Dispatch Down Service Analysis of reference price “stickiness” and
by user
Comments
Transcript
Dispatch Down Service Analysis of reference price “stickiness” and
Dispatch Down Service Analysis of reference price “stickiness” and market participant behaviour in the DDS Market September 17, 2010 Table of Contents 1.0 Executive Summary ........................................................................................... 1 2.0 Purpose ............................................................................................................... 1 3.0 Introduction......................................................................................................... 2 4.0 Reference Price Shelf (price “stickiness”) ....................................................... 2 4.1 4.2 5.0 5.1 5.2 Analysis of price “stickiness” ............................................................................. 2 Symmetry around the reference price and heat rate levels............................... 5 Market Participant Behaviour ............................................................................ 6 Historical Analysis............................................................................................. 6 Spark Spread Analysis...................................................................................... 9 6.0 Summary ........................................................................................................... 12 7.0 Conclusions / Next Steps ................................................................................ 13 1.0 Executive Summary In December 2007 the AESO implemented improvements to the energy market based on the Department of Energy’s Alberta’s Electricity Policy Framework of June 2005. The changes, sometimes referred to as quick hits, included a price adjustment mechanism that negates the downward effect that dispatching Transmission Must Run (TMR) units have on the pool price. The mechanism is known as the Dispatch Down Service Market (DDS). In July 2009 the AESO published a review of the performance of the Quick Hit changes that included a review of the DDS market and its effectiveness. Since that time, stakeholders have raised various issues regarding the functioning of the DDS market and in particular the offer behaviour of market participants. A 2008 MSA review of DDS highlighted a number of these issues, and coined the term Price “stickiness” at the reference price to describe instances where the pool price is set very close to the reference price for a considerable period of time.1 In 2010 the AESO agreed to do further analysis which was presented to the Market Advisory Committee (MAC) and the MSA. The analysis indicated: • • Price “stickiness” exists; 80% of the price “stickiness” is due to the DDS volume associated with the size of the Transmission Must Run (TMR) dispatch (DDS volume) while 20% is due to offer bunching just below the 12.5 reference price heat rate level. The shelf appears to have created asymmetry around the reference price level indicating that the reference price heat rate does not always result in a reference price that is above the typical historical operation of the market as intended. The AESO undertook further analysis to determine if there have been any observable changes in market participant behaviour with the introduction of the DDS market. Spark spread analysis was also conducted and is viewed to be inconclusive in determining if market participants have altered their offer behaviour in any meaningful or specific way as there does not appear to be any relationship between participation in the DDS market and the spark spread. 2.0 Purpose The AESO monitors the market on an ongoing basis and has prepared this report in response to various stakeholder concerns regarding the functioning of the DDS market. The report contains analysis of reference price “stickiness” and market participant behaviour in the DDS market, the two main issues raised by stakeholders. 1 “Quick Hits Review: Dispatch Down Service”, July 10, 2008: http://www.albertamsa.ca/files/DDS_Report_071008(2).pdf Dispatch Down Service September 17, 2010 1 3.0 Introduction During and after the AESO Quick Hits Six Month Review2 report several market participants raised issues regarding the effectiveness of DDS as it is currently used to correct the effect of TMR in the energy market. There are two main issues: 1. The reference price has created a price shelf in the merit order. 2. Due to market participant behaviour DDS does not fully reconstitute the price, i.e., the market participant behaviour in question is characterized by market participants offering lower than what they might have absent DDS At the request of the MAC the AESO undertook analysis to explore these issues. This paper provides the results of that analysis and the AESO’s conclusions. 4.0 Reference Price Shelf (price “stickiness”) 4.1 Analysis of price “stickiness” The analysis previously presented to the MAC indicated that there has been some evidence that price has been stuck around the reference price. The analysis indicated that reference price “stickiness” can be characterized by the amount of time prices are just below the reference price. There are two issues that contribute to the “stickiness”: 1. The DDS volume and the shelf that it creates in the merit order. 2. The fact that there is a known shelf in the merit order, results in the probability of being dispatched just below a 12.5 heat rate is significantly higher than just above a 12.5 heat rate. Due to this, market participants are incented to offer just below the 12.5 heat rate, resulting in offers “bunching up”, further contributing to the shelf just below the reference price. Figure 1 shows duration curves of annual heat rates pre and post quick hits. It is noted that in both 2008 and 2009 there has been a significant amount of time where the SMP has been set near a 12.5 heat rate. 2 “Quick Hits, A Six Month Review”, July 3, 2009. Dispatch Down Service September 17, 2010 2 Figure 1 Annual Heat Rate Duration Curves 20 15 18 14 13.5 13 16 Heat Rate (SMP/Bid-Week Gas Price, GJ/MWh) Heat Rate (SMP/Bid-Week Gas Price, GJ/MWh) 14.5 14 12 10 8 6 4 2 12.5 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% % of Time 2006 12 2007 2008 2009 YTD 11.5 11 10.5 10 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% % of Time 2006 2007 2008 2009 The DDS volume impact can be measured in part by the amount of time the SMP has been set immediately below the reference price (i.e. within 1¢). This is illustrated in figure 2. The amount of time the SMP has been within $1 and $5 is also included, as the amount of time within $1 is indicative of the impact of offers bunching up just below the reference price. Table 1 provides annual percentages for the amount of time the SMP has been set close to the reference price. It is noted that both 2008 and 2009 have seen increases in the amount of time the SMP has been set within $1 and 1¢ of the reference price. As seen in Figure 2, in 2006 and 2007 there appears to have been symmetry around the derived reference price. Post quick hits the shelf appears to have created asymmetry around the reference price level, suggesting that the reference price heat rate may not be set above the normal operation of the market as intended in the initial design. An analysis of historical symmetry at differing heat rates can be seen in section 4. Dispatch Down Service September 17, 2010 3 Table 1 Percent of Time the SMP is Set Relative to the Reference Price 2006 2007 2008 2009 (To Q3) 1.825% 2.881% 4.092% 7.616% -$5 ≤ SMP < -$1 0.290% 0.941% 4.951% 5.595% -$1 ≤ SMP < -$0.01 0.010% 0.022% 1.349% 4.626% -$0.01 ≤ SMP ≤ Ref. Price 0.001% 0.018% 0.232% 0.003% $Ref. Price < SMP ≤ $0.01 0.395% 0.691% 1.628% 0.258% $0.01 < SMP ≤ $1 1.783% 2.450% 0.574% 1.222% $1 < SMP ≤ $5 95.696% 92.996% 87.174% 80.680% Other Price Figure 2 Stickiness of the SMP to the Reference Price 30% 25% % of Time 20% 15% 10% 5% 0% 2006 2007 2008 2009 Year Five Dollars Less One Dollar Less One Cent Less One Cent More One Dollar More Five Dollars More Other Price The tendency for offers to bunch just below the reference price is illustrated by comparing the merit order curves pre and post quick hits implementation. Figure 3 shows aggregated merit order curves by the heat rate of the offers. Heat rates are calculated based on the bid-week gas price as it is consistent across the month. The figure illustrates that a shelf has developed at the 12.5 heat rate level. More information regarding the 12.5 heat rate shelf is provided in Table 2 which presents the percentage of overall offers in between a 12 heat rate and a 13 heat rate. Even after accounting for the DDS volume there is evidence that there is some amount of bunching below the 12.5 heat rate and this result has been found to be statistically significant, as represented in Table 2 below. The degree of bunching is illustrated in Figure 3 which visually displays the 12.5 heat rate shelf in relation to the rest of the time period. Figure 3 also shows that 1% of the time the price is at the 12.5 heat rate when the DDS volume is excluded compared to 5% with the DDS volume present. Thus 20% of the Dispatch Down Service September 17, 2010 4 shelf is due to market participant behaviour and 80% of the shelf is due to the DDS volume. Table 2 Heat Rate 12 - 12.5 12.5 - 13 Pre Quick Hits 0.16% 0.17% Post Quick Hits 1.58% 0.12% Post Quick Hits with DDS Volume Removed 0.45% 0.12% Figure 3 Percent of Non-Zero Offers vs. Offered Heat Rate Pre Quick Hits: June 2006 - May 2007 Post Quick Hits: Jun 2008 - May 2009 25 15 20 14 Heat Rate (GJ/MWh) Offered Heat Rate (Offer Price/Bid Week Gas Price, GJ/MWh) 16 15 13 12 11 10 9 10 8 60% 61% 62% 63% 64% 65% 66% 67% 68% 69% 70% % of Non-Zero Offers 5 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% % of Non-Zero Offers Pre Quick Hits Offers Post Quick Hits Offers Post Quick Hits Offers with DDS Volume Removed In summary, the analysis of the reference price shelf showed that there is indeed a shelf at or near the reference price. The TMR heat rate is at or near the reference price 8% to 9% of the time creating quite a pronounced shelf. Approximately 80% of this shelf is due to the DDS volume itself and the other 20% is due to market participant offers. 4.2 Symmetry around the reference price and heat rate levels The AESO analyzed the symmetry of offers around the reference price both pre and post quick hits implementation in order to determine if there had been a change in offer behaviour caused by the placement of the reference price. The reference price heat rate was intended to provide a reference price that was above the normal operation of the market. Figures 4 shows the % of time SMP is set below the reference price at varying heat rates for 2005, 2006 & 2007. Dispatch Down Service September 17, 2010 5 Figure 4 % of Time System Marginal Price (SMP) is less than the Transmission Must Run (TMR) Reference Price at Varying Heat Rates 2005 - 2007 100% 98% 96% 94% % of Time 92% 90% 88% 86% 84% 82% 80% 78% 12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 17 17.5 18 18.5 Heat rate Level 2005 2006 2007 As seen above, in all 3 years there is some market activity occurring both above and below the heat rate. It appears that most market activity occurs below a 15 heat rate as illustrated by a flattening of the curves, thus setting the heat rate at this level may reduce offer bunching and therefore price “stickiness”. 5.0 Market Participant Behaviour Several market participants have raised issues regarding the effectiveness of DDS as it is currently used to reconstitute price in the energy market. Stakeholders felt that due to changes in market participant behaviour since implementation DDS does not fully reconstitute the price. Further analysis was conducted and is shown below. 5.1 Historical Analysis Average zero dollar offers pre and post quick hits implementation were compared to determine if there has been any change in the amount of energy offered at $0/MWh. An increase in $0 offers might suggest that market participants are offering at a lower price than before the quick hits implementation to be eligible to provide DDS. As seen in figure 5, there does not appear to be any significant change in the amount of energy offered at $0/MWh. Dispatch Down Service September 17, 2010 6 Figure 5 Average Zero Dollar Offers Pre and Post Quick Hits 2007 - 2009 6,000 Average Offers (MW) 5,000 4,000 3,000 Data invalid due to Quick Hits Implementation in December 2,000 1,000 Dec-09 Oct-09 Nov-09 Sep-09 Jul-09 Aug-09 Jun-09 Apr-09 May-09 Mar-09 Jan-09 Feb-09 Dec-08 Oct-08 Nov-08 Sep-08 Jul-08 Aug-08 Jun-08 Apr-08 May-08 Mar-08 Jan-08 Feb-08 Dec-07 Oct-07 Nov-07 Sep-07 Jul-07 Aug-07 Jun-07 Apr-07 May-07 Mar-07 Jan-07 Feb-07 - Month In figure 6, an increase in average hourly generation can be observed from 2007 to 2009. Generation including DDS has also appeared to increase over 2007, indicating that there may have been a change in behaviour. However it is unclear if this change is due to the DDS market. Figure 6 also exhibits net revenue, total DDS payments and metered volumes of the selected gas generators. Net revenue represents the difference between total revenue and the cost of fuel derived from the unit heat rate and gas price. The decrease observed in revenues from 2008 to 2009 can be partly explained by the lower gas prices observed in 2009, as well as lower pool prices. Some change in market participant behaviour is to be expected with the introduction of a new market. Dispatch Down Service September 17, 2010 7 Figure 6 Yearly Net Revenue, DDS Payments and Average Hourly Metered Volumes for Select Gas Units Average Hourly Gas Generation $250 500 400 450 $200 Net Revenue and Total DDS Payments ($ millions) 400 350 100 50 - 2007 Average Gas Generation 150 Average Gas Generation Average Gas Generation 200 Average Gas Generation incl DDS 250 Average Gas Generation incl DDS MWh 300 2008 380 370 $150 360 $100 350 340 $50 Average Hourly Generation of Select Gas Units (MWh) 390 330 $- 320 2009 2007 Year Net Revenue 2008 2009 Total DDS Payments Average Metered Volumes DDS dispatch share by fuel type has generally displayed a high concentration of gas units due to the fact the gas units generally offer deeper discounts in the DDS market. Figure 7 displays DDS dispatch share by fuel type and by market participant. Figure 7 DDS Dispatch Share by Fuel Type DDS Dispatch Share by Participant 100% 100% 90% 80% 80% 60% 60% % of DDS Disipatched % of DDS Dispatched 70% 50% 40% 40% 30% 20% 20% 10% Gas Nov-09 Dec-09 Oct-09 Sep-09 Jul-09 Aug-09 Jun-09 May-09 Apr-09 Mar-09 Jan-09 Feb-09 0% Dec-08 Dec-09 Nov-09 Oct-09 Sep-09 Aug-09 Jul-09 Jun-09 Apr-09 Coal May-09 Mar-09 Feb-09 Jan-09 Dec-08 0% Hydro Dispatch Down Service September 17, 2010 8 5.2 Spark Spread Analysis The previous analysis undertaken by the AESO looked at DDS provider offer behaviour before and after quick hits was implemented. The analysis suggested that offer behaviour may have changed but it was not clear that DDS contributed to that change in behaviour or was the only contributor. Other possible contributing factors include the must offer must comply rules and gas price. The AESO agreed to further examine gas unit participation in the DDS market by performing a spark spread analysis for the period 2007 to 2009. The spark spread analysis would indicate whether or not it was profitable for various gas-fired units to run. The spark spread was calculated as follows: Spark Spread = Price of Electricity ($/MWh) – [Cost of Gas ($/GJ) * Unit Heat Rate (GJ/MWh)] For the analysis, the NGX AB-NIT Same Day Index was used to represent the cost of natural gas. As seen in figure 7 below, the cumulative spark spread in 2008 displayed an increase over 2007. 2009 cumulative spread has declined over 2008 & 2007 due to lower pool prices. Large increases in spark spread generally occur in response to higher pool price or lower gas price. Figure 8 Cumulative Spark Spread Margin for Select Gas Units 2007 - 2009 $2,500,000 Cumulative Margin ($) $2,000,000 $1,500,000 $1,000,000 $500,000 2007 2008 31-Dec 17-Dec 3-Dec 19-Nov 5-Nov 22-Oct 8-Oct 24-Sep 10-Sep 27-Aug 13-Aug 30-Jul 2-Jul 16-Jul 18-Jun 4-Jun 21-May 7-May 23-Apr 9-Apr 26-Mar 12-Mar 26-Feb 29-Jan 12-Feb 15-Jan 1-Jan $- 2009 Weekly hours of positive spark spread was plotted against DDS dispatched to determine if the two variables are correlated. The results are provided in Figures 9 & 10 below. Dispatch Down Service September 17, 2010 9 Figure 9 Weekly Total Spark Spread vs. DDS Dispatched for Select Gas Units 2008 $160,000 Spark Spread Correlation: -0.33306 25,000 DDS Dispatched $140,000 20,000 $100,000 15,000 $80,000 10,000 $60,000 DDS Dispatched (MWh) Spark Spread ($) $120,000 $40,000 5,000 $20,000 $- 1 3 5 7 9 11 13 15 17 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 2008 weekly average pool price $300 $/MWh 19 $200 $100 $1 3 5 7 9 11 13 15 17 19 21 23 25 27 Week 29 31 33 35 37 39 41 43 45 47 49 51 Figure 10 Weekly Total Spark Spread vs. DDS Dispatched for Select Gas Units 2009 $80,000 weekly total spark spread Spark Spread for week 4: ~$150,000 DDS dispatched 25,000 Correlation: -0.23746 20,000 15,000 $40,000 10,000 DDS Dispatched (MWh) Spark Spread ($) $60,000 $20,000 5,000 $- 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 33 35 37 39 41 43 45 47 49 51 2009 weekly average pool price $300 $200 $100 $1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 39 41 43 45 47 49 51 The analysis indicated that there does not appear to be any relationship between DDS dispatched and the quantity of positive spark spread. The correlation coefficient is slightly negative for both years therefore indicating a slight negative relationship. Thus there is some indication that when spark spread is low participation in the DDS market increases. However, the monthly data also indicates that this relationship is inconsistent over time. Dispatch Down Service September 17, 2010 10 Figure 11 # of Generating Unit Hours where a DDS Dispatch Occurred During Negative & Positive Spark Spread Hours for Select Gas Units 12 1,600 1,400 10 8 # of hours 1,000 6 800 600 4 Average Gas Price (GJ/MWh) 1,200 400 2 200 Dispatch hours where spark spread is negative Dispatch hours where spark spread is positive Dec-09 Nov-09 Oct-09 Sep-09 Aug-09 Jul-09 Jun-09 May-09 Apr-09 Mar-09 Feb-09 Jan-09 Dec-08 Nov-08 Oct-08 Sep-08 Aug-08 Jul-08 Jun-08 May-08 Apr-08 Mar-08 Feb-08 0 Jan-08 - NGX AB-NIT same day index The AESO also examined the relationship between the hours in which DDS was dispatched and the spark spread in those hours. As seen in figure 11, a DDS dispatch can occur in both positive and negative spark spread hours.3 Based on the analysis in figure 11 it is difficult to determine if market participants are participating in the DDS market when it is less profitable for them to participate in the energy market. Figure 12 provides the percent of dispatch type during negative spark spread hours. This analysis was completed to determine if there has been more participation in the DDS market when it is less profitable in the energy market. The analysis was done a year prior to the implementation of quick hits to provide a comparison of market participant behaviour pre and post the inclusion of the DDS market. 3 Please note that in figure 11 generating unit hours where a DDS dispatch occurred were summed by month, thus the number of unit hours will not equal the number of hours in a month. Dispatch Down Service September 17, 2010 11 Figure 12 DDS & Energy Dispatched Compared to Negative Spark Spread Hours 2007 - 2009 100% Post Quick Hits 80% % of hours 60% 40% 20% DDS ONLY ENERGY ONLY Dec-09 Oct-09 Nov-09 Sep-09 Jul-09 Aug-09 Jun-09 Apr-09 May-09 Mar-09 Jan-09 Feb-09 Dec-08 Oct-08 Nov-08 Sep-08 Jul-08 Aug-08 Jun-08 Apr-08 BOTH ENERGY & DDS May-08 Mar-08 Jan-08 Feb-08 Dec-07 Oct-07 Nov-07 Sep-07 Jul-07 Aug-07 Jun-07 Apr-07 May-07 Mar-07 Jan-07 Feb-07 0% NO DISPATCH As seen above, energy dispatches have occurred fairly consistently during negative spark spread hours both pre and post quick hits. DDS is rarely dispatched without an accompanying energy dispatch that is greater than zero MW during a negative spark spread hour. This indicates that for the assets considered, there appears to be no attempt to participate in the DDS market and not in the energy market. There does not appear to be any relationship between participation in the DDS market and spark spread. Offer behaviour during negative spark spread hours since the quick hits implementation appears to be consistent with the behaviour before quick hits. It should also be noted that there are other market dynamics and technology limitations which can make it difficult for gas generators to accurately predict prices and consistently be in a position to capture positive spark spreads which may explain why generators are participating in the markets during negative spark spread hours. 6.0 Summary The AESO has undertaken further analysis of DDS offer behaviour and the reference price shelf. AESO analyses revealed no conclusive evidence of inappropriate offer behaviour and while a shelf has been created in the merit order around the reference price it is largely due to the DDS volume itself and not offer behaviour. Further explanation is provided below. Dispatch Down Service September 17, 2010 12 Reference Price “Stickiness” The emergence of a price shelf at the reference price was anticipated but the impact was expected to be mitigated by setting the reference price outside the typical, historical operation of the market. The AESO analysis indicates that: • There is evidence of a price shelf near a 12.5 heat rate. This price “stickiness” is mainly due to the DDS volume associated with the size of the TMR dispatch (DDS volume) and offer bunching just below the 12.5 heat rate. 80% of the price “stickiness” is due to the DDS volume, and 20% is due to offer bunching. • The imposed shelf appears to have created asymmetry around the reference price level suggesting that the 12.5 heat rate used to determine the reference price is not always above the typical historical operation of the market. While recognizing the impact that the reference price level may have had on the market, the AESO notes that the reference price is currently above the typical operation of the market and there are issues with any price reconstitution mechanism. In that context, the AESO is not convinced that major changes are required or would be efficient and effective in resolving the issue. Market Participant Behaviour Various analyses were carried out to examine market participant behaviour with the introduction of the DDS market. These include analysis of historical merit order distribution, DDS dispatch share by market participant, gas generation versus total net revenue for select gas units, and spark spread. These analyses have been inconclusive in determining if market participants have altered their behaviour in any meaningful or specific way. In particular, there does not appear to be any relationship between participation in the DDS market and the spark spread. Specifically, the AESO notes that DDS is rarely dispatched without an accompanying energy dispatch. This implies that market participants do not appear to be positioning their assets to inappropriately participate in the DDS market and not the energy market which would negate the price reconstitution intent of the DDS rules. The AESO analysis has not found evidence of a significant or inappropriate change in market participant behaviour related to the DDS rules. The AESO has also provided the results of its analysis to the MSA for their consideration and notes that the MSA also independently actively monitors the market for signs of inappropriate behaviour. Again the analysis has not convinced the AESO that major changes are required to the DDS market. 7.0 Conclusions / Next Steps The quick hits rules implemented some important and significant changes to Alberta’s electricity market design and the AESO agreed to report on the market after six months of operating under the quick hits rules. Subsequent to that six month review the AESO has undertaken further analysis specifically with respect to the DDS market rules which Dispatch Down Service September 17, 2010 13 incorporate a full 18 month analysis and address specific DDS market issues raised by various stakeholders. The AESO maintains its view that the DDS rules and the DDS market appears to be operating as designed. While there is evidence that some of the anticipated design issues occasionally arise, the DDS rules are technically sound and the DDS market is operating in a FEOC manner and in the public interest. The AESO does not believe that further work on rule changes are necessary or appropriate at this time. The AESO regularly reviews market metrics internally as part of its normal market monitoring function and has a regulatory duty to ensure that all ISO rules continue to promote the fair, efficient and openly competitive operation of the electricity market. The AESO will monitor the DDS market to ensure market conditions do not change in a manner that inappropriately impacts the DDS market operation. Dispatch Down Service September 17, 2010 14