EFFICIENT PRICING OF NETWORK INFRASTRUCTURE Harry COLEBOURN & Chris AMOS EnergyAustralia - Australia

EFFICIENT PRICING OF NETWORK INFRASTRUCTURE Harry COLEBOURN & Chris AMOS EnergyAustralia - Australia hcolebourn@energy.com.au, camos@energy.com.au INTRODUCTION In recent years, demand growth on Australian networks has been high, requiring additional capital expenditure to maintain acceptable standards of supply. In EnergyAustralia s case, capital expenditure over the 2004-2009 period of the current regulatory Determination is expected to increase three-fold over the previous Determination, with the major factor driving this increase being demand growth. There is thus increased emphasis by Networks on managing demand, and increasing encouragement by Australian Regulators for them to do so. Efficient infrastructure pricing serves to directly influence customer behaviour and forms a foundation for other demand management options. NETWORK PRICING IN PERSPECTIVE A factor which is relevant to the development of both infrastructure pricing and demand management initiatives is the structure of the organisations which form the national Grid, supplying electricity to the Eastern portion of Australia. The formerly integrated organisations were split up during the mid and late 1990 s, to facilitate competition in a National Electricity Market (NEM). Contestability has been introduced wherever feasible, to provide incentives for efficient operation. The significance of this industry disaggregation is that although pricing and demand management initiatives may have beneficial effects throughout the whole supply chain, the incentive driving each business is to improve its financial position within its business sector. No one industry participant is in a position to capture the full benefit arising from pricing or demand management measures and thereby prepared to invest up to the economic cost of implementation. Incentives are thus required to stimulate participation in economic investment. The price signals imposed by network businesses are passed on to the customer via retailers, who may not choose to directly reflect network price signals in their retail tariffs. For example, retail bill smoothing plans may serve to mute network price signals. Despite this, the network price signals are expected to elicit some response by retailers in order to mitigate the risks that would otherwise be imposed on them. PRICING AND DEMAND MANAGEMENT Network price structures can play a significant role in encouraging sustainable infrastructure development, sustainable energy usage and the development of energy efficient technology. The forms of pricing and demand management can be considered as three categories, as follows [1]: Environmentally driven. Mainly aimed at encouraging reduced overall energy consumption or fuel substitution. Energy Market driven. Aimed at reducing energy purchase costs by reducing energy consumption in high price periods, possibly by transferring consumption to lower price periods. Network driven. Aimed at reducing capital expenditure on the network by reducing demand during periods of network congestion, possibly by transferring consumption to periods of lower network loading. The three forms of pricing and demand management and their interrelationships are depicted in Figure 1. Figure 1 Network Driven Peak Demand Reduction Environmentally Driven Energy Efficiency or Demand Side Abatement Types of pricing and demand management Energy Market Driven Demand Response The essential points to be made concerning Figure 1 are as follows: Environmental options, which encourage an overall reduction in consumption, are unlikely to have a great effect either on high energy market prices or network congestion, both of which have a duration of a few tens of hours per annum. The converse also applies, in that measures targetted to reduce consumption during network and market peak periods are unlikely to have much effect on overall consumption. Furthermore, as will be demonstrated, there is a poor degree of correlation in the Australian market between network congestion periods and periods of high energy market price, which restricts the available opportunities for a combined solution. WHY HAVE COST RELECTIVE PRICING? Changes in the pattern of network loading Overall, the growth in energy consumption in EnergyAustralia s region (which includes two thirds of Sydney) has been 2.7% per annum during the period since 1996 (which includes the Olympics build up). Of more significance, however, has been the disproportionate growth in summer demand, which has averaged 4.3% per annum. This is significantly greater than the corresponding winter growth of 1.8%. These statistics reflect the impact of a significant change in the penetration and use of air conditioning. It is estimated that 48% of residential customers premises were air conditioned in 2003/04, compared with 31%

in 1997/98. The air conditioning load is predominantly in the domestic and small business sectors. Figure 2 illustrates the effect of air conditioning usage on the average customer demand profile. As might be expected, there is little discernible change in the pattern of summer consumption in non-air conditioned premises in hot weather. In contrast, the average air conditioned premise tends to be somewhat larger and have a higher base consumption, but the peak consumption increases dramatically in hot weather. Consumption Figure 2 AC, avg workday AC, hot workday No AC, avg workday Air conditioning usage Time of day No AC, hot workday Effect of air conditioning on domestic consumption Summer consumption is at its highest from 14:00 to 18:00 hours, the period when high summer ambient temperatures adversely affect the capacity of network equipment. Moreover, as the load factor of domestic air conditioners is less than 10%, their owners do not pay an equitable share of network augmentation costs through tariffs based on anytime energy. Efficiency objectives of network infrastructure pricing The pricing of network infrastructure is a significant component of the cost of delivered energy to customers, averaging about 40% but sometimes having a greater impact than the energy component. This section explains the principal network pricing objectives. It is important to distinguish between the fixed and marginal costs of operating and maintaining an electricity network. Fixed costs are independent of the load, whilst marginal costs are associated with carrying the next increment of load. It is also necessary to distinguish between short and long run marginal costs. Short Run Marginal Cost (SRMC). Applies to a period where no new infrastructure investment is provided. Network short run marginal costs are very small. The infrastructure is in place and it costs little extra to carry a marginal increase in energy or demand, although sometimes operating expenditures can be increased through impaired access for equipment maintenance. Increased short run costs in the supply chain are limited to electrical losses and the cost of out-of merit generation, where the capacity of the network constrains the lowest cost generation from being dispatched. However, both of these costs are traded within the Australian NEM as part of the energy cost, rather than being reflected in the network price. Long Run Marginal Cost (LRMC). Is evaluated over a period including future augmentation of the network. The Net Present Value (NPV) of the future investment and its associated operating cost are spread over the accompanying increase in demand or energy. This results in an average LRMC, as follows: NPV(Growthrelatedcapitalcos t + Incremental operatingcos t) NetworkLRMC = NPV(Incremental energy or demandconsumption) In the case of EnergyAustralia, this LRMC is some 80% of the average network price, which represents the revenue deemed necessary by the regulator to sustain the network business. The percentage varies between customer classes. Network assets have long lives, generally in excess of 30 years, and often, lengthy construction times. Network augmentation is thus extremely lumpy, as there are large, high cost investments at irregular intervals. Pricing of network service on a long run basis is appropriate to this situation. This approach provides a relatively stable price signal to customers on which they can base their plant and appliance investment decisions. This marginal cost approach is important in achieving allocative efficiency but there is also a need to recover fixed costs, as the marginal cost alone would not recover sufficient revenue. In summary, an efficient network price would have at least two parts and should: Pass on the LRMC of network costs to the customers as a price-signalling component; Recover the balance of network costs in the least distortionary possible way (preferably as a fixed cost, but failing that, through anytime energy); and Be structured in the simplest possible way, so as to facilitate customers consumption and investment decisions being related to their bill. Whilst basic network tariffs should be structured in this way, there are potentially short-term pricing options available to target specific high cost events and so lead to further overall economies in the supply chain. PRICING AND DEMAND MANAGEMENT An analysis of the potential effect on customers of delivering demand management through pricing follows. In determining network costs, the LRMC has been allocated to periods of peak loading. The relative capacity limitations of network equipment in winter and summer periods have also been taken into account. Figure 3 charts the network load, pool price and allocated network cost over the five-year period to June 2004 (the lower two charts are on a logarithmic scale). Network cost RRP Load, percentage of maximum 80% 60% 40% 20% $10,000 $10 $1 10% 1.0% 0.1% 0.01% Jul-99 Figure 3 W S W S W S W S W S W Oct-99 Jan-00 Apr-00 Jul-00 Oct-00 Jan-01 Apr-01 Jul-01 Oct-01 Month Network load, pool price and network cost The following points may be drawn from Figure 3: The weekly and seasonal variation, including the incidence of public holidays at Christmas and Easter, may be identified. There is greater variability and higher demand in Jan-02 Apr-02 Jul-02 Oct-02 Jan-03 Apr-03 Jul-03 Oct-03 Jan-04 Apr-04 Jul-04

summer (Dec-Jan-Feb in the southern hemisphere), and to a lesser extent, winter (June-July-Aug); Hourly average pool prices also display great volatility. The cap on NSW pool prices in any five-minute settlement period was $10,000/MWh, whereas the average price was $A37.70/MWh. This great range of pool price is a feature of the Australian energy-only market settlement arrangements. The network cost driver is strongly seasonal, largely reflecting the summer and winter periods of higher demand, with significant volatility. It is also evident from Figure 3 that there is limited correlation between the pool price (which is driven by the interplay of generation and network availability, bid prices and customer demand) and periods of high network cost (driven solely by high demand). This poor correlation leads to an R² value of 0.29 between the daily peak load and pool price. Australia s gross pool market framework and companion contract market, presents significant challenges to launching a successful coordinated pricing demand response program along the lines of international pioneers. Studies have shown that customer demand response capability is driven by predictable pricing patterns, and that Australian wholesale energy pricing patterns will make developing any demand response program a challenge. Time of Use/Seasonal pricing. The rationale for ToU and Seasonal pricing for networks may be understood by reference to Figure 4, which is based on the network cost allocation. Cost 250% 200% 150% 50% 0% Figure 4 0:00 1:00 2:00 Winter Spring/Autumn Annual Summer ToU Average 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 Network cost and ToU pricing In Figure 4, the daily seasonal pattern of network cost is displayed. Also, on this diagram shows EnergyAustralia s ToU structure. It may be seen that the ToU shoulder and peak periods substantially align with periods of higher network cost. There is some level of thermal lag associated with network capacity which has not been modelled in the costing. ToU pricing does not directly take into account the strongly seasonal cost of meeting demand and, as may be seen from Figure 4, summer loading conditions are predominantly driving the need to augment the network. Thus the cost reflectivity of pricing would be significantly enhanced with seasonal pricing. It is EnergyAustralia s intention to introduce seasonal pricing for ToU customers in 2005/06. The left hand side of Figure 5 presents a comparison of the peak period charges which might arise from conventional tariff options. Prices are scaled to be cost reflective. It may be seen that retail costs vary less than network costs between peak and off peak and significantly mute the network price signals. Notwithstanding this, there is a clear improvement in cost 11:00 12:00 Hour 13:00 14:00 15:00 Peak 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 reflectivity through ToU pricing, even though the 11:1 ratio of peak to off peak network costs becomes a 3:1 ratio when retail and network charges are bundled. Seasonal pricing would enable a further large improvement in cost reflectivity and could impose a bundled price signal with a ratio of around 4:1 between the peak and off peak cost. Seasonal pricing would redefine the peak period to apply during Summer and Winter. $/MWh $250 $200 $150 $50 Figure 5 Network Retail Peak Flat Block 2 rate 3 rate ToU ToU Tariff options Seasonal Critical peak pricing. Four types of critical peak pricing have been considered. Apart from requiring interval metering, this form of pricing is dependent upon the availability of a communications channel and display unit, to signal periods of high price to the customer. The options chosen for this analysis are: Critical peak 30, in which the peak period would target the top 30 hours of: - network peak loading; - pool price; and - both network peak loading and pool price; and Critical peak 50, in which the peak period would target the top 50 hours of both network loading and high pool price. These pricing options are inset in the right hand side of Figure 5. For critical peak prices, the peak rates are significantly less than the cost reflective rates, which were considered to be unacceptably high. The cost reflective Critical peak 30 rate of up to $A3,700/MWh was moderated to $A2,000/MWh. Electrical losses. An aspect of cost arising from the use of the network, which has also been taken into account in the analysis is the variation of electrical network losses with loading. Distribution loss factors are used to uplift energy measured at the customer s meter to the points of market settlement, which are at the connection to the transmission network. The variation in distribution losses thus affects the energy purchase quantities. EFFECT OF PRICING ON CUSTOMER BEHAVIOUR The effect of pricing on customer consumption patterns is governed by the elasticity of demand the effect that a price increase will have in reducing demand, or conversely a price reduction will have in increasing demand. There is some level of uncertainty associated with the price elasticity of customer demand. It is also intuitive, and apparent from some studies, that the elasticity of customer demand can vary, with significant differences between long and short run N NR R Critical Critical 30 50 $2,250 $2,000 $1,750 $/MWh $1,500 $1,250 $750

effects. This occurs as customer short-term preferences are translated into longer-term decisions on appliance and plant stock, building standards and other conservation measures. Australian studies. The National Institute of Economic and Industry Research (NIEIR) has assessed the long run price elasticity of electricity consumption for use by the National Electricity Market Management Company (NEMMCO) in planning the transmission network. Their assessment is that the long term price elasticity for New South Wales has a range of - 0.22 to -0.52, with a mean of -0.37 [2]. Other available NSW documentation relates to the manufacturing sector, where processes and plant are generally long term in nature, but where inter fuel substitution may be more readily available. In a 1998 study, the then Department of Energy concluded that the elasticity across a range of business sectors ranged from 0.52 to -1.05 [3]. Overseas studies. There is a significant body of published work dealing with the topic of demand elasticity. This is referenced in the bibliography [4] - [17]. None of these studies is believed to assess long run elasticities adequately for the current situation in Australia, particularly given the significant recent penetration of air conditioning. The main issues with respect to demand elasticities in electricity may be summarised as follows: There is a level of demand elasticity which can be exploited using appropriate price signals; The amount of load that can be reduced during peak periods is not just dependent on the elasticity, but also the size of the price spike used to elicit that demand response; and Education is important in eliciting demand response in electricity in the small end user market, more so than with other commodities. In many studies it is not clear that customers have been furnished with sufficient information to maximise demand response. Assumed price elasticity of demand. For the purpose of the analysis in this report, an initial value of -0.12 was assumed in the first year, increasing to NIEIR s average of -0.37 in the fifth and subsequent years of a 10-year analysis. It was also assumed that the differential in pricing between high and low price periods resulted in some reduction in demand and some load shifting behaviour. This was modelled by assuming the elasticity for off peak periods was 80% of that of the higher priced periods. Modelling of pricing outcomes. The tariff structures described above have been combined with the price elasticities to estimate the effect on customers usage patterns. The impact on supply industry cost structures was estimated in four categories: Network, based on the reduction in capital expenditure or network LRMC; Pool, based on avoidance of high price periods. The pool price is not necessarily related to the average cost of energy generation (fuel, operating and capital) but rather the bid prices submitted by the operators of available plant. It should be noted that there is a flow on effect where a reduction in load at times of high pool price would lower the price for all customers. This effect has not been taken into account; Long run generation costs. For small NSW customers, the hedging rate between generators and retailers is pre determined. This arrangement is termed the Electricity Tariff Equalisation Fund (ETEF) and the associated Regulated Energy Cost (REC) has ToU rates set with regard to the long run costs of new entrant generation [18],[19]. Greenhouse, based on a cost associated with Carbon Dioxide emissions. An average rate of $15/Tonne of CO 2 was used, together with an average conversion rate of 1 MWh per Tonne. The rates were applied to energy conserved, through direct reduction in consumption or through peak shifting, with energy consumed at periods of lower network losses [20]. The overall reduction in costs per customer arising from the pricing options (compared with a single rate tariff) is illustrated in Figure 6. The analysis is based upon an average customer annual consumption of 15 MWh, which corresponds to an annual retail bill of approximately $A1,600. This corresponds with the lower limit of customer size currently proposed by EnergyAustralia for its meter and ToU roll out. In the case of the critical peak options, it has conservatively been assumed that 50% of the available benefit could be captured, because of the difficulty of forecasting high price periods in advance. For comparison with the cost reductions shown here in NPV terms, the NPV of the customer s bill over the same 10-year period would be $A11,700, so that the Critical peak 30 tariff would provide a potential reduction in average costs of service in the vicinity of 40%. NPV of benefit $900 $800 $700 $600 $400 $300 $200 Figure 6 ETEF benefit Pool benefit G'house benefit Network benefit Block 2 rate ToU CONCLUSIONS 15 MWh customer ToU Seasonal Critical 30 Network price Total benefit from tariff options $4,500 $4,000 $3,500 $3,000 $2,500 $2,000 $1,500 Critical 50 The following observations are made concerning the network infrastructure pricing options that have been analysed in this report. Demand management effects of network pricing options. Pricing options that target network constraint periods to reduce network capital expenditure deliver much smaller benefits in terms of pool price reduction or reduced energy purchase costs. Cost reflective network pricing options also deliver very small benefits in terms of reduced overall consumption, and therefore greenhouse emissions. ToU is a relatively straightforward form of pricing and on the basis of the assumed customer elasticities and costs, is expected on average to return the cost of investment in interval meters for customers of at least 15 MWh annual consumption. It can be observed from Figure 6 that a 3-rate ToU price is much more effective than 2-rate ToU. Seasonal pricing is a refinement of ToU, which for little additional complexity can double the expected benefits. N NR R NPV of benefit

Critical peak pricing appears to offer the most favourable option for targetting consumption during network constraint periods. Whilst this option requires at least a one-way communications path (which might most economically be provided on a wholeof-network basis) and introduces additional complexity, it appears to be worthy of further investigation. The Demand Management effects of cost reflective network pricing options can be summarised as follows: Simple price structures like the inclining block are limited in effectiveness and target overall consumption. There is a limited seasonal effect in that average consumption patterns are greater in Summer and Winter. This type of tariff does serve as a useful adjunct to more cost reflective pricing for larger customers. Time of Use and Seasonal pricing are expected to directly affect customers consumption, leading mainly to reduced consumption during peak periods. These cost reflective forms of pricing are also expected to support economical load management options such as the use of time switches and efficient appliance and building construction standards. Demand and Capacity charges have the potential to influence customer awareness and consumption patterns more directly. The monthly billing associated with these tariffs provides a more frequent reminder to customers of the cost of peak period consumption. Furthermore, kva demand tariffs provide an incentive for customers to install power factor correction equipment. The critical peak pricing options most closely target peak periods and appears to offer the potential of very significant benefits in terms of both network cost and pool purchase energy costs. There is a trade-off between the network and pool benefit and the optimum outcome would target the high price periods of both. The Critical peak 50 pricing delivers lower gains than the Critical peak 30 option. This is because the peak rate is lower (spread over a greater number of hours) and customer response correspondingly reduced. A pricing experiment. The analysis in this report indicates the desirability of conducting a pilot program to evaluate the systems and effectiveness of the critical peak form of pricing. One of the biggest advantages would be that through experimentation, a clearer view of the customers elasticity of demand would be obtained. Regulatory considerations. It may be seen that about one quarter of the benefit from ToU pricing arises from energy cost savings, which are not captured by the network. Whilst it is clear that the major beneficiary of pricing would be the network, there is a strong case for the necessary investment in metering and systems to receive a regulatory allowance in order to facilitate it. The retail cost savings accrue within the contestable portion of the supply chain and in this competitive environment may reasonably be expected to flow through to customers. ACKNOWLEDGEMENT The authors are indebted to EnergyAustralia for permission to publish this paper. BIBLIOGRAPHY [1] Independent Pricing and Regulatory Tribunal, October 2002, Inquiry into the Role of Demand Management and other Options in the Provision of Energy Services - Final Report. [2] National Institute of Economic and Industry Research, June 2002, The price elasticity of demand for electricity in NEM regions - A report for the National Electricity Market Management Company. [3] NSW Department Of Energy, September 1998, Analysis of Energy Use in the NSW Manufacturing Sector. [4] Oren S S, May 2000, Capacity Payments and Supply Adequacy in Competitive Electricity Markets, University of California at Berkeley, USA. [5] Bidwell M and Henney A, April 2004, Will NETA ensure generation adequacy, Power UK Issue 122. [6] Armstead H C H, December 1969, Allocating Fixed Costs, Energy International. [7] Li Y and Flynn P C, 2004, Deregulated Power Prices: Changes Over Time, University of Alberta Study. [8] New York Independent System Operator, 2002 PRL evaluation, Chapter 5 Implicit Price Elasticities of Demand for Electricity and Performance Results. [9] Stevens B and Lerner L, July 17 1996, California Energy Commission, Testimony on The Effect of Restructuring on Price Elasticities of Demand and Supply, Prepared for the August 14, 1996 ER 96 Committee Hearing. [10] Hawdon D, 1992, Is Electricity Consumption Influenced by Time of Use Tariffs? A Survey of Results and Issues, Energy Demand: Evidence and Expectations. [11] Filippini M, July 1995, Electric Demand by Time of Use: An Application of the Household Aides Model Energy Economics, (17), pp197-204. [12] Tischler A, 1991, Complementarity-Substitution Relationships in the Demand for Time Differentiated inputs under Time of Use Pricing, Energy Journal, (12), pp. 137-148. [13] Tischler A, 1998, The Bias in Price Elasticity Rates: An Application to Israel, Energy Journal (19), pp. 217-235. [14] Charles River Associates, January 16 2004, Statewide Pricing Pilot Summer 2003 Impact Analysis. [15] Charles River and Associates for the Essential Services Commission of South Australia, August 2004, Assessment of Demand Management and Metering Strategy Options [16] Charles River and Associates, October 2004, Statewide Pricing Pilot, Summer 2003 Impact Analysis, Final Report, Californian Energy Commission Website. [17] Aigner D J, Newman J and Tischler A, 1994, The Response of Small and Medium-Size Business Customers to Time-of- Use (ToU) Electricity Rates in Israel, Journal of Applied Econometrics, v 9, pp 283-304 [18] NSW Treasury, December 2000, Electricity Tariff Equalisation Fund: Information Paper. [19] Cap Gemini Ernst Young, September 2000, Assessing the Long Run Marginal Costs of Generation in New South Wales - Report for IPART. [20] NSW Department of Energy, February 1999, Greenhouse Gas Emissions from Electricity Supplied in NSW: Framework for the Calculation of Electricity Sales Foregone.