100 MW (3%) Requirement + 3% Notes: Utility average supply exceeds target because utilities face shocks to supply and demand and we assume they build or buy when short Utility buffer is the amount of excess supply utilities will hold rather than offer forward 23 brattle.com

25 Results Sensitivity to Alternative Assumptions Reliability with Candidate Curve most sensitive to utility offers and shock size. If utilities offer without any buffer, or if shock sizes are twice as large as historically observed, reliability may suffer. Net CONE Average Standard Deviation Price Quantity Reliability Frequency at Cap Frequency of Price Separation Total Demand (LRR or PRMR) Local Merchant Demand Supply (LCR) Cleared Utility Supply Cleared Total Supply (Merchant Cleared plus Utility Total) Excess (Deficit) above LCR or PRMR Standard Deviation of Excess (Deficit) LOLE (Zones Incremental to System) Frequency Frequency Below LCR Below or PRMR 1-in-5 ($/MW-d) ($/MW-d) ($/MW-d) (% of years) (% of years) (MW) (MW) (MW) (MW) (MW) (MW) (MW) (events/yr) (% of years) (% of years) Forward Auction Base Case $185 $185 $83 39% n/a 101,651 n/a 9,294 1, , % 9% No Buffer $185 $185 $88 46% n/a 101,651 n/a 7,944 2, ,645 (160) 1, % 26% 200 MW Buffer $185 $185 $69 23% n/a 101,651 n/a 10, , % 1% 2x Shocks $185 $164 $102 43% n/a 101,651 n/a 7,223 2, ,229 (223) 1, % 31% 1/2x Shocks $185 $185 $51 7% n/a 101,651 n/a 10, , % 0% Alternative Correlations $185 $185 $80 36% n/a 101,651 n/a 9,179 1, , % 9% Prompt Auction Base Case $185 $106 $79 13% n/a 101,651 n/a 9,360 92, ,078 1,426 1, % 2% No Buffer $185 $190 $69 46% n/a 101,651 n/a 8,076 93, , , % 25% 200 MW Buffer $185 $63 $65 3% n/a 101,651 n/a 10,094 91, ,846 2,195 1, % 0% 2x Shocks $185 $113 $98 28% n/a 101,651 n/a 7,414 94, ,420 1,819 3, % 16% 1/2x Shocks $185 $93 $59 0% n/a 101,651 n/a 10,455 91, ,003 1, % 0% Alternative Correlations $185 $99 $79 13% n/a 101,651 n/a 9,201 92, ,082 1,431 1, % 2% With 2x shocks, no merchant supply outside zones. Utility offers suppress FRA prices. Price volatility ranges from -10% to +20% compared to base case. Volatility increases with shock size and utility offers. High (and volatile) utility offers in No Buffer and 2x Shocks cases drive out merchants. Reliability negatively affected by aggressive (and volatile) utility offers, with no buffer, but still close to 1-in-1024 brattle.com

26 Contents Executive Summary Modeling Approach Results Findings and Next Steps 25 brattle.com

27 Findings and Next Steps Summary of Findings Design Status Quo CRS With Candidate Curve and Base Case Assumptions Utility Buffer = 100 MW (3%) CRS With Candidate Curve and Alternative Assumptions Performance Status quo does not meet design objectives Poor reliability driven by inability to attract merchant supply to support competitive retail until reliability is unacceptably low Meets reliability objectives, improves volatility performance 1,800 MW of additional merchant supply supported compared to status quo Average FRA cleared quantity exceeds requirement by 120 MW, making competitive retail loads self-sufficient relative to their own reliability requirements Forward reliability at 1-in-10 on behalf of competitive retail loads, prompt exceeding objectives based on utility self-supply Forward price volatility improves slightly (drops 6-15% compared to status quo), but volatility is still a challenge based on swings in utility supply offered into the auction Size of utility buffer is the most substantial uncertainty: reliability with Candidate Curve drops below 1-in-10 with no utility buffer System reliability ranges from 0.08 (1/2x Shocks) to 0.18 (2x Shocks) Additional merchant supply relative to status quo ranges from 400 MW (No Utility Buffer) 2,500 MW (200 MW Utility Buffer) Price volatility ranges from no reduction compared to Status Quo (No Utility Buffer) to a 22% reduction (200 MW Utility Buffer) 26 brattle.com

28 Findings and Next Steps Stakeholder Input Requested Demand Curve Parameters Demand curves should achieve 1-in- 10 system LOLE, but exact parameters may vary: What should the price cap be? What should the quantity at the price cap be? What should the foot quantity be? Utility Modeling Utility model parameters can be adjusted: What size should the utility buffer be? At what price should utilities prefer building vs. buying when short? How much excess supply should utilities hold on average? Demand Curves Considered 1. Candidate Curve 2. Cap at 1x CONE 3. MISO-Wide Forward Market 4. Prompt LOLE Tuning 5. Prompt LOLE Tuning, 100% Excess Offered 27 brattle.com

29 Findings and Next Steps Input Requested on Model Parameters Assumption Base Assumption Range Tested Impact on Results Utility Behavior Utility Excess Utility buffer Build vs. Buy Monte Carlo Shocks Correlations Demand Curve Individual utilities face variable net supply and build/buy when short, resulting in owned supply 3% above requirement Each 3,000 MW utility offers when reserve margin exceeds 100 MW (3% of requirement, or offering 50% of excess) When short, individual utilities buy bilaterally when forward price less than 75% of Net CONE, else build Based on historical PJM and MISO shock sizes Based on PJM relationship between zone size and shock size Cap Quantity 99% of system PRMR, corresponds to 1- in-5 LOLE Foot Quantity 111% of system PRMR, corresponds to 1-in-10 if all supply and demand participated in FRA N/A MW (Offering % of Excess) N/A 1/2x 2x shock sizes High cross-zone correlation case Utility excess supports reliability, but only when not offered forward 200 MW: reliability better than 1-in-10 0 MW: reliability below 1-in-10 Bilateral buys slightly increase and builds slightly decrease if utilities willing to buy at prices up to 100% Net CONE, but no impact on reliability 1/2x: reliability better than 1-in-10 2x: reliability below 1-in-10 Minimal impact on reliability 94% - 99% Right shift attracts additional supply and mitigates low-reliability events 103% - 136% Right shift attracts additional supply and improves average reliability 28 brattle.com

30 Contact Information SAMUEL NEWELL Principal Cambridge, MA KATHLEEN SPEES Principal Cambridge, MA DAVID LUKE OATES Associate Cambridge, MA The views expressed in this presentation are strictly those of the presenter(s) and do not necessarily state or reflect the views of The Brattle Group, Inc. 29 brattle.com

31 About The Brattle Group The Brattle Group provides consulting and expert testimony in economics, finance, and regulation to corporations, law firms, and governmental agencies worldwide. We combine in-depth industry experience and rigorous analyses to help clients answer complex economic and financial questions in litigation and regulation, develop strategies for changing markets, and make critical business decisions. Our services to the electric power industry include: Climate Change Policy and Planning Rate Design and Cost Allocation Cost of Capital Regulatory Strategy and Litigation Support Demand Forecasting Methodology Renewables Demand Response and Energy Efficiency Resource Planning Electricity Market Modeling Retail Access and Restructuring Energy Asset Valuation Risk Management Energy Contract Litigation Market-Based Rates Environmental Compliance Market Design and Competitive Analysis Fuel and Power Procurement Mergers and Acquisitions Incentive Regulation Transmission 30 brattle.com

32 Offices CAMBRIDGE NEW YORK SAN FRANCISCO WASHINGTON, DC TORONTO LONDON MADRID ROME SYDNEY 31 brattle.com

33 Results Sensitivity to Alternative Assumptions Zone 4 Sensitivity of results with Candidate Curve to alternative assumptions Net CONE Average Standard Deviation Price Quantity Reliability Frequency at Cap Frequency of Price Separation Total Demand (LRR or PRMR) Local Demand (LCR) Merchant Supply Cleared Utility Supply Cleared Total Supply (Merchant Cleared plus Utility Total) Excess (Deficit) above LCR or PRMR Standard Deviation of Excess (Deficit) LOLE (Zones Incremental to System) Frequency Below LCR or PRMR Frequency Below 1-in-5 ($/MW-d) ($/MW-d) ($/MW-d) (% of years) (% of years) (MW) (MW) (MW) (MW) (MW) (MW) (MW) (events/yr) (% of years) (% of years) Forward Auction Base Case $195 $195 $72 39% 16% 11,789 3,781 4, ,819 1,971 1, % 0% No Buffer $195 $195 $79 47% 14% 11,789 3,781 4, ,895 2,133 1, % 0% 200 MW Buffer $195 $195 $62 25% 23% 11,789 3,781 4, ,614 1,715 1, % 0% 2x Shocks $195 $195 $92 53% 31% 11,778 3,895 5, ,232 2,264 2, % 8% 1/2x Shocks $195 $195 $45 8% 23% 11,794 3,785 3, , % 0% Alternative Correlations $195 $195 $71 37% 19% 11,789 3,781 4, ,651 1,801 1, % 0% Prompt Auction Base Case $195 $114 $79 15% 12% 11,789 3,781 4, ,824 2,043 1, % 1% No Buffer $195 $190 $66 46% 9% 11,789 3,781 4, ,957 2,176 1, % 1% 200 MW Buffer $195 $83 $78 9% 18% 11,789 3,781 4, ,586 1,804 1, % 3% 2x Shocks $195 $146 $101 41% 28% 11,778 3,895 5,158 1,095 6,263 2,369 2, % 17% 1/2x Shocks $195 $105 $65 3% 13% 11,794 3,785 3, ,790 1, % 1% Alternative Correlations $195 $110 $81 16% 14% 11,789 3,781 4, ,658 1,877 1, % 3% 32 brattle.com

34 Results Sensitivity to Alternative Assumptions Zone 7 Sensitivity of results with Candidate Curve to alternative assumptions Net CONE Average Standard Deviation Price Quantity Reliability Frequency at Cap Frequency of Price Separation Total Demand (LRR or PRMR) Local Merchant Demand Supply (LCR) Cleared Utility Supply Cleared Total Supply (Merchant Cleared plus Utility Total) Excess (Deficit) above LCR or PRMR Standard Deviation of Excess (Deficit) LOLE (Zones Incremental to System) Frequency Frequency Below LCR Below or PRMR 1-in-5 ($/MW-d) ($/MW-d) ($/MW-d) (% of years) (% of years) (MW) (MW) (MW) (MW) (MW) (MW) (MW) (events/yr) (% of years) (% of years) Forward Auction Base Case $195 $195 $71 39% 16% 24,347 20,831 2, , % 0% No Buffer $195 $195 $79 46% 14% 24,347 20,831 2, , % 0% 200 MW Buffer $195 $195 $59 24% 25% 24,347 20,831 2, , % 0% 2x Shocks $195 $195 $90 51% 38% 24,323 20,811 2, , , % 0% 1/2x Shocks $195 $195 $47 10% 30% 24,360 20,841 2, , % 0% Alternative Correlations $195 $195 $74 37% 26% 24,347 20,831 2, , % 0% Prompt Auction Base Case $195 $122 $77 15% 21% 24,347 20,831 2,390 19,115 21, % 0% No Buffer $195 $193 $62 47% 12% 24,347 20,831 2,399 19,267 21, % 0% 200 MW Buffer $195 $96 $76 8% 39% 24,347 20,831 2,292 19,035 21, % 0% 2x Shocks $195 $147 $97 37% 38% 24,323 20,811 2,193 19,643 22,051 1,239 1, % 4% 1/2x Shocks $195 $119 $64 4% 34% 24,360 20,841 2,221 18,898 21, % 0% Alternative Correlations $195 $119 $82 17% 30% 24,347 20,831 2,218 19,133 21, % 0% 33 brattle.com