California ISO Report Regional Surplus Allocation Impact Study October 6, 2010
Regional Surplus Allocation Impact Study Table of Contents Executive Summary... 3 1 Issue and Background... 3 2 Study Framework... 5 3 Study Approach... 7 3.1 Surplus Determination... 7 3.2 MLS Allocation to Measured Demand... 8 3.3 Regional MLS Allocation - No Path 26 Adjustment versus Path 26 Adjustment... 8 4 Results...10 4.1 Results...10 5 CAISO Proposal and Recommendations...12 CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 2 of 12
Regional Surplus Allocation Impact Study Prepared for Discussion at Market Performance and Planning Forum on Executive Summary October 21, 2010 The California ISO currently allocates the Surplus (MLS) pro-rata to all Measured Demand (internal demand plus exports) and the filed methodology was accepted by FERC in the MRTU Order dated September 21, 2006. In filed comments on the ISO MRTU Tariff, PG&E suggested an alternative approach to allocate MLS first based on actual contributions to marginal losses surplus in each of two regions within the ISO balancing authority area, namely, northern California and southern California before allocating to metered demand in each region. The ISO conducted a study in April 2007 to assess the distributional impacts of both methods. 1 That study found that the results based on both allocation methods were not significantly different, and concluded to keep the FERC filed and accepted methodology. Because the earlier study was based on market simulation data and may not be representative of actual market outcomes, the ISO previously committed to stakeholders to update that study using one year of market data after the startup of the new market on April 1, 2009. This report provides an update of the April 2007 study. The study compared the impact of MLS allocation based on the regional contribution to actual marginal losses (regional approach) versus system-wide allocation of MLS to measured demand (filed methodology) in two regions - Northern California and Southern California. Two alternative regional approaches were calculated: Regional approach1: Regional actual losses Regional approach 2: Regional actual losses accounting for losses in one regional supporting service of load in another region based on Path 26 actual flow adjustment 1 In the Se0ptember 21, 2006 MRTU Order, the Commission has stated that the method for disbursing the amounts of any over collections should not directly reimburse customers for their marginal losses payments, as such a reimbursement would interfere with the goal of basing prices on marginal losses and would undermine price signals to investors and load, which has been set forth in the Commission s early rulings on the excess losses revenue refund. In this Order, the Commission has rejected PG&E s proposal due to potential arbitrariness in the selection of a reference location and inconsistency with FERC s early rulings. (Order Conditionally Accepting the California ISO's Electric Tariff Filing to Reflect Market Redesign and Technology Upgrade in Docket Nos. ER06-615-000 and ER02-1656-027, et al. September 21, 2006, PP 94-95.) CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 3 of 12
The premise of the study is that if the filed MLS allocation methodology results in losses allocation close to or between the two regional approaches then the filed methodology would be assumed to be sufficient. However, if the filed MLS allocation methodology results fall substantially outside of the regional approaches then further consideration of an alternative regional allocation may be warranted. The latest study found that the allocation based on the filed methodology for the Northern region was lower than the estimated range indicated by the two regional approaches in the study. Results indicate that for the first year of operation of the new market, the filed methodology resulted in an $18.8 million (22%) and $13.8 million (18%) lower MLS allocation to northern load than regional approach 1 and approach 2, respectively. Conversely, the results indicate that MLS allocated to southern regional load using the filed methodology exceeds the regional approach 1 and regional approach 2 by $18.8 million (18%) and $13.8 million (13%),, respectively. Since the ISO did not utilize the scheduled path 26 flows from the actual Day-Ahead Market because of the lack of availability of such data, the study utilized actual path 26 data for the analysis of regional approach 2. In order to confirm the impact of using actual path 26 data instead of scheduled path 26 data from the Day-Ahead Market, the ISO has obtained path 26 scheduled flows for 30 selected days by rerunning the day ahead market, and has re-estimated the results by replacing the metered flows with the scheduled flows for these 30 days. Although the re-estimated results reduced the difference between the filed rate methodology and the regional approaches, the difference was insignificant. While any MLS allocation method has a degree of arbitrariness due to selection of the reference location as well as the lack of specific balanced association between a specific participants supply and demand, the difference between the filed MLS methodology and alternative regional approaches are significant enough to justify further consideration. As a result, the ISO proposes to consider potential changes in the MLS allocation methodology through a stakeholder policy review initiative. The ISO proposes to begin a stakeholder initiative on the MLS allocation methodology in the 1 st quarter of 2011. In the meantime, the ISO intends to analyze data covering the period after April 1, 2010, which will further inform the stakeholder process. 1 Issue and Background The California ISO (ISO) currently allocates marginal losses surplus (MLS) pro-rata to all Measured Demand (internal demand plus exports). This methodology was filed by the ISO and accepted by FERC in the MRTU Order dated September 21, 2006. In filed comments on the ISO MRTU Tariff, PG&E raised concern that the ISO proposal may fail to recognize the regional differences in actual losses costs and proposed an alternative approach to allocate MLS based on CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 4 of 12
actual contributions to marginal losses surplus in each region. 2 To address PG&E s concern, the ISO conducted and published a study in April 2007 to assess the distributional impacts of both methods (the April 2007 study). 3 That study found that the results based on both allocation methods were not significantly different, and concluded to keep the current methodology. Because the previous study was based on market simulation data and may not be representative of actual market outcome, the ISO has committed to stakeholders to update that study using one year of market operation data after the startup of the New Market. This report provides an update of the April 2007 study using market operation data from April 1, 2009 to March 31, 2010. 2 Study Framework This study adopts the same approach as described in the April 2007 study. An overview of the approach is provided here and more details can be found in the previous report. This paper compares the impact of MLS allocation based on regional contribution to actual marginal losses versus system-wide allocation of MLS to Measured Demand. Measured Demand is metered Demand plus Real-Time interchange export schedules 4. Two regions - Northern California (NP15 plus ZP26, collectively designated as NP26) and Southern California (SP26) are used. Specifically, the study computes the MLS in each of these two regions and compares them to the MLS allocation to these regions based on metered Demand. For the purpose of this study, actual and marginal losses costs in Northern and Southern California were calculated. Transmission losses costs on Path 26 are either included in the 2 3 4 In the September 21, 2006 MRTU Order, the Commission accepted the CAISO s proposed methodology and found that it allows the market participants to pay the marginal cost of energy. The Commission stated that the method for disbursing the amounts of any over collections should not directly reimburse customers for their marginal losses payments, as such a reimbursement would interfere with the goal of basing prices on marginal losses and would undermine price signals to investors and load, which has been set forth in the Commission s early rulings on the excess losses revenue refund. In this Order, the Commission has rejected PG&E s proposal due to potential arbitrariness in the selection of a reference location and inconsistency with FERC s early rulings. (Order Conditionally Accepting the California ISO's Electric Tariff Filing to Reflect Market Redesign and Technology Upgrade in Docket Nos. ER06-615-000 and ER02-1656-027, et al. September 21, 2006, PP 94-95.) Regional Allocation Impact Study, April 2007. http://www.caiso.com/2781/27817949719e0.pdf For purposes of the allocation of marginal losses surplus, the Real-time interchange export schedule at each inter-tie scheduling point is the non-negative net export (i.e., the export minus the import, if the net is an export; otherwise zero) at that scheduling point. This is to avoid potentially perverse scheduling incentives that could result from allocating marginal losses surplus based on exports alone, whereby large quantities of otherwise cancelling import and export schedules could be submitted by an SC at a scheduling point to increase the SC s share of marginal losses surplus. CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 5 of 12
Northern region or the Southern region depending the direction of flow on Path 26. The MLS is determined by the difference between marginal and actual losses costs in each region. Regional MLS are estimated in two scenarios: adjusted (referred below as Path 26 Adjustment) versus unadjusted regional MLS (or No Path 26 Adjustment). Since transmission losses incurred in one region cannot be attributed to demand in that region alone, the actual and marginal losses costs in each region are adjusted to reflect the impact of demand in one region on the losses in the other region. For example, a share of the transmission losses costs within the Northern or the Southern region is deemed to be caused by demand 5 in the other region, and should be allocated to the other region accordingly. In this study, the amount of transmission losses costs in each region incurred to serve demand in the other region was estimated based on the direction and magnitude of Path 26 flow. A portion of the actual losses and marginal losses costs in the region on the exporting side of Path 26 was allocated to the region on the importing side of Path 26. For example, when Path 26 flow is in the north to south direction, some of the losses in the Northern California region are incurred to serve demand in the Southern California region; therefore, a portion of the losses costs in the Northern region were allocated to the Southern region. This applies to both actual and marginal losses. The difference between the adjusted marginal and actual losses costs in each region is the adjusted MLS. To determine what fraction of the losses in one region should be allocated to the other region given the direction and magnitude of the Path 26 Flow, a Path 26 adjustment factor (defined in the next section) was used to estimate the share of losses in one region attributable to demand in the other region. Finally, the regional MLS resulting from the two bookends, No Path 26 Adjustment and Path 26 Adjustment, are compared to the regional MLS derived by allocating system-wide MLS to Measured Demand in each region. If the latter falls into the range of the two bookends, then the difference between these two allocation methodology is not considered to be significant. As an equivalent measure, an average marginal losses surplus rate per MWh of demand was also calculated for both regions and for each of the three scenarios discussed above. 5 The demand in each region includes load in that region as well as net exports from that region to other control areas. CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 6 of 12
3 Study Approach 3.1 Surplus Determination Actual losses cost is calculated as total transmission losses (MW) evaluated at the energy component of the system LMPs. Regional Cost ($/h) = Regional (MW) System Marginal Energy Cost ($/MWh) on Path 26 were included in either one or the other region depending on the direction of the Path 26 flow. When the Path 26 flow is in N-S direction, losses on Path 26 are included in the Southern California, and vice versa. on the ties are included in the relevant region regardless of the direction of the flow on the ties since those either serve internal regional demand (if in the import direction) or exports (if in the export direction), both of which were included in the relevant regional Measured Demand. The Cost at each resource was calculated as net load 6 at that resource multiplied by the LMP Component at that resource. The Regional Marginal Cost is the sum of marginal losses cost at each resource in that region. The Marginal Cost on Path 26 7 was included in either the Northern or the Southern region depending on the direction of the flow on Path 26. If Path 26 flow is N-S, marginal losses cost on Path 26 is included in the Southern region and excluded from the Northern region, and vice versa. Regional Cost ($/h) = j { Net Load at resource j (MW) LMP Component at resource j ($/MWh)} The difference between the marginal and actual losses cost determines the marginal losses surplus. 6 Net load as defined here is the difference between load and generation (i.e., load generation) at the 7 specific location, and could be positive or negative. The Cost of Path 26 is the flow into Path 26 multiplied by the LMP Component of terminal where flow is entering Path26 minus the flow exiting Path 26 multiplied by the LMP Component of the terminal where flow is existing Path26. CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 7 of 12
3.2 MLS Allocation to Measured Demand The system-wide MLS was allocated to the Measured Demand in each region on an hourly basis. The sum of the regional load and tie exports from each region for each hour is the relevant Demand for the region. 3.3 Regional MLS Allocation - No Path 26 Adjustment versus Path 26 Adjustment As discussed in section 2, regional actual and marginal losses costs were calculated for two regional approaches, Path 26 Adjustment and No Path 26 Adjustment. In the Path 26 Adjustment scenario, a portion of the actual losses and marginal losses costs in the region on the exporting side of Path 26 was allocated to the region on the importing side of Path 26. A Path 26 Factor was defined as follows to determine how much to allocate from one region to the other given the direction and magnitude of Path 26 flow. If Path 26 Flow is N-S, P26 N-S Factor = Path 26 Flow/(Northern Region Demand + Path 26 Flow) If Path 26 Flow is S-N, P26 S-N Factor = Path 26 Flow/(Southern Region Demand + Path 26 Flow). The adjustment to actual and marginal losses costs based on the Path 26 Factor is summarized in the following Table. CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 8 of 12
Table 1A. Surplus Allocation (Path 26 Flow is North to South). NP26 No Path 26 Adjustment Path 26 Adjustment = NP26 Actual (excluding Path 26 Actual ) = NP26 - Path 26 Marginal * (1 - Path 26 NS Factor) * (1-Path 26 NS Factor) SP26 = SP26 + Path 26 Actual = SP26 + Path 26 Marginal + *Path 26 NS Factor + Marginal *Path 26 NS Factor Table 1B. Surplus Allocation (Path 26 Flow is South to North). NP26 No Path 26 Adjustment Path 26 Adjustment = NP26 Actual + Path 26 Actual = NP26 +Path 26 Marginal + *Path 26 NS Factor + Marginal *Path 26 SN Factor SP26 = SP26 (excluding Path 26 ) = SP26 - Path 26 Marginal * (1 - Path 26 SN Factor) * (1 - Path 26 SN Factor) CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 9 of 12
The difference between the regional marginal and actual losses cost determines the marginal losses surplus contribution in each region. The marginal losses surplus was calculated for each of the two regional approaches, No Path 26 Adjustment and Path 26 Adjustment, for each region on an hourly basis. The hourly values were then summed up for each month, and for the 12-month study period. 4 Results 4.1 Proxy for Path 26 Flow Due to lack of data for day-ahead path 26 scheduled flow, metered flow on Path 26 are used as a proxy for the study period. To validate the use of metered flow, the ISO has obtained path 26 scheduled flow by rerunning the day-ahead save cases in the Network Application environment for selected 30 days (two days per month including a weekday and weekend plus six days when metered path 26 flow are not available). For the selected 30 days, the metered flow and scheduled flow on path 26 demonstrate same trend and pattern, however there are some differences in magnitude. Results 4.2.1 12-Month Results The MLS allocation results are presented in Table 2. The corresponding results on MLS allocations rates are shown in Table 3 respectively. The results indicate that, from April 2009 to March 2010, the share of MLS in the Northern region was $86 million (or $0.79 per MWh of load) based on the demand ratio allocation method (i.e., the filed methodology). On the other hand, using a regional-based MLS allocation scheme would have resulted in an estimated share of MLS between $99.2 and $104.3 million (or $0.92 to $0.97 per MWh of load). Similarly, the share of MLS in the Southern region was $104 million (or $0.80 per MWh of load) based on the demand ratio allocation method. Using a regional-based MLS allocation scheme would result in an estimated share of MLS between $85.5 and $90.5 million (or $0.66 to $0.69 per MWh of load). CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 10 of 12
Table 2. Surplus Allocation, April 2009 ~ March 2010. System MLS System Load Ratio Method (Filed Methodology) No Path 26 Adjustment Bookend Path 26 Adjustment Bookend NP26 SP26 NP26 SP26 NP26 SP26 Apr-09 $10.9 $5.0 $5.9 $6.5 $4.4 $6.2 $4.7 May-09 $15.8 $7.2 $8.6 $10.1 $5.7 $9.6 $6.2 Jun-09 $13.2 $6.2 $7.0 $7.9 $5.3 $7.6 $5.7 Jul-09 $18.4 $8.1 $10.2 $11.4 $6.9 $10.6 $7.8 Aug-09 $16.8 $7.4 $9.4 $9.2 $7.6 $8.9 $7.9 Sep-09 $16.1 $6.9 $9.2 $9.0 $7.1 $8.8 $7.3 Oct-09 $17.0 $7.6 $9.4 $7.1 $9.8 $7.2 $9.8 Nov-09 $13.9 $6.3 $7.6 $6.9 $6.9 $6.9 $7.0 Dec-09 $22.0 $10.1 $11.9 $11.0 $11.1 $10.7 $11.3 Jan-10 $17.0 $7.7 $9.3 $8.8 $8.2 $8.3 $8.7 Feb-10 $14.3 $6.5 $7.8 $7.9 $6.4 $7.1 $7.2 Mar-10 $14.4 $6.5 $7.9 $8.4 $6.0 $7.5 $6.9 Total $189.8 $85.5 $104.3 $104.3 $85.5 $99.2 $90.5 Table 3. Surplus Allocation Rates ($/MWh Load), April 2009 ~ March 2010. System MLS System Load Ratio Method (Filed Methodology) No Path 26 Adjustment Bookend Path 26 Adjustment Bookend NP26 SP26 NP26 SP26 NP26 SP26 Apr-09 $0.60 $0.60 $0.60 $0.78 $0.45 $0.75 $0.48 May-09 $0.78 $0.77 $0.78 $1.08 $0.52 $1.02 $0.57 Jun-09 $0.65 $0.66 $0.65 $0.83 $0.50 $0.80 $0.53 Jul-09 $0.78 $0.78 $0.78 $1.09 $0.53 $1.01 $0.60 Aug-09 $0.72 $0.72 $0.72 $0.89 $0.59 $0.86 $0.61 Sep-09 $0.72 $0.72 $0.72 $0.94 $0.56 $0.92 $0.57 Oct-09 $0.88 $0.88 $0.89 $0.83 $0.93 $0.83 $0.92 Nov-09 $0.76 $0.75 $0.76 $0.84 $0.69 $0.83 $0.70 Dec-09 $1.13 $1.13 $1.13 $1.22 $1.05 $1.19 $1.08 Jan-10 $0.91 $0.91 $0.91 $1.04 $0.80 $0.98 $0.85 Feb-10 $0.86 $0.86 $0.86 $1.04 $0.70 $0.94 $0.78 Mar-10 $0.78 $0.78 $0.79 $1.01 $0.59 $0.90 $0.68 Total $0.80 $0.79 $0.80 $0.97 $0.65 $0.92 $0.69 CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 11 of 12
4.2.2 30-Day Sample Results To assess the impact of using actually metered path 26 flows instead of scheduled dayahead path 26 flows, the total marginal losses surplus for the selected 30 days where actual scheduled day-ahead path 26 flow data were obtained through rerunning the day-ahead market instead of using actual metered flow on Path 26. The results for these selected days were $14.3 million system-wide, and the MLS for NP26 was $6.5 million, but it could be between $7.4 million and $7.6 million based on regional allocation method. The results of these selected days indicate that for the first year of operation of the new market, the filed methodology resulted in an a 17% and 13% difference in MLS allocation to northern load than regional approach 1 and approach 2, respectively. These results are not significantly lower in terms of percentage difference as the results observed using actual measured Path 26 flows for the entire period of April 1, 2009-March 31, 2010. Table 4. Surplus Allocation for 30 Selected Days with Path26 Rerun Data. System Load Ratio No Path 26 No. of System Method Adjustment Path 26 Adjustment Selected MLS (Filed Methodology) Bookend Bookend Days NP26 SP26 NP26 SP26 NP26 SP26 30 $14.3 $6.5 $7.8 $7.6 $6.7 $7.4 $6.9 5 ISO Proposal and Recommendations The study results comparing the filed MLS methodology with alternative regional allocation methodologies are sufficiently different that the ISO proposes to consider the matter of MLS allocation further. The ISO proposes to start a stakeholder policy initiative in the 1 st quarter of 2011. In the meantime, the ISO intends to analyze data covering the period after April 1, 2010, which will further inform the stakeholder process. CAISO/M&ID/MA&D/Holly Liu 10/06/2010, page 12 of 12